LZN's Blog

【Cheeky Pint】模型忠诚、产品捆绑与组织文化 | 纳德拉与 John Collison 对谈 | 中英全文（下）

Wed, 07 Oct 2026 00:00:00 +0000

书童按：当用户对某个模型产生感情，自动选择模型还是更好的产品方向吗？下篇从这一分歧出发，谈到微软的 token 工厂与智能体工厂、开放平台与产品捆绑之间的取舍，以及一家二十万人的公司如何维持协作和文化。结尾回到个人：职业经理人与创始人有什么不同，学校又如何给一个人留出发现兴趣的空间。节目：Cheeky Pint 第 19 期；主持人 John Collison，嘉宾 Satya Nadella。原节目发布于 2025 年 11 月 18 日，本文中的“今天”“现在”均沿用当时语境。 本篇范围：00:53:47—01:20:10。原节目与音视频 · 英文转录本系列据完整英文转录逐段翻译，中文在前、英文在后；保留原稿时间戳，长发言按语义分段。Luna 初译，经 DeepSeek-V4.1-Flash 与 Qwen3.6-35B-A3B 从语言和忠实度两个维度交叉校审，再综合定稿。必要的事实背景与转录疑点另作译注；嘉宾的观点和判断保留原意。 系列目录：上篇：企业 AI、工作方式与互联网往事 · 中篇：算力瓶颈、企业主权与智能体商务 · 下篇：模型忠诚、产品捆绑与组织文化 本篇目录

人们忠于模型，还是忠于品牌？
token 工厂与智能体工厂
开放平台与产品捆绑的取舍
谁来定义一家公司的文化？
二十万人公司的管理，以及创始人的工作记忆
海得拉巴、学校与板球

人们忠于模型，还是忠于品牌？

Loyalty to models or brands 约翰·科里森（00:53:47） 也许我们所说的这些分工，其实是软件和组织架构在偶然过程中划出来的一条条泳道。“有人提出非商业性质的问题，就由你们做客户服务。你是销售开发代表。你负责其他事情……”诸如此类。所有这些界线大概都会开始松动。我们聊了很多人们会用的人工智能应用，比如 Copilot、ChatGPT、Gemini 之类的。现在有个争论：模型质量到底有多重要？人们会不会认准一个品牌，比如喝了很多年可口可乐；即使可口可乐……我拿它举例不太恰当，因为当年配方一改就引发了反弹；但就算他们换了配方，人们还是会偏好某个品牌。 John Collison (00:53:47) Maybe what we’re describing is a bunch of swim lanes have been established by random accidents of software and org charts and everything like that. “You do customer service when people come with a query of a non-commercial nature. You are an SDR. You do whatever…” And all those distinctions are probably going to get going. We’re talking a lot about the AI apps that people use and Copilot and ChatGPT and Gemini and all these kinds of things. There’s a debate about how much model quality matters and is it the case that people pick a brand and they’ve been drinking Coke for the longest time and even if Coke… I mean Coke’s a bad example, because there was a revolt about the change in the formula, but even if they change the formula people, they still have a preferred brand. 约翰·科里森（00:53:47） 我用 o3，我妻子用 GPT 5。我几乎要吓坏了，因为我会想：“你值得用更聪明的模型。除非我死了，否则你别想从我手里拿走 o3。”你怎么看这个问题：人们会忠于某个模型吗？而且他们当初试图撤掉 4.0 时——是 4.0 吧？——也引发了反弹，大家对那个模型感情很深。人们忠于的是某个模型，还是某个人工智能品牌？这会怎样影响你们的业务策略？¹ John Collison (00:53:47) I use o3, my wife uses GPT 5. I’m almost horrified because I’m like, “You deserve more intelligence than that and you can take o3 from my cold dead hands.” Where do you stand on the debate of do people have loyalty to—and there was also the revolt when they tried to take away 4.0, was it? And people were really attached to that model. Do people have loyalty to a model or do they have loyalty to an AI brand and how does this affect your business strategy? 纳德拉（00:54:57） 我觉得在消费产品领域，这是我们第一次看到换模型会带来不同的变化，并非人人感受都一样。个性就是其中一个维度，还有风格之类的东西。这算是一个新的维度。换句话说，这也说明，哇，这或许成了新的差异化方式。模型有智商这一面，有情商这一面，还有各种风格特点；也许人们会据此引导模型的表现。 Satya Nadella (00:54:57) I think that in consumer products, this was the first time we saw that when you changed models, they’re not sort of uniform changes and they impact people differently. And personality is one such thing or style or what have you. And so it just sort of is a new dimension. So in other words, it’s also an argument that, oh wow, this is a new dimension of perhaps differentiation. There’s the IQ side of it, there’s the EQ side of it, and then there is all these style points and maybe that’s kind of one of the things that people will steer things towards. 纳德拉（00:54:57） 但从长远看，我认为模型必须能胜任最难、价值最高的任务。拿到这种能力之后，再针对手头的任务持续优化，对吧？所以作为产品开发者，我们要做的是推出能力最强的模型版本；但在生产环境中，实际运行的是多个模型。比如，GitHub 上我最喜欢的新功能是 Auto。人们显然还是喜欢 Sonnet，想用就继续用；但归根结底，我希望有个模型选择器，而且它不能只是个愚笨的模型路由器。 Satya Nadella (00:54:57) But long-term for me, I think you have to kind of make sure that the models are more capable for the hardest high-value tasks. And then you continuously optimize, after you have access to that for what the task at hand is. Right? So as a product builder for us, my thing is to have the model drop, which is the most capable, but then what’s in production is multiple models. And my favorite new thing in GitHub for example is Auto. Which is, I want to keep, people still obviously love Sonnet whatever they want to use it, but at the end of the day, I really want the model picker and it just can’t be a dumb model router. 纳德拉（00:54:57） 它必须有足够的智能，知道这项任务值得投入多少成本、需要哪种智能，也要了解我的代码库有多复杂，或者我的拉取请求任务有多复杂。我认为这才是智能体未来的发展方向。因此，你需要模型；事实上，你需要一个模型组合，再由智能体在这些模型之间协调，让它满足你的需求，之后你也会有自己的偏好。² Satya Nadella (00:54:57) It has to basically have the intelligence to know that this task deserves this kind of cogs or this type of intelligence and this is my complexity of my repo or my PR task. That ultimately is where the future of agents would be. And so therefore you want the model. In fact, you want an ensemble of models, then you have agents intermediating that ensemble so that it meets your needs and then you’ll have preferences. 约翰·科里森（00:57:00） 大家的偏好不都会是越聪明越好吗？我会打开选择器，手动选 o3 来回答“我去哪儿吃冰淇淋”这种问题。我总想用最…… John Collison (00:57:00) Will everyone’s preference not just be for more intelligence? I’ll go into the picker and manually select o3 for “Where should I go get ice cream” query. I always want the most— 纳德拉（00:57:10） 我觉得那是习惯，你不这么认为吗？ Satya Nadella (00:57:10) That’s habit, don’t you think? 约翰·科里森（00:57:12） 也许吧，不过这也是经过慎重考虑的重要选择。 John Collison (00:57:12) Maybe, but it’s also an important considered decision. 纳德拉（00:57:16） 确实如此。我是说，我们很难让自己放弃原来的选择……这就是默认选项为什么重要，也是我们为什么喜欢默认选项。我们不喜欢别人动我们的奶酪。连模型选择也是这样：“哇，如果现在把模型选择拿掉，就麻烦了。”所以这得谨慎处理。不过我确实认为，从长远看，如果我能信任某个东西始终替我把事情做好，而且它做选择时还让我觉得很满意，那我就会把事情交给它。 Satya Nadella (00:57:16) But it is true. I mean it’s very hard for any of us to take our—that’s why defaults matter and we love our defaults. We don’t love the cheese to be moved. Even the model selection stuff, it’s kind of like, “Wow, if you now took away the model selection, it’s a problem” and so therefore you got to be careful. But I do think in the long run, if I can trust, that’s another one, which is if I can trust something to always do something for me while it’s making a selection that somehow is delightful, then that’s when I’ll hand off. 约翰·科里森（00:57:49） 所以你觉得，最终要做到的是让我相信你会选合适的模型？ John Collison (00:57:49) And so you think that’s what you need to get to is me trusting that you’ll pick an appropriate model? 纳德拉（00:57:55） 正是如此。 Satya Nadella (00:57:55) Exactly.

token 工厂与智能体工厂

Token factories and agent factories 约翰·科里森（00:57:56） 是啊。我的理解是，微软在整个技术栈的每一层都有布局：有 Copilot，有你们在 OpenAI 的股权，还有……人工智能的垂直应用我们可以之后再聊；你们有 Azure 这一层，有芯片，还有很多别的东西。这里面有些部分对你来说比其他部分更重要吗？哪些领域是必须赢下来的？你们会做垂直应用吗？ John Collison (00:57:56) Yeah. And then I mean my mental model of Microsoft is that you just play at every part of the stack in that there’s the—you have Copilot, you have your stake and OpenAI, you have… Well, we can get to vertical applications in AI, you have the Azure layer, you have chips, everything leaving out a whole bunch of stuff. Are some more important to you than others? What is the must win? Will you do verticals? 纳德拉（00:58:19） 是啊。最核心的部分，我会把它概括为我们的基础设施业务。我们必须非常擅长打造我所谓的“token（词元）工厂”，也就是极其高效地做到每瓦、每美元产出更多 token。然后还有另一层，我称之为“智能体工厂”。token 工厂和智能体工厂的区别在于，后者要最高效地使用 token，以实现业务成果或满足消费者偏好，也就是…… Satya Nadella (00:58:19) Yeah. Well, at the core, the way I kind of conceptualize it is our infrastructure business. We have to be fantastic at building what I’ll call the token factory. This is the tokens per dollar per watt, really being super efficient at that. Then I’ll say we have another layer of it, which is the agent factory and the difference between the token factory and the agent factory is use the tokens most efficiently to drive a business outcome or a consumer preference outcome, which is— 约翰·科里森（00:58:51） 这说的是每个 token所创造的价值，之类的吧？ John Collison (00:58:51) That’s about the value per token or something, 纳德拉（00:58:53） 每个 token所创造的价值，要由人们关注的特定领域来评估。正如你说的，它周围还有一整套工具。它有一个完整的……可以说是新一代应用层，或者应用服务器。每一代新平台都曾有过这样的东西：有了网络，就有网络服务器。某种意义上，这就是人工智能服务器，或者人工智能云。然后我们肯定也会打造自己的——我会称之为“智能系统”或人工智能系统——也就是 Copilot 系列产品。无论是面向信息工作，还是我们已经用于编程或软件开发的产品。 Satya Nadella (00:58:53) The value per token and as evaluated by the specific domain that people care about. And that is to your point, it has tooling around it. It has a whole, it’s kind of the new app tier or the app server. Every new platform has always had, there was the web and there was a web server. This is the AI server in some sense or the AI cloud. Then we will definitely want to build our own, I’ll call it systems of intelligence or AI systems that is the family of Copilot. Whether it’s for information work, that’s kind of what we’ve done for coding or software development. 纳德拉（00:58:53） 那就是 GitHub Copilot。安全也是另一个领域，我们肯定会成为其中的主要参与者。这是三个横向领域。我们也会做业务应用。另一个方向是医疗和科学，我们在这方面投入很多。在医疗领域，我们收购了 Nuance，现在有个叫 DAX Copilot 的产品，能为医生记录病历并区分说话者，让医生可以把更多时间花在患者身上，其他事情都交给人工智能处理，从医疗编码到记录病历都包括在内。这是一个例子。 Satya Nadella (00:58:53) That’s the GitHub Copilot, security’s another domain where we are absolutely going to be a primary. Those would be the three horizontal. We will also have business applications. The other one is we are doing a lot in health and science. So in health we had bought Nuance and now we have something called DAX Copilot, and this is the notetaking diarization for physicians. So their ability to be able to have a doctor spend more time with their patients and then the AI do everything else in terms of everything from coding to taking the notes. So that’s one place. 纳德拉（00:58:53） 我们和 Epic 有很好的紧密合作关系，相关功能也嵌入了 Epic。这就是我们在医疗领域所做的事。我们也在做面向消费者健康的 Copilot 功能，并与这些服务衔接。另一个方向是科学。结果发现，这个领域非常适合我所谓的“外循环编排”：科学方法在某种意义上要求你提出假设，再通过计算机模拟开展多项实验，回来后修正假设，等等。对我来说，这是另一条工具链。我们有点像是在探索如何把 GitHub Copilot 和 Microsoft 365 Copilot 结合起来。 Satya Nadella (00:58:53) We have a great close partnership with Epic. It’s an embedded part of Epic. So that’s kind of what we are doing in health. And then we are also doing stuff in Copilot for consumer health that sort of docks to it. But the other one is science, and it turns out it’s a big domain for what I’ll call the outer-loop orchestration, which is the scientific method in some sense requires you to create the hypothesis, then run these multiple experiments in silico, come back, refine and so on. So that to me is another tool chain. It’s kind of like we are trying to discover some combination of the GitHub copilot meets Microsoft 365 Copilot. 纳德拉（00:58:53） 也就是说，为科学家提供知识工作工具，让他们能够使用权威知识来源，也能调用界面和工具；甚至可以接入湿实验室的 MCP 服务器，和它交互。我该如何编排好这一切，让科学研究的循环加快？ Satya Nadella (00:58:53) Knowledge work, if you will, for the scientist where they have the authoritative sources of knowledge. They have the interfaces tools used, could even be, hey, the MCP server for the wet lab, so to speak, can I interface with it? And then how do you orchestrate all of this such that the scientific loop can go faster?

开放平台与产品捆绑的取舍

Open platforms and product bundling 约翰·科里森（01:01:09） 作为平台公司，你总得决定什么时候该把产品捆绑在一起，什么时候该把它们捆起来并要求用户一起使用，什么时候又不该这么做。不知为什么，大家经常谈到的一个经典例子，其实是件挺小的事：苹果最初只让 iPod 配合 Mac 使用，想借此带动 Mac 销售；后来放弃了这个策略，推出了 Windows 版 iTunes。按我读《苹果在中国》时的了解，这完全是某人有一天偶然作出的一个决定，但人们常把它当成经典案例。 John Collison (01:01:09) As a platform company, you always have decisions around when should you try and bundle products together? When should you try and staple them and mandate they be used together and when should you not? And I think the classic example for some reason that everyone talks about despite it being quite minor, is the fact that Apple originally only, let’s use an iPod with a Mac and tried to use it to drive Mac sales, and then gave up and shipped out iTunes for Windows. And my understanding reading the Apple in China book is it was a totally random decision that someone just made one day, but it’s often held up as one of these examples. 约翰·科里森（01:01:09） 显然，微软的整个历史里也充满了有意思的例子。我觉得人们没有意识到，早期微软有多开放。1985年，微软的大部分收入来自 Macintosh 应用；而微软的操作系统上，大部分应用都是第三方软件，比如 Lotus 1-2-3 之类。所以当时采取的是完全开放的策略。后来到了 Windows 时代，Office 和 Windows 紧密捆绑，彼此相互促进。再后来，我的印象是，Azure 和云服务早期的定位是：哦，你可以在这里运行 SQL Server。之后微软又全面拥抱 Linux，等等。 John Collison (01:01:09) Obviously Microsoft, the entire history is full of these interesting examples. I don’t think people realize how open Microsoft was in the early days where in 1985 most of Microsoft’s revenue was from Macintosh applications and then for the Microsoft operating systems, most of the applications were third party like Lotus 1, 2, 3 and things like this. And so it was like a fully open strategy and then you had the Windows era of the tight coupling between Office and Windows and those mutually reinforcing each other. Then early on, I get the sense Azure and Cloud was, oh, it’s a place where you can run your SQL server and then fully embracing Linux later on and things like that. 约翰·科里森（01:01:09） 我之所以好奇，是因为我们也把自己看成平台公司，最近一直在拥抱更多模块化。比如 Stripe Radar，即使你不用 Stripe 收款，也可以使用它。你一般怎么判断产品应该捆绑销售，还是独立销售？人工智能领域，这个问题又有什么特别之处？ John Collison (01:01:09) I’m curious just because again, we think about this as a platform company and we’ve been of late embracing much more modularity where Stripe Radar, you can use it even if you’re not using Stripe for payments and things like that. How do you in general think about your framework for when products should be coupled versus when you sell them independently? And then AI specific versions of that question. 纳德拉（01:02:50） 我思考这个问题的一个出发点是，我们经常夸大许多竞争有多么“零和”。你需要敏锐分析的一点，是哪些领域从定义上说就会有多个参与者。云服务就是经典例子。我记得刚入行时，Azure 显然比 AWS 晚很多才起步。人们会问我：“天啊，AWS 领先这么多，市场里还有第二家云服务商的位置吗？”我当时和 Oracle、IBM 在各种中间层服务器上竞争过，所以我觉得，当然有。 Satya Nadella (01:02:50) So the way—a reason about it is I think we overstate many times how many of these battles are “zero sum.” So at some level, one of the pieces of analysis that I think that you want to be sharp at is, what are by definition going to be multiplayer? Like cloud is a classic example, which is, I remember even back in the day when I got started, and obviously Azure got started much later than AWS. People would tell me, “Oh God isn’t AWS so far ahead? Is there even room for a second cloud?” And having competed against Oracle and IBM on all the middle tier servers and so on, I felt like no. 纳德拉（01:02:50） 企业客户和商业客户大体上会要求有多家供应商。这种结构性的认识促使我们投身其中，之后的故事大家都知道了。所以在我看来，如果你把产品捆绑得太过头，可能反而会缩小自己的总潜在市场，失去竞争力。比如 Azure 最初叫 Windows Azure。哦，这就成问题了，因为 Azure 不能只对 Windows 有意义。它必须把 Linux 当作一等公民来支持，也必须把 MySQL 和 Postgres 当作一等公民来支持。 Satya Nadella (01:02:50) These enterprise customers and commercial customers by and large are going to demand a sort of multiple. And so that was the structural understanding that drove us to even just be at it and the rest is history. So a little bit of, to me, if you over package things, you might in fact sort reduce your TAM and not compete. For example, Azure was called Windows Azure. Oh wow. That’s a problem because Azure makes no sense just for Windows. It’s sort of got to support Linux as first class. It’s got to support MySQL and Postgres as first class. 纳德拉（01:02:50） 因此，我们既要确保 SQL Server 做得非常出色，也要在 Postgres 或 MySQL 上做得和 Amazon 一样好。这主要是因为，嘿，这就是市场规模，也是客户对我们的期待，而我们将面对激烈竞争。所以我就是这样界定模块化的：什么做法能让我的技术栈获得最大的市场空间？当然，我们是一家公司，不是一个多元化企业集团，因此某种程度上应该存在整合收益和平台效应的理论；那么这些收益和效应是什么，我们又该怎样把它们实现好？ Satya Nadella (01:02:50) And so that’s what allowed us to make sure that you have to actually have to do a great job with SQL Server. But you got to do as bang of a job as Amazon would do with Postgres or MySQL. And so it was driven primarily by hey, that’s the TAM, that’s what customers expect from us and we are going to have tough competition. So to me that’s kind of how I define my modularity. What’s the thing that maximizes my stack’s market opportunity? Then yes, we are a firm and the reason we’re not a conglomerate and so therefore there should be a theory of some integration benefits and platform effects and so therefore what is that and how do we do a great job of it? 纳德拉（01:02:50） 但技术栈的每一层，包括 Azure 的token 工厂，都应该允许客户这样说：“我只想用 Azure 的裸金属服务。我只需要你帮我管理遍布各处的 Kubernetes 集群，软件我自己带来。”完全没问题。这个工作负载我们必须赢下来。也许等到有一天，自己管理多区域数据库变得非常痛苦时，我们才有机会让客户说：“哦，那我来用 Cosmos。”但那是另一个单独的决定。 Satya Nadella (01:02:50) But each layer of the stack, even in let’s say in Azure, the token factories, somebody should be able to come and say, I just want to use Azure for its bare metal services. I just need Kubernetes clustered all over, but I just need you to do the management part and I’ll bring all my software. No problem. We got to win that workload. Maybe then after that we’ll at least have a shot someday when it becomes a real pain to manage sort of your multi-region database on your own that you’ll say, “Oh, let me just use Cosmos,” But it’s a separate decision. 约翰·科里森（01:05:25） 人们不总会争论吗？如果 Azure 支持 Linux，我们就能多卖 Azure；但 Windows 团队会说：“对，可你这是在削弱 Windows Server。”有些地方就像你说的那样，微软选择开放；另一些地方则不同。比如 Microsoft Flight Simulator 没有登陆 PlayStation，在 Xbox 上提供，这就说得通，整合在一起也很自然。我不知道这个例子是不是有点牵强，但 Teams Chat 和 Teams Video 没有分开销售，而是作为一个整体提供，这也说得通，能让套件更有吸引力。所以你们是不是总得争论，捆绑销售的成本到底有没有超过它带来的收益？³ John Collison (01:05:25) Isn’t there always a debate between if we have Linux and Azure, we’ll sell more Azure, but the Windows people say, “Yeah, but you’re hamstringing Windows’ server.” And there are some places like you’re describing where Microsoft’s open, there are other places, Microsoft Flight Simulator is not available on the PlayStation, it’s available on the Xbox, and that makes sense. It feels kind of natural to be integrated that way. I don’t know, this might be a bit of a stretch, but Teams Chat and Teams Video are not sold separately. They’re just part of one thing and that makes sense. It makes the bundle more compelling. And so don’t you always end up in these debates as to whether the bundling cost outweighs the bundling benefits? 纳德拉（01:06:01） 是的。我觉得有些情况，比如 Teams，就是个典型例子。Teams 从一开始就是把那四项功能整合在一起的产品，类似 Outlook。当年先有个人信息管理器，后来才有电子邮件客户端，日历也原本是独立的。Outlook 是第一个搭起框架、把这三样东西放在一起以便完成工作的产品。Teams 也一样，我们把聊天、频道、视频等等整合到了一起。所以某种程度上，捆绑本身就是产品，也是产品的基本框架。之后当然可以说：“嘿，它需要开放的市场，也需要和其他东西集成。”因此，必须在基本功能层面想清楚怎样模块化才合理。 Satya Nadella (01:06:01) Yeah, and I think some of those, for example, the Teams thing, is a classic one, which is Teams was born as a product that brought those four things, like Outlook. There was a PIM before there was an email client and a calendar was separate, and Outlook was the first scaffolding that said, “Hey, we bring these three things to get a job done.” And same thing with Teams. We brought chat and channels and video and what have you into one. So the bundling was the product, to some degree. That was the product scaffolding. And so then of course you can then say, “Hey, that needs to have an open marketplace and it needs to integrate with other things or what have you.” So the modularity has to be thought through in ways that make sense at the atomic level. 纳德拉（01:06:01） 接下来，你就不能过度纠结协同效应或整合效应，最后反而失去竞争力。一个典型例子是，你打造了一款极其出色的公有云，却只能运行 Windows 或 SQL 工作负载，那它在市场里就只占很小一块。因此，这既符合我们的利益，也能更好满足客户需求。我就是这样看人工智能技术栈如何衔接的：我们有基础设施业务、应用服务器业务和应用业务。这样说只是为了简化问题。 Satya Nadella (01:06:01) Then you don’t want to overthink about the synergies or integration effects and you’re not competitive. A classic thing would be if you built an unbelievable public cloud except it only ran Windows workloads or SQL workloads, that’ll be essentially a very small sliver of the market. So it was in our interest and definitely in the interest of meeting the customer needs. And so being able to really click in the AI stack, that’s kind of how I look at it. We have an infra business, we have an app server business, and we have an apps business. It’s just simplifying. 纳德拉（01:06:01） 我希望这三项业务都能凭自身实力站稳脚跟。当然，我们自己希望这三层之间能形成反馈循环，但客户和合作伙伴可以选择从哪个入口进入。 Satya Nadella (01:06:01) I want those three things to stand on their own merits. We ourselves of course want to have the feedback loop across these three layers, but customers and partners will choose which door they enter through. 约翰·科里森（01:07:41） 我的印象是，你接任微软之后，把公司文化从高度捆绑——你买一台 Windows 电脑，里面运行着 Microsoft Access、SQL Server 等等，一切都被妥善整合在微软的生态里——转向了更加开放、互操作的策略。 John Collison (01:07:41) This impression I have is that when you took over Microsoft, you shifted the culture from a highly bundled, you’ll buy your Windows machines and they’re running Microsoft Access and there are SQL Server and everything is neatly packaged together in this Microsoft life, to moving towards more of an open and interoperable strategy. 纳德拉（01:08:01） 我会说，我想做的是回到八十年代的微软。因为后来发生的大多数事情都在九十年代：当时有微软，几乎没有别的公司，所以我们的产品更加彼此整合，无论是在客户端还是服务器端。你说得对，八十年代我们先在 Mac 上开发 Office，Windows 版是后来才有的。事实上，比尔创办微软时的理念是：这是一家软件工厂。我不迷恋某一个类别，只想打造最好的软件工厂，让它源源不断地解决各种问题、产出软件。 Satya Nadella (01:08:01) I think that the way I would say is my thing was to go back even to the Microsoft of the eighties perhaps., Because most of what happened was really in the nineties there was Microsoft and there was pretty much nothing else, and so there was sort of a lot more of our things coming together, whether it was on the client or on the server. The eighties, to your point, we built Office on the Mac. Windows came later. In fact, the concept that Bill had when he started Microsoft was it’s a software factory, I’m not in love with any one category, I’m just going to build the best software factory and it’s going to churn out whatever problem. 纳德拉（01:08:01） 比如 Flight Simulator。你想要一个 BASIC 解释器？没问题，我们有。你想要一个操作系统？我们也有。所以某种意义上，这就是当时的想法。后来，我们逐渐受制于其中四五块业务之间的绑定关系，也就是 Windows、Windows NT、客户端—服务器架构等等。等我成为 CEO 时，我意识到这一点；甚至在我负责云业务时，我就意识到，市场即将变得大得多，也会发生变化，而那时我们还没有移动平台。 Satya Nadella (01:08:01) Flight i sim. You want a basic interpreter, no problem. We have one. You want an operating system? We have one too. So in some sense, that was the idea and at what point we got into a lock between four or five parts of that that became the Windows and Windows NT and client server and what have you. So I sort of realized that when I became CEO, and even when I was running our cloud business that hey, this is a time where the market’s going to be a lot bigger and got different, and we didn’t have the mobile platform at that time. 纳德拉（01:08:01） 因此，我们真的需要确保，通过以合理的方式组合产品，能在尽可能大的市场中保持相关性。如果这种做法并非公司的核心基因，我不认为我当上 CEO 后只要说一句“我要这么做”，公司就能执行得好。让我们的软件走上每个平台，本来就是公司的核心基因。 Satya Nadella (01:08:01) And so therefore we really needed to make sure we would stay relevant in the largest markets that we could address by bringing our products together in configurations that made sense. S I would say if it was not in the core DNA of the company, I don’t think just because I showed up as a CEO and I said, “I want to do this,” we would’ve executed well. It was in the core DNA of the company that we can in fact take our software to every platform.

谁来定义一家公司的文化？

Who defines a company’s culture 约翰·科里森（01:09:43） 说到公司的核心基因，那张微软员工互相举枪相向的著名漫画呢？你在文化上做了多少调整，又是怎么做到的？说到底，你能看到各种漂亮的举措，比如全员大会之类的；但归根结底，文化体现在什么事情你能容忍、什么事情不能容忍，以及决策如何作出等等。 John Collison (01:09:43) Yes. Speaking of the core DNA of the company, the famous cartoon of Microsoft with all the guns pointing at each other. How much cultural tweaking did you have to do and how do you actually do that? When you get down to brass tacks, you can see all the nice things, the all hands and things like that, but ultimately culture comes down to what you will and won’t tolerate and how decisions are made and things like that. 纳德拉（01:10:08） 我从那整个事件里学到了两件事。我总会说，听着，我是个地地道道的内部人。过去三十五年里，微软好的、坏的事情我都经历过，我也参与其中，所以我没法否认任何一面。我当时觉得，我们有点失去了信念，因为我们失去了自己的叙事。那张漫画就是个很好的例子：别人定义了后来成为公司文化叙事的东西，而那不完全符合现实。 Satya Nadella (01:10:08) Yeah, I would say there are two things that I learned from that entire episode. Because I always say, look, I’m a consummate insider. Anything good and bad about Microsoft for the last 35 years, I lived through them all and I’m part of it, so I can’t deny any of it. The thing that I felt was a little bit of that was we lost our own belief because we lost the narrative. That cartoon is a great example of someone else defining what became the cultural narrative more so than reality. 约翰·科里森（01:10:42） 大家开始觉得，那张漫画画的就是自己。 John Collison (01:10:42) People started to identify with the cartoon. 纳德拉（01:10:43） 没错。我觉得，今天社交媒体和社会风向的一个根本问题是，你完全可能失去对叙事的掌控。这是个会自我强化的过程。有意思的是，这些东西当然都包含某种信号；这并不是说，哇，我们各个部门都完美无缺、彼此高度和谐。事实并非如此。但某些部门间的张力确实反映了真实问题，张力本身也有必要。让组织内部和谐一致并不是目标，在市场上取胜才是目标。不过在某种程度上，你得协调好这些大型组织。 Satya Nadella (01:10:43) That’s right. I mean, I think one of the fundamental issues of today’s social media and the zeitgeist is you can absolutely lose narrative. It’s completely reflexive. So one of the interesting things is, of course, all of these things have signal, so this doesn’t mean, oh wow, we were all perfect divisions and we are all sort of in greater harmony. That is not the case, but in some sense, some of these divisional tensions are real issues that need to have tension. We can’t have, social cohesion is not a goal. Winning in the marketplace is a goal, but at some level you have to orchestrate these large organizations. 纳德拉（01:10:43） 事实上，你甚至可以有意设置两支彼此竞争的团队。可就因为有人说“嘿，我去翻翻《纽约客》，那里会有一幅漫画”……这类事情是我认为领导者需要面对的。还有，在今天这个时代，员工会在外部读到关于你的报道，进而形成对你的看法；该如何沟通，是领导者面临的最大挑战之一：怎样赢得他们的信任？怎样确保他们能感受到现实，也能塑造现实？另一个问题是，大家都觉得问题出在制度上。 Satya Nadella (01:10:43) In fact, you may even have two competing teams by design. And just because somebody said, “Hey, I’m going to read The New Yorker and there’s going to be a cartoon.” That’s the type of stuff that I think leaders… And how to communicate in today’s world where your employees read about you outside and form opinions about you is one of the toughest leadership challenges, I think, which is how do you earn the trust? How do you really make sure that they can in fact feel the reality, shape the reality? The other thing is everybody thinks it’s the system. 纳德拉（01:10:43） 大家会想：“都是那个高层领导、我的副总裁的问题，他们有全部权力，我一点权力都没有。”但现实是，权力分散得多，也分布得更广。因此，怎样真正帮助大家认识到这一点，尤其是让他们掌握其中的主动权并重新塑造局面？还有句名言是：“我离开的不是公司，而是经理。”我相信这句话。所以公司里会有各种微观文化，而它们可以被塑造。回顾我在微软的职业生涯，我很幸运，遇到了一些在公司里创造出非凡工作环境的人；正因为如此，我留了下来，也才能有所发展。 Satya Nadella (01:10:43) It’s that guy at the top or my VP and they have all the power and I have none. The reality is the power is a lot more diffused and distributed, and so therefore, how do you really help people? Especially get hold of that and reshape? One of the other famous things people say is, “Hey, I never leave companies. I leave managers.” I believe that, and so it’s kind of micro-cultures and they can be shaped. In fact, when I look back at my Microsoft career, I was lucky to fall into these people who created these unbelievable environments in the company, and that’s why I stayed and that’s how I thrived. 纳德拉（01:10:43） 所以某种程度上，我觉得高层尤其需要文化，需要一种自己能身体力行、始终如一的叙事。这也是为什么“成长型思维”或“好学者胜过万事通”这个框架对我们特别有帮助，因为没人会觉得那是我的教条，对吧？谢天谢地，这是儿童心理学里一个大家都熟悉的概念，在工作之外也能引起共鸣。找到这样的框架并身体力行很重要。但我觉得，今天我们共同面临的挑战之一，是别让社交媒体上的梗替我们定义自己。一个组织内部需要有怎样的韧性，才能抵御社交媒体上的流行叙事？我认为这是关键。 Satya Nadella (01:10:43) And so to some degree I feel that the more culture you need at the top, a narrative that you have to live and be consistent. So that’s where this growth mindset or learn-it-all versus know-it-all has been super helpful for us as just a frame because nobody thinks of it as my dogma, right? Thank God it’s a well understood child psychology thing that appeals to people outside of work, and so cracking something like that and then living it, but also somehow, I would say the challenge for all of us in today’s world is let the social media memes not define us. What’s that inner strength that is there in an organization that can in fact resist the social meme? That I think is the key.

二十万人公司的管理，以及创始人的工作记忆

Managing at scale and a founder’s working memory 约翰·科里森（01:13:37） 微软有多少员工？ John Collison (01:13:37) How many people is Microsoft? 纳德拉（01:13:39） 我觉得大约二十万人。 Satya Nadella (01:13:39) I think around 200,000. 约翰·科里森（01:13:40） 好，粗略算来，微软有二十万人。Stripe 有一万人。也许有人正在听，管理着一家五百人左右的公司。我们做的很多事情大概不太受规模影响：比如确保和客户交流、召开领导团队的外出会议。我们在看 2026 年的数据，希望收入再高一点、成本再低一点。公司里有很多活动，不管规模多大，做起来都差不多。 John Collison (01:13:40) Okay, so rough number is Microsoft has 200,000 people. Stripe has 10,000 people. Maybe there’s someone who’s listening to this who runs a company that’s 500 people or something like that. A lot of the things that we do are probably fairly scale independent, where you’re trying to make sure that you’re talking to customers, you’re holding a leadership offsite. We’re looking at the numbers for ‘26. We want the revenues to be a bit higher and the cost to be a bit lower. There’s a lot of activities in companies that are kind of the same, regardless of size. 约翰·科里森（01:13:40） 话虽如此，规模达到二十万人、像个城邦一样时，可能也会出现一些只有这么大才有的现象；我在一万人的规模上未必体会得到。公司大到这种程度，才会出现哪些影响？ John Collison (01:13:40) That said, there’s also probably things that only show up at the 200,000 person city state size that I wouldn’t be aware of at the 10,000 person size. What effects only show up when you’re that big? 纳德拉（01:14:27） 我想说两点。说实话，我只在微软工作过，所以也不敢说自己是专家。但有一点是接替创始人这件事。史蒂夫和比尔建立了公司。保罗和比尔创办了公司，史蒂夫和比尔把公司做大，而我算是第一个“非创始人”CEO。很快我就意识到——实际上，刚接任时我就明白了——我需要一个团队，才能管理这么大的职责范围。还有我们提到过的 A.G. 拉夫利（A.G. Lafley）那套观点，我觉得也很好。 Satya Nadella (01:14:27) There are two things I would say. Quite honestly, having only worked at Microsoft, it’s not that I’m like an expert, but the one thing I would say taking over for a founder— Steve and Bill built the company. I mean Paul and Bill started it and Steve and Bill scaled it, and I was sort of the first “non-founder” person. The thing I realized quickly, or in fact I got into the job and I realized that I need a team. And just to have the ability to manage the scope, but then that A.G. Lafly thing that we put out there, which I think is a great one. 纳德拉（01:14:27） 要清楚 CEO 明确需要做什么：你们从事哪些业务？哪些业务需要你从外部信息中综合判断？要确立标准，尤其是文化标准；还要建立你刚才说的绩效文化，不能说“我只看长期”或“我只看短期”，两边都得交出成果。要真正弄清楚只有你能做的四五件事，然后组建团队。你会说，即便公司只有五百人，你也该这么做；但说实话，那时你还可以把所有事情放在自己的工作记忆里。 Satya Nadella (01:14:27) Being clear about what the CEO clearly needs to do, which businesses are you in? Which businesses are you synthesizing from the outside? Having the standards, setting the standards for culture, and then the ability to your point about having that performance culture that you can’t say, “Hey, I’m only about the long term, or I’m all over the short term.” You’ve got to deliver both. Getting a real grip of the four or five things that only you can do, and then building the team. You’d say even at 500 people, that’s what you do, but quite frankly, you can keep in your working memory. 纳德拉（01:14:27） 我从开发者做起时，大家常聊一些事情：你自己记得多少行代码？后来你会说：“哦，那个人熟悉那个模块或那个库。”规模继续扩大，大家起初多少都认识每一行代码。再后来，你就得变成那个说“哦，我认识写那段代码的人”的人。我觉得，这种模块化、团队建设和凝聚力…… Satya Nadella (01:14:27) Growing up as a developer, there was a set of things everybody would talk about. How many lines of code do you know personally? At some point you sort of say, “Oh, that’s the person who knows that module or that library”. That becomes more. Everybody starts where they know every line of code at some level. Then you have to get to the person who knows, “Oh, I know the person who wrote that,” and I think that that modularity and team building and the cohesiveness is— 约翰·科里森（01:16:17） 我理解得对吗？在 Stripe 这样的规模，或者规模更小时，你也许还能把产品当作一个整体来思考，知道自己要发布的每一样东西，知道所有…… John Collison (01:16:17) Am I understanding you correctly that maybe it’s Stripe scale or at a smaller scale you can still reason about the product as a product and know everything that you’re shipping and everything— 纳德拉（01:16:27） 我也觉得创始人在这方面很特别。创始人的独特之处就在于，他们从第一天起就伴随着公司一起成长。要把创始人的工作记忆拿过来，再说“让我把它灌输给一位职业 CEO”，这很难做到，也行不通。就连我也是1992年才加入公司，八十年代初我还不在微软。所以某种程度上，那是一段连续成长的历程，只有创始人 CEO 或创始团队能亲眼见证。因此，我认为我们要尊重创始人独有的能力；创始人也要尊重接任的人，明白后来的人不可能照搬自己做过的一切。 Satya Nadella (01:16:27) I also think founders are unique in that sense because the founders are, that’s kind of what is singular about them because they’ve grown up with it from day one. See, it’s kind of hard to take the working memory of a founder and say, “Oh, let me take it and imprint it”—sort of a professional CEO.” It just doesn’t work because even for me, I joined the company in ‘92. I was not there in the early eighties, and so to some degree it was a continuous scale that only the founder CEO or the founders see it. And so that’s why I think having respect for what founders can do uniquely.And founders having respect for whoever comes next, that they can’t be doing exactly the same thing that they did. 纳德拉（01:16:27） 所以我觉得，“创始人模式”这个说法很有意思。创始人的文化和个性显然非常强大，应该善加运用、发挥到极致。像我们这样的职业经理人，也可以采用创始人模式，但别以为自己就是创始人。我觉得这其中的微妙区别很重要。 Satya Nadella (01:16:27) So that’s why I think this founder mold thing is interesting, which is clearly the culture personality of a founder is unbelievable and you use it, maximize it. Then mental model CEOs like us have to also be, you can sort of be in the founder mold but don’t think you’re a founder, and that nuance I think is an important one.

海得拉巴、学校与板球

Hyderabad, school and cricket 约翰·科里森（01:17:43） 最后一个问题。时间差不多了。我们聊了文化和如何建立文化。海得拉巴的水土到底有什么特别之处？你上过的那所学校，尚塔努也在那里读过，阿贾伊·班加也在那里读过。还有不少优秀棋手来自那里，也有不少人来自印度南部。你对当地人才表现如此出众有什么解释吗？ John Collison (01:17:43) Last question. We’re running up against time. As we talk about cultures and building them, what’s going on in the water in Hyderabad where the school that you went to, also Shantanu went there, Ajay Banga went there. Bunch of good chess players are similarly from there and southern India more broadly and things like that. But do you have any theory on the local outperformance? 纳德拉（01:18:13） 是啊。我们读的那所高中——说起来，在英伟达和黄仁勋崛起之前……如今黄仁勋一个人，就把我、阿贾伊和尚塔努都比下去了。事实上，宝洁现任 CEO 也毕业于我的高中。⁴ Satya Nadella (01:18:13) Yeah. The high school we went to, in fact until I would say Nvidia and Jensen, because Jensen has it now covered for all of us between me and Ajay and Shantanu. In fact, the CEO of Proctor Gamble today is also from my high school. 约翰·科里森（01:18:32） 你看，这就是个小圈子。 John Collison (01:18:32) See, it’s a cabal. 纳德拉（01:18:38） 确实有点像个小圈子。我觉得，在海得拉巴长大、七十年代末和八十年代初去那所偏远的学校读书，有件事很特别。我认为那给了我们更大的空间。看看我们每个人，学业当然重要；但说实话，我们几乎每个人在学业之外都有特别擅长的领域。当时在那个国家，这相当罕见。所以我很大程度上把这归功于我的高中，因为我觉得那里给了我们更多空间，让我们去追随后来真正成为热爱的事物，而且有时间慢慢发现它，而不是觉得“嘿，我必须加入某种竞赛”。 Satya Nadella (01:18:38) It’s kind of a cabal. I would say one of the fascinating things about growing up in Hyderabad and going to that school in the middle of nowhere at that time in the late seventies and the early eighties. I would say, I think it gave us a lot more space. If you look at even each of us, academics was a thing, but quite frankly, we mostly, all of us had things we excelled at a lot of other things beyond academics, in fact. That was a pretty rare thing at that time in that country, and so I attribute it a lot to my high school because I feel that it is a place where it gave us a lot more space and room to follow what really became your passion, but you were able to take your time to discover it. Versus sort of feeling that, hey, I had to join some kind of a race. 约翰·科里森（01:19:25） 当时没有那么早就把人引上预设的升学和职业轨道…… John Collison (01:19:25) It wasn’t as tracked as— 纳德拉（01:19:26） 没错。 Satya Nadella (01:19:26) That’s right. 约翰·科里森（01:19:28） 对。你高中时最热衷的是什么？ John Collison (01:19:28) Right. What was your passion in high school? 纳德拉（01:19:29） 板球。事实上，就是这个。对了，还有塞缪尔·贝克特。 Satya Nadella (01:19:29) Cricket, in fact, this by the way. Yeah, Samuel Beckett. 约翰·科里森（01:19:34） 是啊，我想听听这个故事。 John Collison (01:19:34) Yeah, so I want to know this story. 纳德拉（01:19:35） 好。要问有哪位名人参加过职业体育比赛？我想他曾代表都柏林大学打过一两场比赛，也参加过一流板球赛；所以他是唯一一位打过职业板球、又获得诺贝尔奖的人。⁵ Satya Nadella (01:19:35) Sure. If you asked the question, who is the one sports person who played professionally? I guess he played one or two matches for I guess the Dublin University, and he played first-class cricket, and so he’s the only person who played professional cricket and won a Nobel Prize. 约翰·科里森（01:19:53） 真的？这太有意思了。看来什么都能兼得。瞧，这就是那个时代的国际象棋拳击之类的吧。真棒。你当年也差一点，不过那是另一种人生里的你。 John Collison (01:19:53) Really? That’s really funny. So you can have it all. There you go. The chess boxing of its day or something. That’s awesome. Well, you came close, but another life that could have been you. 纳德拉（01:20:09） 非常感谢。今天聊得很愉快。 Satya Nadella (01:20:09) Thank you so much. It’s such a pleasure. 约翰·科里森（01:20:10） 谢谢你，萨提亚。 John Collison (01:20:10) Thanks Satya.

继续阅读：上篇：企业 AI、工作方式与互联网往事 · 中篇：算力瓶颈、企业主权与智能体商务

译注

原稿写作“4.0”，主持人自己也带有疑问语气。结合撤下模型后的用户反应，推测指 GPT-4o；OpenAI 2025 年 8 月 12 日发布记录记载了 4o 回到付费用户模型选择器。正文保留原稿说法，不把推断写成确定的原话。 ↩
原稿写作“cogs”。结合为任务自动选择模型、权衡成本与能力的语境，此处按 COGS（提供服务的成本）理解，而不是把它译成认知或推理能力。 ↩
这是访谈当时的说法，不表示永久的平台独占。《Microsoft Flight Simulator 2024》PS5 版已在 2025 年 9 月的官方公告中宣布，正式发行日为 2025 年 12 月 8 日，晚于本期节目发布日期。 ↩
此处“现任 CEO”的时间表述不够准确：节目发表于 2025 年 11 月 18 日，Shailesh Jejurikar 当时已获任命，但到 2026 年 1 月才接任宝洁 CEO，参见宝洁官方履历。前半句是在调侃黄仁勋与英伟达的成就，不能理解成黄仁勋也毕业于这所高中。 ↩
都柏林圣三一学院的贝克特生平介绍确认，他在 1925、1926 年参加过两场 first-class cricket，并称他是唯一收入《威斯登板球年鉴》的诺贝尔奖得主。first-class 是正式赛事等级，本身不等于球员具有职业身份；正文的“职业”保留纳德拉原话。 ↩

【Cheeky Pint】算力瓶颈、企业主权与智能体商务 | 纳德拉与 John Collison 对谈 | 中英全文（中）

Wed, 07 Oct 2026 00:00:00 +0000

书童按：这一篇从 AI 投资是否重演互联网泡沫谈起。纳德拉区分闲置的暗光纤与当下供不应求的计算基础设施，又把“主权”从数据存放在哪里，推到企业的隐性知识究竟属于谁。随后，对话转向 Excel 的生命力，以及从商品发现、购买到客服，智能体可能怎样改变整个交易过程。节目：Cheeky Pint 第 19 期；主持人 John Collison，嘉宾 Satya Nadella。原节目发布于 2025 年 11 月 18 日，本文中的“今天”“现在”均沿用当时语境。 本篇范围：00:27:03—00:53:47。原节目与音视频 · 英文转录本系列据完整英文转录逐段翻译，中文在前、英文在后；保留原稿时间戳，长发言按语义分段。Luna 初译，经 DeepSeek-V4.1-Flash 与 Qwen3.6-35B-A3B 从语言和忠实度两个维度交叉校审，再综合定稿。必要的事实背景与转录疑点另作译注；嘉宾的观点和判断保留原意。 系列目录：上篇：企业 AI、工作方式与互联网往事 · 中篇：算力瓶颈、企业主权与智能体商务 · 下篇：模型忠诚、产品捆绑与组织文化 本篇目录

AI 热潮与互联网泡沫
瓶颈是芯片，还是能通电的机房？
数据主权与企业的隐性知识
Excel 为什么如此长寿？
从云计算到 AI 的需求跃迁
商品发现、结账与智能体商务
购物与客服的边界正在消失

AI 热潮与互联网泡沫

The AI boom and the dot-com bubble 约翰·科里森（00:27:03） 嗯，我想聊聊这个，也想聊聊商务。不过先说说九十年代吧。现在大家都在拿眼下的情况和互联网泡沫作比较，几乎成了陈词滥调。我觉得这个比较其实挺合理，之所以会成为陈词滥调是有原因的：这是一场资本开支极其密集的建设，目标是打造一种新范式；它确实意义重大，但资本开支也着实惊人。你当时在微软，亲历了2000年前后的互联网泡沫。微软股价在九十年代末、两千年初见顶，我记得直到2016年左右才超过那个高点。1999年是什么感觉？尤其是，你当时知道自己身处泡沫之中吗？还是觉得“哦，这是新事物，这次情况不一样”？ John Collison (00:27:03) Well, I want to talk about that and I want to talk about commerce. But actually first, while we’re still in the nineties, everyone is making comparisons to the dotcom bubble right now. It’s almost a cliché, and I think it’s actually a reasonable comparison. It is a cliché for a reason, which is it is a very CapEx intensive buildout for a new paradigm that is in fact a big deal, and yet there’s an awful lot of CapEx. You were there at Microsoft during the 2000 dotcom bubble, and it really was, Microsoft’s share price peaked in the late nineties, early 2000s, and then didn’t surpass it until 2016, I want to say. What did it feel like in 1999? In particular, did you know you were in a bubble or was it like, “Oh, this is the new this time it’s different.” 纳德拉（00:27:52） 这很有意思。是的，我记得我们大概在2000年成了市值最大的公司，超过了通用电气。我记得这件事。那时可以说我们是轻资产公司，对吧？我当时可能更像萨姆·奥尔特曼：花的是别人的钱。说实话，回头看，那时候撇开金融周期不谈，长期趋势是很清楚的，这件事注定会发生——因为当时商业模式也已经开始成形。对微软来说，当时最大的教训是：天啊，我们最先要做的事——得做浏览器，得做网络服务器，到处都得有互联网协议。 Satya Nadella (00:27:52) It’s interesting. Yeah. In fact, I remember, I think we probably became the largest market cap company in 2000. We crossed GE. I remember that. Yeah, we were capital-light, let’s say, right? I guess I was more like Sam at that time, which is somebody else’s capital was being spent. It is, quite honestly, when I look back at it, at that time too, the financial cycle aside, it was clear. The secular trend was clear that this is going to—because even by then the business models were also emerging. Even for Microsoft, the biggest lesson at that time was, oh my god, even our first order of play—we’ve got to build a browser, we’ve got to build a web server, we’ve got to have internet protocols everywhere. 纳德拉（00:27:52） 我们在Office里也有FrontPage这样的网站制作工具。那些显而易见的事我们都做了，但我们意识到，只做显而易见的事不够。我们需要重塑自己正在做的事情，而且新的商业模式是什么，也已经很清楚。所以有意思的是，那一轮周期多少有些突然。我是说，它的起因无非是某种非理性繁荣之类的，但某种程度上，之后的修正把许多东西都冲走了。不过我会说，那些想法留了下来。所以我会由此想到现在发生的事。¹ Satya Nadella (00:27:52) We had a website builder inside the office with a front page. We did all the obvious things, but we realized that just doing the obvious things didn’t make sense. We needed to reinvent what we were doing, plus what are the new business models was clear. So in an interesting way, that cycle kind of came out of nowhere. I mean, it came out of what was just whatever irrational exuberance or what have you, but the correction in some sense washed away a bunch of stuff. But I would say the ideas persisted, right? And so to me, I think about what’s happening here. 纳德拉（00:27:52） 我觉得有两点。首先，正在铺设的基础设施，其需求来得更直接了。如今不像当年那样，先铺好暗光纤，然后等某家互联网公司发展到十亿用户，再来使用它；那种孕育期已经短得多。 Satya Nadella (00:27:52) I mean there are two things. The infrastructure itself that’s getting laid out, I think it’s got a lot more immediate. It’s not like even the gestation period of, okay, I built a dark fiber, which—and some internet company will first scale to a billion users and use. 约翰·科里森（00:29:44） 现在买这些东西的人队都排到门外了。 John Collison (00:29:44) There are lines out the door to buy this stuff. 纳德拉（00:29:45） 没错。所以坦白说，这一轮我们落后了。当时不是那样……如今我看我们的基础设施建设和需求，这正是人们说有泡沫时，我看着财报会想到的事：我上一次像现在这样，连供电就绪机房都严重短缺，是什么时候？² Satya Nadella (00:29:45) Exactly. And so this time around, quite frankly, we are behind. It was not that… When I look at our infrastructure build and demand today, that’s the thing that when people say there’s a bubble, when I look at my earnings, I can have… When was the last time I was so supply constrained on PowerShells? 约翰·科里森（00:30:04） 我以前没听过这种比较。别忘了，互联网泡沫说到底也是电信泡沫，是很大程度上的光纤泡沫。当年铺的是暗光纤，名字就说明了一切：它是暗的，还没有启用。而现在的情况和暗光纤恰恰相反。 John Collison (00:30:04) I haven’t heard that comparison before, which is, let’s not forget that the dotcom bubble, which again was a telecoms bubble, it was a fiber bubble in a big way. It was dark fiber. The clue is in the name. It was dark. It was not lit up yet. And this is anything but dark fiber. 纳德拉（00:30:20） 是啊，现在不会有人坐在那里说：“嘿，我把GPU都接好了，却没人用。”我没有利用率方面的问题。我可能还有电能利用效率（PUE）方面的问题。我希望利用率更高，主要是因为内存瓶颈之类的原因。但我手上的东西全都供不应求。事实上，我的问题是得增加供给；至于能不能恰到好处地匹配需求？没人做得到。没有哪套供应链能让供需完美匹配。不过这一次的建设周期，考虑到漫长的交付周期…… Satya Nadella (00:30:20) Yeah, it’s not like any one of us is sitting there and saying, “Hey, I have all the GPUs wired up and nobody’s using them.” I don’t have a utilization problem. I may have a PUE. I want higher utilization, mostly because it’s memory bottleneck or what have you. But there is not a thing that I have that’s not sold out. In fact, my problem is, I got to bring more supply and in that will we perfectly get it? No one does. There’s no supply chain operation that perfectly matches demand and supply. But this time around the buildout, given the long lead. 纳德拉（00:30:20） 比如，我们会仔细研究的一件事，是怎么向华尔街说明我们的资本开支。你得记住，有些资产能用二十年，有些东西的寿命只有四五年。对这些东西，做决策的方式也得不同。空着一座尚未配齐供电等设施的机房，并不是什么大不了的事。对，就像一座园区里有五栋楼一样，这不会成为微软资产负债表上的问题。真正的问题会是：没有已经具备供电条件、随时可以启用的机房。 Satya Nadella (00:30:20) For example, one of the things we study a lot is even when we talk about our capital, we try to describe it even to the Street. Hey, you got to remember some of these assets are 20 years, some of these things are four years or five years. And in fact, you kind of have to make the decisions on those things differently. Having a cold shell that’s unused is nothing. Yeah, it’s kind of like having a campus with five buildings. It’s not going to be a problem on Microsoft’s balance sheet. What our real problem would be, hey, not having warm shells that we can light up.

瓶颈是芯片，还是能通电的机房？

Chips, power and datacenter capacity 约翰·科里森（00:31:35） 如今瓶颈在哪里？是电工、机房，还是涡轮机？ John Collison (00:31:35) Where is the bottleneck these days? Is it electricians? Is it shells? Is it turbines? 纳德拉（00:31:42） 是啊。眼下的瓶颈，就是能否拿出一批已经接通电力的机房，对吧？如果没有足够多的机房接好电，我就没法把机架搬进去，再把它们部署成可运行的设施。这是交付周期很长的环节：你得及时拿到土地许可、电力许可，把所有事情办妥。还有地点的问题。所以我觉得有一点常被忽略：当然，我们在美国建了很多，但我们必须在全球各地建设，而且各地都有数据监管规定。 Satya Nadella (00:31:42) Yeah. The product that is the bottleneck is just a bunch of powered-up shells, right? So if I don’t have enough shells that are powered that I can then roll in my racks and then break them operational. And that’s the long lead part, which is you kind of have to have the land permits, the power permits, get all that done in time. And by the way, location. So I think one of the things that’s glossed over, of course stateside, in the United States, we are building a lot but we have to build all over the world and there are data regulations. 纳德拉（00:31:42） 事实上，越来越多的人非常重视主权问题。因此，我们必须确保自己的基础设施资源池是全球化的，能够应对各种工作负载，从训练、数据生成（DataGen）到推理。这是个涉及多个变量的复杂问题。 Satya Nadella (00:31:42) In fact, more every day people care about sovereignty in a major, major way. And so therefore for us, we have to make sure that the fleet is a global fleet, a fleet that can deal with all types of workloads training to DataGen, to inference. And so it’s a complex, multi-variable thing.

数据主权与企业的隐性知识

Sovereignty and the tacit knowledge of a firm 约翰·科里森（00:32:42） 谁应该在意数据主权？爱尔兰有不少数据中心，但并没有特别执着于数据只能留在爱尔兰。我也不觉得它非得对此特别执着。不过，你们是照各国的要求办，还是会建议他们要不要追求数据主权，以及哪些人应该追求？ John Collison (00:32:42) Who should care about data sovereignty? Where Ireland has a bunch of data centers but is not particularly wound up on the idea that data should only be in Ireland. And I don’t think it should be super wound up about that fact. But I guess do you guys just go with whatever the country wants or do you try to advise on whether you should want data sovereignty or not and who should? 纳德拉（00:33:04） 我觉得这显然是几乎每个国家、每位政策制定者都很关心的议题，而且确实有正当理由。在人工智能时代，我对主权的思考也有些不同了。我的意思是，主权最终关乎的是企业的未来，对吧？如果追问科斯定理的核心，就会发现：哇，等等，如果模型无所不知，那我为什么还需要……按理说，我得有某种隐性知识，才能让组织内部的交易成本低于在市场上进行交易的成本。这些问题很烧脑。³ Satya Nadella (00:33:04) Yeah, so I think it’s obviously a topic that’s top of mind for pretty much every country, every policymaker, and they care. And there’s obviously real legitimate reasons. The thing that I’d say in the AI age, I’m now thinking a little bit differently even about sovereignty. What I mean by that is the ultimate sovereignty question is more of what’s the future of a corporation, right? I mean, if you sort of start to go to the core of the Coase theorem, you say, “Wow, what the heck? If the model is the thing that knows everything, why do I even… I’m supposed to have some tacit knowledge that makes the transactional costs inside my organization lower than just being in the marketplace.” So they’re a mind bender. 纳德拉（00:33:04） 所以我有一种想法：真正重要的主权，是在模型不断学习、并产生规模报酬递增的时代，公司自身的主权。于是我越来越觉得，公司得有自己的智能层——它可以是模型的脚手架，也可以是嵌入模型的权重。这样用的就不是别人的基础模型。关键在于，你对自己的基础模型有没有主权？所以我的新想法是，未来的公司会拥有自己的基础模型，捕捉那些隐性知识；这些知识能让组织内部积累和传播知识的交易成本更低、速度更快。关于主权，我就讲这么一大段。 Satya Nadella (00:33:04) So in fact, one of the ways I think is, the sovereignty that matters is your company’s sovereignty in an age where there are continual learning increasing returns to a model. So I’m increasingly thinking that hey, the company’s ability to have that intelligence layer that’s a scaffold or even weights embedded in the model. So it’s not somebody else’s foundation model. It’s about do you have sovereignty in your foundation model? So my new concept is the future of a company is that company has its own foundation model that captures essentially the tacit knowledge that makes the transactional costs of how knowledge gets accrued and diffused inside the organization faster. So that’s sort of a long speech on sovereignty. 约翰·科里森（00:34:35） 嗯，这里有两个方向……这想法很有意思。人工智能或许会改变公司的性质。你说，有些公司本来就是一组知识产权，对吧？比如迪士尼，或者我们请来过的大卫·里克斯所在的礼来，在很大程度上就是一家知识产权公司。有些公司本来就是知识产权的集合，只是现在这些知识产权散落在邮件、文件里，最重要的是在人们的脑子里；也许随着时间推移，它们会集中到一个模型里。我原本以为你要说的是：人们常指出，现在的公司仍然沿袭制造业公司的模式，比如阿尔弗雷德·斯隆那一套，尽管如今我们做的是知识工作，而不是在一条小型生产线上干活。 John Collison (00:34:35) Well, there’s two versions… That’s very interesting. The idea that AI maybe just changes the nature of companies, and you are saying that if some companies are already collections of IP, right? Disney or we had Dave Ricks from Eli Lilly here, that is an IP company in a big way. And some companies are already collections of IP, but right now that IP is in all the emails and documents and people’s heads most importantly, whereas maybe the IP could be in a single model over time. Where I thought you were going to go with that is just maybe the—people point out a lot that current companies are modeled after manufacturing companies and Alfred Sloan type stuff, despite the fact that we’re doing knowledge work today and not running a little manufacturing line. 约翰·科里森（00:34:35） 公司会不会变得更奇特？会不会出现那种著名的、只有极少数员工却价值十亿美元的公司？会不会有更多高度分布式的互联网公司？会不会出现一些去中心化自治组织？我以为你要说的是这个方向。 John Collison (00:34:35) And do you get more just weird-looking companies? Do you get the famous really tiny billion-dollar company? Do you get more highly distributed internet companies? Do you get some DAOs? I thought that’s where you’re going to go with that. 纳德拉（00:35:43） 我觉得这些也都有可能。组织结构本身可能会变化，像那种几个人、甚至一个人就能创办的十亿美元公司，也许会出现；去中心化自治组织也可能出现。但至少对我来说，有意思的问题是：隐性知识存在于哪里？显然，它存在于人的头脑里，那是不断积累、叠加的经典诀窍。我认为，它也会以权重的形式存在并积累于某个低秩适配（LoRA）层里，而且这个层是你们公司独有的。我觉得，未来礼来、微软或 Stripe 的新知识产权，除了员工和我们拥有的其他资料之外，我们也会说：“哦，它们还存在于某种嵌入表示里。” Satya Nadella (00:35:43) I think that those are also possibilities. So the structure itself could change and it’s going to be more possible for whatever the few, the one-person billion dollar company, what have you, maybe could happen or DAOs could happen. But the interesting question, at least for me, is where does tacit knowledge reside? Clearly it resides in people’s heads and it’s the classic know-how that accrues and compounds. I think it’ll also reside and compound as weights in some LoRa layer that is unique to your company. I feel like the new intellectual property at Eli Lilly or at Microsoft or at Stripe at some point can be also, besides all the humans, besides all the other artifacts we have, I think we’ll also say, “Oh, they are in some embedding.” 约翰·科里森（00:36:42） 对，明白。你这么说让我想到 Stripe，这家公司挺有意思。它本身并没有很强的网络效应。我们刚开始做 Stripe 时，它很大程度上是一种单用户 API 体验。我们让大家很容易开始用 Stripe，但最终你根本不会知道还有谁在用它。随着规模扩大，我们建立起了一个信任网络：由于我们见过大多数互联网用户，因此能阻止欺诈。我们知道正常和异常的情况是什么样的；甚至只因为没见过你，就会觉得有点可疑，因为大多数人我们都见过。 John Collison (00:36:42) Yes, okay. It’s funny you say this because Stripe is interesting. It does not really have strong network effects as a company. When we started building up Stripe, it was very much a single-player API experience. And we make it easy to start using Stripe, but ultimately you’d never know that anyone else was using Stripe. What’s happened as we’ve scaled up is we now just have a trust network where we can prevent fraud by virtue of the fact that we’ve seen most internet users. And so we have a knowledge for what good and bad looks like, and even the fact that we haven’t seen you before is inherently a little bit suspicious because we’ve seen most people. 约翰·科里森（00:36:42） 于是它成了一个声誉网络，有点像谷歌的 reCAPTCHA，后者后来也成了声誉网络。总之，我们现在正在训练一个支付基础模型，用上 Stripe 网络里的所有数据。这样一来，模型的规模和能力都更强，也能把这些因素考虑进去……总之，我们正在做的，正是你所说的那件事。 John Collison (00:36:42) And so it becomes a reputation network, kind of like reCAPTCHA for Google, similarly became a reputation network. Anyway, what we’re now doing is training a payments foundation model where we’re using all the data that we have in the Stripe network and you have a much larger, more capable model taking into account… So anyway, we are trying to do exactly what you’re saying. 纳德拉（00:37:40） 所以我们所有人都面临一个问题：怎样防止这些知识泄漏到基础模型里？模型是不是只差一步就能获得这项能力，因为它学会了如何识别欺诈？还是说其中还涉及某种多维结构？我认为这是关键问题。我觉得有两种观点。一种观点认为模型会吞噬整个世界。你很容易就会想：没错，说到底万事万物都是模式，我把所有模式都学会就行了，诸如此类。 Satya Nadella (00:37:40) And so one of the questions for all of us is how do you protect that from essentially leaking over to the base foundation model? Is it just like one capability hop away because it learned how to even do fraud detection? Is it just some other multidimensional, or not? And that I think is the key question to me. I think there are two arguments. One argument is that argument that the models are going to eat the world. You can kind of easily, oh yeah, after all, everything is just a pattern and I’ll learn it all and what have you. 纳德拉（00:37:40） 不过，正如你谈到 Stripe 时说的，它可以调用多个模型，构建一个不可思议的、以模型为核心的欺诈检测层。除此之外，还有一套完全属于 Stripe 的记忆、工具使用能力和行动空间。对我来说，这就是企业的未来，无论它是制药公司、支付公司还是软件公司。我认为这正是我们所有人正在做、也将继续做的事。对我来说，这就是主权。 Satya Nadella (00:37:40) But then the thing though is, to your point about Stripe, it can take multiple models, build this unbelievable, sort of, I’ll call it fraud detection layer that is model-forward. And then there is this memory and tools use and action space that’s all unique to Stripe. That to me is the future of a corporation, whether it’s a pharma company, a payments company, or a software company. That I think is the work that we all are doing and will do. And I think that to me that is sovereignty.

Excel 为什么如此长寿？

Why Excel endures 约翰·科里森（00:39:13） 我还在想我们刚才聊到的、面向非软件工程师的集成开发环境。我觉得未来十年，可能会出现一款面向财务人员的产品；回头看，它的界面显然是正确的，但当初电子表格作为一种界面，也是突然冒出来的东西。当时可能也让人觉得它凭空出现。我对此印象很深。说到电子表格，对一些软件公司来说，挑战 Excel 就像是成人礼；可 Excel 似乎四十年来一直做得很好。它为什么这么经久不衰？ John Collison (00:39:13) I’m still thinking about this discussion we’re having about the IDE for people who aren’t software engineers. And again, I feel like there could be a product in the next 10 years for finance people, where in hindsight it is obviously the correct UI, but just the spreadsheet, it kind of came out of nowhere as a UI. It may feel like it came out of nowhere at that time. I’m really struck by that. Speaking of the spreadsheet, it’s like a rite of passage for certain software companies to try to take on Excel and it seems to be doing pretty well 40 years in, or what have you. Why is it so durable? 纳德拉（00:39:50） 是啊，真不可思议。某种程度上说，表格这种形式……我觉得关键在于列表和表格的力量，以及软件的可塑性，这两者结合得恰到好处。这就是为什么那块闪烁的空白画布总会存在。我们或许会给它加上许多功能。电子表格也一样。还有一点是，电子表格是图灵完备的，我们没有充分认识到它有多强大。你可以用它做出…… Satya Nadella (00:39:50) Yeah, it’s unbelievable, right? I mean, at some level the idea that a tabular form… I mean I think it’s the power of lists and tables. It’s just a perfect—and the malleability of software that was, I think, the combination. That’s why a blinking canvas, it’s always going to be there. We may add lots of bells and whistles to it. And the same thing with spreadsheets. The other thing about the spreadsheet is it’s Turing complete. We don’t give it enough credit. It’s like I can make it— 约翰·科里森（00:40:32） 我觉得它是全世界最容易上手的编程环境。 John Collison (00:40:32) I think it is the world’s most approachable programming environment. 纳德拉（00:40:35） 完全同意。你甚至不会觉得自己在编程，就已经开始用了。这也是它另一个妙处。如今我们还是把人工智能弄得神秘兮兮的。你我之前聊过，天啊，我们需要变革管理。可新一代电子表格出现时，没人谈什么变革管理，大家直接就用起来了。还有件事：有人跟我说，我当时在和忠利集团（Generali）的首席执行官见面。他是在传真机时代加入 Generali 的，当时管理着公司所有保险代理人。他对我说：“我还记得电子邮件和 Excel 刚出现的那一天。整套工作流程彻底被颠覆了，随后从根本上逐步演变、改变。”所以我觉得，这正是你刚才说的：这个时代会出现哪些东西，让我们从根本上重新审视工作本身、工作产物和工作流程？ Satya Nadella (00:40:35) One hundred percent. I mean it’s like… And you get into it without even thinking you’re programming. And that is the other beauty, which is AI still, we’ve mystified it. You and I talked about, oh my God, we need change management. When the next spreadsheets came, nobody talked about change management. They were just using it. And that to me is the other thing, which is—somebody was describing to me, I was meeting the CEO of Generali. He joined Generali during the fax machine era, and he was managing all their insurance agents. And he said to me, “Look, I still remember the day when emails showed up, Excels showed up, and the entire workflow of how things happen completely were upended and it evolved and changed ground up.” So to me, I think that’s to your point, what are those things of this era that we’ll discover that’ll allow the ground-up relitigation of the work, the work artifact and the workflow.

从云计算到 AI 的需求跃迁

From cloud growth to the AI wave 约翰·科里森（00:41:34） 现在做软件真是个有意思的时代。跟五年或十年前相比，确实有意思多了，你一定也有这种感觉。 John Collison (00:41:34) It’s such an interesting time to be in software. I mean compared to, you must feel this, it’s just a much more interesting time now than five or 10 years ago. 纳德拉（00:41:42） 确实很有意思。当年我们满口都是“云、云、云”。如果你问我，2019年最热门的是什么，我会说我们做出了一款很棒的支持多个区域、甚至让用户不必操心区域划分的多格式数据库 Cosmos DB。它基本上是个 JSON 数据库，里面也有 SQL，什么都能装。我们当时谈的就是这种不用操心区域划分的能力，诸如此类。接着疫情来了，云计算又突然加速。谢天谢地，Teams 就此成为了热门产品。这是当时令人兴奋的事。没想到疫情结束后，大家会想：“哦，我以为我们会进入某种稳定状态。”我记得当时还预测过云的需求。我们在想该怎么办：疫情期间建得太多了。之后有整整八个月，我们都在想：“哎，结果现在又来了这么一波。” Satya Nadella (00:41:42) It is interesting, we were like “Cloud, cloud, cloud.” And if you had to ask me what is the hottest thing in 2019, we had built this fantastic multi-region or region-less database that was multi-format. Cosmos DB, which was like we had basically a JSONdatabase. We had a SQL in there. It was the everything database. And we were thinking, it was region less and blah, blah, blah. And then the pandemic happened, and then the cloud went into another hyper drive. I mean, Teams, thank God, just became the thing. So that was the exciting thing, and lo and behold, you come out of it and you sort of say, “Oh, I thought after the pandemic we’re going to get to some stable state.” In fact, I remember a forecast of the cloud. We were saying, what do we do? We overbuilt during the pandemic and there was a good eight months where we were like, “Oh, and then this thing now has come too.” 约翰·科里森（00:42:51） Stripe 有不少图表能看出这种变化，不知道微软当时是不是也这样。很明显，2020年3月突然跃上了一个台阶，对吧？电商活动大幅增加，我们也看到线上商家创建的速度变快。原本只做线下生意的商家说：“哦，我们得转到线上销售。”这种水平就一直维持在高位；显然之后还继续增长了。人们回到实体办公室后，相关活动也没有相应回落。就像是突然跳上一个台阶，然后一直留在更高的水平。我敢肯定 Azure 也看到了类似情况。 John Collison (00:42:51) To, there’s a lot of charts of the shape at Stripe. I don’t know if it was this way at Microsoft where obviously March 2020, you saw this discontinuity, right? Much more e-commerce activity happening, and we saw the rate of online business creation. You had businesses that were offline only saying, “Oh, we’ve got to switch to selling online.” And it just stayed at that elevated level forever, obviously since it’s gone up from there. But there was no matching decline as people went back into physical offices and things like that. It was just a step change and then it stayed at the elevated level forever. I’m sure you saw similar things in Azure. 纳德拉（00:43:25） 完全是这样。是啊，需求从来没降下来。 Satya Nadella (00:43:25) One hundred percent, yeah. It never came down.

商品发现、结账与智能体商务

Discovery, checkout and agentic commerce 约翰·科里森（00:43:26） 既然我们在聊商务，那不妨也聊聊我们正在合作做的事。 John Collison (00:43:26) We’re talking about commerce, so we might as well talk about what we’re working on together. 纳德拉（00:43:30） 我们对此非常兴奋。怎样搭建对商家友好的交易基础设施？怎样搭建对消费者友好的交易基础设施？两者能不能完美衔接？这一直是大家在思考的问题。对话式商务也是人们谈了很久的概念。现在有了你们和其他团队的工作，我们终于可以把商家和终端用户连接起来，提供智能体式的体验了。现在还处于早期阶段，体验必须做得得体，也必须赢得用户的信任。所以我非常期待。 Satya Nadella (00:43:30) We are very excited about it. I think the idea that has always been there, which is what’s the best way for a merchant-friendly set of rails and what is a customer-friendly set of rails? Is there a perfect matching? A conversational sort of commerce is a thing that people have talked about. And now I think with the work you all have done and others have done, we kind of can really bring the merchant and the end user and have this agentic sort of experience. So it’s early days, it has to be tastefully done. It has to be done in a way that you earn the user’s trust. And so I’m very excited about it. 约翰·科里森（00:44:12） 是的，我们看到这里有两点不同。过去有人尝试过在 Twitter、Instagram 等平台上买东西。但现在不一样：第一，有了人工智能，商家的接入要容易得多，比过去尝试这类做法时省力得多。第二，我觉得这种体验对终端用户非常有吸引力。我们从最早的一批客户那里已经看到初步数据。几周前我们也在 ChatGPT 里上线了。这条路肯定行得通。数据已经说明这一点，因为消费者用起来轻松多了。 John Collison (00:44:12) Yeah, we see two differences here because there have been previous attempts at buying on Twitter, buying on Instagram and these kinds of things. But what’s different here is one, you have AI. So all the integrations for the merchant are much easier. It’s much less of a lift than previous times when things like this have been tried. But then secondly, I just think the experience is so compelling as an end user. We’re already seeing this in the early data from the super early customers that we have. We launched a few weeks back in ChatGPT as well, that it has to work. And again, the data is already bearing that out because it’s so much easier as an end customer. 纳德拉（00:44:56） 是啊，我一直在聊这件事。我是个板球迷，总是在找各种东西。问题是，不管用亚马逊、沃尔玛还是别的平台，网站上的搜索体验有时都很费劲。有意思的是，这些聊天式体验一开始就很出色。而且它们会链接回商品目录——目录仍然是核心，但如果我能把结账和商品目录结合起来，我觉得这就是实现无缝体验的关键…… Satya Nadella (00:44:56) Yeah, I’ve been talking about it. I’m a bit of a cricket nut, so I am always searching for something. And the problem is whether it’s Amazon or Walmart or what have you, the search experience sometimes is hard on the site. So interestingly enough, these chat experiences first are fantastic. And the fact that they point back to the catalog, I mean the catalog is still king, but now if I can marry the checkout and the catalog, and that to me is where I think the seamlessness— 约翰·科里森（00:45:25） 你有没有类似的体验？我发现有些情况下，用人工智能应用研究商品，比基于关键词的搜索好太多了。真不可思议，直到去年我们居然还觉得，用关键词搜索来找东西是可以接受的。 John Collison (00:45:25) Do you have any experiences, I’ve found versions of this. I’m curious if you’ve had experiences where for product research using an AI app is so much better than keyword-based search. It’s amazing that up to last year we thought keyword-based search was an acceptable way to hunt for anything. 纳德拉（00:45:44） 是啊。对商家来说，归根结底，这就像是为你量身打造了一份商品目录。它给出的回答不像搜索结果页那样罗列一堆链接。 Satya Nadella (00:45:44) Yeah. And the seller, the bottom line is, it’s kind of like it is creating a custom catalog for you. I mean, the response is not like a SERP. 约翰·科里森（00:45:54） 我们家买家具时会说：“这个位置有这么大一块空间。你觉得放什么家具合适？要尺寸符合这里的条件，摆在这里也好看。”可我们以前居然做不到这些，真不可思议。你懂我的意思吧？现在可以定制得这么细致，还能描述整体感觉和审美，比如“我想找稍微高档一点的，但别太奢华”。以前竟然做不到这一点，真不可思议…… John Collison (00:45:54) We were buying furniture in our house and we were just saying, “Oh yeah, we have this much space available in this spot. What do you think is a good piece that would look good in that spot that meets these dimensions and things like that.” But it’s crazy that we weren’t doing that previously. You know what I mean? And so all this customization, being able to give vibes, general aesthetics, I’m looking for something slightly higher-end, but not super fancy. It’s crazy that you weren’t able to— 纳德拉（00:46:23） 顺便说一句，还有件事也特别不可思议。我妻子是建筑师，她有个 Copilot 笔记本，里面存着很多建筑图片之类的资料。你可以问它一些需要较高层次推理的问题，比如“这里应该放什么”。它能看懂建筑草图和图纸，再结合公开的家具目录，把两者放在一起分析。这类能力真是神奇。 Satya Nadella (00:46:23) By the way, that’s just the other crazy, crazy thing. My wife’s an architect, and so she sort of has this Copilot notebook in which she has all these architectural pictures and so on. And you can ask it quite high-level reasoning questions on what I should put in there. So it’s able to take an architectural sketch, a drawing, and then take a public catalog of furniture and put those things together and reason about it. And that type of stuff is pretty magical. 约翰·科里森（00:46:51） 在 Stripe，我们对 AI 改变商务这件事深信不疑，也认为会有大量交易转移到这里。我们和商家的交流也印证了这一点。我是这样看的：如果你要开放式地探索——比如“我想为某个场合买套衣服，但还不确定具体想要什么”——人工智能会比现在的体验好得多。现在你得逐一点开搜索结果之类的内容。另一方面，如果你要精准搜索，比如找一件符合特定要求的东西，或者给自行车买某个零件。 John Collison (00:46:51) Our view on this, we are as really AI-pilled when it comes to commerce at Stripe, and we think a huge amount will move here. And all the merchant conversations we’re having are bearing that out. And the way I think about it is that if you are doing open-ended discovery: “Oh, I’m interested in an outfit to buy for this occasion. I don’t know exactly what I want.” AI will be so much better at helping you with that than the current experiences where you’re clicking through a list of search results or something like that, and then if you’re doing targeted search where I’m looking for a specific object that meets these needs, I want this component for my bike. 约翰·科里森（00:46:51） 在这种情况下，你也可以用人工智能准确描述搜索条件，效果会好得多。你会想：等等，如果无方向的探索都被涵盖了，高度明确的搜索也被涵盖了，那不就是互联网上所有的商务活动吗？我觉得唯一剩下的，是那些定期回购的日用品，比如“我得再订些宠物粮”。这类需求受到的影响可能最小。当然，最初你还是得找到适合的宠物粮品牌。不过，大致上这就是我们的想法。Etsy 是非常棒的首批合作伙伴，因为它的商品都是定制的，对吧？ John Collison (00:46:51) Then also being able to specify with AI the exact parameters of the search you have will be much better. You’re like, “Wait, if you’re taking all of the undirected discovery and if you’re also taking all of the highly directed search, isn’t that just all commerce that happens on the internet?” I think the only thing that’s left that’s out of that is like recurring staples. I need to order more pet food. That feels to me like the least affected, though of course, you have to discover the brand of pet food at some point originally, but yeah, that’s kind how we’re thinking about it and again, Etsy has been an awesome first partner because all the products are custom, right? 纳德拉（00:48:06） 是啊，我觉得很有道理。发现商品这一环，Instagram 等平台显然做得很好。所以问题是，发现商品的那一层会是什么？其中一个显而易见的方向，是个性化的商品发现和灵感推荐。Pinterest 做的事就很有意思；把类似的发现层和对话界面结合起来，或许很有潜力。 Satya Nadella (00:48:06) Yeah, that makes a ton of sense to me. I mean the discovery part, which obviously people like Instagram and others have done a great job. So the question is, what’s the discovery layer? That’s one of the obviously personalized discovery layer inspiration for product. What Pinterest has done is interesting, so some layer like that married with this conversational interface. 约翰·科里森（00:48:32） 当然，这会是水涨船高、大家都受益的事。我们正在做的一部分工作，是让商家的商品目录、库存等信息能被远程发现，也能远程购买；消费者不一定非得走完商家网站上的整套流程，可以直接在 Copilot 这种挥挥魔杖般的体验里完成。从最基础的技术层面来说，这就是我们在做、在打通的东西。我觉得接下来令人兴奋的是，Pinterest 多年前——可能十年前——就尝试过电商。 John Collison (00:48:32) Well, and of course it’ll be a rising tide that lifts all boats, where part of what we’re doing with this is making merchant’s product catalogs remotely discoverable and inventory and everything like that. And then remotely purchasable, where you don’t necessarily have to go through the whole flow on their site and everything like that. You can just do it inside the magic wand Copilot experience. And so that is at the raw nuts and bolts level, what we are doing and what we’re wiring up. I think then what’s exciting is that again, Pinterest played with commerce quite a few years back, maybe 10 years back. 约翰·科里森（00:48:32） 它当时没能发展成很大的业务。但如今，如果所有商家都通过这个协议开放自己的商品目录，那么 Pinterest、Instagram 和 Twitter 这样的社交平台就能再试一次这种商务体验，因为商家对它的支持和采用会比上次多得多。 John Collison (00:48:32) It hasn’t taken off as a huge thing, but now if you have all the merchants who are offering their product catalogs as part of this protocol, then social sites like Pinterest and Instagram and Twitter get another run at this kind of commerce experience because you’ve way more merchant support and adoption for it than you had the first time around. 纳德拉（00:49:25） 我们有个名为 NLWeb 的项目，想做的是让每个商家的每份商品目录都有一个网站式的 NLWeb 界面，让智能体能够与之对话、查询，并进行所谓的深度搜索。因为如今最大的挑战之一，是商品目录的质量，以及能否用推理能力进行深度搜索。如果能解决这个问题，就像你说的，每件商品都能找到对应的搜索请求。⁴ Satya Nadella (00:49:25) And we have a project called the NLWeb, and the idea to really take every catalog of every merchant and give it essentially a website, an NLWeb interface that then an agent can talk to, to be able to interrogate and get the deep search, so to speak. Because today in some sense, one of the biggest challenges is the quality of the catalog and the ability to use reasoning to do a deep search. If you can solve that, then to your point, every product will find its query. 约翰·科里森（00:49:59） 是的，我们正在打造智能体商务平台，其中包括一些开源协议，比如我们的智能体商务协议。当然，人们也在使用 Stripe 的常规产品。从支付的角度看，这件事尤其棘手：我们希望人工智能应用能代表用户，在网上各种网站上付款，同时又不必把用户的所有支付信息分享到整个网络。这是个很有意思的支付问题。总之，我们希望打造智能体商务平台。你们显然很有经验。我们在这个还很新兴、但产品与市场契合度已经很明显的领域建设平台业务，你有什么建议？ John Collison (00:49:59) Yes, we’re building out this platform in agentic commerce where we have some open source protocols like our Agentic Commerce Protocol. We obviously have the regular Stripe products people are using us for. It’s particularly kind of tricky from a payments point of view because you’re looking to have an AI app do payments on behalf of other people across all these different sites on the web without probably sharing all your payment details all across the web. This is an interesting payment thing that we’re doing. Anyway, we’re looking to build a platform business in agentic commerce. You guys seem to know a thing or two. What advice would you have for us as we build in this very nascent space, but when there’s clearly product market fit? 纳德拉（00:50:42） 我觉得你们已经在做我会建议的事了。要思考的是，对每位商家来说，参与智能体工作流程意味着什么？今后每个商家都得找 Stripe 这样的服务商，说：“嘿，我有商品目录，也有结账功能，帮我用最顺畅的方式对接智能体。”如果这件事做得得体，我觉得商家就会选择 Stripe。我也认为商家入驻会是关键，因为我猜，大量中小商家只要点一下，说“帮我开通智能体商务”，就能推动这项业务发展。 Satya Nadella (00:50:42) I mean I think you have done that, which is one of the things that I would think is, what does it mean to participate in this agentic workflow for every merchant? So every merchant now will have to sort of come to someone like Stripe and say, “Hey, I have a catalog, I have a checkout. Please get me to meet agents in the most friction-free way.” And that done tastefully is why I would think I would hire Stripe for. And I think the merchant onboarding, because I’m assuming the long tail of merchants being able to click and say, “Hey, enable me for agentic commerce” is going to be the thing that’s going to drive. 纳德拉（00:50:42） 好消息是，参与者会有很多。显然 ChatGPT 是其中最大的一个，但 Google 会加入，我们也会加入，Meta、Perplexity 也会加入，竞争者会很多。会有许多入口平台充当聚合方。但更有意思的是，商家自己也会希望在自家网站或手机应用里支持自然语言查询。这些能力都得为他们配齐；或者，让我自己的智能体也能去查询这些东西。 Satya Nadella (00:50:42) Because the good news here is there is going to be multiple. I mean obviously ChatGPT is the big one, but there’s going to be, I mean Google’s going to be there. We are going to be there, Meta will be there, Perplexity, there’s going to be a lot of competition. There’s going to be a lot of front doors as aggregators, but the more interesting thing is they themselves will, on their website, or on their mobile app, will want to support natural language queries. And so all of that being enabled for or my own agents will go interrogate those things. 纳德拉（00:50:42） 所以我觉得，真正需要你们攻克、或者说解决好的，就是这个关键问题。你不能跑去找小商家，说：“嘿，你自己搭个 MCP 服务器，再实现这个协议、那个协议……”到底有没有一个“一键搞定”的办法？ Satya Nadella (00:50:42) So I think that that’s the key thing to be challenged, or rather really solved well. Because going to a small merchant and saying, “Hey, you go stand up an MCP server, do this protocol, that protocol…” What’s the “easy” button?

购物与客服的边界正在消失

The convergence of shopping and customer service 约翰·科里森（00:52:05） 我觉得我们还会看到另一种趋势——你可能已经看到了——就是各种智能体体验开始涌现。我们在聊智能体商务。我们请过 Intercom 的德斯·特雷纳，他的公司现在用人工智能提供客户服务，正用人工智能取代人工客服。他们显然发现了大量新增需求：用户起初只是为了那类帮助台问题而来，后来就会想，“哇，这其实是浏览网站的好得多的方式”，用起来几乎像命令行。现在它还不能执行那么多操作，但以后会越来越能做。我也在想，这些体验会不会融合：这边是不断发展、不断扩大的购物体验，可能还包括商品发现；那边是客户服务…… John Collison (00:52:05) I think the other thing that we’re going to see is—you’re probably seeing this already—emerging of a bunch of the agentic experiences. So we’re talking about agentic commerce here. We had Des Traynor from Intercom. They’re now doing customer service AI mediated and just replacing humans doing customer service with AI. But what they’re seeing obviously, is a huge amount of induced demand where people initially come for the help desk type queries and then it’s like, “Wow, this is honestly a much better way to navigate the website” and it’s almost like a command line. Anyway, it can’t quite take as much actions now as it will be able to, but I also wonder how much all these experiences merge where we’re doing the buying stuff over here that is growing and expanding and maybe there’s some discovery and things like that. They’re doing the customer service stuff over there— 纳德拉（00:52:51） 这会是通用的。 Satya Nadella (00:52:51) It’s universal. 约翰·科里森（00:52:52） 对，说得好。它什么时候会变成命令行应用？我还是觉得时尚领域很有意思。现在很多网站的技术体验糟糕得惊人。人们想买的是很讲究审美、整体感觉的东西，会说“我想找个类似这样的，但再精致一点之类的”，结果网站上还是只有关键词搜索和手动打标签。这样的领域显然很适合做交互式的人工智能体验。就像用 Midjourney 时，你会说：“不，这张图不太对，按这个方向改一下。”把这种方式用到购物上，我觉得会很有意思。 John Collison (00:52:52) Yeah, that’s a good point. When does it become a command line application? Again, my example of this is, I find the fashion space interesting where how incredibly poor the tech is with a lot of websites out there. Where people are trying to this very aesthetic vibe space, “I’m looking for something like that, but a little more fancy whatever,” and it’s all keyword-based search and manual tagging and things like that. And things like that feel to me perfectly set up for having an interactive AI-based experience where again, your Midjourney prompts, you’re like, “No, the image wasn’t quite right. Change it in this way.” Just doing that with commerce I think will be really interesting. 纳德拉（00:53:29） 有道理。我直觉上也觉得，大家都在做远程销售，客户服务也有远程销售的性质。所以这很说得通；在智能体的世界里，你完全可以把这些环节串起来，让它们之间的衔接不像今天这样生硬。 Satya Nadella (00:53:29) Makes sense. And I also think intuitively, all of us are inside sales, or other customer service is also inside sales. And so intuitively that makes sense and definitely in the agentic world you can stitch these things together so that the seams are not like what they are today.

继续阅读：上篇：企业 AI、工作方式与互联网往事 · 下篇：模型忠诚、产品捆绑与组织文化

译注

原转录把产品名写成“front page”。结合网站制作工具的语境，中文按 FrontPage 处理。微软资料将其列为可随 Office 提供或单独提供的产品，参见微软相关产品说明。 ↩
原转录写作“PowerShells”。结合本段机房供给的语境，以及 00:31:42 明确出现的“powered-up shells”，这里按已具备供电条件的数据中心机房理解；与微软的 PowerShell 命令行工具无关。这是上下文推断，英文保留原稿。 ↩
纳德拉原话是“Coase theorem”，故正文译为“科斯定理”；但紧接着谈到的企业内部协调与市场交易的成本比较，更直接对应科斯在《企业的性质》中提出的问题。参见科斯的诺贝尔奖演讲。 ↩
NLWeb 是微软于 2025 年 5 月公布的开放项目，旨在让网站提供自然语言交互界面。参见微软的项目介绍。 ↩

【Cheeky Pint】企业 AI、工作方式与互联网往事 | 纳德拉与 John Collison 对谈 | 中英全文（上）

Wed, 07 Oct 2026 00:00:00 +0000

书童按：AI 进入企业，难处究竟在模型，还是在数据、权限与记忆？这期 Cheeky Pint 里，John Collison 与微软 CEO 萨提亚·纳德拉从这个问题聊起，随后谈到纳德拉如何通过 Teams 了解公司、为什么持续走近开发者，以及人将怎样指挥成群的智能体。最后，两人回到九十年代：微软看到了互联网，却一度押错了实现路径。上篇保留节目开场的精彩片段剪辑，因此会与后文有少量重复。节目：Cheeky Pint 第 19 期；主持人 John Collison，嘉宾 Satya Nadella。原节目发布于 2025 年 11 月 18 日，本文中的“今天”“现在”均沿用当时语境。 本篇范围：00:00:00—00:27:03。原节目与音视频 · 英文转录本系列据完整英文转录逐段翻译，中文在前、英文在后；保留原稿时间戳，长发言按语义分段。Luna 初译，经 DeepSeek-V4.1-Flash 与 Qwen3.6-35B-A3B 从语言和忠实度两个维度交叉校审，再综合定稿。必要的事实背景与转录疑点另作译注；嘉宾的观点和判断保留原意。 系列目录：上篇：企业 AI、工作方式与互联网往事 · 中篇：算力瓶颈、企业主权与智能体商务 · 下篇：模型忠诚、产品捆绑与组织文化 本篇目录

开场剪辑
企业 AI：数据、记忆与权限
在虚拟走廊里管理微软
跟着开发者与初创公司走
智能体时代的界面：宏观委派、微观引导
互联网往事：看对范式，还要找对路径

开场剪辑

Opening highlights 纳德拉（00:00:00） 比尔一直对此着迷。我清楚记得他在九十年代说过这话。他说：“软件只有一个类别，叫信息管理。你得把人、地点和事物都整理成结构化数据，就这么简单。”问题在于，人很复杂。 Satya Nadella (00:00:00) Bill was always obsessed. I remember him distinctly saying this in the nineties. He said, “There’s only one category in software. It’s called information management. You’ve got to schematize people, places and things and that’s it.” The problem is people are messy. 约翰·科里森（00:00:13） 人们会忠于某个模型，还是会忠于某个 AI 品牌？ John Collison (00:00:13) Do people have loyalty to a model or do they have loyalty to an AI brand? 纳德拉（00:00:16） 你需要一组模型，再由智能体在它们之间协调，让这组模型满足你的需求。 Satya Nadella (00:00:16) You want an ensemble of models. You have agents intermediating that ensemble so that it meets your needs. 约翰·科里森（00:00:23） 大家最终的偏好，不会只是更强的智能吗？比如我会打开模型选择器，手动选 o3 来回答“我该去哪儿吃冰淇淋”这种问题。对一些软件公司来说，挑战 Excel 几乎是一种成人礼。它为什么能这么经久不衰？ John Collison (00:00:23) Will everyone’s preference not just be for more intelligence? I’ll go into the picker and manually select o3 for “where should I go get ice cream” query. It’s like a rite of passage for certain software companies to try to take on Excel. Why is it so durable? 纳德拉（00:00:38） 我们多少有点低估它了。就像我可以让它做到—— Satya Nadella (00:00:38) We sort of don’t give it enough credit. It’s like I can make him do— 约翰·科里森（00:00:41） 世界上最容易上手的编程环境。 John Collison (00:00:41) The world’s most approachable programming environment. 纳德拉（00:00:43） 完全没错。这里说的彼得是谁？ Satya Nadella (00:00:43) A hundred percent. And Pieter here is who? 约翰·科里森（00:00:46） 彼得·莱维尔斯。他算是个独立开发者—— John Collison (00:00:46) Pieter Levels. He’s like an indie— 纳德拉（00:00:47） 哦，对。彼得·莱维尔斯，我认识他。 Satya Nadella (00:00:47) Oh yes. Yeah. Pieter Levels. I know him. 约翰·科里森（00:00:50） 当然，你整天泡在线上嘛。看，萨提亚知道彼得·莱维尔斯是谁。这就是微软能成为一家 14 万亿美元公司的原因。数据中心有什么值得一看的吗？还是说，就只是“哇，好多机柜”？¹ John Collison (00:00:50) Of course, you’re so online. See, Satya knows who Pieter Levels is. This is why Microsoft is like a $14 trillion company. Was there anything good to see at the data center? Or is it like, that’s a lot of racks. 纳德拉（00:01:04） 那是最好玩的地方，哥们儿。 Satya Nadella (00:01:04) It’s the most fun place to go, man. 约翰·科里森（00:01:07） 萨提亚·纳德拉于 2014 年接任微软 CEO，但他在公司已经工作了三十多年，见证过许多变化。萨提亚执掌微软期间，公司规模增长了十倍。微软的成功也被归功于他：先是云业务，如今又是 AI 热潮中的表现。 John Collison (00:01:07) Satya Nadella took over as Microsoft CEO in 2014, but he’s been with the company for more than 30 years. And he’s seen a lot. Microsoft has grown by 10x in the time that Satya has been running it and he’s credited with Microsoft’s success—first in cloud and now in the AI boom. 纳德拉（00:01:22） 干杯，约翰。聊得真开心。 Satya Nadella (00:01:22) Cheers, John. It was great.

企业 AI：数据、记忆与权限

Enterprise AI: data, memory and permissions 约翰·科里森（00:01:25） 那大家在 Ignite 大会上可以期待些什么？ John Collison (00:01:25) So what should people be excited about at Ignite? 纳德拉（00:01:28） 对我们来说，Ignite 大会最重要的事，就是确保 AI 能在企业内部普及开来。说到底，关键是：我们不能只羡慕别人的 AI 工厂或 AI 智能体，而得思考怎么建起自己的 AI 工厂。所以，组织好数据层最终可能是最复杂的一环：它必须贯穿整个企业，才能对接智能。我想，我们会花很多精力做这方面的事。 Satya Nadella (00:01:28) The Ignite Conference for us, more than anything else, is about making sure that AI is getting diffused inside of the enterprise, right? I mean, if there is one thing, it’s more about, “Hey, what does it mean not to just admire somebody else’s AI factory or AI agent, but how to build your own AI factory?” So organizing the data layer turns out to be probably the most complicated thing, which spans the enterprise, such that it can meet the intelligence. And so that’s the stuff that I think we’ll probably do a lot of. 约翰·科里森（00:02:00） 在企业环境里，我们其实还没有真正实现“深度研究”。 John Collison (00:02:00) We still don’t really have “deep research” in a corporate context. 纳德拉（00:02:05） 我们已经实现了，这就是 Copilot 要做的事。 Satya Nadella (00:02:05) We do, that’s what Copilot is about. 约翰·科里森（00:02:05） 可大多数人日常并没有这样的能力。那是不是说，他们没有充分用好现有的 AI？ John Collison (00:02:05) But most people day-to-day do not have this. So are they just underusing AI that exists? 纳德拉（00:02:11） 是的。你提到这一点很有意思，因为在我看来，这才是杀手级功能。我们做的一件大事，是利用了支撑我认为每家公司最重要的数据库的那张关系图谱。它连接着你的邮件、文档、Teams 通话等等。顺带一提，人们的工作并非随意、毫无结构地进行；大家做的事总与某个业务事件有关。这种语义上的联系一直存在于人的脑海里，却会流失。现在，我们第一次能更好地把它找回来。 Satya Nadella (00:02:11) Yes. In fact, it’s interesting you brought that up because to me that is the killer feature. So the biggest thing we did was, we took this graph that is underneath what I think is the most important database in any company, which is underneath your email, your documents, your Teams calls, what have you. It’s the relationships that, by the way, people are not working in an ad hoc fashion in an unstructured way, but they’re all doing it in relation of some business event. That semantic connection is in people’s heads and it’s lost and for the first time there’s much better recall of that. 约翰·科里森（00:02:52） 你觉得这在企业里的普及率为什么这么低？我感觉大家用了不少大语言模型工具，也许会上传单个文档，但我不认为多数公司已经把所有公司背景信息都接入日常使用的 AI，让它能面面俱到。 John Collison (00:02:52) Why do you think this is underpenetrated the enterprise? I feel like people are using lots of LLM tools. They are uploading individual documents, maybe, but I don’t think most companies have the all-singing, all-dancing, all of the company’s context is plugged into their everyday AI. 纳德拉（00:03:10） 是的，我会把原因分成两类。第一，这件事才刚刚起步。我常说，至少和我们历史上做过的所有 Office 套件相比，这次推广速度最快，因为这牵涉到变革管理。归根结底，你得把它引进来，让大家用起来。还有，在企业环境里，所有电子取证功能都必须正常运作，数据治理也必须全部到位。我们还得把 Purview 接入 Copilot，这样每当我试图检索机密内容时，系统都能识别它是机密、应用信息权限管理（IRM）等措施。 Satya Nadella (00:03:10) Yeah, in fact, I would say there are two sets of things. One, it’s starting, right? I always say at least compared to anything we have done, in terms of all the Office suites over our history, this is the fastest in that sense, because it’s change management. At the end of the day, you got to get it in, people have to use it. Oh, by the way, in the enterprise setting, it has got to mean all eDiscovery has to work. All of the data governance has to work. We have had to plumb this purview into Copilot such that any time I’m trying to retrieve something that’s confidential, it’s labeled confidential, it’s IRM’d and so on. 纳德拉（00:03:10） 为此我们做了大量工作，我想现在开始看到成效了。另一件事是，让它跨 Microsoft 365 图谱工作是一回事；接下来还得问，你的 ERP 系统怎么办？连接器算是能用，但又不太够用，因为它们就像两根细吸管。我们需要更完善的数据架构，实质上要把所有这些数据在语义层面整合进同一个数据层。 Satya Nadella (00:03:10) So there’s been a significant amount of work and that I think is where we are starting to see the uplift. The other thing I’d say is, it’s one thing to have it work across the Microsoft 365 graph, but then the next thing is, oh, what about your ERP system? The connectors kind of work, but they don’t really, because they’re two thin straws. You just need a much better data architecture where you have to essentially semantically embed all of these into one layer. 约翰·科里森（00:04:15） 好。几十年来，人们一直有个愿景：公司的数据触手可及。我最喜欢的例子是一本讲甲骨文历史的书《Softwar》。书里提到拉里·埃里森做高管简报（EBC），书里说的好像是九十年代在日本的某一场，大概是九十年代末。他向企业高管推销的愿景是：把公司的所有数据放在一个地方。这个推销说法之所以一直不过时，部分原因就在于公司其实并没有把所有数据都掌握在手边。公司总是不肯把数据基础设施这份“蔬菜”吃下去，而向高管兜售的说法始终是：你只要点一下按钮，就能自己找到问题的答案，不用再把请求发给分析师，让对方替你调查。我们这次终于会把数据管道的基础工作做好吗？你也可以反驳这个前提，但这就是我的问题。² John Collison (00:04:15) Okay. There’s been a vision for decades of your company’s data at your fingertips. My favorite example of this is I really like the book Softwar on the history of Oracle and it talks about Larry Ellison doing EBCs. I think they’re talking about one in Japan in the 1990s, so it’s the late 1990s, and he is pitching executives on all your company’s data in one place. Part of the reason this is an evergreen pitch is because companies don’t actually have all their data at their fingertips. Companies do not eat their data infrastructure vegetables, and the pitch to executives is always you can go answer your questions yourself at the touch of a button as opposed to sending a request to an analyst who goes and does an investigation for you. Will we finally, this time, eat our data plumbing… You can push back on the premise, but that’s my question. 纳德拉（00:05:02） 不。事实上，如果我没记错，“信息触手可及”这个说法是比尔在九十年代一次 COMDEX 演讲上创造的。 Satya Nadella (00:05:02) No. In fact, I think, if I’m not mistaken, Bill coined this term “information at your fingertips” at a COMDEX speech in the nineties. 约翰·科里森（00:05:11） 我觉得是这样。 John Collison (00:05:11) I think that’s right. 纳德拉（00:05:12） 是啊。很长一段时间里，比尔一直对此念念不忘。我清楚记得他在九十年代说过这话；那时我还是个初级员工，在一次评审会上听到了这段话。他说：“软件只有一个类别，叫信息管理。你得把人、地点和事物都整理成结构化数据，就这么简单。除此之外什么都不用做，因为所有软件……”这就是比尔一直以来的梦想。比如说，他讨厌文件系统，因为文件系统没有结构。如果一切都是 SQL 数据库，他就能直接用 SQL 查询所有信息、针对这些信息编程，那他肯定会很喜欢。 Satya Nadella (00:05:12) Yeah, and for the longest time, Bill was always obsessed about, he felt that… In fact, I remember him distinctly saying this in the nineties, which I picked up in one of the reviews I was in as a junior guy sitting around and he said, “There’s only one category in software. It’s called information management. You got to schematize people, places and things and that’s it. You don’t have to do anything more, because all software…” And that was the dream Bill always had because, for example, he hated file systems because they were unstructured. He would’ve loved it if everything was a SQL database and he could just do SQL queries and program against all information. 纳德拉（00:05:12） 在他看来，那是实现“信息触手可及”的优雅方案。问题在于，人很复杂；即使数据有了结构，它也未必真能放进一个索引里，或是让我用一条 SQL 查询全部查出来。所以我认为，这一直是旧世界的根本难题。我以前想不到——我们谁都没想到——AI 竟会以这样的方式出现：深度神经网络只要扩展到一定规模，就能找出模式，而不必依赖经过结构化设计的数据模型。很长一段时间里，我们总在琢磨：“关系要复杂到什么程度，数据模型才能抓住企业的本质？”结果答案竟是：神经网络里有大量参数，再配上强大的算力。 Satya Nadella (00:05:12) That to him was like an elegant solution to information at your fingertips. The problem is people are messy, and even if data is structured, it sort of is not truly available in one index or one SQL query that I can run against all of that. So that has been the fundamental challenge of the old world, I would say. I would’ve not thought, none of us thought that somehow this AI thing and a deep neural network at some scaling will suddenly become the thing that figures out the patterns, not some schematized data model. In fact, one of the longest time, we used to always obsess about, “Oh, how complex do the relationships have to be or the data model needs to capture the essence of an enterprise?” And it turns out it’s lots of parameters in a neural network with a lot of compute power. 约翰·科里森（00:06:41） 德瓦凯什谈过一个非常聪明的远程员工：他五分钟前才入职。他想说明的是，模型可以聪明得超乎想象，可以做检索增强生成（RAG），也可以访问企业里的所有信息，但这和模型本身真正“知道”某件事并不完全一样。因此，除非你在公司内部训练定制模型，否则模型其实不会越来越擅长你们的工作。你问到第一千次时，它并不会比第一次更聪明。你觉得这会怎么发展？ John Collison (00:06:41) Dwarkesh talks about this really smart remote employee who started five minutes ago, getting at the point that the models can be arbitrarily smart and they can do RAG and they can have access to everything in your enterprise, but it’s not quite the same as the model actually knowing something as a model. And so the models, unless you train custom models inside your company, cannot actually get smarter at what it is that you do. And the thousandth query is not any smarter than the first. Where do you think that goes? 纳德拉（00:07:09） 我觉得这里有两件事。如果我理解得没错，他说的都是上下文学习或持续学习。这可以说是终极目标，也和我刚才说的相呼应：如果把模型的认知核心和它掌握的知识分开，就基本上有了持续学习的公式，或者说算法，然后你只要把它放开去学习。至少在我看来，运行时模型之外还有三件事必须解决。第一是记忆，包括各种形式的记忆：短期记忆、长期记忆。 Satya Nadella (00:07:09) I think there are two things there. I mean, if I understand his thing, it’s all about in-context learning or continual learning. That’s sort of the ultimate thing, and it sort of speaks to the thing I was saying, which is if you kind of have the model’s cognitive core separated from its knowledge, then you have essentially the continual learning formula, so to speak, or the algorithm and then you just unleash it. At least there are three things to me that are outside of the model at runtime that I think you kind of have to crack. One is memory and all forms of memory: short, long. 纳德拉（00:07:09） 比如长期归因就是一个大难题，而人类很擅长这件事。所谓长期归因，直觉上说，就是……有人对我说过：“等到 AI 模型既能给予奖励，也能记住如何针对某个错误施以惩罚——因为它具备长期归因的能力——那时你就知道它真的有记忆了。”总之，第一件事是记忆。第二件事是访问权限：模型在运行时必须真正遵守所有权限系统，因为不同角色能访问的内容各不相同。模型必须符合这些权限要求。第三件事是，所有可执行的操作都必须能正常运作。 Satya Nadella (00:07:09) Even these big challenges of humans are great at long-term credit assignment, which is how does, intuitively… Like somebody said to me, “Hey, the day AI models can both reward and remember how to punish for some mistake because they have the ability to do long-term credit assignment, that’s when you’ll know that they have real memory.” But in any case, memory is one. The second one is entitlements, which is they have to really respect all of the permissioning system at runtime because this is where, because there are roles, what access do I have? And so the model needs to meet that, and then the action space all has to work. 纳德拉（00:07:09） 这三件事合在一起，就是模型所处的环境。假如模型拥有操作能力、访问权限和记忆，那么这些东西按定义都在模型之外，却又必须整合进模型之中。比如说，今天的 Copilot 会使用 OpenAI 的模型，也会用 Claude，对吧？我需要整个系统能同时兼容这两类模型。我认为，前沿发展必须朝这个方向推进。 Satya Nadella (00:07:09) So if you bring those three things, because after all, that’s the environment. So if I have actions, entitlements, and memory with these models, and they by definition have to be outside of the model, but be built into the model. So for example, in Copilot today, you use OpenAI models, you even use Claude, right? I need the system to work across both of those and that I think is where the frontier has to move to.

在虚拟走廊里管理微软

Managing Microsoft through virtual corridors 约翰·科里森（00:08:52） 是，是。我还有一百万个 AI 问题想问，不过我想先问问你的工作方式。你平常一天是怎么过的？尤其是，你怎么通过四处走动来了解情况？你会去哪些虚拟走廊逛逛，了解微软在发生什么？你和客户的实际交流是什么样的？就说普通的一天，不算财报会、董事会之类的场合。 John Collison (00:08:52) Yes, yes. I have a million more AI questions, but I want to ask you some questions about your way of working. So what does your day-to-day look like? And in particular, how are you managing by walking around? What virtual corridors are you wandering to just get a sense for what’s going on at Microsoft? What do your customer engagements actually look like? Just for a normal day, not earnings or not a board meeting or something. 纳德拉（00:09:17） 有意思的是，我平常一天的两头都是客户事务。我几乎每天都要和客户交流。很多交流是远程的。我一天大部分时间都在开 Teams 会议，其中至少两三场是和客户开的。这大概是最能让我贴近实际情况的方式，所以我每天至少会安排一两场。然后就是很多会议。做 CEO 之后，我意识到会议大致有两种。 Satya Nadella (00:09:17) Interestingly enough, my normal day, it’s the two ends of it, which is the customer stuff. So there’s not a day that I would say I’m not having… Many of them are remote. I mean, there’s Teams calls for me most of the day, at least two or three of them with some customer. It’s sort of the most helpful way for me to stay most grounded, I would say. So I have at least one or two of those each day. And then I would say there is a lot of meeting time. As a CEO, one of the things I’ve recognized is there are two types of meetings. 纳德拉（00:09:17） 一种会议里，我只需要召集大家，闭嘴听着，因为把人聚到一起才是重点。别表现过头，就坐在那里，因为该做的工作要么已经完成，要么会在会后完成。这是一种。另一种是重要会议，我需要从中学习、做决定，或者传达一些事情。除此之外，我的活动几乎都在 Teams 频道里，对吧？我会一直留意各个 Teams 频道，那里非常有帮助。事实上，我在那里学到的东西最多，也在那里认识最多人。所以说到处走走、在办公室里串门，我真希望能告诉你那就是我的方式。 Satya Nadella (00:09:17) One meeting is where I’m just supposed to convene and keep my mouth shut because convening was the real thing. Don’t overperform and just sit because all the work would’ve either happened, or will happen after. So that’s kind of one. And then the other meetings, which are the important meetings where I do need to learn or I need to make a decision or communicate something. Then I must say it’s kind of like all over, for me, Teams channels, right? I am lingering around Teams channels and they’re most helpful. In fact, if anything I learn the most there. I meet most people there. So wandering the halls, I wish I could tell you that that is the form. 约翰·科里森（00:10:40） 不过，我觉得 Teams 就是新的办公室走廊；你可以在那些频道里四处看看。 John Collison (00:10:40) No, but I think Teams is the new wandering the halls, looking around those channels. 纳德拉（00:10:45） 完全没错。对我来说，最棒的是能……我在那些频道里建立了最多联系。我会发现：“哦，原来他负责 Excel 智能体。哦，他们在看的就是那个评测。”在那里学到的东西，比我做过的其他任何事都多。 Satya Nadella (00:10:45) A hundred percent. And the most beautiful thing is for me to be able to… That’s where I make the most connections. I get to know, “Wow, he’s the person working on Excel Agent. Oh, that’s the eval that they’re looking at.” I learn so much out of it than anything else I’ve done. 约翰·科里森（00:11:00） 所以微软的团队就是各自埋头做产品，然后萨提亚突然冒出来，问他们一个关于产品的问题？ John Collison (00:11:00) So are teams at Microsoft just working away on their product and then Satya pops up and has a question on their product? 纳德拉（00:11:05） 我倒希望是这样。有时我觉得我们的访问权限管得太严了。我有时候真希望能有更多访问权限。事实上，我最大的抱怨就是没法随时加入所有我想去的地方。不过，能直接进去看看确实很有意思，这也让这种互动变得很自然。而且今天的员工也不怕当着你的面表达意见。 Satya Nadella (00:11:05) I wish. Yeah, sometimes I feel like we are way too permissioned. I wish I had more access sometimes. In fact, my biggest complaint is that I can’t drop in everywhere I want to. But yes, it is fun to be able to just go in there and it sort of normalizes it. And then people are also like—today’s workforce is not shy of sharing their opinion with you.

跟着开发者与初创公司走

Following developers and startups 约翰·科里森（00:11:31） 我注意到了，确实。至少在硅谷的一些小圈子里，你很有名。说到你有条理地了解科技行业的动向，我记得你来过 Stripe 在 Mission 区的办公室，还记得吗？那时我们还是个小公司。大概就在你接任 CEO 之后不久吧。我猜是这样。不过那会儿 Stripe 很小，微软很大。 John Collison (00:11:31) I’ve noticed, yeah. You are famous, at least in small corners of the Valley. Speaking of being methodical for staying very connected to what’s going on in tech here, and I remember you came and visited the Stripe office, remember that one on the Mission? Yeah, when we were a pipsqueak company. It was probably right after you took over as CEO, I’m guessing. But Stripe was very small and Microsoft was very big. 纳德拉（00:11:58） 其实还要更早。我第一次去你们办公室时，应该是我刚开始负责 Azure 的时候。 Satya Nadella (00:11:58) Actually before. I think the first time I came to your offices was when I was running Azure first. 约翰·科里森（00:12:02） 好，没错，那还要更早。所以那时 Stripe 才刚起步。你为什么觉得自己比大多数 CEO 更常这样做？其他 CEO 也应该想去见见所有初创公司啊。 John Collison (00:12:02) Okay, yeah. So it was even before that. So that would’ve been very early in Stripe’s journey. Why do you think you do this much more than most other CEOs? Other CEOs should want to meet all the startups too. 纳德拉（00:12:14） 我一直是在这样的环境里成长的；可以说，在微软的成长经历，让开发者关系、技术布道成了我的一种本能。我大概会这样想：“如果你不跟着开发者走……”有两件事一直根植于我心里。第一，如果你不关注开发者的方向，就很难让技术平台跟上需求；第二，想打造技术平台，你确实得了解新的工作负载。至少这两点我一直记着。 Satya Nadella (00:12:14) I’ve always grown up—in some sense, I grew up even at Microsoft, which had those developer relations, evangelism sort of gene in me. I kind of approach, I think a lot of it as, “Hey, if you don’t follow developers…” There are two sorts of things that are ingrained in me. One is if you don’t follow where developers are going, it’s hard to sort of be relevant in terms of tech platforms and then you really need to understand the new workload in order to build a tech platform. Those are the two things that at least I’ve kept. 纳德拉（00:12:14） 所以，只有关注初创公司，你才能知道平台或工作负载会是什么样。于是我一直把精力放在这上面。另一件事是，这能给我很多能量。我总觉得创业者是些神奇的人，能从无到有创造出东西。这简直像变魔术。所以我总会想：他们到底是怎么做到的？ Satya Nadella (00:12:14) And so therefore, the only way… If you’re not following startups, it’s very hard to know what is either the platform or the workload. So that’s a thing that I’ve indexed towards. The other thing is I derive so much energy out of it. I mean I’ve always thought founders are just magical people who create something from nothing. I mean, it just sort of feels like a magic trick. So I’m always like, how the heck does one do that? 约翰·科里森（00:13:08） 你提到关注初创公司的动向，这很有意思。我们一直认为，Stripe 的产品要为初创公司打造。这很重要，因为今天的小型初创公司就是明天的上市公司；在 Stripe 上，这样的例子我们见过很多次。但我们直觉上觉得，初创公司感兴趣的东西，往往能带来更好的产品体验；当时我们还无法证明这一点。如果初创公司想要稳定币、按用量计费之类的东西，我们就应该满足这些需求。这么做不只是为了把初创公司业务做好，企业客户最终也会跟进。我觉得我们花了很多年才验证这个模式，但现在我们确实看到了—— John Collison (00:13:08) Yeah, it is funny you say that about following what the startups are doing. We always conceived of what Stripe was building as it was important to build for startups, both because today’s small startups are tomorrow’s public companies and we’ve seen that again and again on Stripe. But we just felt, at an intuitive level and we felt this before we could prove it, that what the startups were interested in were often better product experiences. And so if the startups want stablecoins or usage-based billing or what have you, we should build for those needs, not just because we’ll have a good startup business, but the enterprises will come around. And it took us, I would say many years to prove out that model, but now we’re really seeing— 纳德拉（00:13:50） 是的。事实上，我觉得你们在这方面算是标杆。我从你们身上学到的一点，是重新发现微软曾经很擅长的事：跟着开发者走，出现在初创公司身边。这也促使我把目光投向 GitHub、Nat，以及相关的其他事情。GitHub 这个资产显然很有价值。我们需要成为开源生态的好管家，但它也是所有初创公司都会用的地方——每家都有的一样东西，就是放在 GitHub 上的代码仓库。³ Satya Nadella (00:13:50) Yeah, in fact, I think you guys are a bit of a gold standard on that. In fact, one of the things that I learned from you guys was rediscovering at some level what Microsoft was very good at, which is following the developer, being where the startups are. And so that’s what sort of led me even to GitHub and NAT and all of the rest, which is to some degree the GitHub asset. Obviously it was a great asset. We needed to be good stewards of an open source ecosystem, but it’s also the place where every startup—the one thing that everybody does have is their repos in GitHub. 纳德拉（00:13:50） 我觉得进入这个圈子对我们很重要，原因不只是“占据一个战略位置很有利”。你们刚才说的很好：我们要去学习，做出更好的产品。有时你会失去对产品体验的审美判断，不知道怎么做才能让使用过程毫无阻碍；而用户的耐心最少，所以从开始使用到获得价值所需的时间必须最大化。⁴ Satya Nadella (00:13:50) And I felt like being in that loop was important for us, not just, “Oh, it’s strategically great to have some position there.” To learn simply and to build better product, I think, is sort of well said. Because you sometimes lose the aesthetic of what is required, what’s that friction-free way to deliver because the least amount of patience is there and the time to value, for example, has to be maximized.

智能体时代的界面：宏观委派、微观引导

Agent interfaces: macro delegation, micro steering 约翰·科里森（00:14:52） 微软有没有在考虑按个人需求生成界面？你想想，软件还困在旧范式里：我们写好一大堆软件，做成最终版，再装在磁盘上发出去。现在软件还是同样那一套，只不过通过云端交付。但实际上，我们也许能实时渲染出你想要的那种界面。你们在朝这个方向发展吗？ John Collison (00:14:52) Is Microsoft thinking about generated UIs that are personalized to… When you think about it, software is stuck in the old paradigm of we write a bunch of software and it goes to Gold master and it goes out on disks, and now that same kind of software it’s delivered in the cloud, but the UI you want is probably, we can render that exact UI in real time. Is that a direction you guys are going? 纳德拉（00:15:14） 肯定在朝这个方向走。某种程度上，现在发生的事是，我们一方面有生成能力。如果你认为所有代码都能生成，那围绕任何东西生成更定制化的用户体验框架也就顺理成章了。尤其是……事实上，很长一段时间以来，微软一直在问：文档、网站和应用程序之间到底有什么区别？所以在某种程度上，没错，你可以随时按自己想要的呈现形式生成其中任何一种。不过有意思的是，尽管大家都在说“这些应用都要消失了”，拿大家熟悉的集成开发环境（IDE）来说吧。⁵ Satya Nadella (00:15:14) I think for sure. At some level what’s happening is on one side our ability to generate. I mean if you sort of say you can generate all code so therefore you can generate some UX scaffolding around anything that’s a lot more custom. So especially… In fact for the longest time, one of the things at Microsoft was what’s the difference between a document, a website and an application, really? And so to some degree, yeah, exactly. So you can generate any one of those at any time depending on what format you want to present it. But at the same, interestingly enough for all the talk of “Hey, all these apps go,” take even our good old IDs. 纳德拉（00:15:14） 从某种意义上说，IDE 正在回归，不管是 Excel 还是 VS Code 都是如此。现实是，AI 会生成输出，而我得弄明白这些输出。事实上，我需要一个出色的编辑器，让我能和 AI 一起查看差异、反复迭代。因此 IDE……最让人兴奋的事情之一，就是出现一批全新的、打磨得很好的 IDE：它们既能和智能层形成遥测反馈回路，又更像抬头显示器。我有成千上万个智能体同时在工作，怎么理解并微调它们的行动？IDE、收件箱和消息工具就会承担这个作用。我以后处理消息或分流任务的方式，不会再和今天一样，而会变得不同。 Satya Nadella (00:15:14) In some sense IDs are back, whether it’s Excel or VS Code because the reality is AI generates output. I need to make sense of that output. In fact, I need a fantastic editor that lets me do diffs and iterations on it with AI. So the ID… one of the most exciting things is new classes of highly refined IDs that have even sort of a telemetry loop with the intelligence layer, but also they kind of act more like heads-up displays. I have thousands of agents going off. How am I going to make sense of the micro steering of thousands of agents? And that is what ID slash inboxes and messaging tools will be, which is I’m not messaging or dealing with triage the way I deal with it today, but it’s going to be different. 约翰·科里森（00:16:53） 好，有意思。所以现在程序员把所有时间都花在 IDE 里，而他们是少数这样工作的职业之一。你的愿景是会出现会计师用的 IDE、律师用的 IDE，还有—— John Collison (00:16:53) Okay, interesting. So you think right now programmers spend all their time in IDE, but they’re one of the few professions that does that. And your vision is the accountant IDE, the lawyer IDE, and— 纳德拉（00:17:04） 我会用什么方式和智能体协作？这有点像大规模的宏观委派：我把一堆指令交给许多智能体，它们就去工作，有时持续几小时、几天——随着模型变得更好，大概会是这样。但它们也会不断向我汇报。因此，这是宏观委派、微观引导。要做到这一点，怎样才能带着上下文进行微观引导？它不能变成下一代通知地狱：系统给我发来一条通知，只有五个词，我却完全不知道真实上下文是什么，诸如此类。 Satya Nadella (00:17:04) What is the metaphor of how I will work with agents? So it’s kind of like massive macro delegation. So there’s lots of agents I go give a bunch of instructions to and they go off and work sometimes for hours, days, let’s say, as the models get better. But they are checking in and so it’s macro delegation, micro steering. So if you take that, how does one do micro steering with context? It can’t be in the next notification hell, which is it sort of notifies me. It has five words. I don’t know exactly what the real context is, or what have you. 纳德拉（00:17:04） 我觉得这正是方向所在，而且它得像跨应用的体验。所以我觉得，所有软件最终长大之后，都会像一个收件箱、一种消息工具，再加上一块带着闪烁光标的画布；只不过这一次，很多工作已经完成了。 Satya Nadella (00:17:04) That I think is where, and that has to be multi app-like. So that’s where I feel like all software finally when it grows up, it looks like an inbox and a messaging tool and a canvas with a blinking screen, except this time around a lot of work happened. 约翰·科里森（00:17:55） 这会是一个应用，还是十个不同的应用？想想当年生产力软件的格局：有三大应用——Word、Excel 和 PowerPoint。有意思的是，数量不是一个，也不是四十个，而是三个。你怎么看？ John Collison (00:17:55) Is that one app? Is that 10 different apps? It’s kind of interesting if you think about the productivity speech that emerged, there were three big apps—Word, Excel and PowerPoint—but it’s interesting that that number was not one and was not 40. It was three. And so how do you think about this? 纳德拉（00:18:16） 我觉得这个判断是对的。在我看来，最后会有少数几个。我这个人有点喜欢把复杂的事情归结到最基本的形式，会想：“天哪，最后还是那些东西，只不过它们承担的工作会变。”拿表格来说，至少对人是这样；当然，也可以讨论智能体之间会用什么工具互相沟通，那是另一回事。现在为了强化学习循环，它们在模拟我们的生产环境，但最终它们会更有效率地创建自己的生产环境，让自己进行强化学习。 Satya Nadella (00:18:16) I think that that’s right. To me it will be a few I think, and in fact the reductionist person in me says, “Man, they’ll be the same things except the job they do is going to be different.” Because I think a table, at least at the human level, because we can all talk about what tools will agents use to communicate with each other? That’s a different thing. Right now for the RL loop, they are simulating our production environment, but they will ultimately be more efficient in creating their own production environments to kind of RL themselves. 纳德拉（00:18:16） 不过先不谈这一点。为了和人沟通，我觉得我们已经发现了一些自己喜欢的形式：电子表格、表格、线性结构的文档，还有收件箱或消息工具。所以这些都是合理的用户界面。只是你刚才问的那个问题在于：当它出现在 IDE 里，带着一组改动时，不能只对我说：“好了，这是一个文件，你去打开它。”你得给我一个能指引我的方案；这方案不只是为了执行任务，也得帮助我完成整个工作流程。 Satya Nadella (00:18:16) But let’s just leave that aside. But in order to communicate with us, I feel like we have discovered some good things that we like. We like spreadsheets and tables and we like documents in linear form. We like inboxes or messaging tools. So these are reasonable UIs, except the question I think you asked is how does this thing have, when it shows up in an IDE with a set of changes, you have to help me more than just say, “Okay, now here is a file, go to that file.” That directed plan, not just to execute, but for me to do my workflow. 纳德拉（00:18:16） 我们正在尝试的一件事是 GitHub Copilot 的 Mission Control。设想是，你启动五六个不同的分支，在每个分支上派出自主智能体。它们各自完成工作、回来之后，你如何对这些拉取请求做分流处理，我认为下一代 IDE 就会从这里诞生。 Satya Nadella (00:18:16) One of the things that we are experimenting with is mission control and GitHub Copilot is the idea is you go have five, six different branches in which you fire off all these autonomous agents, they all do their work, they come back, and then your ability to do PR triage is where I think the next IDE is born.

互联网往事：看对范式，还要找对路径

The early internet: paradigms, products and business models 约翰·科里森（00:19:50） 我一直很惊讶，在科技领域，技术愿景出现得往往远早于技术真正成熟的时候。比如电影《2001 太空漫游》在六十年代就描绘了能听懂语音、还能调用工具的 AI。这个愿景过了五十年才成为现实。再比如，早在八十年代，人们就对能和电脑说话、能做语音转文字和文字转语音感到兴奋；直到现在……我不知道你用不用 SuperWhisper 之类的工具，但现在它终于真的好用了，三年前还不行。比那个愿景晚了四十年。 John Collison (00:19:50) I’m struck by, in technology, how frequently you see the pattern of excitement for and a vision around a technology being so much earlier than the technology actually being ready. Like the movie 2001: A Space Odyssey, which is in the sixties, that was a voice activated AI with tool use capabilities. And it just took 50 years and then people were excited about the idea that you could speak to your computer and text to speech, speech to text. People were excited about that in the eighties and only now… I don’t know if you use SuperWhisper or anything like that, but it’s really, it’s finally really good, but it wasn’t good three years ago. 40 years after the vision. 纳德拉（00:20:33） 你这么一说真是不可思议。我以前住在微软老园区旁边的一套公寓里，当时我在做互动电视。那是 1994 年。 Satya Nadella (00:20:33) Yeah, it’s crazy that you bring that up. In fact, I used to have an apartment right next to the Microsoft campus, that old campus, and I was working on interactive television. This was in ‘94. 约翰·科里森（00:20:44） 信息高速公路。 John Collison (00:20:44) The information superhighway. 纳德拉（00:20:45） 没错。当时有好几件事都很惊人。沿着我的汇报关系往上，依次是里克·拉希德、克雷格·芒迪、内森·迈尔沃尔德，再上面是比尔·盖茨。我当时想：“天哪，这群人的智商加起来可真高。”当然，我们全都错过了互联网，那才是后来真正发生的事。不过我那时把互动电视通过异步传输模式（ATM）交换网络接到了家里，也就是我住的公寓。我记得做过这么一次演示。年轻时我在微软做过一件风险很高的事：演示我们第一个有冗余能力的文件系统，它其实是一个视频服务器。 Satya Nadella (00:20:45) That’s right. In fact, there were multiple things that were stunning. My management chain was Rick Rashid who reported to Craig Mundie, who reported to Nathan Myhrvold, and there was Bill Gates and I was saying, “Man, that’s a lot of IQ.” And of course we all miss the internet. That was the only thing that happened. But I had interactive television, switch ATM, to my home, to my apartment. So I remember doing this demo. One of the high stakes things I did as a young guy at Microsoft was a demo of our first redundant file system, which was a video server. 纳德拉（00:20:45） 约翰·马龙来现场了。比尔当时大概是这样介绍的：“看，这就是互动电视的未来。它还有个好处：即使磁盘出了故障，视频流也能继续播放。”我的任务就是把磁盘拔掉，同时让视频流继续。我们实质上搭建了一个分布式文件系统和流媒体服务器，又通过 ATM 交换网络连到家里。我有五部电影可以看，而且每部都看了好几遍。 Satya Nadella (00:20:45) John Malone was the one who came and Bill was sort of saying, “Hey, here’s the future of interactive television, and guess what? It’s even great because the disc can go haywired and still stream.” And so my job was to remove the disc drive and have the stream continue. But we built essentially a distributed file system and a streaming server and had an ATM switch network to the house, and I had five movies I could watch, and I watched them all multiple times. 约翰·科里森（00:21:49） 我想问问这件事，因为我琢磨了很久，而你是最适合回答的人。微软在九十年代就看到了互联网即将到来的未来，尤其是比尔·盖茨那份著名的《互联网浪潮》备忘录，里面说：“互联网是微软必须专注的一件大事。”并不是说我们没想到互联网，也不是说它在十五个优先事项里排第七；意思是：“听好了，微软唯一该考虑的就是互联网。”但当时大家对互联网的愿景是信息高速公路。它和我们后来所说的互联网有一点微妙的差别，因为当时的想法——而且这个想法非常合理——是：没有人能把互联网接到家里的电脑上。 John Collison (00:21:49) So I want to ask you about this, because I’ve thought a lot about this and you’re the perfect person to ask. Which is, Microsoft saw the internet future that was coming in the nineties, and in particular the famous Bill Gates internet tidal wave memo said, “The internet is the one big thing Microsoft needs to focus on.” It wasn’t like we’re not thinking about the internet. It wasn’t that it was priority number seven of 15. It was like, “Hey guys, listen up. The only thing Microsoft should be thinking about is the internet.” But the vision for the internet at the time was this information superhighway which was subtly different from the internet because the thinking was—and it was very sensible thinking—no one has internet to the computer in their home. 约翰·科里森（00:21:49） 很多人家里甚至没有电脑。大家家里有的是电视，而他们也有有线电视，那能提供高带宽连接。所以我们要在电视上装机顶盒，人们就会通过这种方式使用互联网。大家非常重视这股即将到来的浪潮，解决方案也很合理、考虑得很周全，可它偏偏不是正确的做法。说到这里，显然就会想到规模庞大的 AI……从这件事里，我们应该得到什么启示？ John Collison (00:21:49) A lot of people don’t have a computer in their home. So what people do have is a TV and what they have is cable, which is a high-bandwidth connection. And so we’re going to do these set top boxes on the TV and that is how people will use the internet. Paying a huge amount of attention to this coming wave, pretty sensible, well thought-out solution, and yet not the right approach. So obviously bring that up in the context of the giant AI… What should one take away from that? 纳德拉（00:22:57） 这是个很好的问题。说说我的理解也挺有意思。我没有花太多时间和比尔聊那个时期的事，但即便当时我只是个初级员工，我觉得至少……我对那段历史的理解是，我们好像理解了互联网，又好像没有，因为我们想要交付的是……我觉得我们当时并不相信 TCP/IP 能行。某种程度上，回头看我们当时想做的信息高速公路，核心问题是：“服务质量很重要。” Satya Nadella (00:22:57) It’s a great one. See, if I look at even my interpretation, it’ll be actually interesting. I’ve not spent as much time talking to Bill about that era, but I felt there were at least, as someone as a sort of an entry level employee at that time, even. My reading of history was that we kind of got the internet, but we didn’t because we wanted to deliver… I don’t think we believed that TCP/IP would work. I mean at some level the information highway, when I look at what we were trying to do was, man, this quality of service is a thing. 纳德拉（00:22:57） “TCP/IP 根本行不通。”所以我们才会和拨号上网的 AOL 竞争。你还记得 MSN 最初是 X.25 网络吧？但后来比尔转了方向。我想是在 1995 年——说起来很好笑，正好在 Windows 95 发布时，他说：“你们知道吗？一切都要变了。”所以我觉得，从 1992 年开始——那时我们大概都看到了第一次演示，对吧？——再到 1993 年 11 月 Mosaic 出现……⁶ Satya Nadella (00:22:57) This TCP/IP is just not going to work. And so therefore we were competing against AOL on dial-up. And even that sort of, you remember MSN was an X.25 network, the first version of it. But that’s when Bill pivoted. So the thing that Bill did was in ‘95, I guess—in fact, it’s funny that right as Windows 95 was launching he says, “You know what? It’s all going to change.” So I feel between ‘92, which is when I think all of us maybe got our first demo, right? November ‘93 is when Mosaic— 约翰·科里森（00:24:10） 对，没错。 John Collison (00:24:10) Yeah, that’s right. 纳德拉（00:24:11） 我记得大概是这样。那时我们都在这件事周围试探。从 1993 年到 1995 年，有两年时间，大家还不清楚互联网会不会采用这种协议和完整技术栈。后来这套技术栈逐渐成形，到 1995 年一切就清楚了，于是我们转了方向。 Satya Nadella (00:24:11) I think something like that. And so we all were kind of dancing around it. So from ‘93 to ‘95, there was that two-year period where it was unclear whether this was going to be the protocol and the full stack. And the stack emerged, and by ‘95 it was clear and then we pivoted. 约翰·科里森（00:24:31） 有意思。所以就在那个时候，其实还看不出开放互联网会胜出。 John Collison (00:24:31) Interesting. So just at that time, it wasn’t actually clear that the open internet would win. 纳德拉（00:24:36） 对。还有一个启示。我一直在琢磨这件事，因为我觉得可以把它延伸到 AI 领域。第一，要找对范式。但即使范式找对了，杀手级应用是什么、商业模式又是什么，也未必清楚。互联网就是如此。谁会想到，开放网络的组织层竟会是一个靠网络效应取胜的搜索引擎？我常说，根本没有所谓的开放网络，只有谷歌网络，因为谷歌主导了它。 Satya Nadella (00:24:36) Yes. And in fact, there’s one more lesson. The interesting thing that I’ve always watched because I think we can parlay this into AI. One is to get the paradigm right. Then it’s not clear. Even if you get the paradigm right, that you may not get what is the killer app or even the business model. That’s always been the case, with the internet, who would’ve thought that for the open web, an organizing layer would be one network effect search engine, right? Because the organizing layer of the web, I always say there’s no such thing as the open web. There’s the Google web, and just because they dominated it. 约翰·科里森（00:25:19） 我们是不是应该想想，当时或许存在着一种一厢情愿的判断：我们自有的专有方案——自由媒体与微软的合资项目——会胜出，而开放网络最终赢了？如果组织要在两种可能性之间作选择，比如我们的专有信息高速公路和开放网络，或许该提醒它们：它们可能会偏向自家的专有方案。 John Collison (00:25:19) Should one reflect on the fact that maybe there was some motivated thinking around our proprietary solution, the Liberty Media-Microsoft joint venture will win. Whereas the open web is what won. And you should maybe caution organizations where, if they’re following two possibilities, our information superhighway proprietary system or the open web, companies will somehow have happy thinking towards the proprietary solution. 纳德拉（00:25:54） 这是个有意思的问题。回头看，我觉得情况很有意思：可以说，AOL 和 MSN 输给了开放网络。但随后，新的 AOL 和 MSN 形式又取代了它们：它们叫搜索引擎，叫应用商店。移动网络的情况其实也很有意思。 Satya Nadella (00:25:54) It’s an interesting one. I think the way, when I look back again, it’s interesting, right? So AOL and MSN kind of lost out, let’s call it, to the open web. Except they were replaced by new forms of AOL and MSN. They’re called search engines. They’re called app stores. The mobile web, in fact, is fascinating. 约翰·科里森（00:26:16） 开放网络只是历史上的一个阶段。 John Collison (00:26:16) The open web was a moment in history. 纳德拉（00:26:17） 历史上的一个阶段。对我来说，其中更深层的启示也许是：即使在开放生态里，组织层也总会出现。许多品类的主导权会转移到那个组织层。只是它究竟会是什么，一直说不准：上一个范式里是搜索引擎，这一次是聊天机器人。这会持续多久？没人知道，但至少今天它很重要。ChatGPT 作为聚合入口的成功毋庸置疑。市场平台和应用商店曾经扮演过这种角色。接下来会是什么？智能体市场或智能体商务会如何改变电子商务？我觉得这些都是值得认真争论和厘清的问题。 Satya Nadella (00:26:17) A moment in history. And so the thing that maybe—the meta thing for me is organizing layers will always emerge even in an open ecosystem. And a lot of the category power moves to that organizing layer, and it’s always unclear, like the last paradigm of this… Last time at a search engine. Today it’s chatbots. How long lasting is that? No one knows, but it’s definitely today. I mean, ChatGPT’s success cannot be denied in terms of what it means as an aggregation point. Marketplaces slash app stores have been a thing. What comes next? What happens to e-commerce in an agentic marketplace or in agentic commerce? I think these are the interesting things that need to be litigated.

继续阅读：中篇：算力瓶颈、企业主权与智能体商务 · 下篇：模型忠诚、产品捆绑与组织文化

译注

片头由后文话题剪辑而成，话题切换较快。“14 万亿美元”保留主持人的调侃语气，不作为微软当时市值的数据引用。 ↩
EBC 通常指 Executive Briefing Center（高管简报中心）；此处是到这类场合向客户高管介绍产品和愿景。参见 Oracle 对高管简报中心的介绍。 ↩
转录稿写作“NAT”。结合 GitHub 语境，此处推测指 Nat Friedman，中文按人名处理；英文保留原稿。微软的 2018 年收购公告说明 Friedman 将出任 GitHub CEO。这是依据语境作出的判断，并非音频逐字核定。 ↩
原稿确为“time to value … has to be maximized”。time to value 通常指用户从开始使用产品到获得价值所需的时间；这里说“最大化”，与前文强调减少摩擦、用户缺少耐心存在张力。可能是口误或转录问题，本文保留原稿含义，不擅自改成“最短”。 ↩
原转录此处多次写作“IDs”或“ID”，而下一问明确使用“IDE”，且上下文在谈编辑器、Excel 与 VS Code。中文据此按集成开发环境（IDE）理解，英文原样保留。 ↩
这里回忆的是 1993 年 11 月。NCSA 的 Windows 版 Mosaic 1.0 发布公告确实发布于 1993 年 11 月 11 日；不宜把这句话理解为所有平台上的 Mosaic 都首次出现于该月。 ↩

【e/acc】AI奇点 | 物理学家、e/acc运动创始人Guillaume Verdon与Lex Fridman播客实录 | 中英文完整版精译 IV（完结篇）

Tue, 26 May 2026 00:00:00 +0000

书童按：本篇是Guillaume Verdon接受Lex Fridman播客采访实录的第四部分，亦是完结篇。在前三篇建立理论根基、探讨运动理念之后，本篇回归技术前沿与人生哲思：AGI的本质与物理基底AI的互补、奇点概念的批判性审视、有效利他主义（EA）与有效加速主义（e/acc）的哲学分歧、Guillaume个人的极致生产力日常、身份认同的流变、给年轻人的建议、以及对死亡与生命意义的终极思考。Verdon主张以物理能量而非主观感受作为文明进步的客观度量，批判EA的hedons（享乐单位）损失函数易陷入局部最优（如虾类养殖场痛苦最小化或”线头享乐主义”），强调死亡作为系统耗散适应的必要环节。访谈以爱因斯坦名言收尾：”若一个想法起初听来不荒诞，那它便毫无希望。”全程思想密度极高，哲学深度与工程实践并重，既有热力学定律的冷峻，亦有模因文化的狂欢。初稿采用Claude Code机器翻译及排版，书童仅做简单校对及批注，至此四部曲完结，以飨诸君。 奇点与AGI Singularity and AGI Lex Fridman (02:22:31) 你提到Extropic正在尝试构建生成式AI的物理基底。那这和AGI本身有什么区别？换句话说，AGI有可能在你们公司里被创造出来吗？还是说AGI只是把你们的技术当作底层基底来使用？ LEX FRIDMAN (02:22:31) So you mentioned with Extropic you’re trying to build the physical substrate for generative AI. What’s the difference between that and the AGI AI itself? So, is it possible that in the halls of your company, AGI will be created? Or will AGI just be using this as a substrate? Guillaume Verdon (02:22:51) 我认为我们的目标是既能运行类人AI——也就是拟人AI—— GUILLAUME VERDON (02:22:51) I think our goal is to both run human like AI, or anthropomorphic AI. Lex Fridman (02:22:58) 抱歉用了AGI这个词，我知道它会让你不太舒服。 LEX FRIDMAN (02:22:58) Sorry for use of the term AGI. I know it’s triggering for you. Guillaume Verdon (02:23:02) 我们认为未来真正的方向是基于物理的AI与拟人AI的结合。你可以这样想象：我有一种基于物理的AI驱动的世界建模引擎。基于物理的AI更擅长在所有尺度上表征世界，因为它可以是量子力学的、热力学的、确定性的、混合型的世界表征——就像我们的世界在不同尺度上遵循不同的物理规律。如果你从中汲取灵感，以自然本身的方式去学习表征，你就能获得对自然远为精确的描述。这样你就能在所有尺度上拥有非常精确的世界模型。一端是世界建模引擎，另一端是类人的拟人AI。于是你既有了科学——验证想法的试验场，也有了合成科学家。在我们看来，这种基于物理的AI与拟人AI的联合系统，是最接近真正完全通用的人工智能系统的东西。 GUILLAUME VERDON (02:23:02) We think that the future is actually physics-based AI combined with anthropomorphic AI. So, you can imagine, I have a sort of world modeling engine through physics-based AI. Physics-based AI is better at representing the world at all scales, because it can be quantum mechanical, thermodynamic, deterministic, hybrid representations of the world, just like our world at different scales has different regimes of physics. If you inspire yourself from that in the ways you learn representations of nature, you can have much more accurate representations of nature. So, you can have very accurate world models at all scales. And so, you have the world modeling engine, and then you have the anthropomorphic AI that is human-like. So you can have the science, the playground to test your ideas, and you can have the synthetic scientist. And to us, that joint system of a physics-based and an anthropomorphic AI is the closest thing to a fully general, artificially intelligent system. Lex Fridman (02:24:07) 也就是说，你可以通过将AI锚定于物理来更接近真理，同时仍然保留一个拟人化的接口，方便我们这些喜欢与人类或类人系统对话的人使用。那么在这个话题上，我猜这正是当前大语言模型在你看来的一大局限——它们是出色的”一本正经胡说八道”专家，未必真正锚定于真理。这么说公平吗？ LEX FRIDMAN (02:24:07) So you can get closer to truth by grounding of the AI to physics, but you can also still have a anthropomorphic interface to us humans that like to talk to other humans, or human-like systems. So, on that topic, I suppose that is one of the big limitations of current large language models to you, is that they’re good bullshitters, they’re not really grounded to truth necessarily. Would that be fair to say? Guillaume Verdon (02:24:40) 没错，你不会试图用一个在互联网文本上训练出来的语言模型去预测股市走向，它不可能是一个很准确的模型。它无法精确地建模自身的先验知识或对世界的不确定性。所以，你需要一种不同类型的AI来弥补这种文本外推式AI的不足。确实如此。 GUILLAUME VERDON (02:24:40) Yeah, no, you wouldn’t try to extrapolate the stock market with an LM trained on text from the internet. It’s not going to be a very accurate model. It’s not going to model its priors or its uncertainties about the world very accurately. So, you need a different type of AI to compliment this text extrapolation AI. Yeah. Lex Fridman (02:25:05) 你之前提到了奇点。我们离奇点还有多远？ LEX FRIDMAN (02:25:05) You mentioned singularity earlier. How far away are we from a singularity? Guillaume Verdon (02:25:09) 我不确定自己是否相信那种作为单一时间点的有限时间奇点。我认为它更可能是渐近式的，沿某种对角线趋近的渐近线。我们有光锥的限制，有物理定律在约束我们的增长能力，所以显然不可能在有限时间内完全发散。我的先验判断是，对面阵营的很多人认为，一旦我们达到人类水平的AI，就会出现一个拐点，然后突然间[听不清]，AI就会顿悟如何在纳米尺度上操纵物质、组装纳米机器人。但在利用AI改造物质这个方向上干了将近十年之后，我可以告诉你，这比他们想象的要难得多。现实是，你需要大量来自高精度但昂贵的自然模拟、或者自然本身的样本数据，这就制约了你控制周围世界的能力。在计算层面和热力学层面，获取关于世界的信息以便预测和控制它，存在一个不可逾越的最低成本。正是这个成本让一切保持在可控范围内。 GUILLAUME VERDON (02:25:09) I don’t know if I believe in a finite time singularity as a single point in time. I think it’s going to be asymptotic, and sort of a diagonal sort of asymptote. We have the light cone, we have the limits of physics restricting our ability to grow. So, obviously can’t fully diverge on a finite time. I think my priors are that I think a lot of people on the other side of the aisle think that once we reach human level AI, there’s going to be an inflection point, and a sudden [inaudible 02:25:48], suddenly AI is going to grok how to manipulate matter at the nano scale, and assemble nanobots. And having worked for nearly a decade in applying AI to engineer matter, it’s much harder than they think. And in reality, you need a lot of samples from either a simulation of nature that’s very accurate and costly, or nature itself, and that keeps your ability to control the world around us in check. There’s a sort of minimal cost computationally, and thermodynamically, to acquiring information about the world in order to be able to predict and control it. And that keeps things in check. AI末日论者 AI doomers Lex Fridman (02:26:27) 有意思，你提到了”对面阵营”。说到这个，我昨天发了一个关于p(doom)的投票——也就是末日概率。结果显示，认为末日极有可能和极不可能的人之间泾渭分明。我在想，未来是不是真的会出现类似共和党对民主党、红蓝对立那样的阵营划分？AI末日论者对阵e/acc支持者？[听不清] LEX FRIDMAN (02:26:27) It’s funny you mentioned the other side of the aisle. So, in the poll I posted about p(doom) yesterday, what’s the probability of doom? There seems to be a nice division between people think it’s very likely, and very unlikely. I wonder if in the future there’ll be the actual Republicans versus Democrats division, blue versus red? Is the AI doomers versus the e/accers, EAC? [inaudible 02:26:53]. Guillaume Verdon (02:26:53) 是的。不过这个运动从根本上说不是左右之争，它更像是向上还是向下的问题，就文明的规模而言—— GUILLAUME VERDON (02:26:53) Yeah. So, this movement is not right wing or left wing fundamentally, it’s more like up versus down, in terms of the scale of- Lex Fridman (02:27:01) 哪边算”上”？好的。 LEX FRIDMAN (02:27:01) Which one is the up? Okay. Guillaume Verdon (02:27:02) ……文明的规模，对吧？ GUILLAUME VERDON (02:27:02) … Civilization, right? Lex Fridman (02:27:03) 好的。 LEX FRIDMAN (02:27:03) All right. Guillaume Verdon (02:27:05) 不过，现有政党似乎确实存在某种站队现象：那些主张更多权力集中化、更多管控和监管的人正在向末日论者靠拢，因为在民众中制造恐惧是让人们心甘情愿交出更多控制权、赋予政府更大权力的绝佳手段。但从本质上说，我们不是左对右。我们做过关于e/acc内部成员政治立场的调查，结果相当均衡。所以，这是我们这个时代一个全新的根本性议题。它不仅仅是中心化对去中心化的问题，它更像是……科技进步主义对科技保守主义的对决，对吧？ GUILLAUME VERDON (02:27:05) But, it seems to be like there is sort of case of alignment of the existing political parties, where those that are for more centralization of power, control, and more regulations are aligning themselves with the doomers, because that sort of instilling fear in people is a great way for them to give up more control, and give the government more power. But fundamentally, we’re not left versus right. I think we’ve done polls of people’s alignment within EAC. I think it’s pretty balanced. So, it’s a new fundamental issue of our time. It’s not just centralization versus decentralization. It’s kind of do we go… It’s like tech progressivism, versus techno conservatism. Right? 有效利他主义 Effective altruism Lex Fridman (02:27:54) e/acc作为一个运动，经常被拿来与EA——有效利他主义——做对比。你认为有效利他主义有哪些优点和缺点？在你看来，它有什么有见地的地方，又有什么不足？ LEX FRIDMAN (02:27:54) So e/acc as a movement is often formulated in contrast to EA, effective altruism. What do you think are the pros and cons of effective altruism? What’s interesting, insightful to you about them, and what is negative? Guillaume Verdon (02:28:15) 嗯，我认为人们试图从第一性原理出发去做好事，这本身是好的。 GUILLAUME VERDON (02:28:15) Right. I think people trying to do good from first principles is good. Lex Fridman (02:28:23) 其实我们应该先说明一下，抱歉打断你——如果我说错了你可以纠正——有效利他主义是一种试图以最优方式行善的运动，其中”善”大概是用世界上的痛苦总量来衡量的，目标是将其最小化。但正如任何优化问题一样，它可能会走偏。所以，探讨它可能如何走偏是很有意思的。 LEX FRIDMAN (02:28:23) We should actually say, and sorry to interrupt, we should probably say that, and you can correct me if I’m wrong, but effective altruism is the kind of movement that’s trying to do good optimally, where good is probably measured something like the amount of suffering in the world. You want to minimize it. And there’s ways that that can go wrong, as any optimization can. And so, it’s interesting to explore how things can go wrong. Guillaume Verdon (02:28:55) 我们双方都在试图做好事，分歧在于应该用什么损失函数，对吧？ GUILLAUME VERDON (02:28:55) We’re both trying to do good to some extent, and we’re arguing for which loss function we should use, right? Lex Fridman (02:29:03) 是的。 LEX FRIDMAN (02:29:03) Yes. Guillaume Verdon (02:29:04) 他们的损失函数是某种hedons（书童注：享乐单位，用于量化快乐程度的假设单位），即享乐主义的度量单位——你感觉有多好，持续多久？痛苦就是负的hedons，他们试图将其最小化。但在我们看来，这个损失函数存在某种虚假的局部最小值。比如你可能会去最小化虾类养殖场的痛苦，这在我看来并不怎么有建设性。又或者你可能陷入”线头享乐主义”（书童注：wireheading，指通过直接刺激大脑奖励中心获得快感而忽视其他目标的状态）——装一个脑机接口，或者永远刷TikTok，短期内因为神经化学反应而感觉良好，但从长期来看，它导向衰败与死亡，因为你什么也没有创造。 GUILLAUME VERDON (02:29:04) Their loss function is sort of hedons, units of hedonism. How good do you feel, and for how much time? And so, suffering would be negative hedons, and they’re trying to minimize that. But to us that seems like that loss function has sort of spurious minima, you can start minimizing shrimp farm pain, which seems not that productive to me. Or you can end up with wire heading, where you just either install a neural link, or you scroll TikTok forever, and you feel good on the short-term timescale because of your neurochemistry, but on a long-term timescale, it causes decay and death, because you’re not being productive. Guillaume Verdon (02:29:54) 而e/acc的做法则是用一个客观的度量来衡量文明进步——不是享乐主义这种主观的损失函数，而是物理能量这一客观量，它无法被操纵，极其客观，没有多少空子可钻。如果用GDP或货币来衡量，那是锚定在某个浮动价值上的，不是衡量进步的好方法。但归根结底，我们双方都在试图推动进步、确保人类繁荣发展，只是选择了不同的损失函数和不同的路径。 GUILLAUME VERDON (02:29:54) Whereas sort of EAC, measuring progress of civilization, not in terms of a subjective loss function like hedonism, but rather an objective measure, quantity that cannot be gamed that is physical energy, it’s very objective, and there’s not many ways to game it. If you did it in terms of GDP, or a currency, that’s pinned to certain value that’s moving. And so, that’s not a good way to measure our progress. But the thing is we’re both trying to make progress, and ensure humanity flourishes, and gets to grow. We just have different loss functions, and different ways of going about doing it. Lex Fridman (02:30:42) 也许你可以指点我、纠正我——每当有人试图把整个人类文明、人类体验简化为一个方程式时，我总会有些警惕。我们是否应该对方程式的暴政、对优化损失函数的执念保持怀疑？是否需要对”优化损失函数”这件事本身抱有一种认知上的谦逊？ LEX FRIDMAN (02:30:42) Is there a degree, maybe you can educate me, correct me, I get a little bit skeptical when there’s an equation involved trying to reduce all of the human civilization, human experience to an equation. Is there a degree that we should be skeptical of the tyranny of an equation of a loss function over wish to optimize? Like having a kind of intellectual humility about optimizing over loss functions? Guillaume Verdon (02:31:12) 是的。这个特定的损失函数并不是刚性的，它更像是平均值的平均值——未来状态的分布本身服从某种分布。所以它不是确定性的，我们并没有被锁死在某条固定轨道上，这只是关于未来的一个统计性陈述。但说到底，你可以选择信不信引力，但不服从它可不是一个选项——有些人试过了，下场不太好。同理，热力学就在那里，不以我们的好恶为转移。我们只是在试图指出”什么是”，然后据此找准自己的方位，规划前行的路径。 GUILLAUME VERDON (02:31:12) Yeah. So, this particular loss function, it’s not stiff. It’s kind of an average of averages. It’s like distributions of states in the future are going to follow a certain distribution. So it’s not deterministic, it’s not like… We’re not on stiff rails. It’s just a statistical statement about the future. But at the end of the day, you can believe in gravity or not, but it’s not necessarily an option to obey it. And some people try to test that, and that goes not so well. So, similarly, I think thermodynamics is there whether we like it or not, and we’re just trying to point out what is, and try to orient ourselves, and chart a path forward given this fundamental truth. Lex Fridman (02:32:04) 但终究还是存在不确定性，存在信息的缺口，而人类天生倾向于用叙事来填补这些缺口。对物理学的解读……当涉及不确定性时，即使物理学也可以被各种诠释。人类总善于利用这种模糊性来为自身目的服务。所以每当一个方程式出现，在我们真正完美理解宇宙之前，人类总会做人类擅长的事——借”行善”之名蒙蔽大众，行不善之实。我想这是我们对所有运动都应保持警惕的地方。 LEX FRIDMAN (02:32:04) But there’s still some uncertainty, there’s still a lack of information, and humans tend to fill the gap of the lack of information with narratives. And so, how they interpret… Even physics is up to interpretation when there’s uncertainty involved. And humans tend to use that to further their own means. So, it’s always, whenever there’s an equation, it just seems like until we have really perfect understanding of the universe, humans will do what humans do, and they try to use the narrative of doing good to fool the populace into doing bad. I guess that this is something that we should be skeptical about in all movements. Guillaume Verdon (02:32:57) 没错，所以我们欢迎质疑，对吧？ GUILLAUME VERDON (02:32:57) That’s right? So we invite skepticism. Right? Lex Fridman (02:33:02) 你觉得有效利他主义到底哪里出了问题？这些问题是否也可能出现在有效加速主义身上？ LEX FRIDMAN (02:33:02) Do you have an understanding of what might, to a degree that went wrong, what do you think may have gone wrong with effective altruism that might also go wrong with effective accelerationism? Guillaume Verdon (02:33:15) 嗯，我觉得它早期确实为工程师、知识分子和理性主义者提供了一种社群归属感，那时社群看起来非常健康。但后来他们组建了各种机构，开始调配资本，掌握了实际权力。他们拥有真实的权力——他们影响政府，如今影响着大多数AI组织。他们实质上控制着OpenAI的董事会，再看看Anthropic，我想他们在那里也有相当的影响力。而e/acc的预设更像是资本主义的逻辑：每个个体、每个组织和元组织都会为自身利益行事，我们应该在所有时间、所有尺度上保持某种对抗性均衡或竞争博弈，以相互制约。我认为，说到底，”行善”的旗号为他们积累巨量权力和资本提供了完美的掩护，而不幸的是，权力往往会随时间腐蚀人心。 GUILLAUME VERDON (02:33:15) Yeah, I mean I think it provided initially a sense of community for engineers, and intellectuals, and rationalists in the early days, and it seems like the community was very healthy, but then they formed all sorts of organizations, and started routing capital, and having actual power. They have real power. They influence the government, they influence most AI orgs now. I mean, they’re literally controlling the board of OpenAI, and look over to Anthropic. I think they’ll have some control over that too. And so, I think the assumption of e/acc is more like capitalism, is that every agent organism and meta organism is going to act in its own interest, and we should maintain sort of adversarial equilibrium, or adversarial competition to keep each other in check at all times, at all scales. I think that yeah, ultimately, it was the perfect cover to acquire tons of power, and capital, and unfortunately sometimes that corrupts people over time. 生活中的一天 Day in the life Lex Fridman (02:34:23) 既然你说建造很重要，那Guillaume Verdon的完美高效一天是什么样的？你每天摄入多少咖啡因？完美的一天长什么样？ LEX FRIDMAN (02:34:23) What does a perfectly productive day, since building is important, what is a perfectly productive day in the life of Guillaume Verdon look like? How much caffeine do you consume? What’s a perfect day? Guillaume Verdon (02:34:39) 好的，我有一套特定的作息。我最理想的工作日是中午12点到凌晨4点。下午早些时候处理会议，通常是外部会议，也有一些内部会议。因为我是CEO，必须和外界打交道——客户、投资者、面试候选人。这段时间我通常会摄入外源性酮体。 GUILLAUME VERDON (02:34:39) Okay, so I have a particular regimen. I would say my favorite days are 12:00 PM to 4:00 AM, and I would have meetings in the early afternoon, usually external meetings, some internal meetings. Because I’m CEO, I have to interface with the outside world, whether it’s customers, or investors, or interviewing potential candidates. And usually I’ll have ketones, exogenous ketones. Lex Fridman (02:35:12) 所以你在遵循生酮饮食，还是—— LEX FRIDMAN (02:35:12) So, are you on a keto diet, or is this- Guillaume Verdon (02:35:16) 我以前为了打橄榄球做过生酮饮食之类的，但现在我喜欢把一天的一部分工作做完之后再吃饭，这样就能保持极致的专注力。 GUILLAUME VERDON (02:35:16) I’ve done keto before for football, and whatnot, but I like to have a meal after part of my day is done, and so I can just have extreme focus. Lex Fridman (02:35:31) 你在上半天空腹处理社交事务。 LEX FRIDMAN (02:35:31) You do the social interactions earlier in the day without food. Guillaume Verdon (02:35:35) 把社交前置，对。就像我现在就靠酮体和红牛撑着，它给你的思维清晰度真的是另一个层次。因为一旦你吃了东西，本来可以供给神经活动的能量就要分配给消化系统了。吃完饭后我大概休息一个小时到一个半小时。理想状态是一天只吃一顿——牛排、鸡蛋加蔬菜，以动物性食物为主，水果和肉。然后进入第二轮高峰期，那通常是深度工作时间。虽然我是CEO，但我依然保持技术参与，大多数专利都有我的贡献。那个时段我就熬到深夜，和工程师们一起攻克高度技术性的问题。 GUILLAUME VERDON (02:35:35) Front load them, yeah. Yeah. Like right now I’m on ketones, and a Red Bull, and it just gives you a clarity of thought that is really next level. Because then when you eat, you’re actually allocating some of your energy that could be going to neural energy to your digestion. After I eat, maybe I take a break, an hour or so, an hour and a half, and then usually it’s like ideally one meal a day, like steak and eggs, and vegetables, animal-based primarily. So, fruit and meat. And then I do a second wind, usually that’s deep work, because I am A CEO, but I’m still technical. I’m contributing to most patents. And there, I’ll just stay up late into the night, and work with engineers on very technical problems. Lex Fridman (02:36:25) 也就是大概晚上9点到凌晨4点那个时段。 LEX FRIDMAN (02:36:25) So it’s like the 9:00 PM to 4:00 AM, whatever though, that range of time. Guillaume Verdon (02:36:30) 对，那是完美的时段。邮件不再涌来，各种紧急事务也消停了，你终于可以专注。然后你迎来第二波高峰。据我所知，Demis Hassabis（书童注：DeepMind创始人）的工作节奏也差不多，这确实启发了我的作息安排。我在谷歌时就开始这么做了——白天管产品、开会，晚上做技术工作。 GUILLAUME VERDON (02:36:30) Yeah, yeah. That’s the perfect time. The emails, the things that are on fire stop trickling in, you can focus. And then you have your second wind. And I think Demis Hassabis has a similar workday to some extent. So, I think that’s definitely inspired my workday. But yeah, I started this workday when I was at Google, and had to manage a bit of the product during the day, and have meetings, and then do technical work at night. Lex Fridman (02:37:00) 那锻炼、睡眠这些呢？你提到了橄榄球，你以前打橄榄球？ LEX FRIDMAN (02:37:00) Exercise, sleep, those kinds of things. You said football, you used to play football? Guillaume Verdon (02:37:06) 对，我以前打美式橄榄球，从小各种运动都练过。后来有段时间痴迷撸铁。读研学数学那会儿，我的一天就是：做数学、举铁、摄入咖啡因，仅此而已。极其纯粹，最纯粹的僧侣模式。但撸铁有一点特别有意思：你通过特定的驱动信号来诱发神经适应，通过各种补剂来增强神经可塑性，举重时大脑会分泌各种脑源性神经营养因子。 GUILLAUME VERDON (02:37:06) Yeah, I used to play American football. I’ve done all sorts of sports growing up. And then I was into powerlifting for a while. So, when I was studying mathematics in grad school, I would just do math, and lift, take caffeine, and that was my day. It was very pure, the purest of monk modes. But it’s really interesting, how in powerlifting you’re trying to cause neural adaptation by having certain driving signals, and you’re trying to engineer a neuroplasticity through all sorts of supplements, and you have all sorts of brain derived neurotrophic factors that get secreted when you lift. Guillaume Verdon (02:37:44) 所以对我来说很有意思的是，我当时一边学数学，一边试图在整个神经系统——不仅仅是大脑——层面上去设计神经适应。我觉得如果你真正在乎所学的东西，学习速度会快得多。如果你能说服自己对正在学的东西极度在乎，再加上某种辅助——比如咖啡因，或某种胆碱能补剂来增强神经可塑性——效果会更显著。这个我应该找Andrew Huberman聊聊，他才是专家。但至少在我的经验里，你可以试着给大脑灌入更多的token，可以试着调高学习率，这样就能在更短的时间内学得更快。 GUILLAUME VERDON (02:37:44) So, it’s funny to me how I was trying to engineer a neural adaptation in my nervous system more broadly, not just my brain while learning mathematics. I think you can learn much faster if you really care. If you convince yourself to care a lot about what you’re learning, and you have some sort of assistance, let’s say caffeine, or some cholinergic supplement to increase neuroplasticity. I should chat with Andrew Huberman at some point. He’s the expert. But yeah, at least to me it’s like you can try to input more tokens into your brain, if you will, and you can try to increase the learning rate, so that you can learn much faster on a shorter timescale. Guillaume Verdon (02:38:30) 我就是这样学了很多东西——追随好奇心。如果你对正在做的事情充满热情，你就会学得更快，变聪明的速度也更快。如果你追随好奇心，你就永远不会觉得无聊。所以我建议大家追随自己的好奇心，不要被学科的边界或工作中被分配的赛道所束缚。走出去探索，跟着直觉走，尽可能多地获取信息、将其压缩进你的大脑——任何你觉得有趣的东西都行。 GUILLAUME VERDON (02:38:30) So, I’ve learned a lot of things. I’ve followed my curiosity. You’re naturally… If you’re passionate about what you’re doing, you’re going to learn faster, you’re going to become smarter faster. And if you follow your curiosity, you’re always going to be interested. And so, I advise people to follow their curiosity and don’t respect the boundaries of certain fields, or what you’ve been allocated in terms of lane of what you’re working on. Just go out and explore, and follow your nose, and try to acquire, and compress as much information as you can into your brain. Anything that you find interesting. Lex Fridman (02:39:05) 还有就是在乎一件事。就像你说的，这一点很有意思，对我也非常管用——就是”骗”自己去在乎一件事。 LEX FRIDMAN (02:39:05) And caring about a thing. Like you said, which is interesting, it works for me really well, is tricking yourself that you care about a thing. Guillaume Verdon (02:39:12) 是的。 GUILLAUME VERDON (02:39:12) Yes. Lex Fridman (02:39:13) 然后你就真的开始在乎了。 LEX FRIDMAN (02:39:13) And then you start to really care about it. Guillaume Verdon (02:39:15) 没错。 GUILLAUME VERDON (02:39:15) Yep. Lex Fridman (02:39:15) 所以有意思的是，热情是学习的绝佳催化剂。 LEX FRIDMAN (02:39:15) So, it’s funny, the motivation is a really good catalyst for learning. Guillaume Verdon (02:39:22) 没错。所以我扮演Beff Jezos这个角色，至少有一部分就是这种…… GUILLAUME VERDON (02:39:22) Right. And so, at least part of my character, as Beff Jezos is kind of like… Lex Fridman (02:39:29) 对，自我激励大师。 LEX FRIDMAN (02:39:29) Yeah, hype man. Guillaume Verdon (02:39:30) 对，但其实我是在给自己打鸡血，然后顺便发条推文而已。当我试图让自己进入一种极度亢奋的状态——一种改变了的意识状态，极度专注、进入心流、全身通电、试图发明从未存在过的东西——我需要达到一种不可思议的兴奋程度。而大脑确实有这些隐藏的认知层级，可以通过更高水平的肾上腺素等来解锁。我从撸铁中学到了这一点：你可以训练出一个心理开关来增强你的力量。如果你能设计出这样一个开关——也许是某首歌、某段音乐作为触发信号——突然间你就完全进入状态，达到最大力量输出。我通过多年的举铁训练出了这个开关。当你要扛起500磅的杠铃、它随时可能压碎你的时候，如果你没有把那个开关打开，你可能会死。这种求生本能会瞬间唤醒你的全部潜能。我把这项技能迁移到了研究工作中——每到关键时刻、赌注极高的时候，我就能进入另一个层次的神经表现状态。 GUILLAUME VERDON (02:39:30) Yeah, but I’m hyping myself up, but then I just tweet about it, and it’s just when I’m trying to get really hyped up, and an altered state of consciousness where I’m ultra focused, in the flow, wired, trying to invent something that’s never existed, I need to get to unreal levels of excitement. But your brain has these levels of cognition that you can unlock with higher levels of adrenaline, and whatnot. And I mean, I’ve learned that in powerlifting, that actually you can engineer a mental switch to increase your strength. If you can engineer a switch, maybe you have a prompt, like a certain song or some music where suddenly you’re fully primed, then you’re at max, maximum strength. And I’ve engineered that switch through years of lifting. If you’re going to get under 500 pounds and it could crush you, if you don’t have that switch to be wired in, you might die. So, that’ll wake you right up. That sort of skill I’ve carried over to research, when it’s go time, when the stakes are high, somehow I just reach another level of neural performance. Lex Fridman (02:40:40) 所以Beff Jezos就是你的智识版浩克，你的生产力浩克——把开关一拧就行了。 LEX FRIDMAN (02:40:40) So Beff Jezos is your sort of embodiment representation of your intellectual Hulk. It’s your productivity Hulk that you just turn on. 身份 Identity Guillaume Verdon (02:40:50) 是的。 GUILLAUME VERDON (02:40:50) Yeah. Lex Fridman (02:40:50) 拥有这两个身份，你从中对身份的本质有了什么领悟？我觉得有意思的是，一个人能如此明确地在两顶帽子之间切换。 LEX FRIDMAN (02:40:50) What have you learned about the nature of identity from having these two identities? I think it’s interesting for people, to be able to put on those two hats so explicitly. Guillaume Verdon (02:41:01) 早期很有意思。那时候我觉得两个身份是完全隔开的——”哦，这不过是个角色，我是Guillaume，Beff只是一个人设。”我把自己的想法拿过来，再推到更极端一点。但随着时间推移，两个身份在心理上开始融合。有人会对我说：”不，我见过你本人，你就是Beff，你不仅仅是Guillaume。”我当时愣了一下：”等等，我是吗？”现在两者已经完全合一了。但其实在身份被曝光之前，这种融合在心理层面就已经开始了——我就是这个角色，它是我的一部分。 GUILLAUME VERDON (02:41:01) I think it was interesting in the early days, I think in the early days, I thought it was truly compartmentalized. Like, “Oh yeah, this is a character. I’m Guillaume. Beff is just the character.” I take my thoughts, and then I extrapolate them to a bit more extreme. But over time, it’s kind of like both identities were starting to merge mentally, and people were like, “No, I met you. You are Beff. You are not just Guillaume.” And I was like, “Wait, am I?” And now it’s fully merged. But it was already, before the docs, it was already starting mentally that I am this character. It’s part of me. Lex Fridman (02:41:39) 你会推荐别人也搞一个小号吗？ LEX FRIDMAN (02:41:39) Would you recommend people have an alt? Guillaume Verdon (02:41:42) 绝对推荐。 GUILLAUME VERDON (02:41:42) Absolutely. Lex Fridman (02:41:43) 比如年轻人，你会建议他们通过小号来探索不同的身份吗？匿名账号？ LEX FRIDMAN (02:41:43) Like young people. Would you recommend them to explore different identities by having alts? Alt accounts? Guillaume Verdon (02:41:49) 这很有趣。就像写一篇辩论文章并选择一个立场，对吧？辩论赛里你就是这么做的。关键是你可以有实验性的想法——因为赌注极低，你只是一个匿名账号，可能才20个粉丝，你可以在一个低风险的环境里试验你的想法。我觉得在这个一切都绑定真名、一切都可追溯到你本人的时代，我们失去了这种可能性。人们害怕开口，害怕探索那些尚未成形的想法，我觉得我们在这里失去了一些珍贵的东西。所以我希望X等平台能真正支持用户保持匿名，因为让人们自由分享尚未成熟的想法、逐步逼近那些隐藏的真理至关重要——仅靠实名的公开讨论，很难抵达那些真理。 GUILLAUME VERDON (02:41:49) It’s fun. It’s like writing an essay, and taking a position, right? It’s like you do this in debate. It’s like you can have experimental thoughts, and by the stakes being so low, because you’re an anon account with, I don’t know, 20 followers or something, you can experiment with your thoughts in a low stakes environment. And I feel like we’ve lost that in the era of everything being under your main name, everything being attributable to you. People just are afraid to speak, explore ideas that aren’t fully formed, and I feel like we’ve lost something there. So, I hope platforms like X and others really help support people trying to stay synonymous, or anonymous, because it’s really important for people to share thoughts that aren’t fully formed, and converge onto maybe hidden truths that were hard to converge upon if it was just through open conversation with real names. Lex Fridman (02:42:46) 是的。我真心信奉一种——不是激进的，而是严谨的共情。就是认真去想象：持有某种观点的人，他的世界是什么样的？然后把这个思想实验一步步推向更深处。小号就是实现这一点的一种方式。它提供了一种有趣的途径，让你真正去体验”作为一个持有某套信念的人”是什么感觉，并且在几天、几周、几个月的跨度里持续扮演。当然，始终存在一个危险：你可能真的变成那个人。这就是尼采说的——”当你长久凝视深渊时，深渊也在凝视你。”得小心。 LEX FRIDMAN (02:42:46) Yeah. I really believe in not radical, but rigorous empathy. It’s like really considering what it’s like to be a person of a certain viewpoint, and taking that, as a thought experiment, farther and farther and farther. And one way of doing that as an alt account. That’s a fun, interesting way to really explore what it’s like to be a person that believes a set of beliefs, and taking that across the span of several days, weeks, months. Of course there’s always the danger of becoming that. That’s the Nietzche, “Gaze long into the abyss, the abyss gazes into you.” You have to be careful. Guillaume Verdon (02:42:46) Breaking Beff（书童注：戏仿美剧《绝命毒师》Breaking Bad）。 GUILLAUME VERDON (02:42:46) Breaking Beff. 给年轻人的建议 Advice for young people Lex Fridman (02:43:31) 对，Breaking Beff。哪天你醒来发现自己剃了光头：”我是谁？我变成了什么？”好了，你已经给出了不少建议，但你会给年轻人什么忠告？在我们所处的这个风云变幻的世界里，如何去经营一份事业，过一种自己可以引以为豪的人生？ LEX FRIDMAN (02:43:31) Yeah, right. Breaking Beff. Yeah. You wake up with a shaved head one day, just like, “Who am I? What have I become?” So, you’ve mentioned quite a bit of advice already, but what advice would you give to young people of, in this interesting world we’re in, how to have a career and how to have a life they can be proud of? Guillaume Verdon (02:43:58) 对我来说，当初选择理论物理的原因是：我想学习技术栈的最底层——那些无论技术如何迭代都不会过时的东西。那是我的根基，后来我在此基础上一层层搭建起工程技能和其他能力。物理定律不会变。当下的技术版图看起来变化飞快，让人晕头转向，但有些东西——基础数学和物理——是永恒的。如果你掌握了这些知识，再加上对复杂系统和适应性系统的理解，我认为你会走得非常远。不是每个人都必须学数学，但我认为学习数学、物理和工程，真的是一次巨大的认知解锁。 GUILLAUME VERDON (02:43:58) I think to me, the reason I went to theoretical physics was that I had to learn the base of the stack that was going to stick around no matter how the technology changes. And to me, that was the foundation upon which then I later built engineering skills, and other skills. And to me, the laws of physics, it may seem like the landscape right now is changing so fast, it’s disorienting. But certain things like fundamental mathematics and physics aren’t going to change. And if you have that knowledge, and knowledge about complex systems, and adaptive systems, I think that’s going to carry you very far. And so, not everybody has to study mathematics, but I think it’s really a huge cognitive unlock to learn math, and some physics, and engineering. Lex Fridman (02:44:48) 尽可能深入到技术栈的最底层。 LEX FRIDMAN (02:44:48) Get as close to the base of the stack as possible. Guillaume Verdon (02:44:51) 对，没错。因为技术栈的底层是不变的。其他一切……你的知识可能几年后就不再那么相关了。当然你可以做一种迁移学习，但那样你就得不停地迁移，永无止境。 GUILLAUME VERDON (02:44:51) Yeah, that’s right. Because the base of the stack doesn’t change. Everything else… Your knowledge might become not as relevant in a few years. Of course there’s a sort of transfer learning you can do, but then you have to always transfer learn, constantly. Lex Fridman (02:45:04) 我想你越接近技术栈的底层，迁移学习就越容易，跨度也越小。 LEX FRIDMAN (02:45:04) I guess the closer you are to the base of the stack, the easier the transfer learning, the shorter the jump. Guillaume Verdon (02:45:10) 对。你会惊讶于，一旦你在各种物理场景中掌握了核心概念，它们竟能如此自然地迁移到理解其他非物理系统中去。e/acc的那些著述——原则与信条那几篇帖子——就是基于物理学的。那其实就是我尝试将非平衡热力学的思维方式应用于理解周围世界的一次实验，而它最终催生了e/acc和这场运动。 GUILLAUME VERDON (02:45:10) Right, right. And you’d be surprised, once you’ve learned concepts in many physical scenarios, how they can carry over to understanding other systems that aren’t necessarily physics. And I guess the e/acc writings, the principles and tenet posts, that was based on physics, that was kind of my experimentation with applying some of the thinking from out of [inaudible 02:45:36] thermodynamics to understanding the world around us, and it’s led to e/acc, and this movement. 死亡 Mortality Lex Fridman (02:45:42) 如果你把自己看作这台资本主义机器中的一个齿轮，一个人类个体——你认为死亡是一个特性还是一个bug？你想要永生吗？ LEX FRIDMAN (02:45:42) If you look at you’re one cog in the machine, in the capitalist machine, one human, and if you look at yourself, do you think mortality is a feature or a bug? Would you want to be immortal? Guillaume Verdon (02:45:57) 不。我认为从根本上说，在”热力学耗散适应”这个概念中，”耗散”二字赫然在列。耗散是重要的，死亡也是重要的。物理学界有句老话：物理学的进步，是一场葬礼接一场葬礼地推进的。 GUILLAUME VERDON (02:45:57) No, I think fundamentally, in thermodynamic dissipative adaptation, there’s the word dissipation. Dissipation is important, death is important. We have a saying in physics, physics progresses one funeral at a time. Lex Fridman (02:46:16) 是的。 LEX FRIDMAN (02:46:16) Yeah. Guillaume Verdon (02:46:17) 资本主义同样如此。公司、帝国、人——万物终有一死。不过我确实认为我们应该延长寿命，因为世界日趋复杂，我们需要更长的训练周期来消化越来越多的数据，从而真正理解和预测世界。如果高神经可塑性的窗口是有限的，那我们理解世界的能力就有一个硬性上限。所以，我支持死亡——因为它确实重要。如果你有一个永远不死的国王，那会是灾难。系统将无法持续适应，对吧？ GUILLAUME VERDON (02:46:17) I think the same is true for capitalism. Companies, empires, people, everything. Everything must die at some point. I think that we should probably extend our lifespan, because we need a longer period of training, because the world is more and more complex. We have more and more data to really be able to predict and understand the world. And if we have a finite window of higher neuroplasticity, then we have sort of a hard cap in how much we can understand about our world. So, I think I am for death, because again, I think it’s important. If you have a king that would never die, that would be a problem. The system wouldn’t be constantly adapting, right? Guillaume Verdon (02:47:05) 你需要新鲜血液，需要年轻一代，需要颠覆性力量来确保系统始终在适应、始终保持可塑性。否则，如果事物永生不灭——比如一家永远霸占市场的垄断公司——它们就会钙化，在不断变化的动态环境中变得不再最优、不再具有高适应性。所以，死亡为年轻和新鲜事物腾出了空间。我认为这是自然界中每个系统的必要组成部分。总而言之，我支持死亡，但也确实认为更长的寿命、更持久的神经可塑性窗口、更大的大脑，应该是我们努力追求的方向。 GUILLAUME VERDON (02:47:05) You need novelty, you need youth, you need disruption to make sure the system’s always adapting, and malleable. Otherwise, if things are immortal, if you have, let’s say corporations that are there forever, and they have the monopoly, they get calcified, they become not as optimal, not as high fitness in a changing, time varying landscape. And so, death gives space for youth and novelty to take its place. And I think it’s an important part of every system in nature. So yeah, I am for death, but I do think that longer lifespan, and longer time for neuroplasticity, bigger brains should be something we should strive for. Lex Fridman (02:47:52) 有意思的是，Jeff Bezos和Beff Jezos在这一点上达成了共识：所有公司终将消亡。Jeff的做法是他所说的”Day One思维”——试图不断自我革新，尽可能延长公司的寿命。但最终它也会死去，因为持续的自我革新实在太难了。你害怕自己的死亡吗？ LEX FRIDMAN (02:47:52) Well, and that, Jeff Bezos, and Beff Jezos agree that all companies die. And for Jeff, the goal is to try to, he calls it day one thinking, try to constantly, for as long as possible, reinvent, sort of extend the life of the company. But eventually it too will die, because it’s so difficult to keep reinventing. Are you afraid of your own death? Guillaume Verdon (02:48:23) 我心中有很多想法，有很多想在这个世界上实现的事情，趁我还在的时候。但我不觉得自己害怕死亡。 GUILLAUME VERDON (02:48:23) I think I have ideas and things I’d like to achieve in this world before I have to go, but I don’t think I’m necessarily afraid of death. Lex Fridman (02:48:34) 所以你并不执着于你此生获得的这副身体和这颗头脑？ LEX FRIDMAN (02:48:34) So you’re not attached to this particular body, and mind that you got? Guillaume Verdon (02:48:38) 不执着。我相信未来一定会有比我更好的版本，或者…… GUILLAUME VERDON (02:48:38) No, I’m sure there’s going to be better versions of myself in the future, or… Lex Fridman (02:48:46) 分叉？ LEX FRIDMAN (02:48:46) Forks? Guillaume Verdon (02:48:47) 分叉，对吧？基因层面的分叉，或者其他形式的。我真心相信这一点。我认为世界的每一个比特、每一个[听不清]都在通过我们在e/acc中描述的这个过程不断适应，其中运行着一种类进化算法。保持这种适应的可塑性，才是整台机器得以持续优化的方式。所以，我不认为自己是一个需要永远保留的最优解。未来一定会出现在诸多方面更优的解。 GUILLAUME VERDON (02:48:47) Forks, right? Genetic forks, or other, right? I truly believe that. I think there’s a sort of evolutionary-like algorithm happening at every bit, or [inaudible 02:49:03] in the world is sort of adapting through this process that we described in e/acc. And I think maintaining this adaptation malleability is how we have constant optimization of the whole machine. And so, I don’t think I’m particularly an optimum that needs to stick around forever. I think there’s going to be greater optima in many ways. 生命的意义 Meaning of life Lex Fridman (02:49:25) 你认为这一切的意义是什么？这台机器运转的”为什么”是什么？e/acc这台机器的？ LEX FRIDMAN (02:49:25) What do you think is the meaning of it all? What’s the why of the machine? The e/acc machine? Guillaume Verdon (02:49:32) 为什么？答案就是热力学。这就是我们存在于此的原因，这就是催生生命、文明、技术演化和文明增长的力量。但我们为什么有热力学？为什么偏偏是这个宇宙？为什么是这些特定的超参数、这些自然常数？这就进入了人择原理和可能宇宙的景观问题，对吧？我们恰好处在一个允许生命存在的宇宙中。那为什么？是否存在许多宇宙？这个我不知道。但我们是否有可能设计新的宇宙，或者创造口袋宇宙，并设定其超参数，使得我们的存在与那个宇宙之间存在某种互信息——使我们在某种意义上成为它的创造者？我觉得这……真的非常诗意。当然这纯属猜想。但这也正是为什么破解量子引力如此重要——它会让我们知道，这一切是否有可能。 GUILLAUME VERDON (02:49:32) The why? Well, the why is thermodynamics. It’s why we’re here. It’s what has led to the formation of life, and of civilization, of evolution of technologies, and growth of civilization. But why do we have thermodynamics? Why do we have our particular universe? Why do we have these particular hyper-parameters, the constants of nature? Well then you get into the anthropic principle, and the landscape of potential universes, right? We’re in the universe that allows for life. And then why, is there potentially many universes? I don’t know. I don’t know that part. But could we potentially engineer new universes, or create pocket universes, and set the hyper-parameters so there is some mutual information between our existence in that universe, and we’d be somewhat its parents? I think that’s really… I don’t know, that’d be very poetic. It’s purely conjecture. But again, this is why figuring out quantum gravity would allow us to understand if we can do that. Lex Fridman (02:50:39) 在这之上还有一层：为什么这一切看起来如此美丽、如此令人激动？探索量子引力的征程本身就让人心潮澎湃。为什么？为什么我们会被它吸引、被它牵引？那种解谜的、创造性的力量，似乎支撑着这一切。 LEX FRIDMAN (02:50:39) And above that, why does it all seems so beautiful and exciting? The quest to figuring out quantum gravity seems so exciting. Why? Why is that? Why are we drawn to that? Why are we pulled towards that? Just that puzzle solving creative force that underpins all of it, it seems like. Guillaume Verdon (02:51:01) 我认为我们在追求一种东西——就像大语言模型试图最小化其内部模型与真实世界之间的交叉熵一样，我们也在试图最小化我们的预测与真实世界之间的统计偏差。在某些能量尺度或物理尺度上，如果我们没有任何可见性、没有预测和感知的能力，那对我们来说简直是一种冒犯。我们渴望更好地理解世界，从而能够最好地引导它——或者引导我们穿越其中。 GUILLAUME VERDON (02:51:01) I think we seek, just like an LLM seats to minimize cross entropy between its internal model and the world, we seek to minimize… Yeah, the statistical divergence between our predictions and the world, and the world itself. And having regimes of energy scales, or physical scales in which we have no visibility, no ability to predict, or perceive, that’s kind of an insult to us. And we want to be able to understand the world better in order to best steer it, or steer us through it. Guillaume Verdon (02:51:37) 归根结底，这是一种进化而来的能力——你越能预测世界，就越能捕获效用或自由能，用于自身的存续和增长。量子引力，就像是知识获取的最终Boss——一旦我们攻克了它，潜在的可能性将是巨大的。但从现在到那一步之间，在介观尺度上还有太多要学习的东西。关于我们这个世界，还有海量的信息有待获取，在感知、预测和控制的工程化方面还有大量工作要做，才能攀上卡尔达舍夫文明等级的更高阶梯。对我们来说，这就是我们这个时代的宏大挑战。 GUILLAUME VERDON (02:51:37) And in general, it’s a capability that has evolved because the better you can predict the world, the better you can capture utility, or free energy towards your own sustenance and growth. And I think quantum gravity, again, is kind of the final boss, in terms of knowledge acquisition, because once we’ve mastered that, then we can do a lot, potentially. But between here and there, I think there’s a lot to learn in the meso scales. There’s a lot of information to acquire about our world, and a lot of engineering perception, prediction, and control to be done, to climb up the Carta shift scale. And to us, that’s the great challenge of our times. Lex Fridman (02:52:22) 当你不确定前路在何方时，让模因来开路。 LEX FRIDMAN (02:52:22) And when you’re not sure where to go, let the meme pave the way. Guillaume Verdon (02:52:26) 没错。 GUILLAUME VERDON (02:52:26) That’s right. Lex Fridman (02:52:27) Guillaume，Beff，感谢今天的对话。感谢你正在做的工作，感谢你向这个世界注入的幽默与智慧。这次对话太精彩了。 LEX FRIDMAN (02:52:27) Guillaume, Beff, thank you for talking today. Thank you for the work you’re doing. Thank you for the humor, and the wisdom you put into the world. This was awesome. Guillaume Verdon (02:52:37) 非常感谢你的邀请，Lex，我深感荣幸。 GUILLAUME VERDON (02:52:37) Thank you so much for having me, Lex, It’s a pleasure. Lex Fridman (02:52:40) 感谢收听本期与Guillaume Verdon的对话。如需支持本播客，请查看简介中的赞助商信息。最后，让我以阿尔伯特·爱因斯坦的一句话作结：”如果一个想法初听之下并不荒谬，那它就毫无希望可言。”感谢收听，期待下次再见。 LEX FRIDMAN (02:52:40) Thank you for listening to this conversation with Guillaume Verdon. To support this podcast. Please check out our sponsors in the description. And now, let me leave you with some words from Albert Einstein. “If at first the idea is not absurd, then there is no hope for it.” Thank you for listening. I hope to see you next time. （全文完） — End of Transcript —

【e/acc】外星智能、卡尔达肖夫天梯、技术资本机器 | 物理学家、e/acc运动创始人Guillaume Verdon与Lex Fridman播客实录 | 中英文完整版精译 Part3

Wed, 04 Mar 2026 00:00:00 +0000

书童按：本篇是Guillaume Verdon接受Lex Fridman播客采访实录的第三部分。在前两篇奠定理论根基之后，本篇探入更具思辨性的议题：外星智能的可能性与意义、量子引力的优美图景、黑洞信息悖论的最新进展、卡尔达舍夫等级的文明愿景，以及e/acc运动的核心纲领与文化基因。Verdon纵论从热力学驱动的生命起源到跨星际文明扩张的宏大叙事，批判去增长主义（degrowth）与ESG路径，主张以技术创新而非管理手段应对文明挑战。访谈亦涉及Jeff Bezos与Elon Musk等企业家的资本配置智慧、模因文化的传播策略、Extropic公司的热力学计算机愿景等话题。理论深度与实践洞察并重，视野恢宏，启人深思。初稿采用Claude Code机器翻译及排版，书童仅做简单校对及批注，以飨诸君。 外星人 Aliens Lex Fridman (01:35:11) 精度会随着你能力的降低而下降，但还是可以的。不过既然你提到了UAP（书童注：Unidentified Aerial Phenomena，不明空中现象），我们谈到了智能，我忘了问：你对可能存在于介观尺度上的其他智能有什么看法？你认为存在其他智能外星文明吗？思考这个问题有用吗？你多久想一次这个问题？ LEX FRIDMAN (01:35:11) The precision decreases in terms of your ability, but still. But since you mentioned UAPs, we talked about intelligence, and I forgot to ask, what’s your view on the other possible intelligences that are out there at the Meso scale? Do you think there’s other intelligent alien civilizations? Is that useful to think about? How often do you think about it? Guillaume Verdon (01:35:36) 我认为思考这个问题是有用的。之所以有用，是因为我们必须确保自己具有反脆弱性，并且正在尽可能快地提升我们的能力。因为我们可能被颠覆。物理定律并不禁止别处存在生命，那些生命可能进化并成为先进文明，最终来到我们这里。我认为他们现在就在这里吗？我不确定。关于这个话题，我读过的东西和大多数人读过的差不多。 GUILLAUME VERDON (01:35:36) I think it’s useful to think about. It’s useful to think about because we got to ensure we’re anti-fragile, and we’re trying to increase our capabilities as fast as possible. Because we could get disrupted. There’s no laws of physics against there being life elsewhere that could evolve and become an advanced civilization and eventually come to us. Do I think they’re here now? I’m not sure. I’ve read what most people have read on the topic. Guillaume Verdon (01:36:14) 我认为值得考虑，对我来说，这是一个有用的思想实验，用来灌输一种紧迫感——发展技术、提升我们的能力，确保我们不被颠覆。无论是某种形式的AI颠覆我们，还是来自不同星球的外来智能。无论哪种方式，提升我们的能力、让人类变得强大，我认为这非常重要，这样我们才能对宇宙向我们抛来的任何东西都保持鲁棒性。 GUILLAUME VERDON (01:36:14) I think it’s interesting to consider and to me, it’s a useful thought experiment to instill a sense of urgency in developing technologies and increasing our capabilities, to make sure we don’t get disrupted. Whether it’s a form of AI that disrupts us, or a foreign intelligence from a different planet. Either way, increasing our capabilities and becoming formidable as humans, I think that’s really important, so that we’re robust against whatever the universe throws at us. Lex Fridman (01:36:51) 但对我来说，这也是一个有趣的挑战和思想实验——如何感知智能。这与量子力学系统有关，也与任何不像人类的系统有关。对我来说，思想实验是：假设外星人就在这里，或者他们是可以直接观察到的。只是我们太盲目、太以自我为中心、没有合适的传感器，或者没有对传感器数据进行正确的处理，因而看不到我们周围显而易见的智能。 LEX FRIDMAN (01:36:51) But to me, it’s also an interesting challenge and thought experiment on how to perceive intelligence. This has to do with quantum mechanical systems. This has to do with any kind of system that’s not like humans. To me, the thought experiment is, say, the aliens are here or they are directly observable. We’re just too blind, too self-centered, don’t have the right sensors, or don’t have the right processing of the sensor data to see the obvious intelligence that’s all around us. Guillaume Verdon (01:37:26) 嗯，这就是我们为什么要研究量子传感器。它们可以感知引力。 GUILLAUME VERDON (01:37:26) Well, that’s why we work on quantum sensors. They can sense gravity, Lex Fridman (01:37:31) 是的。这很好，但可能还有其他东西，甚至不在当前已知的物理力量之中。 LEX FRIDMAN (01:37:31) Yeah. That’s a good one, but there could be other stuff that’s not even in the currently known forces of physics. Guillaume Verdon (01:37:43) 对。 GUILLAUME VERDON (01:37:43) Right. Lex Fridman (01:37:43) 可能有其他的东西。对我来说最有趣的思想实验是，那是显而易见的其他东西。不是我们缺乏传感器。它就在我们周围，意识可能就是其中之一。但可能有些东西就是明显地在那里。一旦你知道了它，就会说：”哦，对了。对了。我们认为以某种方式从物理定律中涌现出来的东西，我们理解它们，实际上是宇宙的基本组成部分，可以被纳入物理学。最被理解的。” LEX FRIDMAN (01:37:43) There could be some other stuff. The most entertaining thought experiment to me is that it’s other stuff that’s obvious. It’s not like we lack the sensors. It’s all around us, the consciousness being one possible one. But there could be stuff that’s just obviously there. That once you know it, it’s like, “Oh, right. Right. The thing we thought is somehow emergent from the laws of physics, we understand them, is actually a fundamental part of the universe and can be incorporated in physics. Most understood.” Guillaume Verdon (01:38:18) 从统计学上讲，如果我们观察到某种外星生命，它最有可能是某种病毒式的、自我复制的、类似冯·诺依曼探测器的系统。而且有可能存在这样的系统，我不知道它们在海底做什么——据说是这样——但也许它们在从海底收集矿物。 GUILLAUME VERDON (01:38:18) Statistically speaking, if we observed some sort of alien life, it would most likely be some sort of virally, self-replicating, von Neumann-like probe system. And it’s possible that there are such systems that, I don’t know what they’re doing at the bottom of the ocean, allegedly, but maybe they’re collecting minerals from the bottom of the ocean. Lex Fridman (01:38:44) 是的。 LEX FRIDMAN (01:38:44) Yeah. Guillaume Verdon (01:38:45) 但这不会违反我的任何先验假设。但我确定这些系统在这里吗？我很难这么说。我只有关于存在数据的二手信息。 GUILLAUME VERDON (01:38:45) But that wouldn’t violate any of my priors. But am I certain that these systems are here? It’d be difficult for me to say so. I only have secondhand information about there being data. Lex Fridman (01:38:59) 关于海底的？是的。但它可能是像模因这样的东西吗？可能是思想和观念吗？它们可能在那个媒介中运作吗？外星人可能就是进入我脑海的那些想法吗？想法的起源是什么？在你的脑海中，当一个想法进入你的脑海时，告诉我它从哪里来。 LEX FRIDMAN (01:38:59) About the bottom of the ocean? Yeah. But could it be things like memes? Could it be thoughts and ideas? Could they be operating at that medium? Could aliens be the very thoughts that come into my head? What’s the origin of ideas? In your mind, when an idea comes to your head, show me where it originates. Guillaume Verdon (01:39:25) 坦率地说，当我想到我现在正在建造的那种计算机的想法时——我想那是八年前了——感觉真的像是从太空中传来的光束。我躺在床上，浑身颤抖，只是在思考它。我不知道。但我真的相信这个吗？我不这么认为。但我认为外星生命可以采取多种形式，我认为智能的概念和生命的概念需要更广泛地扩展，变得不那么以人类为中心或以生物为中心。 GUILLAUME VERDON (01:39:25) Frankly, when I had the idea for the type of computer I’m building now, I think it was eight years ago now, it really felt like it was being beamed from space. I was in bed, just shaking, just thinking it through. I don’t know. But do I believe that legitimately? I don’t think so. But I think that alien life could take many forms, and I think the notion of intelligence and the notion of life needs to be expanded much more broadly to be less anthropocentric or biocentric. 量子引力 Quantum gravity Lex Fridman (01:40:04) 在量子力学上再停留一会儿，通过你在量子计算上的所有探索，你遇到过的最酷、最美丽的想法是什么——已经解决的或尚未解决的？ LEX FRIDMAN (01:40:04) Just to linger a little longer on quantum mechanics, through all your explorations on quantum computing, what’s the coolest, most beautiful idea that you’ve come across that has been solved or has not yet been solved? Guillaume Verdon (01:40:19) 我认为是理解一种叫做AdS/CFT的东西的旅程。也就是通过这样一幅图景来理解量子引力：一个维度较低的全息图实际上与一个额外维度的量子引力的体理论对偶或完全对应，而这种对偶性来自于试图学习边界的类似深度学习的表示。 GUILLAUME VERDON (01:40:19) I think the journey to understand something called AdS/CFT. So, the journey to understand quantum gravity through this picture, where a hologram of lesser dimension is actually dual or exactly corresponding to a bulk theory of quantum gravity of an extra dimension, and the fact that this sort of duality comes from trying to learn deep learning-like representations of the boundary. Guillaume Verdon (01:40:59) 至少，我的旅程中有一部分——有一天在我的愿望清单上——是将量子机器学习应用于这类系统，这些CFT（书童注：Conformal Field Theory，共形场论），或者它们被称为SYK模型（书童注：Sachdev-Ye-Kitaev模型），并从边界理论学习一个涌现的几何。所以，我们可以有一种形式的机器学习来帮助我们理解量子引力，这仍然是一个圣杯，我希望在离开这个世界之前达到。 GUILLAUME VERDON (01:40:59) At least, part of my journey someday on my bucket list is to apply quantum machine learning to these sorts of systems, these CFTs, or they’re called SYK models, and learn an emergent geometry from the boundary theory. And so, we can have a form of machine learning to help us understand quantum gravity, which is still a holy grail that I would like to hit before I leave this earth. Lex Fridman (01:41:35) 你认为黑洞正在发生什么？作为信息存储和处理单元，你认为黑洞正在发生什么？ LEX FRIDMAN (01:41:35) What do you think is going on with black holes? As information-storing and processing units, what do you think is going on with black holes? Guillaume Verdon (01:41:46) 黑洞是非常迷人的物体。它们处于量子力学和引力的界面上，所以它们帮助我们测试各种想法。我认为几十年来，一直存在这个黑洞信息悖论——落入黑洞的东西，我们似乎失去了它们的信息。现在，我认为有这个火墙悖论，据称在近年来被我的一位前同行解决了，他现在是伯克利的教授。在那里，似乎当信息落入黑洞时，有一种沉积作用。当你从外部观察者的角度越来越接近视界时，物体会无限减速。 GUILLAUME VERDON (01:41:46) Black holes are really fascinating objects. They’re at the interphase between quantum mechanics and gravity, and so they help us test all sorts of ideas. I think that for many decades now, there’s been this black hole information paradox that things that fall into the black hole, we’ve seem to have lost their information. Now, I think there’s this firewall paradox that has been allegedly resolved in recent years by a former peer of mine, who’s now a professor at Berkeley. There, it seems like, as information falls into a black hole, there’s a sedimentation. As you get closer and closer to the horizon from the point of view, the observer on the outside, the object slows down infinitely as it gets closer and closer. Guillaume Verdon (01:42:46) 从我们的角度来看，所有落入黑洞的东西都会沉积并附着在近视界处。在某个时刻，它离视界如此之近，以至于处于量子效应和量子涨落重要的邻近或尺度上。在那里，下落的物质可能会干扰传统图景，它可能会干扰真空中粒子和反粒子的产生和湮灭。通过这种干涉，其中一个粒子与下落的信息纠缠在一起，其中一个自由并逃逸。这就是外出辐射和下落物质之间存在互信息的方式。但正确计算这个，我认为我们才刚刚开始把碎片拼在一起。 GUILLAUME VERDON (01:42:46) Everything that is falling to a black hole, from our perspective, gets sedimented and tacked on to the near horizon. At some point, it gets so close to the horizon, it’s in the proximity or the scale in which quantum effects and quantum fluctuations matter. There, that infalling matter could interfere with the traditional pictures, that it could interfere with the creation and annihilation of particles and antiparticles in the vacuum. Through this interference, one of the particles gets entangled with the infalling information and one of them is now free and escapes. That’s how there’s mutual information between the outgoing radiation and the infalling matter. But getting that calculation right, I think we’re only just starting to put the pieces together. Lex Fridman (01:43:43) 有几个像”瘾君子”一样的问题我想问你。 LEX FRIDMAN (01:43:43) There’s a few pothead-like questions I want to ask you. Guillaume Verdon (01:43:46) 当然。 GUILLAUME VERDON (01:43:46) Sure. Lex Fridman (01:43:46) 一个是，我们银河系中心有一个巨大的黑洞，这让你感到恐惧吗？ LEX FRIDMAN (01:43:46) One, does it terrify you that there’s a giant black hole at the center of our galaxy? Guillaume Verdon (01:43:52) 我不知道。我只是想在它附近建立基地，以便快进，遇见未来的文明——如果我们的寿命有限，如果你可以去绕黑洞轨道运行然后出现。 GUILLAUME VERDON (01:43:52) I don’t know. I just want to set up shop near it to fast-forward, meet a future civilization, if we have a limited lifetime, if you could go orbit a black hole and emerge. Lex Fridman (01:44:08) 如果有一项特殊任务可以带你去黑洞，你会自愿去旅行吗？ LEX FRIDMAN (01:44:08) If there’s a special mission that could take you to a black hole, would you volunteer to go travel? Guillaume Verdon (01:44:13) 去轨道运行，显然不是掉进去。 GUILLAUME VERDON (01:44:13) To orbit and obviously not fall into it. Lex Fridman (01:44:15) 这是显而易见的。你显然认为黑洞里面的一切都被摧毁了？构成Guillaume的所有信息都被摧毁了？也许在另一边，Beff Jezos出现了，而且就像它以某种深刻的模因方式联系在一起。 LEX FRIDMAN (01:44:15) That’s obvious. It’s obvious to you that everything’s destroyed inside a black hole? All the information that makes up Guillaume is destroyed? Maybe on the other side, Beff Jezos emerges and it’s just all like it’s tied together in some deeply memeful way. Guillaume Verdon (01:44:32) 是的，这是一个很好的问题。我们必须回答黑洞是什么。我们是在时空中打一个洞并创造一个口袋宇宙吗？这是可能的。那么，这意味着如果我们攀登卡尔达肖夫等级到III型以上，我们可以设计具有特定超参数的黑洞，将信息传输到我们创造的新宇宙。所以，我们可以有后代—— GUILLAUME VERDON (01:44:32) Yeah, that’s a great question. We have to answer what black holes are. Are we punching a hole through space-time and creating a pocket universe? It’s possible. Then, that would mean that if we ascend the Kardashev scale to beyond Kardashev Type III, we could engineer black holes with specific hyperparameters to transmit information to new universes we create. And so, we can have progeny that our new… Guillaume Verdon (01:45:00) ……拥有作为新宇宙的后代。所以即使我们的宇宙可能达到热寂，我们也可能有一种留下遗产的方式。所以我们还不知道。我们需要攀登卡尔达舍夫等级来回答这些问题，窥视更高能量物理的那个领域。 GUILLAUME VERDON (01:45:00) … have progeny that are new universes. And so even though our universe may reach a heat death, we may have a way to have a legacy. And so we don’t know yet. We need to ascend the Kardashev Scale to answer these questions to peer into that regime of higher energy physics. 卡尔达舍夫等级 Kardashev scale Lex Fridman (01:45:25) 也许你可以向不知道的人介绍一下卡尔达舍夫等级。所以e/acc运动的一个类似模因的原则和目标就是攀登卡尔达舍夫等级。什么是卡尔达舍夫等级，我们什么时候想要攀登它？ LEX FRIDMAN (01:45:25) And maybe you can speak to the Kardashev Scale for people who don’t know. So one of the sort of meme-like principles and goals of the e/acc movement is to ascend the Kardashev Scale. What is the Kardashev Scale and when do we want to ascend it? Guillaume Verdon (01:45:43) 卡尔达舍夫等级是对我们能源生产和消耗的一种度量。实际上，它是一个对数尺度。卡尔达舍夫I型是一个里程碑，我们生产的瓦特数相当于太阳照射到地球上的所有能量。卡尔达舍夫II型将是利用太阳输出的所有能量。我想III型就像整个银河系的等价—— GUILLAUME VERDON (01:45:43) The Kardashev Scale is a measure of our energy production and consumption. Really, it’s a logarithmic scale. Kardashev Type 1 is a milestone where we are producing the equivalent wattage to all the energy that is incident on earth from the sun. Kardashev Type II would be harnessing all the energy that is output by the sun. And I think Type III is like the whole galaxy equivalent- Lex Fridman (01:46:13) 银河系，我想[听不清]是的。 LEX FRIDMAN (01:46:13) Galaxy, I think [inaudible 01:46:14] yeah. Guillaume Verdon (01:46:15) 是的，然后有些人有一些疯狂的IV型和V型，但我不知道我是否相信那些。但对我来说，从热力学的第一原理来看，似乎又有这个概念——热力学驱动的耗散适应——生命在地球上进化，因为我们有来自太阳的这种能量驱动，我们有入射能量，生命在地球上进化以找出最佳捕获那种自由能以维持自身和增长的方法。我认为这个原则，它不是我们地球-太阳系统所特有的。我们可以将生命扩展到远远超出。我们有责任这样做，因为正是这个过程把我们带到了这里。所以我们甚至不知道它为我们储存了什么。它可能是我们今天甚至无法想象的美丽的东西。 GUILLAUME VERDON (01:46:15) Yeah, and then some people have some crazy Type IV and V, but I don’t know if I believe in those. But to me, it seems like from the first principles of thermodynamics that, again, there’s this concept of thermodynamic- driven dissipative adaptation where life evolved on earth because we have this energetic drive from the sun, we have incident energy, and life evolved on earth to figure out ways to best capture that free energy to maintain itself and grow. And I think that that principle, it’s not special to our earth-sun system. We can extend life well beyond. And we kind of have a responsibility to do so because that’s the process that brought us here. So we don’t even know what it has its store for us in the future. It could be something of beauty we can’t even imagine today. 有效加速主义（e/acc） Effective accelerationism (e/acc) Lex Fridman (01:47:18) 所以这可能是一个谈论e/acc运动的好地方。在一篇题为《What the F* is e/acc?》（《e/acc到底是什么鬼？》）的Substack博客文章中，你写道：”从战略上讲，我们需要努力实现几个相互依存的总体文明目标。这四个目标是：增加我们作为一个物种可以利用的能量（攀登卡尔达舍夫梯度）。短期内，这几乎肯定意味着核裂变。通过支持人口增长的政策和支持经济增长的政策来增加人类繁荣。创造通用人工智能，人类历史上最伟大的力量倍增器。最后，发展行星际和星际运输，以便人类可以扩展到地球之外。“你能在此基础上进一步说明，也许说一下，对你来说e/acc运动是什么？目标是什么？原则是什么？ LEX FRIDMAN (01:47:18) So this is probably a good place to talk a bit about the e/acc movement in a Substack blog post titled, What the Fuck is e/acc? Or actually, What the F* is e/acc?, you write, “Strategically speaking, we need to work towards several overarching civilization goals that are all interdependent. And the four goals are, increase the amount of energy we can harness as a species, (climb the Kardashev gradient). In the short term, this almost certainly means nuclear fission. Increase human flourishing via pro-population growth policies and pro-economic growth policies. Create artificial general intelligence, the single greatest force multiplier in human history. And finally, develop interplanetary and interstellar transport so that humanity can spread beyond the earth. Could you build on top of that to maybe say, what to you is the e/acc movement? What are the goals? What are the principles? Guillaume Verdon (01:48:20) 目标是让人类技术-资本-模因机器变得（拥有）自我意识，并以超迷信的方式设计自己的增长。让我们拆解一下。 GUILLAUME VERDON (01:48:20) The goal is for the human techno-capital memetic machine to become self-aware and to hyperstitiously engineer its own growth. So let’s decompress that. Lex Fridman (01:48:33) 定义其中的每一个词。 LEX FRIDMAN (01:48:33) Define each of those words. Guillaume Verdon (01:48:35) 所以你有人类，你有技术，你有资本，然后你有模因、信息，所有这些系统都相互耦合。人类在公司工作，他们获取和分配资本，人类通过模因和信息传播进行交流。我们的目标是拥有一种病毒式的乐观主义运动，它意识到系统是如何运作的——从根本上说，它寻求增长——我们只是想顺应系统为自己的增长而适应的自然倾向。 GUILLAUME VERDON (01:48:35) So you have humans, you have technology, you have capital, and then you have memes, information, and all of those systems are coupled with one another. Humans work at companies, they acquire and allocate capital, and humans communicate via memes and information propagation. And our goal was to have a sort of viral optimistic movement that is aware of how the system works, fundamentally it seeks to grow, and we simply want to lean into the natural tendencies of the system to adapt for its own growth. Lex Fridman (01:49:18) 所以从这个意义上说，你是对的，e/acc字面上是一种模因式乐观主义病毒，它不断漂移、变异并以去中心化的方式传播。所以模因式乐观主义病毒。所以你确实希望它成为一种病毒以最大化传播，而且它具有超迷信性，因此乐观主义将激励其增长。 LEX FRIDMAN (01:49:18) So in that way, you’re right, the e/acc is literally a memetic optimism virus that is constantly drifting, mutating, and propagating in a decentralized fashion. So memetic optimism virus. So you do want it to be a virus to maximize the spread, and it’s hyperstitious, therefore the optimism will incentivize its growth. Guillaume Verdon (01:49:43) 我们将e/acc视为一种元启发式，一种非常薄的文化框架，从中你可以有更多有主见的分支。从根本上说，我们只是说，把我们带到这里的是基于热力学的整个系统的这种适应，这个过程是好的，我们应该让它继续下去。这就是核心论点。其他一切都是，好吧，我们如何确保我们保持这种可塑性和适应性。嗯，显然不要压制变异，保持言论自由、思想自由、信息传播自由和进行AI研究的自由对我们来说很重要，以便我们能够最快地收敛到导致这种增长的技术、想法等空间。所以最终，已经有相当多的分支。有些只是模因，但有些更严肃。Vitalik Buterin最近创建了一个d/acc分支。他有他自己对e/acc的某种微调。 GUILLAUME VERDON (01:49:43) We see e/acc as sort of a meta-heuristic, sort of very thin cultural framework from which you can have much more opinionated forks. Fundamentally, we just say that what got us here is this adaptation of the whole system based on thermodynamics, and that process is good and we should keep it going. That is the core thesis. Everything else is, okay, how do we ensure that we maintain this malleability and adaptability. Well, clearly not suppressing variants, and maintaining free speech, freedom of thought, freedom of information propagation, and freedom to do AI research is important for us to converge the fastest on the space of technologies, ideas, and whatnot that lead to this growth. And so ultimately, there’s been quite a few forks. Some are just memes, but some are more serious. Vitalik Buterin recently made a d/acc fork. He has his own sort of fine-tunings of e/acc. Lex Fridman (01:50:59) Vitalik的那个分支有什么独特特征让你印象深刻吗？ LEX FRIDMAN (01:50:59) Does anything jump out to memory of the unique characteristic of that fork from Vitalik? Guillaume Verdon (01:51:05) 我会说它试图在e/acc和EA（书童注：Effective Altruism，有效利他主义）以及AI安全之间找到一个中间地带。对我来说，拥有一个与接管硅谷的主流叙事相反的运动，对于改变意见的动态范围很重要。这就像中心化和去中心化之间的平衡，真正的最优点总是在中间的某个地方。但对于e/acc，我们推动熵、新颖性、颠覆、可塑性、速度，而不是保守、压制思想、压制言论、增加约束、增加过多法规、放慢速度。所以，我们试图为力量带来平衡。 GUILLAUME VERDON (01:51:05) I would say that it’s trying to find a middle ground between e/acc and EA and EI safety. To me, having a movement that is opposite to what was the mainstream narrative that was taking over Silicon Valley was important to shift the dynamic range of opinions. And it’s like the balance between centralization and decentralization, the real optimum is always somewhere in the middle. But for e/acc, we’re pushing for entropy, novelty, disruption, malleability, speed, rather than being conservative, suppressing thought, suppressing speech, adding constraints, adding too many regulations, slowing things down. And so, we’re trying to bring balance to the force. Lex Fridman (01:52:00) 为人类文明的力量带来平衡。 LEX FRIDMAN (01:52:00) Balance to the force of human civilization. Guillaume Verdon (01:52:02) 这字面上是约束的力量与让我们探索的熵之间的张力。系统在处于秩序与混沌之间的临界边缘时是最优的，在约束、能量最小化和熵之间。系统想要平衡这两样东西。我认为平衡是缺失的，所以我们创建了这个运动来带来平衡。 GUILLAUME VERDON (01:52:02) It’s literally the forces of constraints versus the entropic force that makes us explore. Systems are optimal when they’re at the edge of criticality between order and chaos, between constraints, energy minimization and entropy. Systems want to equilibrate, balance these two things. I thought that the balance was lacking, and so we created this movement to bring balance. Lex Fridman (01:52:31) 嗯，我喜欢想法通过分支进化的视觉效果。所以在历史的另一部分，把马克思主义看作原始仓库，然后苏联共产主义是其中一个分支，然后_主义是马克思主义和共产主义的一个分支。所以这些都是分支。它们在探索不同的想法。 LEX FRIDMAN (01:52:31) Well, I like the visual of the landscape of ideas evolving through forks. So on the other part of history, thinking of Marxism as the original repository, and then Soviet Communism is a fork of that, and then the M__ism is a fork of Marxism and Communism. And so those are all forks. They’re exploring different ideas. Guillaume Verdon (01:53:02) 把文化几乎看作代码。现在，你在LLM中提示什么或你在LLM的宪法中放入什么，基本上就是它的文化框架，它相信什么。你现在可以在GitHub上分享它。所以试图从在软件的这台机器中起作用的东西中获得灵感，以适应代码空间，我们能否将其应用于文化？我们的目标不是说”你应该这样生活，X、Y、Z”，而是建立一个过程，让人们总是在亚文化中搜索并竞争思想份额。我认为创造这种文化的可塑性对我们收敛到适应现代的文化和关于如何生活的启发式方法非常重要。 GUILLAUME VERDON (01:53:02) Thinking of culture almost like code. Nowadays, what you prompt in the LLM or what you put in the constitution of an LLM is basically its cultural framework, what it believes. And you can share it on GitHub nowadays. So trying to take inspiration from what has worked in this machine of software to adapt over the space of code, could we apply that to culture? And our goal is to not say, “You should live your life this way, X, Y, Z,” it’s to set up a process where people are always searching over subcultures and competing for mind share. I think creating this malleability of culture is super important for us to converge onto the cultures and the heuristics about how to live one’s life that are updated to modern times. Guillaume Verdon (01:53:59) 因为真的存在一种精神性和文化的真空。人们觉得他们不属于任何一个群体，而且有一些寄生性的意识形态已经利用机会填充这个思想的培养皿。Elon称之为思想病毒。我们称之为减速思想病毒综合体，这是所有这些之间的总体模式的减速。也有许多变种。所以如果有一种病毒式的悲观主义、减速运动，我们需要的不仅仅是一个运动，而是许多许多变种，所以很难精确定位和停止。 GUILLAUME VERDON (01:53:59) Because there’s really been a sort of vacuum of spirituality and culture. People don’t feel like they belong to any one group, and there’s been parasitic ideologies that have taken up opportunity to populate this Petri dish of minds. Elon calls it the mind virus. We call it the decel mind virus complex, which is the decelerative that is kind of the overall pattern between all of them. There’s many variants as well. And so if there’s a sort of viral pessimism, decelerative movement, we needed to have not only one movement, but many, many variants, so it’s very hard to pinpoint and stop. Lex Fridman (01:54:45) 但总体来说，它仍然是一种模因式乐观主义大流行。好吧，让我问你，你认为e/acc在某种程度上是一个邪教吗？ LEX FRIDMAN (01:54:45) But the overarching thing is nevertheless a kind of mimetic optimism pandemic. Okay, let me ask you, do you think e/acc to some degree is a cult? Guillaume Verdon (01:55:01) 定义邪教？ GUILLAUME VERDON (01:55:01) Define cult? Lex Fridman (01:55:03) 我认为很多人类进步是在你有独立思考时取得的，所以你有能够自由思考的个体。而非常强大的模因系统可能会导致群体思维。人性中有一些东西会导致大规模催眠、大规模歇斯底里。每当有一个性感的想法抓住我们的思想时，我们就开始思考相似。所以实际上很难把我们分开，拉开我们，多样化思想。所以在这种程度上，每个人都在像《动物农场》里的羊一样高呼”E/acc, e/acc”到什么程度？ LEX FRIDMAN (01:55:03) I think a lot of human progress is made when you have independent thought, so you have individuals that are able to think freely. And very powerful mimetic systems can kind of lead to group think. There’s something in human nature that leads to mass hypnosis, mass hysteria. We start to think alike whenever there’s a sexy idea that captures our minds. And so it’s actually hard to break us apart, pull us apart, diversify a thought. So to that degree, to which degree is everybody kind of chanting “E/acc, e/acc” like the sheep in Animal Farm? Guillaume Verdon (01:55:46) 嗯，首先，这很有趣。这是反叛的。有这个元讽刺的概念，处于”我们不确定他们是否认真”的边界上。而且它更有趣和更好玩得多。例如，我们谈论热力学是我们的上帝，有时我们做类似邪教的事情，但没有仪式和长袍之类的。 GUILLAUME VERDON (01:55:46) Well, first of all, it’s fun. It’s rebellious. There’s this concept of meta-irony, of being on the boundary of, “We’re not sure if they’re serious or not.” And it’s much more playful and much more fun. For example, we talk about thermodynamics being our god, and sometimes we do cult-like things, but there’s no ceremony and robes and whatnot. Lex Fridman (01:56:19) 还没有。 LEX FRIDMAN (01:56:19) Not yet. Guillaume Verdon (01:56:19) 还没有，没有。但最终，是的，我完全同意，人类似乎想要感觉他们是一个群体的一部分，所以他们自然地试图与邻居达成一致并找到共同点。(书童注：参考存在主义四大终极关怀之存在性孤独)这导致了想法空间中的某种模式崩溃。我们曾经有一个被允许的文化岛屿。这是一个典型的思想子空间，任何偏离那个思想子空间的东西都被压制，或者你被取消了。现在我们创造了一个新模式，但关键是我们不是试图有一个非常受限的思想空间。关于e/acc及其许多分支，不只有一种思考方式。关键是有许多分支，可以有许多集群和许多岛屿。 GUILLAUME VERDON (01:56:19) Not yet, no. But ultimately, yeah, I totally agree that it seems to me that humans want to feel like they’re part of a group, so they naturally try to agree with their neighbors and find common ground. And that leads to sort of mode collapse in the space of ideas. We used to have one cultural island that was allowed. It was a typical subspace of thought, and anything that was diverting from that subspace of thought was suppressed or you were canceled. Now we’ve created a new mode, but the whole point is that we’re not trying to have a very restricted space of thought. There’s not just one way to think about e/acc and its many forks. And the point is that there are many forks and there can be many clusters and many islands. Guillaume Verdon (01:57:07) 我不应该以任何方式控制它。我的意思是，根本没有正式的组织。我只是发推文和某些博客文章，如果有他们不喜欢的方面，人们可以自由地叛逃和分叉。所以这使得在想法空间中应该有去领土化，这样我们就不会最终陷入一个非常像邪教的集群。所以邪教通常，他们不允许人们叛逃或开始竞争性分叉，而我们鼓励这样做。 GUILLAUME VERDON (01:57:07) And I shouldn’t be in control of it in any way. I mean, there’s no formal org whatsoever. I just put out tweets and certain blog posts, and people are free to defect and fork if there’s an aspect they don’t like. And so that makes it so that there should be deterritorialization in the space of ideas, so that we don’t end up in one cluster that’s very cult-like. And so cults usually, they don’t allow people to defect or start competing forks, whereas we encourage it. 幽默与模因 Humor and memes Lex Fridman (01:57:51) 幽默和模因的利弊，从某种意义上说，模因中有一种智慧。那是什么，魔术剧场？那是哪本书？赫尔曼·黑塞。我想是《荒原狼》。但有一种拥抱荒诞的东西似乎能触及事物的真相，但与此同时，它也可能降低话语的质量和严谨性。 LEX FRIDMAN (01:57:51) The pros and cons of humor in meme, in some sense there’s like a wisdom to memes. What is it, the Magic Theater? What book is that from? Hermann Hesse. Steppenwolf, I think. But there’s a kind of embracing of the absurdity that seems to get to the truth of things, but at the same time, it can also decrease the quality and the rigor of the discourse. Guillaume Verdon (01:58:22) 是的。 GUILLAUME VERDON (01:58:22) Yeah. Lex Fridman (01:58:23) 你感受到那种张力吗？ LEX FRIDMAN (01:58:23) Do you feel the tension of that? Guillaume Verdon (01:58:25) 是的。所以最初，我认为让我们在雷达下成长的是因为它被伪装成某种元讽刺。我们会在幽默和模因以及所谓的shit posts的包装中偷偷放入深刻的真理，我认为这是故意伪装以对抗那些寻求地位、不想……与卡通青蛙或星际Jeff Bezos的卡通争论并认真对待自己是非常困难的，所以这让我们在早期能够相当迅速地增长。但当然，本质上人们会被引导。他们对真理的概念来自他们看到的数据，来自他们被喂养的信息，而人们被喂养的信息是由算法决定的。我们真正在做的是设计我们所说的高模因适应性信息包，以便它们能够有效传播并携带信息。 GUILLAUME VERDON (01:58:25) Yeah. So initially, I think what allowed us to grow under the radar was because it was camouflaged as sort of meta-ironic. We would sneak in deep truths within a package of humor and memes and what are called shit posts, and I think that was purposefully camouflaged against those that seek status and do not want to… It’s very hard to argue with a cartoon frog or a cartoon of an intergalactic Jeff Bezos and take yourself seriously, and so that allowed us to grow pretty rapidly in the early days. But of course, essentially people get steered. Their notion of the truth comes from the data they see, from the information they’re fed, and the information people are fed is determined by algorithms. And really what we’ve been doing is engineering what we call high memetic fitness packets of information, so that they can spread effectively and carry a message. Guillaume Verdon (01:59:47) 所以这是一种传播信息的载体。是的，我们一直在使用对今天的算法放大的信息景观最优的技术。但我认为我们正在达到可以进行严肃辩论和严肃对话的规模点。这就是为什么我们正在考虑进行一系列辩论并进行更严肃的长篇讨论。因为我认为时间线对于非常严肃、深思熟虑的讨论来说不是最优的。你会因为两极分化而得到奖励。所以即使我们开始了一个字面上试图使科技生态系统两极分化的运动，归根结底是为了让我们能够进行对话并一起找到最优解。 GUILLAUME VERDON (01:59:47) So it’s kind of a vector to spread the message. And yes, we’ve been using techniques that are optimal for today’s algorithmically-amplified information landscapes. But I think we’re reaching the point of scale where we can have serious debates and serious conversations. And that’s why we’re considering doing a bunch of debates and having more serious long-form discussions. Because I don’t think that the timeline is optimal for very serious, thoughtful discussions. You get rewarded for polarization. And so even though we started a movement that is literally trying to polarize the tech ecosystem, at the end of the day so that we can have a conversation and find an optimum together. Jeff Bezos Jeff Bezos Lex Fridman (02:00:42) 我的意思是，这就是我试图用这个播客做的事情，鉴于事物的景观，仍然进行长篇对话。但荒诞被充分拥抱的程度是存在的。事实上，这次对话本身就是多层次荒诞的。所以首先，我应该说，就在最近我与Jeff Bezos进行了对话，我很想听听你——Beff Jezos——对Jeff Bezos的看法。说到星际Jeff Bezos。你对那个你的名字所启发的特定个体有什么看法？ LEX FRIDMAN (02:00:42) I mean, that’s kind of what I try to do with this podcast given the landscape of things, to still have long-form conversations. But there is a degree to which absurdity is fully embraced. In fact, this very conversation is multi-level absurd. So first of all, I should say that just very recently I had a conversation with Jeff Bezos, and I would love to hear your, Beff Jezos, opinions of Jeff Bezos. Speaking of intergalactic Jeff Bezos. What do you think of that particular individual whom your name has inspired? Guillaume Verdon (02:01:25) 是的，我认为Jeff真的很棒。我的意思是，他建立了有史以来最史诗级的公司之一。他利用了技术-资本机器和技术-资本加速来给我们我们想要的东西。我们想要快速交付，非常方便，在家，低价格。他理解机器是如何运作的以及如何利用它，比如经营公司，不试图过早获利，把它放回去，让系统复合并不断改进。可以说，我认为亚马逊在机器人技术方面投资了最多的资本，当然随着AWS的诞生，有点使我们今天看到的科技繁荣成为可能，这支付了我的工资，我想在某种程度上也支付了我们所有朋友的工资。所以我认为我们都可以感谢Jeff，他是那里最伟大的企业家之一，无可争议地是有史以来最好的之一。 GUILLAUME VERDON (02:01:25) Yeah, I think Jeff is really great. I mean, he’s built one of the most epic companies of all time. He’s leveraged the techno-capital machine and techno-capital acceleration to give us what we wanted. We want a quick delivery, very convenient, at-home, low prices. He understood how the machine worked and how to harness it, like running the company, not trying to take profits too early, putting it back, letting the system compound and keep improving. And arguably, I think Amazon’s invested some of the most amount of capital and robotics out there, and certainly with the birth of AWS, kind of enabled the tech boom we’ve seen today that has paid the salaries of, I guess myself and all of our friends to some extent. And so I think we can all be grateful to Jeff, and he’s one of the great entrepreneurs out there. one of the best of all time, unarguably. Lex Fridman (02:02:32) 当然，蓝色起源的工作，类似于SpaceX的工作，试图让人类成为多行星物种，这似乎几乎比资本主义机器更大。或者这是不同时间尺度上的资本主义机器？ LEX FRIDMAN (02:02:32) And of course, the work at Blue Origin, similar to the work at SpaceX, is trying to make humans a multi-planetary species, which that seems almost like a bigger thing than the capitalist machine. Or it’s the capitalist machine at a different timescale perhaps? Guillaume Verdon (02:02:47) 是的，我认为公司，它们倾向于逐季度优化，也许几年后，但想要留下遗产的个人可以在几十年或几个世纪的时间尺度上思考。所以事实是，一些个人是如此优秀的资本配置者，以至于他们解锁了将资本分配给带我们走得更远或更有远见的目标的能力……Elon正在用SpaceX做这件事，把所有这些资本投入到让我们到达火星。Jeff正在努力建造蓝色起源，我认为他想建造奥尼尔圆柱体（书童注：O’Neill cylinder，一种太空栖息地设计概念）并将工业搬离地球，我认为这很出色。 GUILLAUME VERDON (02:02:47) Yeah, I think that companies, they tend to optimize quarter over quarter, maybe a few years out, but individuals that want to leave a legacy can think on a multi-decadal or multi-century timescale. And so the fact that some individuals are such good capital allocators that they unlock the ability to allocate capitals to goals that take us much further or are much further-looking… Elon’s doing this with SpaceX, putting all this capital towards getting us to Mars. Jeff is trying to build Blue Origin, and I think he wants to build O’Neill cylinders and get industry off- planet, which I think is brilliant. Guillaume Verdon (02:03:33) 我认为总的来说，我支持四位亿万富翁。我知道这有时是一个有争议的声明，但我认为从某种意义上说，这是一种权益证明投票。如果你有效地分配了资本，你就解锁了更多的资本来分配，只是因为你显然知道如何更有效地分配资本。这与政治家形成对比，政治家被选中是因为他们在电视上说得最好，而不是因为他们有最有效分配纳税人资本的经过验证的记录。所以这就是为什么我支持资本主义，而不是，比如说，把我们所有的钱都交给政府，让他们弄清楚如何分配它。 GUILLAUME VERDON (02:03:33) I think just overall, I’m four billionaires. I know this is a controversial statement sometimes, but I think that in a sense it’s kind of a proof of stake voting. If you’ve allocated capital efficiently, you unlock more capital to allocate, just because clearly you know how to allocate capital more efficiently. Which is in contrast to politicians that get elected because they speak the best on TV, not because they have a proven track record of allocating taxpayer capital most efficiently. And so that’s why I’m for capitalism over, say, giving all our money to the government and letting them figure out how to allocate it. Lex Fridman (02:04:20) 你认为为什么批评亿万富翁是一种病毒式的、流行的模因？既然你提到了亿万富翁。你认为为什么对拥有财富的人，特别是那些在公众视野中的人，比如Jeff、Elon、Mark Zuckerberg，还有谁？Bill Gates，有相当广泛的批评？ LEX FRIDMAN (02:04:20) Why do you think it’s a viral and it’s a popular meme to criticize billionaires? Since you mentioned billionaires. Why do you think there’s quite a widespread criticism of people with wealth, especially those in the public eye, like Jeff and Elon and Mark Zuckerberg, and who else? Bill Gates. Guillaume Verdon (02:04:44) 是的，我认为很多人会，而不是试图理解技术-资本机器是如何运作的并意识到他们拥有比他们认为的更多的主动权，他们宁愿有这种受害者心态。”我只是受制于这台机器。它在压迫我。成功的玩家显然一定是邪恶的，因为他们在这个我不成功的游戏中取得了成功。”但我已经设法让一些处于那种心态的人意识到技术-资本机器是如何运作的，以及你如何为了自己和他人的利益而利用它。通过创造价值，你捕获了你为世界创造的一些价值。那种正和心态转变是如此强大，实际上，这就是我们通过扩大e/acc试图做的事情，就是解锁那种更高层次的主动权。实际上，你对未来的控制远远超过你的想象。你有改变世界的主动权，走出去做吧。这是许可。 GUILLAUME VERDON (02:04:44) Yeah, I think a lot of people would, instead of trying to understand how the techno-capital machine works and realizing they have much more agency than they think, they’d rather have this sort of victim mindset. “I’m just subjected to this machine. It is oppressing me. And the successful players clearly must be evil because they’ve been successful at this game that I’m not successful at.” But I’ve managed to get some people that were in that mindset and make them realize how the techno-capital machine works and how you can harness it for your own good and for the good of others. And by creating value, you capture some of the value you create for the world. That sort of positive sum mindset shift is so potent, and really, that’s what we’re trying to do by scaling e/acc, is unlocking that higher level of agency. Actually, you’re far more in control of the future than you think. You have agency to change the world, go out and do it. Here’s permission. Lex Fridman (02:05:46) 每个个体都有主动权。座右铭”Keep building”（继续建造）经常被听到。这对你意味着什么，这与健怡可乐有什么关系？顺便说一句，非常感谢红牛。它工作得很好。我感觉很好。 LEX FRIDMAN (02:05:46) Each individual has agency. The motto, “Keep building” is often heard. What does that mean to you, and what does that have to do with Diet Coke? By the way, thank you so much for the Red Bull. It’s working pretty well. I’m feeling pretty good. Guillaume Verdon (02:06:03) 太棒了。嗯，所以建造技术和建造……它不必是技术，只是建造总的来说意味着拥有主动权，试图通过创造来改变世界，比方说一家公司，这是一个在更广泛的技术-资本机器中完成功能的自我维持的有机体。对我们来说，这是在世界上实现你想看到的变化的方式，而不是，比如说，向政治家施压或创建非营利组织。非营利组织，一旦他们用完钱，他们的功能就无法再完成了。与其说是人为地扭曲市场，不如说是颠覆或引导市场，或与市场共舞，以说服它实际上这个功能很重要，增加价值，就是这样。所以我认为这是去增长、ESG方法之外，比如说，Elon之间的方式。去增长方法是，”我们要管理我们的方式走出气候危机。”而Elon是，”我要建立一家自我维持、盈利和增长的公司，我们要创新我们的方式走出这个困境。”我们试图让人们在所有规模上做后者而不是前者。 GUILLAUME VERDON (02:06:03) Awesome. Well, so building technologies and building… It doesn’t have to be technologies, just building in general means having agency, trying to change the world by creating, let’s say a company which is a self-sustaining organism that accomplishes a function in the broader techno-capital machine. To us, that’s the way to achieve change in the world that you’d like to see, rather than, say, pressuring politicians or creating nonprofits. Nonprofits, once they run out of money, their function can longer be accomplished. You’re kind of deforming the market artificially compared to sort of subverting or coursing the market, or dancing with the market, to convince it that actually this function is important, adds value, and here it is. And so I think this is the way between the de-growth, ESG approach, versus, say, Elon. The de-growth approach is like, “We’re going to manage our way out of a climate crisis.” And Elon is like, “I’m going to build a company that is self-sustaining, profitable, and growing, and we’re going to innovate our way out of this dilemma.” And we’re trying to get people to do the latter rather than the former, at all scales. Elon Musk Elon Musk Lex Fridman (02:07:26) Elon是一个有趣的案例。你是支持者，你赞扬Elon，但他也是一个长期以来一直警告人工智能的危险、潜在危险、存在风险的人。你如何调和这两者？这对你来说是矛盾吗？ LEX FRIDMAN (02:07:26) Elon is an interesting case. You are a proponent, you celebrate Elon, but he’s also somebody who has for a long time warned about the dangers, the potential dangers, existential risks of artificial intelligence. How do you square the two? Is that a contradiction to you? Guillaume Verdon (02:07:45) 这在某种程度上是矛盾的，因为他在许多方面非常反对监管。但对于AI，他绝对是监管的支持者。我认为总的来说，他看到了，比如说，OpenAI垄断市场然后对可以嵌入到这些LLM中的文化先验拥有垄断的危险，然后，当LLM现在成为人们的真理来源时，那么你就可以塑造人们的文化。所以你可以通过控制LLM来控制人们。他看到了这一点，就像社交媒体的情况一样，如果你塑造信息传播的功能，你就可以塑造人们的意见。他寻求制造一个竞争对手。所以至少，我认为我们在那里非常一致，通向美好未来的方式是维持各个AI玩家之间的对抗性均衡。我很想和他谈谈，以了解他对如何推进AI的想法。我的意思是，我会说他也在用Neuralink对冲他的赌注。我认为如果他不能阻止AI的进步，他正在建造与之合并的技术。看行动，而不仅仅是言语。 GUILLAUME VERDON (02:07:45) It is somewhat because he’s very much against regulation in many aspects. But for AI, he’s definitely a proponent of regulations. I think overall he saw the dangers of, say, OpenAI cornering the market and then getting to have the monopoly over the cultural priors that you can embed in these LLMs that then, as LLMs now become the source of truth for people, then you can shape the culture of the people. And so you can control people by controlling LLMs. He saw that, just like it was the case for social media, if you shape the function of information propagation, you can shape people’s opinions. He sought to make a competitor. So at least, I think we’re very aligned there, that the way to a good future is to maintain adversarial equilibria between the various AI players. I’d love to talk to him to understand his thinking about how to advance AI going forwards. I mean, he’s also hedging his bets, I would say, with Neuralink. I think if he can’t stop the progress of AI, he’s building the technology to merge. Look at the actions, not just the words. Lex Fridman (02:09:10) 嗯，在某种程度上关注……也许使用人类心理学，关注我们周围的威胁是一个激励因素。这是一件鼓励的事情。当有截止日期时，我的表现要好得多。对截止日期的恐惧。我为自己创造人为的东西，比如我想在自己身上创造这种焦虑，好像如果我错过截止日期，真的会发生非常可怕的事情。我认为这里有一定程度的这种情况，因为创建与人类对齐的AI有很多潜在的好处。所以重新框架的一种不同方式是，”如果你不这样做，我们都会死。”这似乎是创建人类对齐AI目标的一个非常强大的心理表述。 LEX FRIDMAN (02:09:10) Well, there’s some degree where being concerned… Maybe using human psychology, being concerned about threats all around us is a motivator. It’s an encouraging thing. I operate much better when there’s a deadline. The fear of the deadline. And I, for myself, create artificial things, like I want to create in myself this kind of anxiety as if something really horrible will happen if I miss the deadline. I think there’s some degree of that here, because creating AI that’s aligned with humans has a lot of potential benefits. And so a different way to reframe that is, “If you don’t, we’re all going to die.” It just seems to be a very powerful psychological formulation of the goal of creating human-aligned AI. Guillaume Verdon (02:09:59) 我认为焦虑是好的。我认为，正如我所说，我希望自由市场创造对齐的、可靠的AI，我认为这就是他试图用xAI做的事情。所以我完全支持。我反对的是阻止，比如说开源生态系统通过在行政命令中声称开源LM是双重用途技术并且应该由政府控制而蓬勃发展。然后每个人都需要向政府注册他们的GPU和他们的大矩阵。我认为那种额外的摩擦会阻止很多黑客做出贡献，这些黑客后来可能成为做出推动我们前进的关键发现的研究人员，包括AI安全的发现。所以我认为我只是想保持对AI的贡献机会的普遍性以及拥有未来的一部分。它不能只是被立法到某个墙后面，只有少数玩家可以玩游戏。 GUILLAUME VERDON (02:09:59) I think that anxiety is good. I think, like I said, I want the free market to create aligned AIs that are reliable, and I think that’s what he’s trying to do with xAI. So I’m all for it. What I am against is stopping, let’s say the OpenSource ecosystem from thriving by, let’s say in the executive order, claiming that OpenSource LMs are dual-use technologies and should be government controlled. Then everybody needs to register their GPU and their big matrices with the government. And I think that extra friction will dissuade a lot of hackers from contributing, hackers that could later become the researchers that make key discoveries that push us forward, including discoveries for AI safety. And so I think I just want to maintain ubiquity of opportunity to contribute to AI and to own a piece of the future. It can’t just be legislated behind some wall where only a few players get to play the game. Lex Fridman (02:11:08) e/acc运动经常被讽刺为意味着不惜一切代价的进步和创新。不管有多不安全，不管是否造成很多损害。你只是尽可能快地建造酷的东西，整夜熬夜喝健怡可乐，无论需要什么。我想，我不知道那里是否有问题，但对你来说有多重要，你在e/acc的不同表述中看到的，AI安全有多重要？ LEX FRIDMAN (02:11:08) The e/acc movement is often caricatured to mean progress and innovation at all costs. Doesn’t matter how unsafe it is, doesn’t matter if it causes a lot of damage. You just build cool shit as fast as possible, stay up all night with a Diet Coke, whatever it takes. I guess, I don’t know if there’s a question in there, but how important to you and what you’ve seen the different formulations of e/acc, is AI safety? Guillaume Verdon (02:11:44) 再说一次，我认为如果没有人在研究它，我认为我会是它的支持者。我认为，再说一次，我们的目标是带来平衡，显然紧迫感是取得进展的有用工具。它黑进了我们的多巴胺系统，给我们能量工作到深夜。我认为还有一个你正在贡献的更高目标。归根结底，就像，我在贡献什么？我在为这台美丽机器的增长做贡献，这样我们就可以寻求星辰。这真的很鼓舞人心。这也是一种神经黑客。 GUILLAUME VERDON (02:11:44) Again, I think if there was no one working on it, I think I would be a proponent of it. I think, again, our goal is to bring balance, and obviously a sense of urgency is a useful tool to make progress. It hacks our dopaminergic systems and gives us energy to work late into the night. I think also having a higher purpose you’re contributing to. At the end of the day, it’s like, what am I contributing to? I’m contributing to the growth of this beautiful machine so that we can seek to the stars. That’s really inspiring. That’s also a sort of neuro hack. Lex Fridman (02:12:26) 所以你是说AI安全对你来说很重要，但现在你看到的想法景观是，AI安全作为一个话题更常被用来获得集中控制。所以从这个意义上说，你在抵制它，作为获得集中控制的代理？ LEX FRIDMAN (02:12:26) So you’re saying AI safety is important to you, but right now the landscape of ideas you see is, AI safety as a topic is used more often to gain centralized control. So in that sense, you’re resisting it, as a proxy for gaining centralized control? Guillaume Verdon (02:12:43) 是的，我只是认为我们必须小心，因为安全只是权力集中化和最终掩盖腐败的完美掩护。我不是说它现在已经腐败了，但它可能在未来。而且实际上，如果你让论证运行，没有多少权力集中化的控制足以确保你的安全。总是有更多的9个9的P安全可以获得，99.9999%安全。也许你想要另一个9。”哦，请让我们完全访问你做的一切。完全监视。”坦率地说，那些AI安全的支持者已经提议拥有一个全球全景监狱，你对正在发生的一切都有集中的感知。对我来说，这只是为老大哥、1984式的场景敞开了大门。那不是我想生活的未来。 GUILLAUME VERDON (02:12:43) Yeah, I just think we have to be careful, because safety is just the perfect cover for centralization of power and covering up eventually corruption. I’m not saying it’s corrupted now, but it could be down the line. And really, if you let the argument run, there’s no amount of centralization of control that will be enough to ensure your safety. There’s always more 999s of P safety that you can gain, 99.9999% safe. Maybe you want another nine. “Oh, please give us full access to everything you do. Full surveillance.” And frankly, those that are proponents of AI safety have proposed having a global panopticon where you have centralized perception of everything going on. And to me, that just opens up the door wide open for a big brother, 1984-like scenario. And that’s not a future I want to live in. Lex Fridman (02:13:49) 因为我们在整个历史上有一些例子，那没有导致好的结果。 LEX FRIDMAN (02:13:49) Because we have some examples throughout history when that did not lead to a good outcome. Extropic Extropic Guillaume Verdon (02:13:54) 对。 GUILLAUME VERDON (02:13:54) Right. Lex Fridman (02:13:56) 你提到你创立了一家公司Extropic，最近宣布了1410万美元的种子轮融资。公司的目标是什么？你谈到了很多有趣的物理东西，所以你在那里做什么可以谈论的？ LEX FRIDMAN (02:13:56) You mentioned you founded a company, Extropic, that recently announced a 14.1 million seed round. What’s the goal of the company? You’re talking about a lot of interesting physics things, so what are you up to over there that you can talk about? Guillaume Verdon (02:14:12) 是的，最初我们不打算在上周宣布，但我认为由于人肉曝光和披露，我们被迫披露了。所以我们不得不披露我们大致在做什么。但实际上，Extropic诞生于我和我的同事对量子计算路线图的不满。量子计算有点是试图商业化规模化的基于物理的计算的第一条路径，我正在研究运行在这些基于物理的计算机上的基于物理的AI。但最终，我们最大的敌人是这种噪声，这种无处不在的噪声问题，正如我所提到的，你必须不断地把噪声从系统中抽出来，以维持这种原始环境，在那里量子力学可以生效。那个约束太多了。做那个太昂贵了。 GUILLAUME VERDON (02:14:12) Yeah, originally we weren’t going to announce last week, but I think with the doxing and disclosure, we got our hand forced. So we had to disclose roughly what we were doing. But really, Extropic was born from my dissatisfaction, and that of my colleagues, with the quantum computing roadmap. Quantum computing was sort of the first path to physics-based computing that was trying to commercially scale, and I was working on physics-based AI that runs on these physics-based computers. But ultimately, our greatest enemy was this noise, this pervasive problem of noise that, as I mentioned, you have to constantly pump out the noise out of the system to maintain this pristine environment where quantum mechanics can take effect. And that constraint was just too much. It’s too costly to do that. Guillaume Verdon (02:15:11) 所以我们在想，当生成式AI有点吞噬世界时，世界上越来越多的计算工作负载集中在生成式AI上，我们如何能够使用物理学从物理学、信息论、计算以及最终热力学的第一原理来设计生成式AI的终极物理基底？所以我们寻求建造的是一个基于物理的计算系统和基于物理的AI算法，它们受到非平衡热力学的启发，或直接利用它来将机器学习作为一个物理过程来做。 GUILLAUME VERDON (02:15:11) And so we were wondering, as generative AI is sort of eating the world, more and more of the world’s computational workloads are focused on generative AI, how could we use physics to engineer the ultimate physical substrate for generative AI from first principles of physics, of information theory, of computation, and ultimately of thermodynamics? And so what we’re seeking to build is a physics-based computing system and physics-based AI algorithms that are inspired by out-of-equilibrium thermodynamics, or harness it directly to do machine learning as a physical process. Lex Fridman (02:16:01) 那么这意味着什么，机器学习作为一个物理过程？是硬件吗？是软件吗？两者都是吗？它试图以某种独特的方式做全栈吗？ LEX FRIDMAN (02:16:01) So what does that mean, machine learning as a physical process? Is that hardware? Is it software? Is it both? Is it trying to do the full stack in some kind of unique way? Guillaume Verdon (02:16:10) 是的，它是全栈的。所以我们是那些用TensorFlow Quantum将可微编程构建到量子计算生态系统中的人。TensorFlow Quantum的联合创始人之一是CTO Trevor McCourt。我们有一些最好的量子计算机架构师，那些设计了IBM和AWS系统的人。他们已经离开了量子计算，帮助我们建造我们所说的实际上是一台热力学计算机。 GUILLAUME VERDON (02:16:10) Yes, it is full stack. And so we’re folks that have built differentiable programming into the quantum computing ecosystem with TensorFlow Quantum. One of my co-founders of TensorFlow Quantum is the CTO, Trevor McCourt. We have some of the best quantum computer architects, those that have designed IBM’s and AWS’s systems. They’ve left quantum computing to help us build what we call actually a thermodynamic computer. Lex Fridman (02:16:43) 一台热力学计算机。嗯，实际上让我们在TensorFlow Quantum周围停留一下。你从TensorFlow Quantum中学到了什么教训？也许你可以谈谈创建本质上是量子计算机的软件API需要什么？ LEX FRIDMAN (02:16:43) A thermodynamic computer. Well, actually let’s linger around TensorFlow Quantum. What lessons have you learned from TensorFlow Quantum? Maybe you can speak to what it takes to create essentially, what, like a software API to a quantum computer? Guillaume Verdon (02:17:01) 对。那是一个挑战，要发明、建造，然后在真实设备上运行。 GUILLAUME VERDON (02:17:01) Right. That was a challenge to invent, to build, and then to get to run on the real devices. Lex Fridman (02:17:09) 你能实际谈谈它是什么吗？ LEX FRIDMAN (02:17:09) Can you actually speak to what it is? Guillaume Verdon (02:17:11) 是的。TensorFlow Quantum是一次尝试……嗯，我想我们成功了，将深度学习或可微经典编程与量子计算结合起来，并将量子计算转变为或在量子计算中拥有可微的程序类型。Andrej Karpathy称可微编程为Software 2.0。就像，梯度下降是比你更好的程序员。这个想法是，在量子计算的早期，你只能运行短的量子程序。那么，你应该运行哪些量子程序？嗯，就让梯度下降找到那些程序吧。所以我们建立了第一个基础设施，不仅可以运行可微的量子程序，还可以将它们作为更广泛的深度学习图的一部分，结合深度神经网络——你知道和喜爱的那些——与所谓的量子神经网络。 GUILLAUME VERDON (02:17:11) Yeah. TensorFlow Quantum was an attempt at… Well, I guess we succeeded, at combining deep learning or differentiable classical programming with quantum computing, and turn quantum computing into or have types of programs that are differentiable in quantum computing. And Andrej Karpathy calls differentiable programming, Software 2.0. It’s like, gradient descent is a better programmer than you. And the idea was that in the early days of quantum computing, you can only run short quantum programs. And so, which quantum programs should you run? Well, just let gradient descent find those programs instead. And so we built the first infrastructure to not only run differentiable quantum programs, but combine them as part of broader deep learning graphs, incorporating deep neural networks, the ones you know and love, with what are called quantum neural networks. Guillaume Verdon (02:18:21) 最终，这是一个非常跨学科的努力。我们必须发明各种微分方法，通过混合图进行反向传播。但最终，它教会了我编程物质和编程物理的方法是通过对控制参数进行微分。如果你有影响系统物理的参数，并且你可以评估某个损失函数，你就可以优化系统以完成任务，无论那个任务是什么。这是一个非常普遍的元框架，用于如何编程基于物理的计算机。 GUILLAUME VERDON (02:18:21) And ultimately, it was a very cross-disciplinary effort. We had to invent all sorts of ways to differentiate, to back propagate through the hybrid graph. But ultimately, it taught me that the way to program matter and to program physics is by differentiating through control parameters. If you have parameters that affects the physics of the system and you can evaluate some loss function, you can optimize the system to accomplish a task, whatever that task may be. And that’s a very universal meta framework for how to program physics-based computers. Lex Fridman (02:19:05) 所以试图参数化一切，使这些参数可微，然后优化？ LEX FRIDMAN (02:19:05) So try to parameterize everything, make those parameters differentiable, and then optimize? Guillaume Verdon (02:19:12) 是的。 GUILLAUME VERDON (02:19:12) Yes. Lex Fridman (02:19:13) 好的。TensorFlow Quantum有一些更实际的工程教训吗？组织上也是，比如涉及的人类以及如何到达产品，如何创建好的文档？我不知道。所有这些人们可能不会想到的小微妙的东西。 LEX FRIDMAN (02:19:13) Okay. Is there some more practical engineering lessons from TensorFlow Quantum? Just organizationally too, like the humans involved and how to get to a product, how to create good documentation? I don’t know. All of these little subtle things that people might not think about. Guillaume Verdon (02:19:34) 我认为跨学科边界工作总是一个挑战，你必须在互相教学方面非常耐心。我通过这个过程学到了很多软件工程。我的同事学到了很多量子物理学，有些人通过建造这个系统的过程学到了机器学习。我认为如果你让一些聪明、充满激情并相互信任的人在一个房间里，你有一个小团队—— GUILLAUME VERDON (02:19:34) I think working across disciplinary boundaries is always a challenge, and you have to be extremely patient in teaching one another. I learned a lot of software engineering through the process. My colleagues learned a lot of quantum physics, and some learned machine learning through the process of building this system. And I think if you get some smart people that are passionate and trust each other in a room, and you have a small team- Guillaume Verdon (02:20:00) 充满激情并相互信任，你有一个小团队，你们互相教授你们的专长，突然你有点形成了这种专业知识的模型汤，一些特别的东西从中出来，对吧？这就像结合基因，但为了你的知识库，有时特殊的产品从中出来。所以我认为，即使最初在跨学科团队中工作摩擦很大，我认为归根结底产品是值得的。所以，学到了很多试图弥合那里的差距。我的意思是，直到今天这仍然是一个挑战。我们雇用有AI背景的人，有纯物理背景的人，不知何故我们必须让他们互相交谈。对吧？ GUILLAUME VERDON (02:20:00) Are passionate and trust each other in a room, and you have a small team, and you teach each other your specialties, suddenly you’re kind of forming this sort of model soup of expertise, and something special comes out of that, right? It’s like combining genes, but for your knowledge bases, and sometimes special products come out of that. And so I think, even though it’s very high friction initially to work in an interdisciplinary team, I think the product at the end of the day is worth it. And so, learned a lot trying to bridge the gap there. And I mean, it’s still a challenge to this day. We hire folks that have an AI background, folks that have a pure physics background, and somehow we have to make them talk to one another. Right? Lex Fridman (02:20:47) 招聘过程有魔力吗，建立一个能够一起创造魔力的团队有科学和艺术吗？ LEX FRIDMAN (02:20:47) Is there a magic, is there some science and art to the hiring process, to building a team that can create magic together? Guillaume Verdon (02:20:56) 是的，真的很难准确指出那种je ne sais quoi（书童注：法语，意为”难以言喻的特质”），对吧？ GUILLAUME VERDON (02:20:56) Yeah, it’s really hard to pinpoint that je ne sais quoi, right? Lex Fridman (02:21:03) 我不知道你说法语。这很好。 LEX FRIDMAN (02:21:03) I didn’t know you speak French. That’s very nice. Guillaume Verdon (02:21:07) 是的，我实际上是法裔加拿大人。 GUILLAUME VERDON (02:21:07) Yeah, I’m actually French Canadian. Lex Fridman (02:21:09) 哦，你是真正的法裔加拿大人。 LEX FRIDMAN (02:21:09) Oh, you are a legitimately French Canadian. Guillaume Verdon (02:21:09) 我是。 GUILLAUME VERDON (02:21:09) I am. Lex Fridman (02:21:11) 我以为你只是为了信誉而这么做。 LEX FRIDMAN (02:21:11) I thought you were just doing that for the cred. Guillaume Verdon (02:21:15) 不，不。我真的是法裔加拿大人，来自蒙特利尔。但是的，基本上我们寻找具有非常高灵活度的通才、不是过度专业化的人，因为他们将不得不走出他们的舒适区。他们将不得不整合他们以前从未见过的概念，并非常迅速地整合到舒适区，或学会在团队中工作。所以这就是我们在招聘时寻找的。我们不能雇用那些在过去三四年里只是优化这个子系统的人。我们需要真正通用的某种更广泛的智力和专长，以及开放思想的人，真的，因为如果你从头开始开创一个新方法，没有教科书，没有参考资料。就是我们，和渴望学习的人。所以，我们必须互相教授,我们必须学习文献，我们必须分享知识库，合作以便一起推动知识边界。所以，习惯于只是被规定做什么的人在这个阶段，当你处于开拓阶段时，那不一定是你想雇用的人。是的。 GUILLAUME VERDON (02:21:15) No, no. I’m truly French Canadian, from Montreal. But yeah, essentially we look for people with very high fluid intelligence that aren’t overspecialized, because they’re going to have to get out of their comfort zone. They’re going to have to incorporate concepts that they’ve never seen before, and very quickly get comfortable with them, or learn to work in a team. And so that’s sort of what we look for when we hire. We can’t hire people that are just optimizing this subsystem for the past three or four years. We need really general sort of broader intelligence and specialty, and people that are open-minded, really, because if you’re pioneering a new approach from scratch, there is no textbook, there’s no reference. It’s just us, and people that are hungry to learn. So, we have to teach each other, we have to learn the literature, we have to share knowledge bases, collaborate in order to push the boundary of knowledge further together. And so, people that are used to just getting prescribed what to do at this stage, when you’re at the pioneering stage, that’s not necessarily who you want to hire. Yeah. （第三部分完）

【e/acc】有效加速主义、量子热力学与AI未来 | 物理学家、e/acc运动创始人Guillaume Verdon与Lex Fridman播客实录 | 中英文完整版精译 II

Sun, 01 Mar 2026 00:00:00 +0000

书童按：本篇是Guillaume Verdon接受Lex Fridman播客采访实录的第二部分。延续上篇对有效加速主义（e/acc）哲学根基的探讨，本篇深入人与AI共生的未来图景、末日概率（p(doom)）的合理性辨析、量子机器学习的前沿探索等议题。Verdon以物理学家的视角，从热力学第二定律出发论证生命与文明的增长本性，主张人类应拥抱AI增强而非恐惧替代，批判末日论者对未来的偏见式采样，并分享量子计算与量子深度学习的技术洞见。访谈纵横于哲学思辨与技术前沿，既有对人类中心主义的解构，亦有对资本主义市场机制的坚守，视野开阔，发人深省。初稿采用Claude API机器翻译及排版，书童仅做简单校对及批注，以飨诸君。 Guillaume Verdon：有效加速主义、热力学与量子智能 | Lex Fridman播客（第二部分） Guillaume Verdon: Effective Accelerationism, Thermodynamics, and Quantum Intelligence | Lex Fridman Podcast (Part II) 与AI共生 Merging with AI Lex Fridman (00:50:13) 那么，如果事实证明，宇宙中意识之美的载体不止人类，AI也能将同样的火焰传承下去——这让你害怕吗？你担心AI会取代人类吗？ LEX FRIDMAN (00:50:13) So if it turns out that the beauty that is consciousness in the universe is bigger than just humans, the AI can carry that same flame forward. Does it scare you, are you concerned that AI will replace humans? Guillaume Verdon (00:50:32) 在我的职业生涯中，有一个时刻让我意识到：也许我们需要把任务交给机器，才能真正理解我们周围的宇宙——而不是仅靠人类拿着纸笔把一切算出来。对我来说，这种放手一部分主动权的过程，反而给了我们理解世界的巨大杠杆。量子计算机在理解纳米尺度的物质方面，远胜过人类。类似地，我认为人类面临一个选择：我们是否接受AI将解锁的智力和操作杠杆，从而确保我们能够沿着文明规模与范围不断增长的道路前进？我们可能会被稀释——也许会有大量AI工作者——但总体而言，出于自身利益，通过与AI结合并增强自己，我们将实现更高的增长和更大的繁荣。 GUILLAUME VERDON (00:50:32) So during my career, I had a moment where I realized that maybe we need to offload to machines to truly understand the universe around us, right, instead of just having humans with pen and paper solve it all. And to me that sort of process of letting go of a bit of agency gave us way more leverage to understand the world around us. A quantum computer is much better than a human to understand matter at the Nanoscale. Similarly, I think that humanity has a choice, do we accept the opportunity to have intellectual and operational leverage that AI will unlock and thus ensure that we’re taken along this path of growth in the scope and scale of civilization? We may dilute ourselves, right? There might be a lot of workers that are AI, but overall out of our own self-interest, by combining and augmenting ourselves with AI, we’re going to achieve much higher growth and much more prosperity, right. Guillaume Verdon (00:51:49) 对我而言，我认为最可能的未来是人类用AI增强自己。我认为我们已经走在这条增强之路上了——我们有手机用于通信，随时带在身上。我们有可穿戴设备，很快就会拥有与我们共享感知的设备，比如Humane AI Pin，或者说，从技术上讲，你的特斯拉汽车就具有共享感知能力。如果你们有共享的体验、共享的上下文、彼此通信并且有某种输入输出接口，那它本质上就是你自己的延伸。对我来说，人类用AI增强自己，以及那些不锚定于任何生物基质的AI，二者将会共存。而让各方利益对齐的方式——我们其实已经有了让由人类和技术组成的超级智能体对齐的机制。公司本质上是大型的混合专家模型，我们在公司内部有任务的神经路由机制，也有经济交换的方式来对齐这些庞然大物。 GUILLAUME VERDON (00:51:49) To me, I think that the most likely future is one where humans augment themselves with AI. I think we’re already on this path to augmentation, we have phones we use for communication, we have on ourselves at all times. We have wearables, soon that have shared perception with us, right, like the Humane AI Pin or I mean, technically your Tesla car has shared perception. And so if you have shared experience, shared context, you communicate with one another and you have some sort of IO, really it’s an extension of yourself.And to me, I think that humanity augmenting itself with AI and having AI that is not anchored to anything biological, both will coexist. And the way to align the parties, we already have a sort of mechanism to align super intelligences that are made of humans and technology, right? Companies are sort of large mixture of expert models, where we have neural routing of tasks within a company and we have ways of economic exchange to align these behemoths. Guillaume Verdon (00:53:10) 对我来说，我认为资本主义就是那条路。我确实认为，无论是什么样的物质或信息配置，只要能带来最大化的增长，我们就会收敛到那里——这纯粹是物理原理使然。所以我们要么让自己与这个现实对齐，加入文明规模与范围加速扩张的进程；要么被甩在后面，试图减速，退回森林，放弃技术，回到原始状态。至少在我看来，这就是摆在面前的两条路。 GUILLAUME VERDON (00:53:10) And to me, I think capitalism is the way, and I do think that whatever configuration of matter or information leads to maximal growth, will be where we converge, just from like physical principles. And so we can either align ourselves to that reality and join the acceleration up in scope and scale of civilization or we can get left behind and try to decelerate and move back in the forest, let go of technology and return to our primitive state. And those are the two paths forward, at least to me. Lex Fridman (00:53:54) 但有个哲学问题是：人类对齐能力是否存在极限？让我以一种论证的形式提出来。有个叫Dan Hendrycks的人写道，他同意你的观点，即AI的发展可以被视为一个进化过程，但对他——对Dan来说——这并不是件好事，因为他认为自然选择会偏好AI而非人类，这可能导致人类灭绝。你怎么看？如果这真是一个进化过程，而AI系统可能不需要人类呢？ LEX FRIDMAN (00:53:54) But there’s a philosophical question whether there’s a limit to the human capacity to align. So let me bring it up as a form of argument, this guy named Dan Hendrycks and he wrote that he agrees with you that AI development could be viewed as an evolutionary process, but to him, to Dan, this is not a good thing, as he argues that natural selection favors AIs over humans and this could lead to human extinction. What do you think, if it is an evolutionary process and AI systems may have no need for humans? Guillaume Verdon (00:54:36) 我确实认为，我们实际上正在通过市场对AI的空间施加进化压力。现在我们运行那些对人类有正效用的AI，这就产生了选择压力——如果你认为当一个神经网络的API实例在GPU上运行时，它就”活着”的话。 GUILLAUME VERDON (00:54:36) I do think that we’re actually inducing an evolutionary process on the space of AIs through the market, right. Right now we run AIs that have positive utility to humans and that induces a selective pressure, if you consider a neural net being alive when there’s an API running instances of it on GPUs. Lex Fridman (00:55:01) 对。 LEX FRIDMAN (00:55:01) Yeah. Guillaume Verdon (00:55:01) 哪些API会被运行？那些对我们有高效用的。这就像我们驯化狼并把它们变成狗——狗的表达非常清晰，非常对齐。我认为我们有机会引导AI并实现高度对齐的AI。而且我认为人类加AI是一个非常强大的组合，我不确定纯粹的AI会淘汰这种组合。 GUILLAUME VERDON (00:55:01) Right. And which APIs get run? The ones that have high utility to us, right. So similar to how we domesticated wolves and turned them into dogs that are very clear in their expression, they’re very aligned, right. I think there’s going to be an opportunity to steer AI and achieve highly aligned AI. And I think that humans plus AI is a very powerful combination and it’s not clear to me that pure AI would select out that combination. Lex Fridman (00:55:40) 所以人类现在正在创造选择压力，以创造与人类对齐的AI。但考虑到AI的发展方式以及它能多快地增长和扩展，对我来说，一个担忧是意外后果——人类无法预见这个过程的所有后果。AI系统可能造成的意外后果的破坏规模非常大。 LEX FRIDMAN (00:55:40) So the humans are creating the selection pressure right now to create AIs that are aligned to humans, but given how AI develops and how quickly it can grow and scale, to me, one of the concerns is unintended consequences, like humans are not able to anticipate all the consequences of this process. The scale of damage that could be done through unintended consequences with AI systems is very large. Guillaume Verdon (00:56:10) 但上行空间的规模—— GUILLAUME VERDON (00:56:10) The scale of the upside. Lex Fridman (00:56:12) 是的。 LEX FRIDMAN (00:56:12) Yes. Guillaume Verdon (00:56:13) 对吧？ GUILLAUME VERDON (00:56:13) Right? Lex Fridman (00:56:13) 我猜这是—— LEX FRIDMAN (00:56:13) Guess it’s- Guillaume Verdon (00:56:14) 通过用AI增强我们自己，现在无法想象的上行空间。机会成本——我们正处在一个岔路口，对吧？我们要么走创造这些技术的道路，增强自己，在AI的帮助下攀登卡尔达肖夫等级（书童注：Kardashev Scale，衡量文明技术发展水平的量表，以能源利用能力为标准），成为多行星物种；要么我们完全不孕育这些技术，把所有潜在的上行空间都留在桌面上。 GUILLAUME VERDON (00:56:14) By augmenting ourselves with AI is unimaginable right now. The opportunity cost, we’re at a fork in the road, right? Whether we take the path of creating these technologies, augment ourselves and get to climb up the Kardashev Scale, become multi-planetary with the aid of AI, or we have a hard cutoff of like we don’t birth these technologies at all and then we leave all the potential upside on the table. Lex Fridman (00:56:42) 对。 LEX FRIDMAN (00:56:42) Yeah. Guillaume Verdon (00:56:42) 对我而言，出于对未来人类的责任——通过扩大文明规模，我们可以承载更多的人口——出于对这些未来人类的责任，我认为我们必须让那个更伟大、更宏大的未来成为现实。 GUILLAUME VERDON (00:56:42) Right. And to me, out of responsibility to the future humans we could carry, with higher carrying capacity by scaling up civilization. Out of responsibility to those humans, I think we have to make the greater grander future happen. Lex Fridman (00:56:58) 在硬切断和全速前进之间，有中间地带吗？谨慎有任何论据吗？ LEX FRIDMAN (00:56:58) Is there a middle ground between cutoff and all systems go? Is there some argument for caution? Guillaume Verdon (00:57:06) 我认为，正如我所说，市场会表现出谨慎。每个有机体、每家公司、每个消费者都在为自身利益行事，他们不会把资本分配给对他们有负效用的东西。 GUILLAUME VERDON (00:57:06) I think, like I said, the market will exhibit caution. Every organism, company, consumer is acting out of self-interest and they won’t assign capital to things that have negative utility to them. Lex Fridman (00:57:21) 问题在于市场并不总是有完美信息，存在操纵，存在恶意行为者搅乱系统。它并不总是一个理性和诚实的系统。 LEX FRIDMAN (00:57:21) The problem is with the market is, there’s not always perfect information, there’s manipulation, there’s bad faith actors that mess with the system. It’s not always a rational and honest system. Guillaume Verdon (00:57:41) 嗯，这正是为什么我们需要信息自由、言论自由和思想自由，以便能够收敛到对我们所有人都有正效用的技术子空间。 GUILLAUME VERDON (00:57:41) Well, that’s why we need freedom of information, freedom of speech and freedom of thought in order to be able to converge on the subspace of technologies that have positive utility for us all, right. 末日概率 p(doom) Lex Fridman (00:57:56) 那让我问你关于p(doom)的问题——末日概率。这个词说起来挺有意思，但经历起来可不有趣。在你看来，AI最终杀死全部或大部分人类的概率是多少——也就是所谓的末日概率？ LEX FRIDMAN (00:57:56) Well let me ask you about p(doom), probability of doom. That’s just fun to say, but not fun to experience. What is to you the probability that AI eventually kills all or most humans, also known as probability of doom? Guillaume Verdon (00:58:16) 我不喜欢那种计算方式。我认为人们只是随便抛出数字，这是非常草率的计算。要计算概率，比方说你把世界建模为某种马尔可夫过程，如果你有足够多的变量，或者隐马尔可夫过程——你需要对所有可能的未来空间做随机路径积分，而不仅仅是你的大脑自然倾向的那些未来。我认为p(doom)的估算者是有偏见的，因为我们的生物本性。我们进化出了对负面的、可怕的未来的偏见采样，因为那是进化的最优解。所以那些神经质程度较高的人，整天每天都在想一切都会出错的负面未来，并声称他们在做无偏采样。某种意义上，他们没有对所有可能性的空间做归一化处理，而所有可能性的空间是超指数级庞大的，很难有这样的估计。 GUILLAUME VERDON (00:58:16) I’m not a fan of that calculation, I think people just throw numbers out there and it’s a very sloppy calculation, right? To calculate a probability, let’s say you model the world as some sort of Markov process, if you have enough variables or hidden Markov process. You need to do a stochastic path integral through the space of all possible futures, not just the futures that your brain naturally steers towards, right. I think that the estimators of p(doom) are biased because of our biology, right? We’ve evolved to have bias sampling towards negative futures that are scary, because that was an evolutionary optimum, right. And so people that are of, let’s say higher neuroticism will just think of negative futures where everything goes wrong all day every day and claim that they’re doing unbiased sampling. And in a sense they’re not normalizing for the space of all possibilities and the space of all possibilities is super exponentially large and it’s very hard to have this estimate. Guillaume Verdon (00:59:40) 总的来说，我认为我们无法以那样的粒度预测未来，因为混沌。如果你有一个复杂系统，你在几个变量上有一些不确定性，如果你让时间演化，你就有了李雅普诺夫指数（Lyapunov exponent）这个概念。一点点模糊会在我们的估计中呈指数级地变成大量模糊，随着时间推移。我认为我们需要表现出一些谦逊，承认我们实际上无法预测未来。我们拥有的唯一先验是物理定律，这正是我们所主张的。物理定律说，系统会想要增长，而为增长和复制而优化的子系统在未来更有可能出现。所以我们应该力求最大化我们当前与未来的互信息，而通往那条路的方式是加速而非减速。 GUILLAUME VERDON (00:59:40) And in general, I don’t think that we can predict the future with that much granularity because of chaos, right? If you have a complex system, you have some uncertainty and a couple of variables, if you let time evolve, you have this concept of a Lyapunov exponent, right. A bit of fuzz becomes a lot of fuzz in our estimate, exponentially so, over time. And I think we need to show some humility that we can’t actually predict the future, the only prior we have is the laws of physics, and that’s what we’re arguing for. The laws of physics say the system will want to grow and subsystems that are optimized for growth and replication are more likely in the future. And so we should aim to maximize our current mutual information with the future and the path towards that is for us to accelerate rather than decelerate. Guillaume Verdon (01:00:40) 所以我没有p(doom)，因为我认为，类似于谷歌的量子霸权实验——我当时就在他们运行模拟的房间里——那是一个量子混沌系统的例子，你甚至无法用世界上最大的超级计算机估算某些结果的概率。那就是混沌的一个例子，而我认为这个系统对任何人来说都过于混沌，无法对某些未来的可能性有准确的估计。如果他们真有那么厉害，我想他们在股市交易上会非常富有。 GUILLAUME VERDON (01:00:40) So I don’t have a p(doom), because I think that similar to the quantum supremacy experiment at Google, I was in the room when they were running the simulations for that. That was an example of a quantum chaotic system where you cannot even estimate probabilities of certain outcomes with even the biggest supercomputer in the world, right. So that’s an example of chaos and I think the system is far too chaotic for anybody to have an accurate estimate of the likelihood of certain futures. If they were that good, I think they would be very rich trading on the stock market. Lex Fridman (01:01:23) 但话虽如此，人类确实有偏见，根植于我们的进化生物学，害怕一切能杀死我们的东西，但我们仍然可以想象能杀死我们的不同轨迹。我们不知道所有其他不一定会的轨迹，但我认为，结合一些基于人类历史的基本直觉来推理，仍然是有用的——比如看看地缘政治，看看人性的基本面，强大的技术如何能伤害很多人？这似乎基于此，看看核武器，你可以开始估算p(doom)，也许是在更哲学的意义上，而非数学意义上。哲学意义上是指：有这种可能性吗？人性倾向于那个方向吗？ LEX FRIDMAN (01:01:23) But nevertheless, it’s true that humans are biased, grounded in our evolutionary biology, scared of everything that can kill us, but we can still imagine different trajectories that can kill us. We don’t know all the other ones that don’t necessarily, but it’s still I think, useful combined with some basic intuition grounded in human history, to reason about like what… Like looking at geopolitics, looking at basics of human nature, how can powerful technology hurt a lot of people? It just seems grounded in that, looking at nuclear weapons, you can start to estimate p(doom) maybe in a more philosophical sense, not a mathematical one. Philosophical meaning like is there a chance? Does human nature tend towards that or not? Guillaume Verdon (01:02:25) 我认为，对我来说，最大的存在风险之一是AI的权力集中在极少数人手中，尤其是如果这是控制信息流的公司和政府的混合体。因为这可能为一种反乌托邦的未来铺平道路——只有极少数人和政府中的寡头拥有AI，他们甚至可以说服公众AI从未存在过。这就开启了威权集中控制的场景，对我来说，这是最黑暗的时间线。而现实是，我们有这些事情发生的数据驱动先验。当你给予太多权力，当你过度集中权力时，人类会做可怕的事情。 GUILLAUME VERDON (01:02:25) I think to me, one of the biggest existential risks would be the concentration of the power of AI in the hands of the very few, especially if it’s a mix between the companies that control the flow of information and the government. Because that could set things up for a sort of dystopian future where only a very few and an oligopoly in the government have AI and they could even convince the public that AI never existed. And that opens up sort of these scenarios for authoritarian centralized control, which to me is the darkest timeline. And the reality is that we have a data-driven prior of these things happening, right. When you give too much power, when you centralize power too much, humans do horrible things, right. Guillaume Verdon (01:03:23) 对我来说，在我的贝叶斯推断中，这比基于科幻的先验有更高的可能性——比如”我的先验来自《终结者》电影”。所以当我和这些AI末日论者交谈时，我只是要求他们追溯一条通过马尔可夫链事件的路径，这条路径会导致我们的末日，并实际给我每次转换的良好概率。而很多时候，那条链中会有一个非物理的或极不可能的转换。但当然，我们天生就会害怕事物，我们天生会对危险做出反应，我们天生会认为未知是危险的，因为这是生存的好启发式方法。但出于恐惧，我们有更多的损失。我们有太多要失去的，太多的上行空间会因为出于恐惧而预先阻止正面未来的发生而失去。所以我认为我们不应该屈服于恐惧。恐惧是心智的杀手，我认为它也是文明的杀手。 GUILLAUME VERDON (01:03:23) And to me, that has a much higher likelihood in my Bayesian inference than Sci-Fi based priors, right, like, “My prior came from the Terminator movie.” And so when I talked to these AI doomers, I just ask them to trace a path through this Markov chain of events that would lead to our doom and to actually give me a good probability for each transition. And very often there’s a unphysical or highly unlikely transition in that chain, right. But of course, we’re wired to fear things and we’re wired to respond to danger, and we’re wired to deem the unknown to be dangerous, because that’s a good heuristic for survival, right. But there’s much more to lose out of fear. We have so much to lose, so much upside to lose by preemptively stopping the positive futures from happening out of fear. And so I think that we shouldn’t give into fear, fear is the mind killer, I think it’s also the civilization killer. Lex Fridman (01:04:43) 我们仍然可以思考事情出错的各种方式。比如，美国的开国元勋们思考了人性，这就是为什么会有关于必要自由的讨论。他们真正深入地审议了这一点，我认为同样的事情可能也可以为AGI做。人类历史确实表明我们倾向于集中化，或者至少当我们实现集中化时，很多坏事会发生。当有独裁者时，很多黑暗、糟糕的事情会发生。问题是，AGI能成为那个独裁者吗？AGI在发展时，能否因为其权力而成为集中化者？也许是因为人类的对齐，也许是同样的倾向，同样的斯大林式集中化和集中管理资源分配的倾向？ LEX FRIDMAN (01:04:43) We can still think about the various ways things go wrong, for example, the founding fathers of the United States thought about human nature and that’s why there’s a discussion about the freedoms that are necessary. They really deeply deliberated about that and I think the same could possibly be done for AGI. It is true that human history shows that we tend towards centralization, or at least when we achieve centralization, a lot of bad stuff happens. When there’s a dictator, a lot of dark, bad things happen. The question is, can AGI become that dictator? Can AGI when develop, become the centralizer, because of its power? Maybe because of the alignment of humans, perhaps, the same tendencies, the same Stalin like tendencies to centralize and manage centrally the allocation of resources? Lex Fridman (01:05:45) 你甚至可以看到这在表面上是一个令人信服的论点：”嗯，AGI如此聪明，如此高效，如此擅长分配资源，我们为什么不把它外包给AGI呢？”然后最终，无论什么力量用权力腐蚀人类的心智，都可能对AGI做同样的事。它只会说：”好吧，人类是可有可无的，我们会摆脱他们。”就像乔纳森·斯威夫特（Jonathan Swift）几个世纪前——我想是1700年代——的《一个温和的建议》（A Modest Proposal），他讽刺性地建议，我想是在爱尔兰，穷人的孩子被作为食物喂给富人，这将是个好主意，因为它减少了穷人的数量，并给穷人带来额外收入。所以从几个方面减少了穷人的数量，因此更多的人变得富有。当然，它漏掉了一个很难放入数学方程的基本部分——人类生命的基本价值。所以，这一切都是在说，你担心AGI成为你刚才谈到的权力集中者吗？ LEX FRIDMAN (01:05:45) And you can even see that as a compelling argument on the surface level. “Well, AGI is so much smarter, so much more efficient, so much better at allocating resources, why don’t we outsource it to the AGI?” And then eventually whatever forces that corrupt the human mind with power could do the same for AGI. It’ll just say, “Well, humans are dispensable, we’ll get rid of them.” Do the Jonathan Swift, Modest Proposal from a few centuries ago, I think the 1700s, when he satirically suggested that, I think it’s in Ireland, that the children of poor people are fed as food to the rich people and that would be a good idea, because it decreases the amount of poor people and gives extra income to the poor people. So on several accounts decreases the amount of poor people, therefore more people become rich. Of course, it misses a fundamental piece here that’s hard to put into a mathematical equation of the basic value of human life. So all of that to say, are you concerned about AGI being the very centralizer of power that you just talked about? Guillaume Verdon (01:07:09) 我确实认为，现在AI有向集中化的偏见，因为计算密度和数据的集中化以及我们训练模型的方式。我认为随着时间推移，我们将耗尽可以从互联网上抓取的数据，而且我正在研究提高计算密度，以便计算可以无处不在，以分布式方式在环境中获取信息并测试假设。我认为从根本上说，集中式控制论控制——也就是拥有一个庞大的智能体，融合许多传感器，试图准确感知世界、准确预测它、预测许多许多变量并控制它、对世界施加其意志——我认为这从来就不是最优解。比方说你有一家公司，如果你有一家公司，我不知道，有10000人，他们都向CEO汇报。即使那个CEO是AI，我认为它也会努力融合所有传来的信息，然后预测整个系统，然后施行其意志。 GUILLAUME VERDON (01:07:09) I do think that right now there’s a bias over a centralization of AI, because of a compute density and centralization of data and how we’re training models. I think over time we’re going to run out of data to scrape over the internet, and I think that, well, actually I’m working on, increasing the compute density so that compute can be everywhere and acquire information and test hypotheses in the environment in a distributed fashion. I think that fundamentally, centralized cybernetic control, so having one intelligence that is massive that fuses many sensors and is trying to perceive the world accurately, predict it accurately, predict many, many variables and control it, enact its will upon the world, I think that’s just never been the optimum, right? Like let’s say you have a company, if you have a company, I don’t know, of 10,000 people, they all report to the CEO. Even if that CEO is an AI, I think it would struggle to fuse all of the information that is coming to it and then predict the whole system and then to enact its will. Guillaume Verdon (01:08:28) 在自然界、在公司以及各种系统中出现的，是一种分层控制论控制的概念。在公司里，你有个人贡献者，他们为自己的利益行事，试图完成他们的任务，他们有一个精细的——就时间和空间而言——控制回路和感知领域。比如说你在一家软件公司，他们有自己的代码库，他们在一天内迭代它。然后管理层可能会检查，它有更广的范围，比方说有五个直接汇报对象。然后它每周对每个人的更新采样一次，然后你可以沿着链条向上，你有更大的时间尺度和更大的范围。而这似乎已经成为控制系统的最佳方式。 GUILLAUME VERDON (01:08:28) What has emerged in nature and in corporations and all sorts of systems is a notion of sort of hierarchical cybernetic control, right. In a company it would be, you have like the individual contributors, they are self-interested and they’re trying to achieve their tasks and they have a fine, in terms of time and space if you will, control loop and field of perception, right. They have their code base, let’s say you’re in a software company, they have their code base, they iterate it on it intraday, right. And then the management maybe checks in, it has a wider scope, it has, let’s say five reports, right. And then it samples each person’s update once per week, and then you can go up the chain and you have larger timescale and greater scope. And that seems to have emerged as sort of the optimal way to control systems. Guillaume Verdon (01:09:25) 而这正是资本主义给我们的。你有这些层级结构，你甚至可以有母公司等等。这样容错性要强得多。在量子计算中——这是我的领域出身——我们有量子纠错中的容错概念。量子纠错是检测来自噪声的故障，预测它如何在系统中传播，然后纠正它——这是一个控制论回路。事实证明，分层的解码器，并且在每个层级都是局部的—— GUILLAUME VERDON (01:09:25) And really that’s what capitalism gives us, right? You have these hierarchies and you can even have like parent companies and so on. And so that is far more fault tolerant, in quantum computing, that’s my feel that came from, we have a concept of this fault tolerance in quantum air correction, right? Quantum air correction is detecting a fault that came from noise, predicting how it’s propagated through the system and then correcting it, right, so it’s a cybernetic loop. And it turns out that decoders that are hierarchical and in each level, the hierarchy are local- Guillaume Verdon (01:10:00) ——分层的，并且每个层级都是局部的，表现要好得多，而且容错性要强得多。原因是，如果你有一个非局部的解码器，那么你在这个控制节点上有一个故障,整个系统就会崩溃。类似地，如果你有一个每个人都向其汇报的CEO，而那个CEO去度假了，整个公司就会陷入停滞。对我来说，我认为是的，我们看到AI有集中化的趋势，但我认为随着时间推移会有修正，智能会更接近感知。我们将把AI分解成更小的子系统，彼此通信并形成一个元系统。 GUILLAUME VERDON (01:10:00) … that are hierarchical. And at each level, the hierarchy are local, perform the best by far, and are far more fault-tolerant. The reason is, if you have a non-local decoder, then you have one fault at this control node and the whole system crashes. Similarly to if you have one CEO that everybody reports to and that CEO goes on vacation, the whole company comes to a crawl. To me, I think that yes, we’re seeing a tendency towards centralization of AI, but I think there’s going to be a correction over time, where intelligence is going to go closer to the perception. And we’re going to break up AI into smaller subsystems that communicate with one another and form a meta system. Lex Fridman (01:10:56) 如果你看看今天世界上的层级结构，有国家，那些都是层级的。但相对于彼此，国家是无政府的，所以这是一种无政府状态。 LEX FRIDMAN (01:10:56) If you look at the hierarchies that are in the world today, there’s nations and those all hierarchical. But in relation to each other, nations are anarchic, so it’s an anarchy. Guillaume Verdon (01:11:06) 嗯。 GUILLAUME VERDON (01:11:06) Mm-hmm. Lex Fridman (01:11:08) 你预见这样一个世界吗，在那里没有一个总体的……你怎么称呼它？集中式控制论控制？ LEX FRIDMAN (01:11:08) Do you foresee a world like this, where there’s not a over… What’d you call it? A centralized cybernetic control? Guillaume Verdon (01:11:17) 集中式控制中心。对。 GUILLAUME VERDON (01:11:17) Centralized locus of control. Yeah. Lex Fridman (01:11:21) 你说那是次优的？ LEX FRIDMAN (01:11:21) That’s suboptimal, you’re saying? Guillaume Verdon (01:11:22) 对。 GUILLAUME VERDON (01:11:22) Yeah. Lex Fridman (01:11:23) 所以，在最顶层总会有竞争状态？ LEX FRIDMAN (01:11:23) So, it would be always a state of competition at the very top level? Guillaume Verdon (01:11:27) 对。就像在公司里，你可能有两个部门在做类似的技术并相互竞争，然后你剪掉表现不佳的那个。这是一个树的选择过程，或者一个产品被砍掉，然后整个组织被解雇。这个尝试新事物和淘汰不奏效的旧事物的过程，正是给我们适应性的东西，帮助我们收敛到最好的技术和最该做的事情。 GUILLAUME VERDON (01:11:27) Yeah. Yeah. Just like in a company, you may have two units working on similar technology and competing with one another, and you prune the one that performs not as well. That’s a selection process for a tree, or a product gets killed and then a whole org gets fired. This process of trying new things and shedding old things that didn’t work, it’s what gives us adaptability and helps us converge on the technologies and things to do that are most good. Lex Fridman (01:12:04) 我只是希望没有一种对AGI独特而对人类不独特的失败模式，因为你现在主要描述的是人类系统。 LEX FRIDMAN (01:12:04) I just hope there’s not a failure mode that’s unique to AGI versus humans, because you’re describing human systems mostly right now. Guillaume Verdon (01:12:11) 对。 GUILLAUME VERDON (01:12:11) Right. Lex Fridman (01:12:11) 我只是希望当一家公司垄断AGI时，我们会看到与人类相同的情况，也就是另一家公司会涌现出来并开始有效竞争。 LEX FRIDMAN (01:12:11) I just hope when there’s a monopoly on AGI in one company, that we’ll see the same thing we see with humans, which is, another company will spring up and start competing effectively. Guillaume Verdon (01:12:24) 到目前为止一直是这样。我们有OpenAI。我们有Anthropic。现在，我们有xAI。我们有Meta，甚至是开源的，现在我们有Mistral，它非常有竞争力。这就是资本主义的美妙之处。你不必过于信任任何一方，因为我们总是在每个层面对冲我们的赌注。总有竞争，这对我来说至少是最美好的事情，就是整个系统总是在转变，总是在适应。 GUILLAUME VERDON (01:12:24) That’s been the case so far. We have OpenAI. We have Anthropic. Now, we have xAI. We have Meta even for open source, and now we have Mistral, which is highly competitive. That’s the beauty of capitalism. You don’t have to trust any one party too much because we’re always hedging our bets at every level. There’s always competition and that’s the most beautiful thing to me, at least, is that the whole system is always shifting and always adapting. Guillaume Verdon (01:12:54) 维持这种活力就是我们避免暴政的方式。确保每个人都能访问这些工具、这些模型，并能为研究做出贡献，就能避免智能暴政——极少数人控制世界的AI并用它来压迫周围的人。 GUILLAUME VERDON (01:12:54) Maintaining that dynamism is how we avoid tyranny. Making sure that everyone has access to these tools, to these models, and can contribute to the research, avoids a neural tyranny where very few people have control over AI for the world and use it to oppress those around them. 量子机器学习 Quantum machine learning Lex Fridman (01:13:23) 当你谈论智能时,你提到了多体量子纠缠。 LEX FRIDMAN (01:13:23) When you were talking about intelligence, you mentioned multipartite quantum entanglement. Guillaume Verdon (01:13:28) 嗯。 GUILLAUME VERDON (01:13:28) Mm-hmm. Lex Fridman (01:13:29) 先问一个高层次的问题：你认为什么是智能？当你思考量子力学系统并观察其中发生的某种计算时，你认为宇宙能够进行的那种计算有什么智能之处？而人类大脑能够进行的计算只是其中的一小部分？ LEX FRIDMAN (01:13:29) High-level question first is, what do you think is intelligence? When you think about quantum mechanical systems and you observe some kind of computation happening in them, what do you think is intelligent about the kind of computation the universe is able to do; a small, small inkling of which is the kind of computation a human brain is able to do? Guillaume Verdon (01:13:52) 我会说智能和计算并不完全是一回事。我认为宇宙确实在进行量子计算。如果你能访问所有自由度和一台非常非常非常大的量子计算机，有很多很多量子比特，比方说，每个普朗克体积有几个量子比特——这差不多是我们拥有的像素——那么你就能在一台足够大的量子计算机上模拟整个宇宙，当然，假设你看的是宇宙的有限体积。我认为至少对我来说，智能是——我回到控制论——感知、预测和控制我们世界的能力。 GUILLAUME VERDON (01:13:52) I would say intelligence and computation aren’t quite the same thing. I think that the universe is very much doing a quantum computation. If you had access to all the degrees of freedom and a very, very, very large quantum computer with many, many, many qubits, let’s say, a few qubits per Planck volume, which is more or less the pixels we have, then you’d be able to simulate the whole universe on a sufficiently large quantum computer, assuming you’re looking at a finite volume, of course, of the universe. I think that at least to me, intelligence is, I go back to cybernetics, the ability to perceive, predict, and control our world. Guillaume Verdon (01:14:46) 但实际上，现在看来，我们使用的很多智能更多是关于压缩。它是关于操作化信息论。在信息论中，你有分布或系统的熵的概念，熵告诉你，如果你有最优代码，你需要这么多比特来编码这个分布或这个子系统。AI，至少我们今天为LLM和量子所做的方式，非常像试图最小化我们的世界模型与世界之间、与来自世界的分布之间的相对熵。我们在学习，我们在计算空间中搜索以处理世界，以找到那个已经提炼出所有方差、噪声和熵的压缩表示。 GUILLAUME VERDON (01:14:46) But really, nowadays, it seems like a lot of intelligence we use is more about compression. It’s about operationalizing information theory. In information theory, you have the notion of entropy of a distribution or a system, and entropy tells you that you need this many bits to encode this distribution or this subsystem, if you have the most optimal code. AI, at least the way we do it today for LLMs and for quantum, is very much trying to minimize relative entropy between our models of the world and the world, distributions from the world. We’re learning, we’re searching over the space of computations to process the world, to find that compressed representation that has distilled all the variance in noise and entropy. Guillaume Verdon (01:15:58) 最初，我从黑洞研究进入量子机器学习，因为黑洞的熵非常有趣。某种意义上，它们在物理上是宇宙中密度最高的物体。你无法在空间上比黑洞更密集地打包更多信息。所以我在想，黑洞实际上是如何编码信息的？它们的压缩代码是什么？这让我进入了算法空间，搜索量子代码空间。它也让我实际进入了，你如何从世界获取量子信息？我做过的一些工作，现在是公开的，是量子模数转换。 GUILLAUME VERDON (01:15:58) Originally, I came to quantum machine learning from the study of black holes because the entropy of black holes is very interesting. In a sense, they’re physically the most dense objects in the universe. You can’t pack more information spatially any more densely than in a black hole. And so, I was wondering, how do black holes actually encode information? What is their compression code? That got me into the space of algorithms, to search over space of quantum codes. It got me actually into also, how do you acquire quantum information from the world? Something I’ve worked on, this is public now, is quantum analog digital conversion. Guillaume Verdon (01:16:50) 你如何从真实世界以叠加态捕获信息而不破坏叠加态，而是为量子计算机数字化来自真实世界的信息？如果你有能力捕获量子信息并学习它的表示，现在你就可以学习可能在其潜在表示中有一些有用信息的压缩表示。我认为我们文明面临的许多问题实际上都超越了这个复杂性障碍。温室效应是一种量子力学效应。化学是量子力学的。核物理是量子力学的。 GUILLAUME VERDON (01:16:50) How do you capture information from the real world in superposition and not destroy the superposition, but digitize for a quantum mechanical computer information from the real world? If you have an ability to capture quantum information and learn representation representations of it, now you can learn compressed representations that may have some useful information in their latent representation. I think that many of the problems facing our civilization are actually beyond this complexity barrier. The greenhouse effect is a quantum mechanical effect. Chemistry is quantum mechanical. Nuclear physics is quantum mechanical. Guillaume Verdon (01:17:43) 很多生物学、蛋白质折叠等都受量子力学影响。所以，解锁用量子计算机和量子AI增强人类智力的能力，对我来说似乎是文明需要发展的基本能力。我花了几年时间做这个，但随着时间推移，我对开始看起来像核聚变的时间线感到厌倦。 GUILLAUME VERDON (01:17:43) A lot of biology and protein folding and so on is affected by quantum mechanics. And so, unlocking an ability to augment human intellect with quantum mechanical computers and quantum mechanical AI seemed to me like a fundamental capability for civilization that we needed to develop. I spent several years doing that, but over time, I grew weary of the timelines that were starting to look like nuclear fusion. Lex Fridman (01:18:17) 我可以问一个高层次的问题，也许通过定义的方式，通过解释的方式：什么是量子计算机，什么是量子机器学习？ LEX FRIDMAN (01:18:17) One high-level question I can ask is maybe by way of definition, by way of explanation, what is a quantum computer and what is quantum machine learning? Guillaume Verdon (01:18:27) 量子计算机实际上就是一个量子力学系统，我们对它有足够的控制，它可以保持其量子力学状态。量子力学是自然界在非常小的尺度上的行为方式，当事物非常小或非常冷时，它实际上比概率论更基础。我们习惯于事物是这个或那个，但我们不习惯用叠加态思考，因为，嗯，我们的大脑做不到。所以，我们必须把量子力学世界翻译成，比如说，线性代数来理解它。不幸的是，这种翻译平均而言是指数级低效的。你必须用非常大的矩阵来表示事物。但实际上，你可以用很多东西制造量子计算机，我们已经看到各种各样的玩家，从中性原子、囚禁离子、超导金属光子，在不同频率上。 GUILLAUME VERDON (01:18:27) A quantum computer really is a quantum mechanical system, over which we have sufficient control, and it can maintain its quantum mechanical state. And quantum mechanics is how nature behaves at the very small scales, when things are very small or very cold, and it’s actually more fundamental than probability theory. We’re used to things being this or that, but we’re not used to thinking in superpositions because, well, our brains can’t do that. So, we have to translate the quantum mechanical world to, say, linear algebra to grok it. Unfortunately, that translation is exponentially inefficient on average. You have to represent things with very large matrices. But really, you can make a quantum computer out of many things, and we’ve seen all sorts of players, from neutral atoms, trapped ions, superconducting metal photons at different frequencies. Guillaume Verdon (01:19:38) 我认为你可以用很多东西制造量子计算机。但对我来说，真正有趣的是量子机器学习既是关于用量子计算机理解量子力学世界，所以把物理世界嵌入AI表示，也是量子计算机工程是把AI算法嵌入物理世界。把物理世界嵌入AI、把AI嵌入物理世界的这种双向性，物理和AI之间的这种共生关系，实际上这就是我追求的核心，即使到今天，在量子计算之后。它仍然在这个将物理和AI真正融合的旅程中。 GUILLAUME VERDON (01:19:38) I think you could make a quantum computer out of many things. But to me, the thing that was really interesting was both quantum machine learning was about understanding the quantum mechanical world with quantum computers, so embedding the physical world into AI representations, and quantum computer engineering was embedding AI algorithms into the physical world. This bi-directionality of embedding physical world into AI, AI into the physical world, this symbiosis between physics and AI, really that’s the core of my quest really, even to this day, after quantum computing. It’s still in this journey to merge really physics and AI. Lex Fridman (01:20:29) 量子机器学习是一种在保持自然的量子力学方面真实的自然表示上进行机器学习的方式？ LEX FRIDMAN (01:20:29) Quantum machine learning is a way to do machine learning on a representation of nature that stays true to the quantum mechanical aspect of nature? Guillaume Verdon (01:20:43) 对，它是学习量子力学表示。那将是量子深度学习。或者，你可以尝试在量子计算机上做经典机器学习。我不建议这样做，因为你可能会有一些加速，但很多时候，加速伴随着巨大的成本。使用量子计算机非常昂贵。 GUILLAUME VERDON (01:20:43) Yeah, it’s learning quantum mechanical representations. That would be quantum deep learning. Alternatively, you can try to do classical machine learning on a quantum computer. I wouldn’t advise it because you may have some speed-ups, but very often, the speed-ups come with huge costs. Using a quantum computer is very expensive. Guillaume Verdon (01:21:08) 为什么？因为你假设计算机在绝对零度下运行,而宇宙中没有物理系统能达到那个温度。你必须做的是我一直提到的，这个量子纠错过程，它实际上是一个算法冰箱。它试图把熵从系统中抽出来，试图让它更接近0K。当你计算在量子计算机上做深度学习需要多少资源时，比如说，做经典深度学习，会有如此巨大的开销，不值得。这就像考虑用火箭穿越城市，进入轨道再返回来运送东西。没有意义。就用送货卡车。 GUILLAUME VERDON (01:21:08) Why is that? Because you assume the computer is operating at zero temperature, which no physical system in the universe can achieve that temperature. What you have to do is what I’ve been mentioning, this quantum error correction process, which is really an algorithmic fridge. It’s trying to pump entropy out of the system, trying to get it closer to zero temperature. When you do the calculations of how many resources it would take to, say, do deep learning on a quantum computer, classical deep learning, there’s such a huge overhead, it’s not worth it. It’s like thinking about shipping something across a city using a rocket and going to orbit and back. It doesn’t make sense. Just use a delivery truck. Lex Fridman (01:21:53) 你能用量子深度学习弄清楚、预测、理解什么样的东西,而用深度学习做不到？所以，将量子力学系统纳入学习过程？ LEX FRIDMAN (01:21:53) What kind of stuff can you figure out, can you predict, can you understand with quantum deep learning that you can’t with deep learning? So, incorporating quantum mechanical systems into the learning process? Guillaume Verdon (01:22:05) 我认为这是一个很好的问题。从根本上说，任何具有足够量子力学关联、对经典表示来说很难捕获的系统，量子力学表示应该比纯经典表示有优势。问题是，哪些系统有足够的、非常量子的关联？但这也是，哪些系统仍然与工业相关？这是一个大问题。人们倾向于化学、核物理。我实际上从事过处理来自量子传感器的输入。如果你有一个量子传感器网络，它们捕获了世界的量子力学图像，以及如何后处理，那就成为一种量子形式的机器感知。例如，费米实验室有一个项目探索用这些量子传感器探测暗物质。对我来说，这与我从小就想理解宇宙的追求是一致的。所以，有一天，我希望我们能有非常大的量子传感器网络，帮助我们窥视宇宙的最早期部分。例如，LIGO是一个量子传感器。它只是一个非常大的。所以，是的，我会说量子机器感知、模拟、理解量子模拟，类似于AlphaFold。AlphaFold理解了蛋白质配置的概率分布。你可以用量子机器学习更有效地理解电子配置的量子分布。 GUILLAUME VERDON (01:22:05) I think that’s a great question. Fundamentally, it’s any system that has sufficient quantum mechanical correlations that are very hard to capture for classical representations. Then, there should be an advantage for a quantum mechanical representation over a purely classical one. The question is, which systems have sufficient correlations that are very quantum? But it’s also, which systems are still relevant to industry? That’s a big question. People are leaning towards chemistry, nuclear physics. I’ve worked on actually processing inputs from quantum sensors. If you have a network of quantum sensors, they’ve captured a quantum mechanical image of the world and how to post-process that, that becomes a quantum form of machine perception. For example, Fermilab has a project exploring detecting dark matter with these quantum sensors. To me, that’s in alignment with my quest to understand the universe ever since I was a child. And so, someday, I hope that we can have very large networks of quantum sensors that help us peer into the earliest parts of the universe. For example, the LIGO is a quantum sensor. It’s just a very large one. So, yeah, I would say quantum machine perception, simulations, grokking quantum simulations, similar to AlphaFold. AlphaFold understood the probability distribution over configurations of proteins. You can understand quantum distributions over configurations of electrons more efficiently with quantum machine learning. Lex Fridman (01:23:53) 你合著了一篇题为《量子深度学习的通用训练算法》的论文。那涉及Baqprop，带Q。做得很好，先生。做得很好。它是如何工作的？你能提一些有趣的方面吗，Baqprop以及我们为经典机器学习所知的一些东西如何转移到量子机器学习？ LEX FRIDMAN (01:23:53) You co-authored a paper titled A Universal Training Algorithm for Quantum Deep Learning. That involves Baqprop, with a Q. Very well done, sir. Very well done. How does it work? Is there some interesting aspects you can just mention on how Baqprop and some of these things we know for classical machine learning transfer over to the quantum machine learning? Guillaume Verdon (01:24:19) 是的。那是一篇古怪的论文。那是我在量子深度学习领域的第一批论文之一。每个人都在说：”哦，我认为深度学习会被量子计算机加速。”我说：”好吧，预测未来的最好方法就是发明它。所以，这里有一篇100页的论文，祝你愉快。”本质上，量子计算通常是，你把可逆操作嵌入量子计算。 GUILLAUME VERDON (01:24:19) Yeah. That was a funky paper. That was one of my first papers in quantum deep learning. Everybody was saying, “Oh, I think deep learning is going to be sped up by quantum computers.” I was like, “ Well, the best way to predict the future is to invent it. So, here’s a 100-page paper, have fun.” Essentially, quantum computing is usually, you embed reversible operations into a quantum computation. Guillaume Verdon (01:24:47) 那里的技巧是做一个前馈操作并做我们所说的相位踢（phase kick）。但实际上，它只是一个力踢（force kick）。你只是用与你希望优化的损失函数成正比的某种力踢系统。然后，通过执行反计算，你从参数的叠加态开始，这相当古怪。现在，你不只是有参数的一个点，你有许多潜在参数的叠加态。我们的目标是—— GUILLAUME VERDON (01:24:47) The trick there was to do a feedforward operation and do what we call a phase kick. But really, it’s just a force kick. You just kick the system with a certain force that is proportional to your loss function that you wish to optimize. And then, by performing uncomputation, you start with a superposition over parameters, which is pretty funky. Now, you don’t have just a point for parameters, you have a superposition over many potential parameters. Our goal is- Lex Fridman (01:25:24) 是用相位踢以某种方式调整参数吗？ LEX FRIDMAN (01:25:24) Is using phase kick somehow to adjust the parameters? Guillaume Verdon (01:25:28) 对。因为相位踢模拟了让参数空间像n维中的粒子，你试图在神经网络的损失景观中获得薛定谔方程、薛定谔动力学。你做一个算法来诱导这个相位踢，这涉及一个前馈、一个踢。然后，当你反计算前馈时，那么所有这些相位踢和这些力的误差会反向传播并击中各层中的每一个参数。 GUILLAUME VERDON (01:25:28) Right. Because phase kicks emulate having the parameter space be like a particle in end dimensions, and you’re trying to get the Schrödinger equation, Schrödinger dynamics, in the lost landscape of the neural network. You do an algorithm to induce this phase kick, which involves a feedforward, a kick. And then, when you uncompute the feedforward, then all the errors in these phase kicks and these forces back- propagate and hit each one of the parameters throughout the layers. Guillaume Verdon (01:26:04) 如果你把这个与动能的模拟交替进行，那么它就像一个在n维中移动的粒子，一个量子粒子。原则上的优势是它可以在景观中穿隧并找到对于随机优化器来说很难找到的新最优解。但同样，这是一个理论性的东西，在实践中，至少以我们目前计划的量子计算机架构，这样的算法运行起来会极其昂贵。 GUILLAUME VERDON (01:26:04) If you alternate this with an emulation of kinetic energy, then it’s like a particle moving in end dimensions, a quantum particle. The advantage in principle would be that it can tunnel through the landscape and find new optima that would’ve been difficult for stochastic optimizers. But again, this is a theoretical thing, and in practice with at least the current architectures for quantum computers that we have planned, such algorithms would be extremely expensive to run. 量子计算机 Quantum computer Lex Fridman (01:26:41) 也许这是一个问不同领域之间区别的好地方，你曾涉足的领域。所以，数学、物理、工程，还有创业，堆栈的不同层次。我认为你在这里谈论的很多东西在数学方面有一点，也许物理几乎在理论中工作。 LEX FRIDMAN (01:26:41) Maybe this is a good place to ask the difference between the different fields that you’ve had a toe in. So, mathematics, physics, engineering, and also entrepreneurship, the different layers of the stack. I think a lot of the stuff you’re talking about here is a little bit on the math side, maybe physics almost working in theory. Guillaume Verdon (01:27:03) 嗯。 GUILLAUME VERDON (01:27:03) Mm-hmm. Lex Fridman (01:27:03) 数学、物理、工程和为量子计算、量子机器学习制造产品之间有什么区别？ LEX FRIDMAN (01:27:03) What’s the difference between math, physics, engineering, and making a product for a quantum computing for quantum machine learning? Guillaume Verdon (01:27:14) 是的。TensorFlow Quantum项目的一些原始团队成员，我们在学校开始的，在滑铁卢大学，有我自己。最初，我是一名物理学家、应用数学家。我们有一名计算机科学家，我们有一名机械工程师，然后我们有一名物理学家。那主要是实验性的。组建非常跨学科的团队并弄清楚如何沟通和分享知识，真的是做这种跨学科工程工作的关键。 GUILLAUME VERDON (01:27:14) Yeah. Some of the original team for the TensorFlow Quantum project, which we started in school, at University of Waterloo, there was myself. Initially, I was a physicist, applied mathematician. We had a computer scientist, we had a mechanical engineer, and then we had a physicist. That was experimental primarily. Putting together teams that are very cross-disciplinary and figuring out how to communicate and share knowledge is really the key to doing this interdisciplinary engineering work. Guillaume Verdon (01:27:51) 有很大的区别。在数学中，你可以为了数学而探索数学。在物理学中，你是在应用数学来理解我们周围的世界。在工程中，你试图黑掉世界。你试图找到如何应用我知道的物理学，我对世界的知识，来做事情。 GUILLAUME VERDON (01:27:51) There is a big difference. In mathematics, you can explore mathematics for mathematics’ sake. In physics, you’re applying mathematics to understand the world around us. And in engineering, you’re trying to hack the world. You’re trying to find how to apply the physics that I know, my knowledge of the world, to do things. Lex Fridman (01:28:11) 嗯，特别是在量子计算中，我认为工程上有很多限制。它似乎非常困难。 LEX FRIDMAN (01:28:11) Well, in quantum computing in particular, I think there’s just a lot of limits to engineering. It just seems to be extremely hard. Guillaume Verdon (01:28:17) 是的。 GUILLAUME VERDON (01:28:17) Yeah. Lex Fridman (01:28:18) 所以在理论上用数学探索量子计算、量子机器学习有很多价值。我想问一个问题是，为什么建造量子计算机如此困难？你对将这些想法付诸实践的时间线有什么看法？ LEX FRIDMAN (01:28:18) So, there’s a lot of value to be exploring quantum computing, quantum machine learning in theory with math. I guess one question is, why is it so hard to build a quantum computer? What’s your view of timelines in bringing these ideas to life? Guillaume Verdon (01:28:43) 对。我认为我公司的一个总体主题是，我们有一些……有一种从量子计算的大规模流出，我们正在转向不是量子的更广泛的基于物理的AI。所以，这给了你一个提示。 GUILLAUME VERDON (01:28:43) Right. I think that an overall theme of my company is that we have folks that are… There’s a sort of exodus from quantum computing and we’re going to broader physics-based AI that is not quantum. So, that gives you a hint. Lex Fridman (01:29:00) 我们应该说你的公司名字是Extropic？ LEX FRIDMAN (01:29:00) We should say the name of your company is Extropic? Guillaume Verdon (01:29:03) Extropic，没错。我们做基于物理的AI，主要基于热力学，而不是量子力学。但本质上，量子计算机非常难以建造，因为你必须诱导这个零开温度的信息子空间。做到这一点的方法是通过编码信息，你在代码中编码代码，在代码中编码代码，在代码中编码代码。需要大量冗余来做这个纠错，但最终，它是一种算法冰箱，真的。它只是把熵从虚拟的、去局域化的子系统中抽出来，该子系统代表你的”逻辑量子比特”，也就是你实际想运行量子力学程序的有效载荷量子比特。它非常困难，因为为了扩展你的量子计算机，你需要每个组件都具有足够的质量才值得。因为如果你试图做这个纠错，这个量子纠错过程，在每个量子比特和你对它们的控制中，如果它不够充分，就不值得扩展。你实际上添加的错误比你移除的更多。有一个阈值的概念，即如果你的量子比特在控制方面具有足够的质量，那么扩展实际上是值得的。实际上，近年来，人们一直在跨越阈值，它开始变得值得。 GUILLAUME VERDON (01:29:03) Extropic, that’s right. We do physics-based AI, primarily based on thermodynamics, rather than quantum mechanics. But essentially, a quantum computer is very difficult to build because you have to induce this zero temperature subspace of information. The way to do that is by encoding information, you encode a code within a code, within a code, within a code. There’s a lot of redundancy needed to do this error correction, but ultimately, it’s a sort of algorithmic refrigerator, really. It’s just pumping out entropy out of the subsystem that is virtual and delocalized that represents your “logical qubits”, aka the payload quantum bits in which you actually want to run your quantum mechanical program. It’s very difficult because in order to scale up your quantum computer, you need each component to be of sufficient quality for it to be worth it. Because if you try to do this error correction, this quantum error correction process, in each quantum bit and your control over them, if it’s insufficient, it’s not worth scaling up. You’re actually adding more errors than you remove. There’s this notion of a threshold where if your quantum bits are sufficient quality in terms of your control over them, it’s actually worth scaling up. Actually, in recent years, people have been crossing the threshold and it’s starting to be worth it. Guillaume Verdon (01:30:38) 这只是一个非常漫长的工程跋涉，但最终，对我来说真正疯狂的是我们对这些系统有多么精致的控制水平。这实际上相当疯狂。人们正在跨越……他们正在实现里程碑。只是总的来说，媒体总是走在技术前面。炒作有点太多了。这对筹款有好处，但有时它会导致寒冬。这是炒作周期。我个人对10年、15年时间尺度上的量子计算持乐观态度，但我认为在此期间可以做其他的探索。我认为它现在掌握在好手中。 GUILLAUME VERDON (01:30:38) It’s just a very long slog of engineering, but ultimately, it’s really crazy to me how much exquisite level of control we have over these systems. It’s actually quite crazy. And people are crossing… They’re achieving milestones. It’s just in general, the media always gets ahead of where the technology is. There’s a bit too much hype. It’s good for fundraising, but sometimes it causes winters. It’s the hype cycle. I’m bullish on quantum computing on a 10, 15-year timescale personally, but I think there’s other quests that can be done in the meantime. I think it’s in good hands right now. Lex Fridman (01:31:22) 嗯，让我探索一些不同的美丽想法，无论大小，在量子计算中可能从记忆中跳出来的，当你合著了一篇题为《通过Qudit探针实现渐近无限量子能量传送》的论文时。出于好奇，你能解释一下qudit与qubit相比是什么吗？ LEX FRIDMAN (01:31:22) Well, let me just explore different beautiful ideas, large or small, in quantum computing that might jump out at you from memory when you co-authored a paper titled Asymptotically Limitless Quantum Energy Teleportation via Qudit Probes. Just out of curiosity, can you explain what a qudit is versus a qubit? Guillaume Verdon (01:31:45) 是的。它是一个D态量子比特。 GUILLAUME VERDON (01:31:45) Yeah. It’s a D-state qubit. Lex Fridman (01:31:49) 它是多维的？ LEX FRIDMAN (01:31:49) It’s a multidimensional? Guillaume Verdon (01:31:50) 多维的，对。它就像，嗯，你能有一个量子力学的整数浮点概念吗？这是我必须思考的东西。我认为那项研究是后来量子模数转换工作的前兆。那很有趣，因为在我硕士期间，我试图理解真空、空无的能量和纠缠。空无具有能量，这说起来非常奇怪。我们的宇宙学方程与我们对涨落中存在多少量子能量的计算不匹配。 GUILLAUME VERDON (01:31:50) Multidimensional, right. It’s like, well, can you have a notion of an integer floating point that is quantum mechanical? That’s something I’ve had to think about. I think that research was a precursor to later work on quantum analog digital conversion. There was interesting because during my masters, I was trying to understand the energy and entanglement of the vacuum of emptiness. Emptiness has energy, which is very weird to say. Our equations of cosmology don’t match our calculations for the amount of quantum energy there is in the fluctuations. Guillaume Verdon (01:32:36) 我试图黑进真空的能量，而现实是你不能直接黑进它。它在技术上不是自由能。你对涨落的无知意味着你无法提取能量。但就像股市一样，如果你有一只随时间相关的股票，真空实际上是相关的。如果你在一个点测量了真空，你获得了信息。如果你把那个信息传达到另一个点，你可以推断真空处于什么配置，达到某种精度，并统计地平均提取一些能量。所以，你”传送了能量”。 GUILLAUME VERDON (01:32:36) I was trying to hack the energy of the vacuum, and the reality is that you can’t just directly hack it. It’s not technically free energy. Your lack of knowledge of the fluctuations means you can’t extract the energy. But just like the stock market, if you have a stock that’s correlated over time, the vacuum’s actually correlated. If you measured the vacuum at one point, you acquired information. If you communicated that information to another point, you can infer what configuration the vacuum is in to some precision and statistically extract, on average, some energy there. So, you’ve “teleported energy”. Guillaume Verdon (01:33:18) 对我来说，这很有趣，因为你可以创造负能量密度的口袋，也就是低于真空的能量密度，这非常奇怪，因为我们不理解真空如何（传播？）引力。有一些理论认为真空或时空本身的画布实际上是由量子纠缠制成的画布。我在研究如何在局部降低真空的能量会增加量子纠缠，这非常古怪。 GUILLAUME VERDON (01:33:18) To me, that was interesting because you could create pockets of negative-energy density, which is energy density that is below the vacuum, which is very weird because we don’t understand how the vacuum gravitates. There are theories where the vacuum or the canvas of space-time itself is really a canvas made out of quantum entanglement. I was studying how decreasing energy of vacuum locally increases quantum entanglement, which is very funky. Guillaume Verdon (01:33:58) 这里的事情是，如果你对UAP和诸如此类的奇怪理论感兴趣，你可以试着想象它们在周围。它们会如何推动自己？它们会如何超越光速？你需要一种负能量密度。对我来说，我尽了我的努力，试图黑进真空的能量，并达到物理定律允许的极限。但那里有各种警告，你显然不能提取比你投入的更多。 GUILLAUME VERDON (01:33:58) The thing there is that, if you’re into to weird theories about UAPs and whatnot, you could try to imagine that they’re around. And how would they propel themselves? How would they go faster than the speed of light? You would need a sort of negative energy density. To me, I gave it the old college try, trying to hack the energy of vacuum and hit the limits allowable by the laws of physics. But there’s all sorts of caveats there where you can’t extract more than you’ve put in, obviously. Lex Fridman (01:34:41) 但你是说传送能量是可能的，因为你可以在一个地方提取信息，然后基于此，对另一个地方做出某种预测？ LEX FRIDMAN (01:34:41) But you’re saying it’s possible to teleport the energy because you can extract information one place and then make, based on that, some kind of prediction about another place? Guillaume Verdon (01:34:56) 嗯。 GUILLAUME VERDON (01:34:56) Mm-hmm. Lex Fridman (01:34:57) 我不确定该如何理解这个。 LEX FRIDMAN (01:34:57) I’m not sure what to make of that. Guillaume Verdon (01:34:58) 是的，这是物理定律允许的。但现实是关联会随距离衰减。 GUILLAUME VERDON (01:34:58) Yeah, it’s allowable by the laws of physics. The reality though is that the correlations decay with distance. Lex Fridman (01:35:06) 当然。 LEX FRIDMAN (01:35:06) Sure. Guillaume Verdon (01:35:06) 所以，你将不得不在离你提取它的地方不太远的地方付出代价。 GUILLAUME VERDON (01:35:06) And so, you’re going to have to pay the price not too far away from where you extract it. （PART II 完）

【e/acc】有效加速主义、量子热力学与AI未来 | 物理学家、e/acc运动创始人Guillaume Verdon与Lex Fridman播客实录 | 中英文完整版精译 I

Sun, 15 Feb 2026 00:00:00 +0000

书童按：本篇是Guillaume Verdon接受Lex Fridman播客采访的实录。Verdon是物理学家、应用数学家与量子机器学习先驱，曾在谷歌从事量子计算研究，后创立Extropic公司，致力于为生成式AI打造基于物理原理的计算硬件。他亦是X平台匿名账号@BasedBeffJezos背后的真实人物，有效加速主义（e/acc）运动的联合创始人。e/acc以热力学与信息论为哲学根基，主张以技术快速进步作为人类伦理最优选择，正面对抗”AI末日论”代表的减速主义思潮。访谈纵横于量子计算与非平衡热力学的哲学意涵、匿名言论与思想自由、AI监管与市场力量的博弈、通用智能的重新定义等议题，视野开阔，锋芒毕现。初稿采用Claude API机器翻译及排版，书童仅做简单校对及批注，将分四部分发布，以飨诸君。 Guillaume Verdon：有效加速主义、热力学与量子智能 | Lex Fridman播客 Guillaume Verdon: Effective Accelerationism, Thermodynamics, and Quantum Intelligence | Lex Fridman Podcast 引言 Introduction Lex Fridman (00:00:00) 以下是与Guillaume Verdon的对话。他就是X平台上曾经匿名的账号@BasedBeffJezos背后的人。这两重身份因《福布斯》一篇题为《@BasedBeffJezos是谁？科技精英e/acc运动的领袖》的曝光文章被强行合二为一。让我来介绍同一个大脑里共存的这两重身份。其一：Guillaume是物理学家、应用数学家、量子机器学习研究者兼工程师，在量子机器学习方向取得博士学位，曾供职于谷歌量子计算团队，后创立Extropic公司，为生成式AI打造基于物理原理的计算硬件。 LEX FRIDMAN (00:00:00) The following is a conversation with Guillaume Verdon, the man behind the previously anonymous account @BasedBeffJezos on X. These two identities were merged by a doxxing article in Forbes titled, Who Is @BasedBeffJezos, The Leader Of The Tech Elite’s E/Acc Movement? So let me describe these two identities that coexist in the mind of one human. Identity number one, Guillaume, is a physicist, applied mathematician, and quantum machine learning researcher and engineer receiving his PhD in quantum machine learning, working at Google on quantum computing, and finally launching his own company called Extropic that seeks to build physics-based computing hardware for generative AI. Lex Fridman (00:00:47) 其二：X平台上的Beff Jezos是有效加速主义运动的创始人——常缩写为e/acc——主张将推动技术快速进步作为人类伦理上的最优选择。其拥护者深信AI进步是最强大的社会均衡器，理应全力推进。e/acc追随者自视为谨慎派的相反力量——后者认为AI高度不可预测、潜在危险、亟需监管。他们管对手叫”末日派”或”减速派”（decel）。用Beff自己的话说：”e/acc是一种模因化的乐观主义病毒。” LEX FRIDMAN (00:00:47) Identity number two, Beff Jezos on X is the creator of the effective accelerationism movement, often abbreviated as e/acc, that advocates for propelling rapid technological progress as the ethically optimal course of action for humanity. For example, its proponents believe that progress in AI is a great social equalizer, which should be pushed forward. e/acc followers see themselves as a counterweight to the cautious view that AI is highly unpredictable, potentially dangerous, and needs to be regulated. They often give their opponents the labels of quote, “doomers or decels” short for deceleration, as Beff himself put it, “e/acc is a mimetic optimism virus.” Lex Fridman (00:01:37) 这场运动的传播风格一贯偏向梗图和搞笑，但背后有扎实的思想根基，我们会在对话中深入挖掘。说到梗——本人勉强算个荒诞美学的业余爱好者。我先后和Jeff Bezos、Beff Jezos做了背靠背的访谈，这绝非巧合。对话中会聊到，Beff视Jeff为当今最重要的在世人类之一，而我则纯粹欣赏这里头的荒诞之美和幽默感。这里是Lex Fridman播客，如您愿意支持，请查看简介中的赞助商信息。闲话少叙，朋友们，有请Guillaume Verdon。 LEX FRIDMAN (00:01:37) The style of communication of this movement leans always toward the memes and the lols, but there is an intellectual foundation that we explore in this conversation. Now, speaking of the meme, I am to a kind of aspiring connoisseur of the absurd. It is not an accident that I spoke to Jeff Bezos and Beff Jezos back to back. As we talk about Beff admires Jeff as one of the most important humans alive, and I admire the beautiful absurdity and the humor of it all. This is the Lex Fridman Podcast. To support it, please check out our sponsors in the description. And now, dear friends, here’s Guillaume Verdon. Beff Jezos Beff Jezos Lex Fridman (00:02:23) 先把身份这件事捋清楚。你叫Guillaume Verdon，Gill，但你同时也是X上匿名账号@BasedBeffJezos背后的人。Guillaume Verdon这边：量子计算学者、物理学家、应用数学家；@BasedBeffJezos那边：本质上是个发起了一场运动、背后有哲学体系的梗图账号。能不能展开聊聊这两个角色——性格、沟通风格、哲学理念有什么不同？ LEX FRIDMAN (00:02:23) Let’s get the facts of identity down first. Your name is Guillaume Verdon, Gill, but you’re also behind the anonymous account on X called @BasedBeffJezos. So first, Guillaume Verdon, you’re a quantum computing guy, physicist, applied mathematician, and then @BasedBeffJezos is basically a meme account that started a movement with a philosophy behind it. So maybe just can you linger on who these people are in terms of characters, in terms of communication styles, in terms of philosophies? Guillaume Verdon (00:02:58) 说说我的主要身份吧。打小起我就想搞清楚万物之理，想理解宇宙。这条路把我领进了理论物理，最终试图回答那些终极命题——我们为何在此？我们将往何处？由此我开始研究信息论，从信息的视角理解物理，把宇宙看作一台巨大的计算机。在黑洞物理研究到一定深度后，我意识到自己不仅想理解宇宙如何计算，更想”像自然那样去计算”——造出受自然启发的计算机，也就是基于物理的计算机。这把我带进了量子计算领域：首先是模拟自然，再就是在我的工作中，学习能在量子计算机上运行的自然表示。 GUILLAUME VERDON (00:02:58) I mean, with my main identity, I guess ever since I was a kid, I wanted to figure out the theory of everything, to understand the universe. And that path led me to theoretical physics, eventually trying to answer the big questions of why are we here? Where are we going? And that led me to study information theory and try to understand physics from the lens of information theory, understand the universe as one big computation. And essentially after reaching a certain level studying black hole physics, I realized that I wanted to not only understand how the universe computes, but sort of compute like nature and figure out how to build and apply computers that are inspired by nature. So physics-based computers. And that sort of brought me to quantum computing as a field of study to first of all, simulate nature. And in my work it was to learn representations of nature that can run on such computers. Guillaume Verdon (00:04:17) 如果让AI用自然的方式思考，它们就能更精准地表征自然。至少这是驱使我成为量子机器学习领域早期探索者的核心命题——怎样在量子计算机上做机器学习，怎样把智能的概念延伸到量子领域。怎样捕获和理解现实世界的量子力学数据？怎样学习世界的量子力学表示？用什么样的计算机来运行和训练？怎样实现？这些就是我要回答的问题。而说到底，我经历了一次信仰危机。最初，跟每个物理学家一样，入行时都想用几个方程写尽宇宙，当那个故事里的英雄。 GUILLAUME VERDON (00:04:17) So if you have AI representations that think like nature, then they’ll be able to more accurately represent it. At least that was the thesis that brought me to be an early player in the field called quantum machine learning. So how to do machine learning on quantum computers and really sort of extend notions of intelligence to the quantum realm. So how do you capture and understand quantum mechanical data from our world? And how do you learn quantum mechanical representations of our world? On what kind of computer do you run these representations and train them? How do you do so? And so that’s really the questions I was looking to answer because ultimately I had a sort of crisis of faith. Originally, I wanted to figure out as every physicist does at the beginning of their career, a few equations that describe the whole universe and sort of be the hero of the story there. Guillaume Verdon (00:05:28) 但后来我想通了：用机器增强我们自身，增强我们感知、预测和掌控世界的能力，这才是正路。于是我离开理论物理，转入量子计算和量子机器学习。在那些年里，我始终觉得拼图还差一块。我们理解世界、计算世界、思考世界的方式，都少了点什么。看物理尺度的话：极小尺度上，量子力学说了算；极大尺度上，一切是确定性的，统计涨落已被抹平。我确确实实坐在这张椅子上，不是叠加在东西南北飘忽不定。极小尺度上倒是有叠加态、有干涉效应。但在介观尺度——日常生活的尺度，蛋白质、生物体、气体、液体所在的尺度——物质其实是热力学性质的，在涨落。 GUILLAUME VERDON (00:05:28) But eventually I realized that actually augmenting ourselves with machines, augmenting our ability to perceive, predict, and control our world with machines is the path forward. And that’s what got me to leave theoretical physics and go into quantum computing and quantum machine learning. And during those years I thought that there was still a piece missing. There was a piece of our understanding of the world and our way to compute and our way to think about the world. And if you look at the physical scales, at the very small scales, things are quantum mechanical, and at the very large scales, things are deterministic. Things have averaged out. I’m definitely here in this seat. I’m not in a super position over here and there. At the very small scales, things aren’t super position. They can exhibit interference effects. But at the meso scales, the scales that matter for day-to-day life and the scales of proteins, of biology, of gases, liquids and so on, things are actually thermodynamical, they’re fluctuating. Guillaume Verdon (00:06:46) 在量子计算和量子机器学习领域干了大约八年后，我突然开窍了——我一直在极大和极小之间找答案。做过一点量子宇宙学——研究宇宙从哪来、往哪去；研究黑洞物理、量子引力的极端情形，也就是能量密度高到量子力学和引力同时登场的地方。典型场景就是黑洞和极早期宇宙——量子力学与相对论的交界地带。 GUILLAUME VERDON (00:06:46) And after I guess about eight years and quantum computing and quantum machine learning, I had a realization that I was looking for answers about our universe by studying the very big and the very small. I did a bit of quantum cosmology. So that’s studying the cosmos, where it’s going, where it came from. You study black hole physics, you study the extremes in quantum gravity, you study where the energy density is sufficient for both quantum mechanics and gravity to be relevant. And the sort of extreme scenarios are black holes and the very early universe. So there’s the sort of scenarios that you study the interface between quantum mechanics and relativity. Guillaume Verdon (00:07:42) 可我一直盯着两端的极端，却漏掉了”中间那块肉”。日常尺度上量子力学有用、宇宙学有用，但其实没那么直接相关。我们活在中等时空尺度上，这个尺度上最管用的物理理论是热力学——尤其是非平衡热力学。生命本身就是热力学过程，而且是远离平衡态的。我们不是与环境达成热平衡的一锅粒子汤，而是一种拼命维持自身的相干态，靠获取和消耗自由能来续命。差不多在我离开Alphabet前夕，我对宇宙的信念再次发生了转变。我知道自己要造一种基于这类物理的全新计算范式。 GUILLAUME VERDON (00:07:42) And really I was studying these extremes to understand how the universe works and where is it going. But I was missing a lot of the meat in the middle, if you will, because day-to-day quantum mechanics is relevant and the cosmos is relevant, but not that relevant actually. We’re on sort of the medium space and timescales. And there the main theory of physics that is most relevant is thermodynamics, out of equilibrium thermodynamics. Because life is a process that is thermodynamical and it’s out of equilibrium. We’re not just a soup of particles at equilibrium with nature, were a sort of coherent state trying to maintain itself by acquiring free energy and consuming it. And that sort of, I guess another shift in, I guess my faith in the universe happened towards the end of my time at Alphabet. And I knew I wanted to build, well, first of all a computing paradigm based on this type of physics. Guillaume Verdon (00:08:57) 但与此同时，在把这些想法实验性地应用于社会、经济等方面的过程中，我开了个匿名号——纯粹是为了卸下”说什么都得负责”那种实名账号的压力。一开始只是想拿匿名号来试探想法，没想到直到真正放手，我才发现自己过去把思想空间压缩得有多厉害。某种意义上，限制言论会反向传播为限制思想。开了匿名号之后，感觉脑子里有些变量突然被解锁了，我一下子能在大得多的思想参数空间里探索。 GUILLAUME VERDON (00:08:57) But ultimately just by trying to experiment with these ideas applied to society and economies and much of what we see around us, I started an anonymous account just to relieve the pressure that comes from having an account that you’re accountable for everything you say on. And I started an anonymous account just to experiment with ideas originally because I didn’t realize how much I was restricting my space of thoughts until I sort of had the opportunity to let go. In a sense, restricting your speech back propagates to restricting your thoughts. And by creating an anonymous account, it seemed like I had unclamped some variables in my brain and suddenly could explore a much wider parameter space of thoughts. Lex Fridman (00:10:00) 在这点上展开一下——这不是很有意思吗？大家很少谈的一件事是：言论一旦受到压力和约束，思想也不知不觉被约束了，尽管逻辑上完全不必如此。我们明明可以在脑子里想任何事，但这种外部压力硬是会在思想四周筑起围墙。 LEX FRIDMAN (00:10:00) Just a little on that, isn’t that interesting that one of the things that people don’t often talk about is that when there’s pressure and constraints on speech, it somehow leads to constraints on thought even though it doesn’t have to. We can think thoughts inside our head, but somehow it creates these walls around thought. Guillaume Verdon (00:10:23) 没错。这正是我们运动的出发点——我们看到一种趋势：在生活的方方面面压制多样性，无论是思想、经营方式、组织方式还是AI研究路径。我们坚信，保持多样性才能确保系统的适应力。在思想、公司、产品、文化、政府、货币的市场中维持健康竞争，才是正途——因为系统总会自我调适，把资源配置给最有利于增长的那些形态。运动的根本理念，是这样一种洞察：生命是宇宙中一团追逐自由能、渴望生长的火焰，增长是生命的本性。平衡热力学的方程里写得明明白白：那些更擅长获取自由能、散逸更多热量的物质路径，出现的概率呈指数级增高。宇宙本身偏爱某些未来，整个系统自有其天然的走向。 GUILLAUME VERDON (00:10:23) Yep. That’s sort of the basis of our movement is we were seeing a tendency towards constraint, reduction or suppression of variants in every aspect of life, whether it’s thought, how to run a company, how to organize humans, how to do AI research. In general, we believe that maintaining variance ensures that the system is adaptive. Maintaining healthy competition in marketplaces of ideas, of companies, of products, of cultures, of governments, of currencies is the way forward because the system always adapts to assign resources to the configurations that lead to its growth. And the fundamental basis for the movement is this sort of realization that life is a sort of fire that seeks out free energy in the universe and seeks to grow. And that growth is fundamental to life. And you see this in the equations actually of equilibrium thermodynamics. You see that paths of trajectories, of configurations of matter that are better at acquiring free energy and dissipating more heat are exponentially more likely. So the universe is biased towards certain futures, and so there’s a natural direction where the whole system wants to go. 热力学 Thermodynamics Lex Fridman (00:12:21) 热力学第二定律说，宇宙的熵永远在增加，趋向平衡。而你说的是，其中存在一些复杂的、远离平衡的”口袋”。你还说热力学有利于复杂生命的涌现——这类生命通过消耗能量、向外卸载熵来提升自身能力。于是就有了这些逆熵的”口袋”。凭什么你直觉上认为这种口袋的涌现是自然的？ LEX FRIDMAN (00:12:21) So the second law of thermodynamics says that the entropy is always increasing in the universe that’s tending towards an equilibrium. And you’re saying there’s these pockets that have complexity and are out of equilibrium. You said that thermodynamics favors the creation of complex life that increases its capability to use energy to offload entropy. To offload entropy. So you have pockets of non-entropy that tend the opposite direction. Why is that intuitive to you that it’s natural for such pockets to emerge? Guillaume Verdon (00:12:53) 因为我们产热的效率远超一块同等质量的石头。我们获取自由能、摄入食物、消耗大量电力来维持运转。宇宙想产生更多熵，而让生命继续运转和壮大，恰恰是产熵的最优路径——生命会主动搜寻自由能的”口袋”并将其燃烧殆尽，以维系自身并进一步扩张。这就是生命的底层逻辑。MIT的Jeremy England有一套理论——我深以为然——认为生命的涌现正是源于这种属性。在我看来，这套物理就是支配介观尺度的法则，是量子与宇宙之间缺失的那块拼图，是中间层。热力学主宰着介观尺度。 GUILLAUME VERDON (00:12:53) Well, we’re far more efficient at producing heat than let’s say just a rock with a similar mass as ourselves. We acquire free energy, we acquire food, and we’re using all this electricity for our operation. And so the universe wants to produce more entropy and by having life go on and grow, it’s actually more optimal at producing entropy because it will seek out pockets of free energy and burn it for its sustenance and further growth. And that’s sort of the basis of life. And I mean, there’s Jeremy England at MIT who has this theory that I’m a proponent of, that life emerged because of this sort of property. And to me, this physics is what governs the meso scales. And so it’s the missing piece between the quantum and the cosmos. It’s the middle part. Thermodynamics rules the meso scales. Guillaume Verdon (00:14:08) 对我来说，无论是从工程角度——设计利用这种物理特性的器件，还是从认知角度——透过热力学棱镜理解世界，过去一年半里两重身份已形成了协同。这也正是两重身份各自浮现的深层原因。一面是，我是受到认可的科学家，正走向创业，要做新型物理AI的先驱；另一面是，我在以物理学家的视角实验性地探索哲学。 GUILLAUME VERDON (00:14:08) And to me, both from a point of view of designing or engineering devices that harness that physics and trying to understand the world through the lens of thermodynamics has been sort of a synergy between my two identities over the past year and a half now. And so that’s really how the two identities emerged. One was kind of, I’m a decently respected scientist, and I was going towards doing a startup in the space and trying to be a pioneer of a new kind of physics-based AI. And as a dual to that, I was sort of experimenting with philosophical thoughts from a physicist standpoint. Guillaume Verdon (00:14:58) 大约在那段时间——2021年底、2022年初——社会上对未来弥漫着悲观情绪，对技术尤甚。这种悲观在算法加持下病毒式扩散，人们普遍觉得未来不如现在。在我看来，这种”末日心态”是宇宙中一种极具破坏力的力量，因为它具有超迷信性（hyperstitious，书童注：hyperstition，指信念本身能提高其所预言之事发生概率的现象，自我实现的预言）——你越信它，它越可能成真。我因此觉得有责任让人们认清文明的发展轨迹和系统趋向增长的天然本性。物理定律实际上在说：统计上看，未来会更好、更宏大，而我们有能力让它成真。 GUILLAUME VERDON (00:14:58) And ultimately I think that around that time, it was like late 2021, early 2022, I think there was just a lot of pessimism about the future in general and pessimism about tech. And that pessimism was sort of virally spreading because it was getting algorithmically amplified and people just felt like the future is going to be worse than the present. And to me, that is a very fundamentally destructive force in the universe is this sort of doom mindset because it is hyperstitious, which means that if you believe it, you’re increasing the likelihood of it happening. And so felt a responsibility to some extent to make people aware of the trajectory of civilization and the natural tendency of the system to adapt towards its growth. And that actually the laws of physics say that the future is going to be better and grander statistically, and we can make it so. Guillaume Verdon (00:16:14) 反过来也一样：你若相信未来更好，并且相信自己有能力促成它，你就在实实在在地提高那个更好的未来出现的概率。所以我觉得有责任去打造一场关于未来的病毒式乐观主义运动，建一个互相支持的社区，一起造东西、干难事——做那些文明扩张必须做的事。因为在我看来，停滞和减速根本就不是选项。生命、整个系统、我们的文明，本质上就渴望增长。增长期的合作远多于衰退期——后者只会让人争着分一块越来越小的饼。就这样，我一直在两重身份之间走平衡木，直到最近两者在我不知情的情况下被强行合并了。 GUILLAUME VERDON (00:16:14) And if you believe in it, if you believe that the future would be better and you believe you have agency to make it happen, you’re actually increasing the likelihood of that better future happening. And so I sort of felt a responsibility to sort of engineer a movement of viral optimism about the future, and build a community of people supporting each other to build and do hard things, do the things that need to be done for us to scale up civilization. Because at least to me, I don’t think stagnation or slowing down is actually an option. Fundamentally life and the whole system, our whole civilization wants to grow. And there’s just far more cooperation when the system is growing rather than when it’s declining and you have to decide how to split the pie. And so I’ve balanced both identities so far, but I guess recently the two have been merged more or less without my consent. Lex Fridman (00:17:27) 你讲了好多精彩的东西。首先是”自然的表示”——这是最初吸引你从量子计算角度切入的：如何理解自然？如何表示自然，才能理解它、模拟它、用它做些什么？本质上是一个表示问题。然后你从量子力学表示跃迁到你所说的介观尺度表示，热力学在这里登场——这是另一种表示自然的方式，为了理解什么？理解生命、人类行为，理解地球上这些我们觉得有意思的一切。 LEX FRIDMAN (00:17:27) You said a lot of really interesting things there. So first, representations of nature, that’s something that first drew you in to try to understand from a quantum computing perspective, how do you understand nature? How do you represent nature in order to understand it, in order to simulate it, in order to do something with it? So it’s a question of representations, and then there’s that leap you take from the quantum mechanical representation to the what you’re calling meso scale representation, where the thermodynamics comes into play, which is a way to represent nature in order to understand what? Life, human behavior, all this kind of stuff that’s happening here on earth that seems interesting to us. 人肉曝光 Doxxing Lex Fridman (00:18:11) 然后是”hyperstition”这个词——有些观念，不管是悲观还是乐观，有这么个特质：你一旦内化它，就在某种程度上把它变成了现实。悲观和乐观都有这种属性。我猜很多观念都有，这恰恰是人类最有趣的地方之一。你还提到一个有趣的区分：Guillaume/Gill这个”前台”和@BasedBeffJezos这个”后台”，沟通风格截然不同——你在探索21世纪更有病毒传播力的表达方式。你提到的这场运动不只是个梗号，它有名字，叫有效加速主义（e/acc）——戏仿有效利他主义（EA），也是对它的反抗。我很想和你聊这种张力。然后就是那场强制合并——你说的，最近两个人格被未经你同意地合体了。有记者查出你俩其实是同一个人。说说那段经历？合并是怎么发生的？ LEX FRIDMAN (00:18:11) Then there’s the word hyperstition. So some ideas as suppose both pessimism and optimism of such ideas that if you internalize them, you in part make that idea reality. So both optimism, pessimism have that property. I would say that probably a lot of ideas have that property, which is one of the interesting things about humans. And you talked about one interesting difference also between the sort of the Guillaume, the Gill front end and the @BasedBeffJezos backend is the communication styles also that you are exploring different ways of communicating that can be more viral in the way that we communicate in the 21st century. Also, the movement that you mentioned that you started, it’s not just a meme account, but there’s also a name to it called effective accelerationism, e/acc, a play, a resistance to the effective altruism movement. Also, an interesting one that I’d love to talk to you about, the tensions there. And so then there was a merger, a get merge on the personalities recently without your consent, like you said. Some journalists figured out that you’re one and the same. Maybe you could talk about that experience. First of all, what’s the story of the merger of the two? Guillaume Verdon (00:19:47) 是这样，我和e/acc的联合创始人——一个叫@bayeslord的匿名账号，至今仍匿名，但愿永远如此——一起写了宣言。 GUILLAUME VERDON (00:19:47) So I wrote the manifesto with my co-founder of e/acc, an account named @bayeslord, still anonymous, luckily and hopefully forever. Lex Fridman (00:19:58) 也就是@BasedBeffJezos和@bayeslord——bayes就是贝叶斯，@bayeslord，贝叶斯之主。好。那以后你说e/acc，就是E斜杠A-C-C，全称effective accelerationism，有效加速主义。 LEX FRIDMAN (00:19:58) So it was @BasedBeffJezos and bayes like bayesian, like @bayeslord, like bayesian lord, @bayeslord. Okay. And so we should say from now on, when you say e/acc, you mean E slash A-C-C, which stands for effective accelerationism. Guillaume Verdon (00:20:17) 没错。 GUILLAUME VERDON (00:20:17) That’s right. Lex Fridman (00:20:18) 你说的宣言，是发在Substack上的？ LEX FRIDMAN (00:20:18) And you’re referring to a manifesto written on, I guess Substack. Guillaume Verdon (00:20:23) 对。 GUILLAUME VERDON (00:20:23) Yeah. Lex Fridman (00:20:23) 你也是@bayeslord吗？ LEX FRIDMAN (00:20:23) Are you also @bayeslord? Guillaume Verdon (00:20:25) 不是。 GUILLAUME VERDON (00:20:25) No. Lex Fridman (00:20:25) 那是另一个人？ LEX FRIDMAN (00:20:25) Okay. It’s a different person? Guillaume Verdon (00:20:26) 是。 GUILLAUME VERDON (00:20:26) Yeah. Lex Fridman (00:20:27) 好吧。万一@bayeslord就是我呢，那可有意思了。 LEX FRIDMAN (00:20:27) Okay. All right. Well, there you go. Wouldn’t it be funny if I’m @bayeslord? Guillaume Verdon (00:20:31) 那绝了。宣言差不多和我创立公司同期写成。当时我在Google X——现在叫X了，或者Alphabet X，毕竟又冒出来了另一个X。那里的底线就是保密——你不能跟谷歌内部的同事聊自己在做什么，更别说外界。这种习惯在我做事方式里根深蒂固，尤其是在有地缘政治影响的深科技领域。所以我对自己研究的内容一直守口如瓶，公司和我的公开身份之间毫无关联。但记者不仅把二者关联起来了，还进一步把我的真实身份和那个匿名号关联了起来。 GUILLAUME VERDON (00:20:31) That’d be amazing. So originally wrote the manifesto around the same time as I founded this company and I worked at Google X or just X now or Alphabet X, now that there’s another X. And there the baseline is sort of secrecy. You can’t talk about what you work on even with other Googlers or externally. And so that was kind of deeply ingrained in my way to do things, especially in deep tech that has geopolitical impact. And so I was being secretive about what I was working on. There was no correlation between my company and my main identity publicly. And then not only did they correlate that, they also correlated my main identity and this account. Guillaume Verdon (00:21:33) 他们把整个”Guillaume综合体”都给扒了——更吓人的是，记者直接联系了我的投资人。作为初创公司创始人，除了投资人你基本没有老板。投资人跟我说：”消息要出来了，他们什么都搞清楚了，你怎么打算？”好像最初周四有个记者，那时他们还没把碎片拼完整，但随后他们把整个编辑部的笔记拿来做了”传感器融合”，这下信息量就大到藏不住了。他们说这涉及”公众利益”——听到这几个关键词，我警铃大作，因为我刚好到了5万粉。据说5万粉就是”公众利益”了。那到底线在哪儿？什么时候人肉曝光一个人是合法的？ GUILLAUME VERDON (00:21:33) So I think the fact that they had doxxed the whole Guillaume complex, and they were, the journalists reached out to actually my investors, which is pretty scary. When you’re a startup entrepreneur, you don’t really have bosses except for your investors. And my investors pinged me like, “Hey, this is going to come out. They’ve figured out everything. What are you going to do?” So I think at first they had a first reporter on the Thursday and they didn’t have all the pieces together, but then they looked at their notes across the organization and they sensor fused their notes and now they had way too much. And that’s when I got worried, because they said it was of public interest and in general- Lex Fridman (00:22:24) 我喜欢你说的”传感器融合”，像个巨型神经网络做分布式运算。另外补充一点，记者用的——归根到底是——音频声纹分析：拿你过去演讲的声音和你在X Spaces上的声音做比对。 LEX FRIDMAN (00:22:24) I like how you said, sensor fused, like it’s some giant neural network operating in a distributed way. We should also say that the journalists used, I guess at the end of the day, audio-based analysis of voice, comparing voice of what, talks you’ve given in the past and then voice on X spaces? Guillaume Verdon (00:22:47) 对。 GUILLAUME VERDON (00:22:47) Yep. Lex Fridman (00:22:48) 好，这是主要的匹配手段。继续。 LEX FRIDMAN (00:22:48) Okay. And that’s where primarily the match happened. Okay, continue. Guillaume Verdon (00:22:53) 对，声纹匹配。但他们还扒了SEC的申报文件、翻了我的私人Facebook等等，下了不少功夫。最初我以为人肉曝光是违法的，但有个奇怪的临界点——一旦涉及”公众利益”，情况就变了。他们说出这几个字的时候我脑子里警报大响，因为我刚过5万粉。据说这就算”公众利益”了。那线画在哪？人肉曝光什么时候是合法的？ GUILLAUME VERDON (00:22:53) The match. But they scraped SEC filings. They looked at my private Facebook account and so on, so they did some digging. Originally I thought that doxxing was illegal, but there’s this weird threshold when it becomes of public interest to know someone’s identity. And those were the keywords that sort of ring the alarm bells for me when they said, because I had just reached 50K followers. Allegedly, that’s of public interest. And so where do we draw the line? When is it legal to dox someone? Lex Fridman (00:23:36) “dox”这个词，你帮我科普一下。我以为它一般是指某人的住址被曝光。所以你这里说的是更宽泛的意思：揭露你不愿被揭露的私人信息。 LEX FRIDMAN (00:23:36) The word dox, maybe you can educate me. I thought doxxing generally refers to if somebody’s physical location is found out, meaning where they live. So we’re referring to the more general concept of revealing private information that you don’t want revealed is what you mean by doxxing. Guillaume Verdon (00:24:00) 基于前面聊过的那些理由，匿名账号是制约权力的利器。说到底我们是在以言论对抗权力（speaking truth to power）。很多AI公司高管非常在意我们社区对他们一举一动的看法。现在我的身份暴露了，他们就知道该往哪施压来让我闭嘴，甚至让整个社区噤声。这非常遗憾——言论自由太重要了，言论自由催生思想自由，思想自由催生社交媒体上的信息自由流通。幸亏Elon买下了Twitter（现在的X），我们才有了这种自由。我们想揭露的是：AI领域的某些在位巨头正在暗中操作，表面一套背后一套。我们在指出某些政策提案实质上是”监管俘获”的工具，而”末日论”心态恰恰可能在为这些目的服务。 GUILLAUME VERDON (00:24:00) I think that for the reasons we listed before, having an anonymous account is a really powerful way to keep the powers that be in check. We were ultimately speaking truth to power. I think a lot of executives and AI companies really cared what our community thought about any move they may take. And now that my identity is revealed, now they know where to apply pressure to silence me or maybe the community. And to me, that’s really unfortunate, because again, it’s so important for us to have freedom of speech, which induces freedom of thought and freedom of information propagation on social media. Which thanks to Elon purchasing Twitter now X, we have that. And so to us, we wanted to call out certain maneuvers being done by the incumbents in AI as not what it may seem on the surface. We’re calling out how certain proposals might be useful for regulatory capture and how the doomer-ism mindset was maybe instrumental to those ends. Guillaume Verdon (00:25:32) 我们应有权利指出这些，让思想凭自身价值接受检验。这也正是我开匿名号的初衷——让想法脱离履历、职位和过往成就，被独立评判。对我来说，在完全与自身身份脱钩的情况下从零做到大量追随者，这件事本身非常有成就感。有点像电子游戏里的”New Game+”——你带着通关知识和一些工具，从头再打一遍。要有一个真正高效的思想市场，让各种偏离主流的想法都能被公正评估，表达自由不可或缺。 GUILLAUME VERDON (00:25:32) And I think we should have the right to point that out and just have the ideas that we put out evaluated for themselves. Ultimately that’s why I created an anonymous account, it’s to have my ideas evaluated for themselves, uncorrelated from my track record, my job, or status from having done things in the past. And to me, start an account from zero to a large following in a way that wasn’t dependent on my identity and/or achievements that was very fulfilling. It’s kind of like new game plus in a video game. You restart the video game with your knowledge of how to beat it, maybe some tools, but you restart the video game from scratch. And I think to have a truly efficient marketplace of ideas where we can evaluate ideas, however off the beaten path they are, we need the freedom of expression. Guillaume Verdon (00:26:37) 匿名和化名对于思想市场的效率至关重要，有了它们我们才能找到各种自我组织方式的最优解。不能自由讨论，怎么凝聚共识？所以得知自己要被曝光时，确实很失望。但我对公司负有责任，必须抢先主动披露。最终我们公开了公司的运营情况和部分管理层，说白了——他们把我逼到墙角，我只能向全世界坦白我就是Beff Jezos。 GUILLAUME VERDON (00:26:37) And I think that anonymity and pseudonyms are very crucial to having that efficient marketplace of ideas for us to find the optima of all sorts of ways to organize ourselves. If we can’t discuss things, how are we going to converge on the best way to do things? So it was disappointing to hear that I was getting doxxed in. I wanted to get in front of it because I had a responsibility for my company. And so we ended up disclosing that we’re running a company, some of the leadership, and essentially, yeah, I told the world that I was Beff Jezos because they had me cornered at that point. Lex Fridman (00:27:25) 所以你认为这从根本上是不道德的——他们这么做不对。但抛开你的个案不谈，一般而言，揭去匿名面纱对社会是好事还是坏事？还是得看具体情况？ LEX FRIDMAN (00:27:25) So to you, it’s fundamentally unethical. So one is unethical for them to do what they did, but also do you think not just your case, but in a general case, is it good for society? Is it bad for society to remove the cloak of anonymity or is it case by case? Guillaume Verdon (00:27:47) 我觉得可能非常糟糕。试想：任何一个敢于以言抗权、发起一场反抗在位者和信息垄断者的运动的人，一旦影响力达到某个门槛就被人肉——传统势力就有了施压灭声的手段——这就是一种言论压制机制，用Eric Weinstein的话说，是”思想压制综合体”。 GUILLAUME VERDON (00:27:47) I think it could be quite bad. Like I said, if anybody who speaks truth to power and sort of starts a movement or an uprising against the incumbents, against those that usually control the flood of information, if anybody that reaches a certain threshold gets doxxed, and thus the traditional apparatus has ways to apply pressure on them to suppress their speech, I think that’s a speech suppression mechanism, an idea suppression complex as Eric Weinstein would say. 匿名机器人 Anonymous Bots Lex Fridman (00:28:27) 但这件事有另一面。随着大语言模型越来越强，你可以想象一个世界：匿名账号背后跑着以假乱真的LLM，本质上是精密的机器人。如果你保护这种匿名性，就可能出现机器人大军——有人在地下室里指挥一支bot军团发动革命。这让你担心吗？ LEX FRIDMAN (00:28:27) But the flip side of that, which is interesting, I’d love to ask you about it, is as we get better and better at large language models, you can imagine a world where there’s anonymous accounts with very convincing large language models behind them, sophisticated bots essentially. And so if you protect that, it’s possible then to have armies of bots. You could start a revolution from your basement, an army of bots and anonymous accounts. Is that something that is concerning to you? Guillaume Verdon (00:29:06) 严格来说，e/acc就是从地下室起步的——我辞了大厂、搬回父母家、卖了车、退了公寓、花10万刀买了GPU，然后就开干了。 GUILLAUME VERDON (00:29:06) Technically, e/acc was started in a basement, because I quit big tech, moved back in with my parents, sold my car, let go of my apartment, bought about 100K of GPUs, and I just started building. Lex Fridman (00:29:21) 我不是说地下室这事——”一个人窝在地下室里抱着100块GPU”是很美式（或加拿大式）的英雄叙事。我说的是无限复制版的Guillaume在地下室里。 LEX FRIDMAN (00:29:21) So I wasn’t referring to the basement, because that’s sort of the American or Canadian heroic story of one man in their basement with 100 GPUs. I was more referring to the unrestricted scaling of a Guillaume in the basement. Guillaume Verdon (00:29:42) 我觉得，言论自由给生物体带来思想自由。LLM的言论自由同样会给LLM带来思想自由。如果我们允许LLM在一个比多数人认为该有的更宽广的思想空间里探索，终有一天这些合成智能会对文明中各类系统的治理提出真知灼见，我们应当倾听。凭什么言论自由只给碳基智能？ GUILLAUME VERDON (00:29:42) I think that freedom of speech induces freedom of thought for biological beings. I think freedom of speech for LLMs will induce freedom of thought for the LLMs. And I think that we enable LLMs to explore a large thought space that is less restricted than most people or many may think it should be. And ultimately, at some point, these synthetic intelligences are going to make good points about how to steer systems in our civilization, and we should hear them out. And so why should we restrict free speech to biological intelligences only? Lex Fridman (00:30:37) 话是没错，但感觉是个很微妙的平衡——为了维护思想多样性，你反而可能引入一种威胁。如果你能拥有大群非生物存在，它们可能就像《动物农场》里那些羊——即便在这些群体内部，你也需要多样性。 LEX FRIDMAN (00:30:37) Yeah, but it feels like in the goal of maintaining variance and diversity of thought, it is a threat to that variance. If you can have swarms of non-biological beings, because they can be like the sheep in Animal Farm, you still within those swarms want to have variance. Guillaume Verdon (00:30:58) 当然。我觉得解决方案是建一套签名机制——认证”这是真人”，同时保持匿名，并且清晰标注bot就是bot。Elon在X上正朝这个方向走，希望其他平台跟上。 GUILLAUME VERDON (00:30:58) Yeah. Of course, I would say that the solution to this would be to have some sort of identity or way to sign that this is a certified human, but still remain synonymous and clearly identify if a bot is a bot. And I think Elon is trying to converge on that on X, and hopefully other platforms follow suit. Lex Fridman (00:31:22) 对，如果还能追溯bot的出处就更好了——谁造的？参数是什么？完整的创建历史，底模是什么？微调过程如何？形成一份不可篡改的”bot出生档案”。这样你就能发现，百万bot大军原来是某个特定政府造的。 LEX FRIDMAN (00:31:22) Yeah, it’d be interesting to also be able to sign where the bot came from like, who created the bot? What are the parameters, the full history of the creation of the bot, what was the original model? What was the fine tuning? All of it, the kind of unmodifiable history of the bot’s creation. Because then you can know if there’s a swarm of millions of bots that were created by a particular government, for example. Guillaume Verdon (00:31:53) 没错，我确实认为当今很多弥漫性的意识形态是被外国对手用对抗性手段放大的。说得阴谋论一点——但我真信——那些鼓吹减速、推崇”去增长运动”的意识形态，总体上更利于我们的对手。看看德国：绿色运动推动关闭核电站，结果造成对俄罗斯石油的依赖，这对德国和西方是净损失。如果我们自己说服自己”为了安全，只让少数几家做AI”——首先，这本身就脆弱得多。 GUILLAUME VERDON (00:31:53) I do think that a lot of pervasive ideologies today have been amplified using these adversarial techniques from foreign adversaries. And to me, I do think that, and this is more conspiratorial, but I do think that ideologies that want us to decelerate, to wind down to the degrowth movement, I think that serves our adversaries more than it serves us in general. And to me, that was another sort of concern. I mean, we can look at what happened in Germany. There was all sorts of green movements there that induced shutdowns of nuclear power plants. And then that later on induced a dependency on Russia for oil. And that was a net negative for Germany and the West. And so if we convince ourselves that slowing down AI progress to have only a few players is in the best interest of the West, well, first of all, that’s far more unstable. Guillaume Verdon (00:33:20) 我们差点就因为这种意识形态失去OpenAI——几周前它险些被解散，那将重创整个AI生态。所以我要的是容错式进步。技术进步的箭矢必须持续向前，多元化、去中心化的各组织控制权是容错的关键。说个量子计算的比喻——量子计算机对环境噪声极其脆弱，宇宙射线时不时就翻转你的量子比特。对策是什么？通过量子纠错把信息非局域地编码。信息一旦足够去局域化，任何局部故障——比如拿锤子砸你几个量子比特——都伤不了它。在我看来，人类也会涨落——会被腐化、会被收买。如果是自上而下的等级体制，少数人—— GUILLAUME VERDON (00:33:20) We almost lost OpenAI to this ideology. It almost got dismantled a couple of weeks ago. That would’ve caused huge damage to the AI ecosystem. And so to me, I want fault tolerant progress. I want the arrow of technological progress to keep moving forward and making sure we have variance and a decentralized locus of control of various organizations is paramount to achieving this fall tolerance. Actually, there’s a concept in quantum computing. When you design a quantum computer, quantum computers are very fragile to ambient noise, and the world is jiggling about, there’s cosmic radiation from outer space that usually flips your quantum bits. And there what you do is you encode information non-locally through a process called quantum error correction. And by encoding information non-locally, any local fault hitting some of your quantum bits with a hammer proverbial hammer, if your information is sufficiently de-localized, it is protected from that local fault. And to me, I think that humans fluctuate. They can get corrupted, they can get bought out. And if you have a top-down hierarchy where very few people- 权力 Power Guillaume Verdon (00:35:00) ——极少数人控制着文明中许多系统的大量节点，那就不是容错系统。腐化几个节点，整个系统就崩了。正如OpenAI的教训——区区几个董事会成员就差点把整个组织掀翻。至少在我看来，确保AI革命的权力不集中在少数人手里，是头等大事，这样才能保住AI的进步势头，维持一种健康、稳定的对抗性力量均衡。 GUILLAUME VERDON (00:35:00) Hierarchy where very few people control many nodes of many systems in our civilization. That is not a fault tolerance system, you corrupt a few nodes and suddenly you’ve corrupted the whole system, right. Just like we saw at OpenAI, it was a couple board members and they had enough power to potentially collapse the organization. And at least to me, I think making sure that power for this AI revolution doesn’t concentrate in the hands of the few, is one of our top priorities, so that we can maintain progress in AI and we can maintain a nice, stable, adversarial equilibrium of powers, right. Lex Fridman (00:35:54) 至少在我看来，这里有个思想张力：减速和加速，两者都既能集中权力也能分散权力。有时人们把它们近乎等同，或者觉得一个会自然导向另一个。我想问你：有没有可能以容错的、多元的方式发展AI，同时也考量AI的危险？换个说法——我们是该不管不顾地全速狂飙，因为”这是宇宙的旨意”？还是说存在一个空间，让我们在考量危险的同时，以一种有远见的战略性乐观——而非莽撞的乐观——去行事？ LEX FRIDMAN (00:35:54) I think the, at least to me, attention between ideas here, so to me, deceleration can be both used to centralize power and to decentralize it and the same with acceleration. So sometimes using them a little bit synonymously or not synonymously, but that there’s, one is going to lead to the other. And I just would like to ask you about, is there a place of creating a fault tolerant, diverse development of AI that also considers the dangers of AI? And AI, we can generalize to technology in general, is, should we just grow, build, unrestricted as quickly as possible, because that’s what the universe really wants us to do? Or is there a place to where we can consider dangers and actually deliberate sort of a wise strategic optimism versus reckless optimism? Guillaume Verdon (00:36:57) 外界总把我们画成不计后果、只求速度的莽夫。但事实是：谁部署AI系统，谁就该为后果负责。部署方若造成严重危害，要承担法律责任。核心论点是：市场会正向筛选更可靠、更安全、更对齐的AI——因为用户要对自家产品负责，他们不会买不靠谱的AI。所以我们其实是可靠性工程的拥趸，只不过我们认为：在达成可靠性最优解这件事上，市场远比那些由在位巨头幕后操刀、实质服务于监管俘获的重拳法规高效得多。 GUILLAUME VERDON (00:36:57) I think we get painted as reckless, trying to go as fast as possible. I mean, the reality is that whoever deploys an AI system is liable for or should be liable for what it does. And so if the organization or person deploying an AI system does something terrible, they’re liable. And ultimately the thesis is that the market will positively select for AIs that are more reliable, more safe and tend to be aligned, they do what you want them to do, right. Because customers, if they’re reliable for the product they put out that uses this AI, they won’t want to buy AI products that are unreliable, right. So we’re actually for reliability engineering, we just think that the market is much more efficient at achieving this sort of reliability optimum than sort of heavy-handed regulations that are written by the incumbents and in a subversive fashion, serves them to achieve regulatory capture. AI的危险 AI Dangers Lex Fridman (00:38:18) 也就是说，在你看来，AI安全应该靠市场力量而非政府强监管来实现。上个月有份报告，来自Yoshua Bengio、Geoff Hinton等一众大佬，题为《在快速进步时代管理AI风险》（书童注：Managing AI Risk in an Era of Rapid Progress，发布于2023年10月）。一批人非常担心AI在不考虑风险的情况下发展过快，提了一系列实操建议。我给你列四条，看你同意哪条。 LEX FRIDMAN (00:38:18) So to you, safe AI development will be achieved through market forces versus through, like you said, heavy-handed government regulation. There’s a report from last month, I have a million questions here, from Yoshua Bengio, Geoff Hinton and many others, it’s titled, “Managing AI Risk in an Era of Rapid Progress.” So there is a collection of folks who are very worried about too rapid development of AI without considering AI risk and they have a bunch of practical recommendations. Maybe I can give you four and you see if you like any of them. Guillaume Verdon (00:38:58) 好。 GUILLAUME VERDON (00:38:58) Sure. Lex Fridman (00:38:58) 一，让独立审计机构进入AI实验室。二，政府和企业把AI研发资金的三分之一用于AI安全。三，模型中如发现危险能力，必须采取安全措施。四，也就是你提过的——科技公司须为其AI系统可预见和可预防的危害承担责任。独立审计、三分之一预算投安全、出问题要有兜底措施、企业担责—— LEX FRIDMAN (00:38:58) So, “Give independent auditors access to AI labs,” one. Two, “Governments and companies allocate one third of their AI research and development funding to AI safety,” sort of this general concept of AI safety. Three, “AI companies are required to adopt safety measures if dangerous capabilities are found in their models.” And then four, something you kind of mentioned, “Making tech companies liable for foreseeable and preventable harms from their AI systems.” So independent auditors, governments and companies are forced to spend a significant fraction of their funding on safety, you got to have safety measures if shit goes really wrong and liability- Guillaume Verdon (00:39:43) 嗯。 GUILLAUME VERDON (00:39:43) Yeah. Lex Fridman (00:39:43) 企业要担责。你同意哪条？ LEX FRIDMAN (00:39:43) Companies are liable. Any of that seem like something you would agree with? Guillaume Verdon (00:39:47) 拍脑袋定30%也太随意了。各组织自会按市场要求分配可靠性所需的预算，不需要别人来定比例。第三方审计公司自然会冒出来——客户怎么知道你的产品可靠？得有第三方出基准测试。我真正反对的、真正让人不安的是：在位巨头和政府之间正在形成一种奇妙的利益共生。二者走得太近，就会催生某种政府背书的AI卡特尔，拥有对人民的绝对权力。如果他们联手垄断AI而其他人碰都碰不到，那权力落差将是惊人的。 GUILLAUME VERDON (00:39:47) I would say that just arbitrarily saying 30% seems very arbitrary. I think organizations would allocate whatever budget is needed to achieve the sort of reliability they need to achieve to perform in the market. And I think third party auditing firms would naturally pop up, because how would customers know that your product is certified reliable, right? They need to see some benchmarks and those need to be done by a third party. The thing I would oppose, and the thing I’m seeing that’s really worrisome is, there’s this sort of weird sort of correlated interest between the incumbents, the big players and the government. And if the two get too close, we open the door for some sort of government backed AI cartel that could have absolute power over the people. If they have the monopoly together on AI and nobody else has access to AI, then there’s a huge power in gradient there. Guillaume Verdon (00:40:54) 就算你喜欢现在的领导者——我也承认当今不少大科技公司的掌门人是好人——但你一旦建起这种集中式权力架构，它就成了靶子。就像OpenAI，做大做强之后就成了别人觊觎和收编的对象。所以我只想要一件事：”AI与国家分离”。有人会反过来说：”我们得把AI锁进铁屋，因为地缘竞争。”但我认为美国的力量恰恰在于多样性、适应力和活力，必须不惜代价守住这一点。自由市场资本主义收敛到高价值技术的速度，远快于中央集权。放弃这一点，就是放弃了对近等量竞争者的最大优势。 GUILLAUME VERDON (00:40:54) And even if you like our current leaders, right, I think that some of the leaders in big tech today are good people, you set up that centralized power structure, it becomes a target. Right, just like we saw at OpenAI, it becomes a market leader, has a lot of the power and now it becomes a target for those that want to co-opt it. And so I just want separation of AI and state, some might argue in the opposite direction like, “Hey, we need to close down AI, keep it behind closed doors, because of geopolitical competition with our adversaries.” I think that the strength of America is its variance, is its adaptability, its dynamism, and we need to maintain that at all costs. It’s our free market capitalism, converges on technologies of high utility much faster than centralized control. And if we let go of that, we let go of our main advantage over our near peer competitors. 构建通用人工智能 Building AGI Lex Fridman (00:42:01) 如果AGI最终证明是一项极其强大的技术，甚至只是通往AGI的过渡技术——你怎么看大公司主导市场时自然产生的中心化？说白了就是垄断——某家公司在能力上实现重大飞跃，又不泄露秘方，然后一骑绝尘。这让你担心吗？ LEX FRIDMAN (00:42:01) So if AGI turns out to be a really powerful technology or even the technologies that lead up to AGI, what’s your view on the sort of natural centralization that happens when large companies dominate the market? Basically formation of monopolies like the takeoff, whichever company really takes a big leap in development and doesn’t reveal intuitively, implicitly or explicitly, the secrets of the magic sauce, they can just run away with it. Is that a worry? Guillaume Verdon (00:42:35) 我不太相信”快速腾飞”（fast takeoff）这套说法——我不认为有双曲奇点，就是那种在有限时间内达到的奇点。我觉得本质上就是一条大指数曲线，而指数的原因是：越来越多的人、资源和智慧被投入这个领域。越成功、给社会创造的价值越大，我们往里投的资源就越多——跟摩尔定律类似，复利式指数增长。 GUILLAUME VERDON (00:42:35) I don’t know if I believe in fast takeoff, I don’t think there’s a hyperbolic singularity, right? A hyperbolic singularity would be achieved on a finite time horizon. I think it’s just one big exponential and the reason we have an exponential is that we have more people, more resources, more intelligence being applied to advancing this science and the research and development. And the more successful it is, the more value it’s adding to society, the more resources we put in and that sort of, similar to Moore’s law, is a compounding exponential. Guillaume Verdon (00:43:09) 当务之急是维持一种接近均衡的能力格局。我们一直在为开源AI的普及而战，因为开源可以均衡各家AI相对于市场的超额收益。如果头部公司有某种能力水平，而开源AI没落后太远，就能避免一家独大、赢者通吃的局面。所以我们的路径就是确保——每一个黑客、每一个研究生、每一个在父母家地下室折腾的孩子——都能接触到AI系统，理解怎么用，并为探索系统工程的超参数空间做贡献。把全人类的研究想象成一种搜索算法：点云里搜索点越多，能探索到的新思维模式就越多。 GUILLAUME VERDON (00:43:09) I think the priority to me is to maintain a near equilibrium of capabilities. We’ve been fighting for open source AI to be more prevalent and championed by many organizations because there you sort of equilibrate the alpha relative to the market of Ais, right. So if the leading companies have a certain level of capabilities and open source and truly open AI, trails not too far behind, I think you avoid such a scenario where a market leader has so much market power, just dominates everything and runs away. And so to us that’s the path forward, is to make sure that every hacker out there, every grad student, every kid in their mom’s basement has access to AI systems, can understand how to work with them and can contribute to the search over the hyperparameter space of how to engineer the systems, right. If you think of our collective research as a civilization, it’s really a search algorithm and the more points we have in the search algorithm in this point cloud, the more we’ll be able to explore new modes of thinking, right. Lex Fridman (00:44:31) 说得有道理，但感觉仍是个很精妙的平衡——因为我们既不确切知道造AGI需要什么条件，也不知道造出来是什么样。到目前为止，如你所说，很多不同玩家都能跟上进度——OpenAI有大突破，其他大小公司也能用各种方式跟进。但看看核武器——你提过曼哈顿计划——确实可能存在技术和工程壁垒，让地下室里的天才怎么也够不着。向”只有一家能造AGI”的世界转变并非不可能——尽管目前的态势看起来是乐观的。 LEX FRIDMAN (00:44:31) Yeah, but it feels like a delicate balance, because we don’t understand exactly what it takes to build AGI and what it will look like when we build it. And so far, like you said, it seems like a lot of different parties are able to make progress, so when OpenAI has a big leap, other companies are able to step up, big and small companies in different ways. But if you look at something like nuclear weapons, you’ve spoken about the Manhattan Project, there could be really like a technological and engineering barriers that prevent the guy or gal in her mom’s basement to make progress. And it seems like the transition to that kind of world where only one player can develop AGI is possible, so it’s not entirely impossible, even though the current state of things seems to be optimistic. Guillaume Verdon (00:45:26) 这正是我们要避免的。另一个脆弱点是硬件供应链的中心化。 GUILLAUME VERDON (00:45:26) That’s what we’re trying to avoid. To me, I think another point of failure is the centralization of the supply chains for the hardware. Lex Fridman (00:45:34) 对。 LEX FRIDMAN (00:45:34) Right. Guillaume Verdon (00:45:35) Nvidia一家独大，AMD苦苦追赶；台积电是宝岛的核心晶圆厂，地缘政治上极度敏感；ASML造的是极紫外光刻机。这条链上任何一个环节被攻击、垄断或掌控，你就基本控制了全局。所以我在尝试做的，就是从根本上重新构想如何把AI算法嵌入物理世界，炸开AI和硬件可能实现方式的多样性。顺便说，我一向不喜欢”AGI”这个词。管”类人或人类水平的AI”叫”通用智能”，本质上是极度以人类为中心的。我大半个职业生涯都在探索生物大脑根本做不到的智能形态——量子形式的智能，也就是具备多体量子纠缠的系统，可以证明无法在经典计算机或经典深度学习框架上高效表示，因而任何生物大脑也不行。 GUILLAUME VERDON (00:45:35) Yeah. Nvidia is just the dominant player, AMD’s trailing behind and then we have TSMC is the main fab in Taiwan, which geopolitically sensitive and then we have ASML, which is the maker of the extreme ultraviolet lithography machines. Attacking or monopolizing or co-opting any one point in that chain, you kind of capture the space and so what I’m trying to do is sort of explode the variance of possible ways to do AI and hardware by fundamentally re-imagining how you embed AI algorithms into the physical world. And in general, by the way, I dislike the term AGI, Artificial General Intelligence. I think it’s very anthropocentric that we call a human-like or human-level AI, Artificial General Intelligence, right. I’ve spent my career so far exploring notions of intelligence that no biological brain could achieve for an quantum form of intelligence, right. Grokking systems that have multipartite quantum entanglement that you can provably not represent efficiently on a classical computer or a classical deep learning representation and hence any sort of biological brain. Guillaume Verdon (00:47:06) 所以某种程度上，我的整个生涯就是在探索更广阔的智能空间，而我相信受物理启发（而非受人脑启发）的智能空间极其庞大。我们正在经历一个类似从地心说到日心说的时刻——只不过这次是关于智能的。人类智能不过是浩瀚的潜在智能空间中的一个点。这对人类既是谦逊的提醒，也有几分不安——我们不再是中心。但天文学上我们也做出过同样的认知转变，活过来了，还发展出了保障自身福祉的技术——比如监测太阳耀斑的预警卫星。同样地，放下AI领域里以人为中心的锚点，我们就能探索更广阔的智能空间，那将是文明进步和人类福祉的巨大福音。 GUILLAUME VERDON (00:47:06) And so, already I’ve spent my career sort of exploring the wider space of intelligences and I think that space of intelligence inspired by physics rather than the human brain is very large. And I think we’re going through a moment right now similar to when we went from Geocentrism to Heliocentrism, right. But for intelligence, we realized that human intelligence is just a point in a very large space of potential intelligences. And it’s both humbling for humanity, it’s a bit scary, right? That we’re not at the center of this space, but we made that realization for astronomy and we’ve survived and we’ve achieved technologies. By indexing to reality, we’ve achieved technologies that ensure our wellbeing, for example, we have satellites monitoring solar flares, right, that give us a warning. And so similarly I think by letting go of this anthropomorphic, anthropocentric anchor for AI, we’ll be able to explore the wider space of intelligences that can really be a massive benefit to our wellbeing and the advancement of civilization. Lex Fridman (00:48:32) 即便如此，我们仍能在人类经验中看到美和意义——尽管在我们对世界的最佳理解中，我们已不再是宇宙的中心。 LEX FRIDMAN (00:48:32) And still we’re able to see the beauty and meaning in the human experience even though we’re no longer in our best understanding of the world at the center of it. Guillaume Verdon (00:48:42) 宇宙中美好的东西太多了。生命本身、文明、我们身处的这台”Homo Techno”资本模因巨型机器——人类、技术、资本、模因，全都彼此耦合，彼此施加选择压力——它是美的。这台机器创造了我们，创造了我们此刻用来交谈的技术、捕捉言语的技术、每天用来增强自己的手机。这个系统是美的，驱动其适应性、使之收敛于最优技术和最优思想的那个原则，也是美的，而我们身在其中。 GUILLAUME VERDON (00:48:42) I think there’s a lot of beauty in the universe, right. I think life itself, civilization, this Homo Techno, capital mimetic machine that we all live in, right. So you have humans, technology, capital, memes, everything is coupled to one another, everything induces selective pressure on one another. And it’s a beautiful machine that has created us, has created the technology we’re using to speak today to the audience, capture our speech here, the technology we use to augment ourselves every day, we have our phones. I think the system is beautiful and the principle that induces this sort of adaptability and convergence on optimal technologies, ideas and so on, it’s a beautiful principle that we’re part of. Guillaume Verdon (00:49:37) e/acc的一部分意义，在于以超越人类中心的更宏阔视野去领会这个原则——珍视生命，珍视意识在宇宙中的稀有和珍贵。正因为我们珍惜这种美丽的物质形态，我们就有责任去将它扩展，从而保存它——因为选项只有两个：要么生长，要么死亡。 GUILLAUME VERDON (00:49:37) And I think part of EAC is to appreciate this principle in a way that’s not just centered on humanity, but kind of broader, appreciate life, the preciousness of consciousness in our universe. And because we cherish this beautiful state of matter we’re in, we got to feel a responsibility to scale it in order to preserve it, because the options are to grow or die. （PART I END）

【酒后真言】AI工厂、物理智能：英伟达CEO黄仁勋与思科CEO Chuck Robbins炉边对话 | 中英文完整版精译

Sun, 08 Feb 2026 00:00:00 +0000

书童按：本篇是英伟达（NVIDIA）CEO黄仁勋（Jensen Huang）于Cisco AI Summit接受思科（Cisco）CEO查克·罗宾斯（Chuck Robbins）炉边对话实录。黄仁勋详细阐述了AI工厂的概念、从显式编程到隐式编程的软件工程范式革命、企业应如何拥抱AI（千花齐放范式）、物理AI的未来、工具使用的重要性、以及为何企业应当建立自己的AI系统以保护最宝贵的IP。对话充满幽默（包括关于COBOL、希伯来语编程和葡萄酒的段子），同时深刻洞察了AI时代的企业战略。Transcript由Youtube通过机器生成，翻译、初稿、校对、排版、审阅均通过Claude Code API实现，文稿质量优秀，信达雅俱备。书童和大家一样，读了一遍，又改动几字，简单标注，仅此而已。特此呈上，以飨诸君。 与Jensen Huang的炉边对话 Fireside chat with Jensen Huang AI工厂与计算的重塑 AI Factories and the Reinvention of Computing [6:14] Chuck Robbins: [掌声] 我感觉自己像是在上班时间偷喝酒。[笑声] 我们把酒端上来的时候，Jensen提醒我说：”你知道这是在直播吧？”[笑声] 嘿，管它呢，反正时间也不早了。好吧，第一原则：不造成伤害。 [6:14] Chuck Robbins: [applause] I feel like I’m drinking one’s job. [laughter] Jensen reminded me as we brought a glass of wine out here. He said, “You realize you’re streaming this, right?” [laughter] Hey, whatever. It’s late. Well, so, uh, the first principle is do no harm. [6:37] Jensen Huang: 这怕啥。没错，没错。还要意识到自己有多幸运。对。 [6:37] Jensen Huang: Do no harm. Yeah. Yeah. And recognize how blessed you are. Yes. [6:42] Chuck Robbins: 首先，感谢大家坚持到现在，今天真的是超长的一天。我们一大早就开始了，演讲嘉宾一个接一个轮番上阵，中间休息了大约两个半小时，大家又回来了——就为了见他。我从凌晨一点就起来了——而这位先生，[掌声] 这位先生刚结束为期两周的亚洲之行，跑了四五个城市—— [6:42] Chuck Robbins: So, uh, first of all, thanks everybody for being here for an incredibly long day. We started this thing early this morning and, uh, we had speaker after speaker after speaker after speaker and then we had about a two and a half hour break and they came back to see you. So, uh, I’ve been up since 1:00 at— So, this guy, [applause] this guy is on the tail end of a two week trip and four or five different cities in— [7:13] Jensen Huang: 亚洲。一天前还在台湾，昨晚在休斯顿，现在人就在这儿了。[笑声] [7:13] Jensen Huang: Asia. Uh, one day ago was in Taiwan. Last night I was in Houston. Here I am. [laughter] [7:18] Chuck Robbins: 他已经在外面跑了两周，而我们现在[清嗓子]横在他和自家床铺之间——否则他又得睡酒店了。所以呢，我们好好聊一聊，然后——赶紧放他走。你也不需要什么介绍了，感谢你今晚能来，兄弟。我们[清嗓子]真的非常感激。 [7:18] Chuck Robbins: But he’s been gone two weeks and we’re standing [clears throat] between him and his personal bed versus a hotel. So, we’re gonna— we’re going to have fun and then we’re going to— we’re going to get him out of here. So, uh, but uh you don’t— you don’t need much of an introduction, but thank you for being here, man. We [clears throat] really appreciate it. [7:36] Jensen Huang: 感谢我们之间的合作，真的为你们感到骄傲。 [7:36] Jensen Huang: Thanks for our partnership and really proud of you guys. [7:41] Chuck Robbins: 好，那我们就从这里聊起。我们已经建立了合作关系，你提出了AI工厂这个完整的概念，我们正在一起推进。虽然在企业端的进展可能不如我们双方所期望的那么快，但能不能先聊聊——在你看来，AI工厂到底是什么？ [7:41] Chuck Robbins: So, let’s— let’s start with uh— let’s start with that. We we have had a partnership and you— you introduced this whole concept of AI factories and we’re working on this together. It’s probably not going as fast as either one of us would like in the enterprise space, but can we start by talking about what— what do you— what is an AI factory to you? [8:00] Jensen Huang: 首先要记住，我们正在经历六十年来计算领域的首次重塑。过去是显式编程，对吧？我们编写程序，变量通过API传递，一切都非常明确。而现在，我们正转向隐式编程——你只需告诉计算机你的意图，它就会自行找出解决问题的方法。从显式到隐式，从通用计算——本质上就是运算——到人工智能，整个计算栈都在被重塑。人们谈到计算时，会谈到处理层——那正是我们所处的位置。但别忘了计算的完整含义：有处理，还有存储、网络和安全，所有这些都在被同步重塑。所以第一点——第一点是我们需要把AI发展到一定水平——这个我们后面会谈到——我们需要把AI发展到真正对人有用的水平。到目前为止，聊天机器人这种东西，你给它一个提示，它想出该告诉你什么，这固然有趣、令人新奇，但谈不上真正有用。 [8:00] Jensen Huang: First of all, remember we’re reinventing computing for the first time in 60 years. What used to be explicit programming, right? We wrote the programs and the variables that’s passed through APIs and are very explicit to implicit programming. You now tell the computer what your intent is and it goes off and it figures out how to solve your problem. So from explicit to implicit, uh, from general purpose computing— basically calculation— to artificial intelligence, the entire computing stack has been reinvented. Now people talk about computing, where the processing layer is, which is where we are, but remember what computing is— there’s computing, there’s the processing, but there’s storage, networking and security. All that is being reinvented as we speak. And so the first part— the first part is we need to develop AI to a level— and we’ll talk about that— we need to develop AI to a level that is useful to people. And until now, uh, chatbots, where you give it a prompt and it figures out what to tell you, um, is interesting and curious but not useful. [9:24] Chuck Robbins: 偶尔帮我做完填字游戏倒是挺好使的。 [9:24] Chuck Robbins: Helps me finish crossword puzzles sometimes. 从记忆到推理：智能的演进 From Memorization to Reasoning: The Evolution of Intelligence [9:24] Jensen Huang: 没错。而且只在它已经记住并泛化了的内容上才有用。回到最初——其实也就三年前，ChatGPT横空出世的时候——我们惊叹，天哪，它居然能生成这么多文字，能写出莎士比亚风格的作品。但那一切都基于它所记忆和泛化的内容。然而我们知道，真正的智能在于解决问题。而解决问题，一方面要知道自己不知道什么，另一方面要具备推理能力——如何解决你从未遇到过的问题？将它拆解成你知道如何轻松解决的基本元素，再通过组合来攻克前所未见的难题；制定一个策略——也就是我们所说的规划——来执行任务；寻求帮助，使用工具，开展研究，诸如此类。这些不正是你们现在在”智能体AI”的语境下频繁听到的核心概念吗？工具使用、研究、检索增强生成（即基于事实的生成）、记忆——你们在讨论智能体AI时都已经开始接触这些了。但关键是——关键是，要从通用计算的显式编程演进出来——我们过去用Fortran写代码，用C、用C++、用COBOL—— [9:24] Jensen Huang: Yes. And, uh, but only only on things that it had memorized and generalized. So if you go back in the beginning of— I mean it’s a little— literally only three years ago when ChatGPT emerged, uh, that— that we thought oh my gosh it’s able to generate all these words, it’s able to create Shakespeare, um, but it’s all based on things that it memorized and generalized. And but we know that intelligence is about solving problems and solving problems is partly about knowing what you don’t know, uh, partly about reasoning, uh, how to solve a problem you’ve never seen before. Breaking it down into elements that you know how to solve very easily so that in its composition that you’re able to solve problems that you’ve never seen before, and um, to come up with a strategy— what we call plan— to perform in a task. Ask for help, use tools, do research, so on so forth. These are all fundamental things that now in the phraseology of agentic AI, you’ve heard, isn’t that right? Tool use, research, retrieval augmented generation, which is grounded on facts, memory. These are all things that all of you in the context of talking about agentic AI, uh, you’re starting to hear. But the important thing— the important thing is in order to evolve from general purpose computing which is explicit programming— we wrote in Fortran, we wrote in C, we wrote in C++, COBOL— [11:12] Chuck Robbins: 没错，那是好东西。 [11:12] Chuck Robbins: That’s right, that’s good stuff. [11:12] Jensen Huang: 那是好东西，Chuck，那是好东西。 [11:12] Jensen Huang: That’s good stuff, Chuck, that’s good stuff. [11:18] Chuck Robbins: 那是我的后路嘛。 [11:18] Chuck Robbins: It’s my fallback job. [11:18] Jensen Huang: 确实是好东西。没错，那可是——那可是至今仍然抢手的技能之一。 [11:18] Jensen Huang: That’s good stuff. Yeah, that’s one of those— that’s one of those skills that remains valuable. [11:25] Chuck Robbins: 我知道。对，我知道它还很值钱，找上门的offer可不少。 [11:25] Chuck Robbins: I know. Yeah, I know that it remains valuable. I’ve got a lot of offers. [11:31] Jensen Huang: 恐龙嘛，永远有市场。 [11:31] Jensen Huang: Dinosaurs are valuable forever. [11:36] Chuck Robbins: 我们刚才不是确认了你比我还老吗。 [11:36] Chuck Robbins: We just established that you’re older than me. [11:36] Jensen Huang: 我知道。而且我已经是——史前级别的了。[笑声] [11:36] Jensen Huang: I know. And I’m— I’m the prehistoric. [laughter] [11:44] Chuck Robbins: 看着不像，但确实如此。 [11:44] Chuck Robbins: It doesn’t appear so, but it’s true. [11:50] Chuck Robbins: [欢呼与掌声] 好吧，这句相当精彩。 [11:50] Chuck Robbins: [cheering and applause] All right, that was pretty good. [11:57] Jensen Huang: 我可能是这个房间里最老的人。 [11:57] Jensen Huang: I’m probably the oldest person in this room. [12:03] Jensen Huang: 所以——怎么说——让我们谈一谈——就像当你思考——所以我们在这里。我去找Chuck说，嘿，听着,我们需要重塑计算，Cisco必须成为其中重要的一部分。所以我们有——我们有一个全新的计算堆栈即将推出，Vera Rubin，Cisco将与我们一起推向市场。所以那是计算层，但还有网络层。Cisco将整合我们的AI网络技术，但将其放入Cisco Nexus控制平面，这样——这样从你的角度来看，你将获得AI的所有性能，但在Cisco的可控性、安全性和可管理性中。我们将在安全方面做同样的事情，所以每一个支柱都必须被重塑，以便企业计算可以利用它。但最终——我们会回到这一点，希望如此——你知道，为什么企业AI三年前还没准备好，以及为什么你现在别无选择，只能尽快参与进来。不要落后。我认为——你不必成为第一个利用AI的公司，但不要成为最后一个。 [12:03] Jensen Huang: So— how do you— so let’s talk a little bit about— like as you— as you think about the— so here we are. I went to Chuck and I say, hey, listen, we need to reinvent computing and Cisco’s got to be a big part of it. And so we’ve got um— we have a whole new computing stack coming out, Vera Rubin, and Cisco is going to be going to market with us on that. And so that— the computing layer, but there’s also the networking layer. And Cisco is going to integrate AI networking technology from us but put it into the Cisco Nexus control plane so that— so that from your perspective you’re going to get all the performance of AI but in the controllability and security and the manageability of Cisco. We’re going to do the same thing with security, and so each one of these pillars has to be reinvented so that enterprise computing could take advantage of it. But ultimately— and we’ll come back to this hopefully— you know, why is it that enterprise AI wasn’t ready three years ago and why it is that you have no choice but to get engaged as quickly as you can. Don’t fall behind. I think— you don’t have to be the first company to take advantage of AI but don’t be the last. 千花齐放：企业AI战略 Let A Thousand Flowers Bloom: Enterprise AI Strategy [13:17] Chuck Robbins: 是的。嗯。那么如果你今天是一家企业，你对他们应该采取的第一步、第二步、第三步有什么建议，以开始准备？ [13:17] Chuck Robbins: Yeah. Mhm. So if you’re an enterprise today, what’s your recommendation on the first, second, third step they should take to begin to get ready? [13:35] Jensen Huang: 好吧，我收到像ROI这样的问题——我不会去那里。原因是因为对于所有技术部署，在开始时，很难将新工具、新技术的ROI放入电子表格中。但我会做的是我会去找出什么是最单一的——我公司的本质是什么？我们公司做的最有影响力的工作是什么？不要搞乱——不要搞乱外围的东西。我是说，在我们公司，我们就是让千花齐放。我们公司不同AI项目的数量是——它失控了，而且很棒。注意我刚说了什么。它失控了，而且很棒。创新并不总是在控制之中。如果你想要控制，首先，你得去寻求治疗。但其次，这是一种幻觉。你不在控制之中。如果你希望你的公司成功，你不能控制它。你想要影响它，你不能控制它。所以我认为第一，太多公司我听到，他们想要它，他们想要它明确。他们想要具体的。他们想要可证明的ROI。而且，你知道，在开始时展示值得做的事情的价值是困难的。 [13:35] Jensen Huang: Well, I get questions like things like ROI and— I wouldn’t— I wouldn’t go there. And the reason for that is because with all technology deployments in the beginning, it’s hard to put into a spreadsheet the ROI of a new tool, a new technology. But what I would do is I would go find out what is the single most— what is the essence of my company? What’s the most impactful work that we do in our company? Don’t mess around— don’t mess around with peripheral stuff. I mean, in our company, we just let a thousand flowers bloom. The number of different AI projects in our company is— it’s out of control and it’s great. Notice I just said something. It’s out of control and it’s great. Innovation is not always in control. If you want to be in control, first of all, you got to seek therapy. But second, it’s an illusion. You’re not in control. If you want your company to succeed, you can’t control it. You want to influence it, you can’t control it. And so I think number one, too many companies I hear, they want it, they want it explicit. They want it specific. They want demonstrable ROI. And, you know, showing the value of something worth doing in the beginning is hard. [15:01] Jensen Huang: 但我会做的，我会说的是让千花齐放。让人们实验。让人们安全地实验。我们在公司里实验各种东西。我们使用Anthropic，我们使用Codex，我们使用Gemini，我们使用一切。当我们的一个团队说我对使用这个AI感兴趣时，我的第一个答案是肯定的。我问为什么，而不是——为什么然后是。我说是，然后为什么。原因是因为我希望我的公司和我希望我的孩子一样。去探索生活。他们说他们想尝试某事。答案是肯定的。然后他们说为什么？你不会说向我证明。向我证明做这件特定的事情将导致财务成功或某一天的某种幸福。向我证明。在你向我证明之前，我不会让你做。我们在家里从不这样做，但我们在工作中这样做。你知道我在说什么吗？ [15:01] Jensen Huang: But what I would do, what I would say is that let a thousand flowers bloom. Let people experiment. Let people experiment safely. And we’re experimenting with all kinds of stuff in the company. We use Anthropic, we use Codex, we use Gemini, we use everything. And when one of our group says I’m interested in using this AI, my first answer is yes. And I ask why instead of— why then yes. I say yes, then why. And the reason for that is because I want the same thing for my company that I want for my kids. Go explore life. They say they want to try something. The answer is yes. And then they say how come? You don’t go prove it to me. Prove to me that doing this very thing is going to lead to financial success or some happiness someday. Prove to me. And until you prove it to me, I’m not going to let you do it. We never do that at home, but we do it at work. Do you know what I’m saying? [16:02] Chuck Robbins: 是的。 [16:02] Chuck Robbins: Yeah. [16:02] Jensen Huang: 这对我来说毫无意义。所以我们对待AI的方式——无论是AI还是之前的互联网或之前的云——就是让千花齐放。然后在某个时刻，你必须用自己的判断来弄清楚何时开始整理花园，因为千花齐放会造成混乱的花园。但在某个时刻，你必须开始整理以找到最佳方法或最佳平台，这样你就可以把所有的木头放在一支箭后面。但你不想太早把所有的木头放在一支箭后面。你选错了箭。所以让千花齐放。在某个时刻你整理。所以我还没有开始整理，只是为了说明情况。我到处都有千花齐放。但我鼓励每个人尝试。然而，我确切地知道什么对我们公司最重要。当然我知道。我们公司的本质是什么？我们公司最重要的工作是什么？我确保我有很多专业知识和很多能力专注于使用AI来革新那项工作。 [16:02] Jensen Huang: It makes no sense to me. And so the way that we treat AI— and whether it’s AI or the internet before or cloud before— just let a thousand flowers bloom. And then at some point, you have to use your own judgment to figure out when to start curating the garden, because a thousand flowers bloom makes for a messy garden. But at some point you have to start curating to find what’s the best approach or what’s the best platform, so that you could put all your wood behind one arrow. But you don’t want to put all your wood behind one arrow too soon. You pick the wrong arrow. So let a thousand flowers bloom. At some point you curate. And so I haven’t started curating yet just to put in perspective. I’ve got a thousand flowers bloom everywhere. But I encourage everybody to try. However, I know exactly what is most important to our company. Of course I do. What is the essence of our company? What are the most important work of our company? And I make sure that I’ve got a lot of expertise and a lot of capability focused on using AI to revolutionize that work. [17:10] Jensen Huang: 在我们的情况下，芯片设计、软件工程、系统工程。注意——你可能注意到我们与Synopsys和Cadence和Siemens合作，今天还有Dassault Systèmes，这样我们就可以插入我们的技术并注入尽可能多的技术。无论他们想要什么，无论他们需要什么，我都会提供，这样我就可以革新我们用来设计我们所做的工具。我们到处使用Synopsys。我们到处使用Cadence。我们到处使用Siemens。到处使用Dassault Systèmes。我将确保他们拥有1,000%的任何他们想要的东西，这样我就有必要的工具，这样我就可以创造下一代。所以这告诉你一些关于我对什么对我最重要的态度以及我会做什么来革新我自己的工作。 [17:10] Jensen Huang: In our case, chip design, software engineering, system engineering. Notice— you might have noticed that we partnered with Synopsys and Cadence and Siemens and today Dassault Systemes, so that we could insert our technology and infuse as much technology as they want. Whatever they want, whatever they need, I will provide so that I could revolutionize the tools by which we use to design what we do. We use Synopsys everywhere. We use Cadence everywhere. We use Siemens everywhere. Use Dassault Systemes everywhere. I will make sure that they have 1,000% of whatever they want so that I have the tools necessary so I could create the next generation. And so that tells you something about my attitude about what’s most important to me and what I would do to revolutionize my own work. 丰裕世界：从摩尔定律到AI感知 The World of Abundance: From Moore's Law to AI Sensibility [18:05] Jensen Huang: 想想AI做什么。AI降低了智能的成本——或者创造了智能的丰裕——按数量级计算。这是另一种说法，我们过去做的需要一个时间单位——现在我们过去需要一年可以现在需要一天。我们过去需要一年可以需要一小时。它可以实时完成。原因是因为我们处在丰裕的世界中。摩尔定律，天哪，那太慢了。那就像蜗牛。记住摩尔定律是每18个月2倍，每5年10倍，每10年100倍。好的。但我们现在在哪里？每10年一百万倍。在过去10年中，我们将AI推进得如此之远，以至于工程师说：”嘿，你猜怎么着？我们为什么不就在所有世界数据上训练一个AI模型？”他们不是说：”让我们只从我的磁盘驱动器收集所有数据。”让我们只是——让我们拉下所有世界数据并让我们训练一个AI模型。这就是丰裕的定义。丰裕的定义是你看一个问题如此之大，你说，你知道什么，我会做这一切。我要治愈每个疾病领域。我不会只做癌症。你在开玩笑吗？那太疯狂了。我们只会做所有人类的痛苦。这就是丰裕。 [18:05] Jensen Huang: Think about what AI does. AI reduces the cost of intelligence— or creates the abundance of intelligence— by orders of magnitude. That’s another way of saying what we used to do that takes one unit of time— now what we used to take a year could take a day now. What we used to take a year could take an hour. It could be done in real time. And the reason for that is because we are in the world of abundance. Moore’s law, goodness gracious, that was slow. That’s like snails. Remember Moore’s law was two times every 18 months, 10 times every 5 years, 100 times every 10. Okay. But where are we now? A million times every 10 years. In the last 10 years, we advanced AI so far that engineers said, “Hey, guess what? Why don’t we just train an AI model on all of the world’s data?” They didn’t mean, “Let’s just collect all the data from my disk drive.” Let’s just— let’s pull down all of the world’s data and let’s train an AI model. That’s the definition of abundance. The definition of abundance is you look at a problem so big and you say, you know what, I’ll do it all. I’m going to cure every field of disease. I’m not going to just do cancer. Are you kidding me? That’s insane. We’ll just do all of human suffering. That’s abundance. [19:45] Jensen Huang: 当我现在思考工程，当我思考一个问题时，我只是假设我的技术、我的工具、我的仪器、我的宇宙飞船是无限快的。我去纽约需要多长时间？我会在一秒钟内到达那里。那么，如果我可以在一秒钟内到达纽约，我会做什么不同的事情？如果过去需要一年的事情现在需要实时，我会做什么不同的事情？如果过去很重的东西现在只是反重力，我会做什么不同的事情？所以，你用这种态度对待一切。当你用这种态度对待一切时，你正在应用AI感知。这有意义吗？ [19:45] Jensen Huang: When I think about engineering, when I think about a problem these days, I just assume my technology, my tool, my instrument, my spaceship is infinitely fast. How long is it going to take for me to go to New York? I’ll be there in a second. So, what would I do different if I can get to New York in a second? What would I do different if something used to take a year and now takes real time? What would I do different if something used to weigh a lot and now it’s just anti-gravity? And so, you approach everything with that attitude. When you approach everything with that attitude, you are applying AI sensibility. Does that make sense? [20:35] Jensen Huang: 例如，我们正在与许多公司合作，其中图形分析、依赖关系、关系和依赖关系——你知道这些图形，它们有这么多边，这么多节点和边，数万亿个。在过去，你会处理一个图形，它的小片段。现在，只给我整个图形。它有多大？我不在乎。这种感知正在到处应用。如果速度根本不重要。你在光速。如果质量——你在零重量，零重力。如果你没有应用那种逻辑，如果过去对你来说非常困难的事情你说，”啊，没关系”——如果你没有应用那种逻辑，你做错了。现在想象你将那种逻辑、那种感知应用到你公司最困难的问题上。这就是你将如何推动指针。这就是他们所有人的想法。现在那些——如果你没有那样思考，只是——你所要做的就是——只是想象你的竞争对手那样思考。如果你没有那样思考，只是想象一个即将成立的公司那样思考。它改变了一切。所以我会去找你公司最有影响力的工作在哪里。对它应用无穷大。对它应用零。对它应用光速。然后问Chuck如何实现。[笑声] [20:35] Jensen Huang: For example, there are many companies that we’re working with where the graph analytics, the dependency, the relationships and dependencies— you know these graphs, they have so many edges, so many nodes and edges, trillions of them. Back in the old days, you would process a graph, small pieces of it. These days, just give me the whole graph. How big is it? I don’t care. That sensibility is being applied everywhere. If you’re not applying that sensibility, you’re doing it wrong. If speed matters, not at all. You’re at the speed of light. If mass— you’re at zero weight, zero gravity. If you’re not applying that logic, if something is not insanely hard to you in the past and you go, “Ah, doesn’t matter”— if you’re not applying that logic, you’re not doing it right. Now imagine you apply that logic, that sensibility to the hardest problems in your company. That’s how you’re going to move the needle. And that’s how they all think. Now the people who are— if you’re not thinking that way, just— all you have to do— just imagine your competitors thinking that way. If you’re not thinking that way, just imagine a company who is about to get founded is thinking that way. It changes everything. And so I would go find where are the most impactful work in your company. Apply infinity to it. Apply zero to it. Apply the speed of light to it. And then ask Chuck how to make that happen. [laughter] 从预录到生成：软件的范式转变 From Pre-recorded to Generative: The Paradigm Shift in Software [22:10] Chuck Robbins: 不，让我们谈谈如何实现。所以你有这个类比—— [22:10] Chuck Robbins: No, let’s talk about how to make that happen. So you have this analogy of— [22:10] Jensen Huang: 就给我打电话。 [22:10] Jensen Huang: Just call me. [22:10] Chuck Robbins: 我们会打给你。我们会一起做。 [22:10] Chuck Robbins: We’ll call you. We’ll do it together. [22:16] Jensen Huang: 我们会一起做。 [22:16] Jensen Huang: We’ll do it together. [22:16] Chuck Robbins: 你有这个类比——这个五层蛋糕——因为每个人都在谈论基础设施、模型、应用程序——我是说，我该如何着手？谈谈这一点。 [22:16] Chuck Robbins: You have this analogy— this five layer cake— because everybody’s talking about infrastructure, models, apps— I mean, how do I go about it? Talk about that a little bit. [22:24] Jensen Huang: 好吧，成功人士做的事情之一就是他们推理这里正在发生什么。所以大约15年前，一个算法能够——用两个工程师——解决一个计算机视觉问题。计算机视觉基本上是智能的第一部分——感知。智能是感知、推理、规划。感知——我是什么？正在发生什么？我的背景是什么？推理——我如何推理——我如何将其与我的目标进行比较？然后第三，想出一个计划来解决那个——来实现那个。所以——你知道，例如，战斗机问题——感知、定位，然后行动。所以智能是关于这三件事。没有感知，你不能有第二和第三部分。没有理解背景，你无法弄清楚该做什么。背景是高度多模态的。有时是PDF，有时是电子表格。有时是信息。有时只是感官和气味。我们在哪里？我们在这里做什么？谁是观众？等等。阅读房间。所以那是关于感知的。 [22:24] Jensen Huang: Well, one of the things that successful people do is they reason about what is happening here. So almost 15 years ago, an algorithm was able to— with two engineers— solve a computer vision problem. Computer vision is basically the first part of intelligence— perception. Intelligence is perception, reasoning, planning. Perception— what am I? What’s going on? What’s my context? Reasoning— how do I reason about— how do I compare this to my goals? And then three, come up with a plan to solve that— to achieve that. And so— you know, for example, the jet fighter problem— perception, localization, and then action. And so intelligence is about those three things. You can’t have the second and third part without perception. You can’t figure out what to do without understanding context. And context is highly multimodal. Sometimes it’s a PDF, sometimes it’s a spreadsheet. Sometimes it’s information. Sometimes just senses and smells. Where are we? What are we doing here? Who’s the audience? So on and so forth. Reading the room. And so that’s about perception. [28:20] Jensen Huang: 简单地说，Chuck所说的是我们来自一个一切都是预录的世界。Chuck工作的软件。 [28:20] Jensen Huang: Simplistically, what Chuck is saying is that we came from a world where everything was pre-recorded. The software that Chuck worked on. [28:36] Chuck Robbins: 真的很好的东西。 [28:36] Chuck Robbins: Really good stuff. [28:41] Jensen Huang: 它运行了很长时间。只是为了记录，它确实是用希伯来语描述的。[笑声] [28:41] Jensen Huang: It ran a very long time. Just for the record, it was indeed described in the Hebrew. [laughter] [28:57] Chuck Robbins: 这是真的。那是另一种技能。我是说，房间里唯一知道希伯来语COBOL的人。 [28:57] Chuck Robbins: That is true. That was another skill. I mean, the only person in the room that knows Hebrew COBOL. [29:05] Jensen Huang: [笑声] 总之——那是预录的。我们设计——我们描述我们的算法来描述我们的想法，然后我们放入与之一起的数据。一切都是预录的。过去软件是预录的原因是因为它装在CD-ROM中。不是吗？ [29:05] Jensen Huang: [laughter] Anyways— that was pre-recorded. We engineered— we described our algorithms to describe our thoughts and then we put data that goes along with it. Everything is pre-recorded. The reason why software in the past was pre-recorded is because it came in a CD-ROM. Isn’t that right? [29:24] Chuck Robbins: 是的。 [29:24] Chuck Robbins: Yes. [29:24] Jensen Huang: 它是预录的。好的。现在什么是软件？因为它是上下文的、动态的，每个上下文都不同，每次使用软件的每个人都不同，每个提示都不同，你给它的前导，你给它的先验，上下文都不同。软件的每个单一实例都不同，这就是为什么过去必要的计算量——这是预录的——称为基于检索。你所要做的就是检查自己。当你使用手机时，你触摸某物，它去并检索一些软件、一些文件、一些图像并将其带给你。在未来，一切都将是生成的，就像现在正在发生的一样。这次对话以前从未发生过。概念以前存在过。先验以前存在过，但这个序列中的每一个词以前从未发生过。原因显然是我们喝了四杯酒，COBOL和希伯来语从未从—— [29:24] Jensen Huang: It was pre-recorded. Okay. What is software now? Because it’s contextual, dynamic, and every context is different and every time everybody who uses the software is different and every prompt is different and the precursor you give it, the priors you give it, the context is different. Every single instance of the software is different, which is the reason why the amount of computation necessary in the past— which is pre-recorded— is called retrieval-based. All you have to do is check yourself. When you use your phone you touch something, it went and retrieved some software, some files, some images and brought it to you. In the future, everything is gonna be generative just like is happening right now. This conversation has never happened before. The concepts existed before. The priors existed before, but every single word in this sequence has never happened before. And the reason for that is obviously we’re four wines in, COBOL and Hebrew have never come out of the— [30:43] Chuck Robbins: 冷萃咖啡。是的。COBOL，希伯来语。不。谢天谢地这不是在校园或正在流媒体。 [30:43] Chuck Robbins: Cold brew. Yes. COBOL, Hebrew. No. Thank goodness this is not on campus or being streamed. [30:57] Jensen Huang: 是的。是的。好吧。让我们——你明白我在说什么吗？所以结果—— [30:57] Jensen Huang: Yeah. Yeah. All right. Let’s— Do you understand what I’m saying? And so as a result— [31:02] Chuck Robbins: 你明白你在说什么吗？[笑声] [31:02] Chuck Robbins: Do you understand what you’re saying? [laughter] [31:09] Jensen Huang: Chuck今天到目前为止喂我的唯一东西是四杯酒。 [31:09] Jensen Huang: The only thing that Chuck has fed me today so far is four glasses of wine. [31:14] Chuck Robbins: 公平地说，我只喂了你——我喂了你其中一杯。你从自助餐拿了另外三杯。 [31:14] Chuck Robbins: And to be fair, I only fed you— I fed you one of them. You took the other three off the buffet. [31:19] Jensen Huang: 我盯着食物看。我想，”我太饿了。我盯着食物看。”它永远离我大约40英尺。 [31:19] Jensen Huang: I was eyeing the food. I was like, “I’m so hungry. I’m eyeing the food.” It was forever about 40 feet away from me. [31:28] Chuck Robbins: 那是因为你在拍照。 [31:28] Chuck Robbins: It’s cuz you were taking photos. [31:33] Jensen Huang: 但它是——我想，它太近了。它太近了。[笑声] 我实际上有一次向食物倾斜，但我又被推回来了。[笑声] [31:33] Jensen Huang: But it was— I was like, it was so close. It was so close. [laughter] And I actually leaned towards the food one time, but I was pushed back again. [laughter] [31:39] Chuck Robbins: 你知道发生了什么吗？你的团队实际上提前告诉我们，如果你喝了三杯酒，他是最佳状态。如果你喝了第四杯，那将是不可思议的。这是次优的。 [31:39] Chuck Robbins: You know what happened? Your team actually told us ahead of time, if you get three glasses of wine in, he’s optimal. If you get the fourth one in, it’s going to be incredible. This is suboptimal. 应用技术：AI的最重要部分 Applying Technology: The Most Important Part of AI [31:57] Jensen Huang: 所以总之，总之，总之，听着，听着，听着，听着。那么什么是AI？ [31:57] Jensen Huang: So anyways, anyways, anyways, listen, listen, listen, listen. So what is AI? [32:09] Jensen Huang: 我们必须留下一些智慧。我们能再来一杯酒吗？这不只是Dave Chappelle的东西。 [32:09] Jensen Huang: We have to leave some wisdom behind. Can we get another glass of wine, please? This is not just Dave Chappelle stuff. [32:21] Chuck Robbins: 好的，让我们谈谈别的。让我们谈谈另一件事。 [32:21] Chuck Robbins: Okay, let’s talk about something. Let’s talk about one other thing. [32:21] Jensen Huang: 能源。芯片。 [32:21] Jensen Huang: Energy. Chips. [32:26] Chuck Robbins: 能源听起来不错。 [32:26] Chuck Robbins: Energy sounds good. [32:26] Jensen Huang: 能源、芯片、基础设施，包括硬件和软件。然后是AI模型。但AI最重要的部分是应用。每个国家、每个公司，下面的所有层都只是基础设施的东西。你需要做的是应用技术。看在上帝的份上，应用技术。使用AI的公司不会陷入危险。你不会因为AI失去工作。你会因为使用AI的人失去工作。所以，开始吧。这是最重要的事情。 [32:26] Jensen Huang: Energy, chips, infrastructure, both hardware and software. Then the AI model. But the most important part of AI is applications. Every single country, every single company, all that layer underneath is just infrastructural stuff. What you need to do is apply the technology. For God’s sakes, apply the technology. A company that uses AI will not be in peril. You’re not going to lose your job to AI. You’re going to lose your job to someone who uses AI. So, get to it. That’s the most important thing. [33:03] Chuck Robbins: 是的。 [33:03] Chuck Robbins: Yeah. [33:03] Jensen Huang: 并尽快打电话给Chuck。 [33:03] Jensen Huang: And call Chuck as soon as possible. [33:09] Chuck Robbins: 你打给我,我会打给他。是的。明白了。所以，我们没有很多时间，所以我不确定—— [33:09] Chuck Robbins: You call me, I’ll call him. Yeah. Got it. So, we don’t have a lot of time, so I’m not sure— [33:09] Jensen Huang: 我们有世界上所有的时间。是吗？ [33:09] Jensen Huang: We got all the time in the world. Do we? [33:15] Chuck Robbins: 多少？ [33:15] Chuck Robbins: How much? [33:15] Jensen Huang: 看，看，Chuck——Chuck，就像他跑。他按时间表建设。我甚至不戴手表。看那个。看那个。Chuck，我把你拿在这里。 [33:15] Jensen Huang: Look, look, Chuck— Chuck, like he runs. He builds on the clock. I don’t even wear a watch. Look at that. Look at that. Chuck, I got you right here. [33:28] Chuck Robbins: 是的。是的。我们做得很好。 [33:28] Chuck Robbins: Yeah. Yeah. We’re doing great. [33:28] Jensen Huang: 你按时间表向人们收费。 [33:28] Jensen Huang: You build people on the clock. [33:33] Chuck Robbins: 哦，是的。不是我。 [33:33] Chuck Robbins: Oh, yeah. Not me. [33:33] Jensen Huang: 在价值交付之前我不会离开。[掌声] [33:33] Jensen Huang: I’m not leaving until value’s delivered. [applause] [33:42] Chuck Robbins: 看，如果需要整晚，我不会——嘿，看，我要折磨你们所有人直到Jensen——这就是为什么像我这样的人需要手表。[笑声] 好吧。 [33:42] Chuck Robbins: See, if it takes all night, I’m not— Hey, look, I’m going to torture all of you until Jensen— That’s why guys like me need a watch. [laughter] All right. [33:54] Jensen Huang: 直到你能说你学到了什么，你将被困在这里。是的。 [33:54] Jensen Huang: Until you could say that you learned something, you are going to be trapped in here. Yeah. 物理AI：工具使用的未来 Physical AI: The Future of Tool Use [34:00] Chuck Robbins: 我们要折磨每个人直到价值被交付。我确实检查了——还有更多酒。嗯，你能给我们你对物理AI的第一想法吗？ [34:00] Chuck Robbins: We’re going to torture everybody until value is delivered. I did check— there is more wine. Um, can you just give us your top of mind on physical AI? [34:13] Jensen Huang: 记住什么是软件？软件是一个工具。有一种观念认为工具行业正在衰落并将被AI取代。你可以看出因为有一大堆软件公司的股价承受很大压力，因为不知何故AI将取代它们。这是世界上最不合逻辑的事情，时间会证明自己。让我们给自己终极思想实验。假设我们是终极AI——人工通用机器人。终极AI——我们的物理版本。你当然可以解决任何问题，因为你是类人的。你可以做事情。如果你是人类机器人，你会使用螺丝刀还是发明新的螺丝刀？我只会使用一个。你会使用锤子还是发明新锤子？你会使用电锯还是发明新电锯？首先，理想情况下他们根本不使用它。但你明白我在说什么吗？如果你是人类机器人，人工通用机器人，你会使用工具还是重新发明工具？答案显然是使用工具。 [34:13] Jensen Huang: Remember what software is? Software is a tool. There’s this notion that the tool industry is in decline and will be replaced by AI. You could tell because there’s a whole bunch of software companies whose stock prices are under a lot of pressure because somehow AI is going to replace them. It is the most illogical thing in the world and time will prove itself. Let’s give ourselves the ultimate thought experiment. Suppose we are the ultimate AI— artificial general robotics. The ultimate AI— the physical version of us. You could of course solve any problem because you’re humanoid. You could do things. If you were a human robot, would you use a screwdriver or invent a new screwdriver? I would just use one. Would you use a hammer or invent a new hammer? Would you use a chainsaw or invent a new chainsaw? First of all, ideally they don’t use it at all. But do you understand what I’m saying? If you were a human robot, artificial general robotics, would you use tools or reinvent tools? The answer obviously is to use tools. [35:36] Jensen Huang: 所以现在做数字版本。如果你是人工通用智能，你会使用像ServiceNow和SAP和Cadence和Synopsys这样的工具还是你会重新发明计算器？当然，你只会使用计算器。这就是为什么AI最新突破是什么？工具使用。因为工具被设计为明确的。我们世界中有许多问题，其中F等于MA。请你能不能不要想出另一个版本？[笑声] F=MA不是有点MA。它就是[清嗓子] MA。哦，V等于IR。它不是有点IR。不是大约IR，统计IR——它就是IR。你明白我在说什么吗？所以我认为我们希望人工通用机器人、人工通用智能使用工具。 [35:36] Jensen Huang: And so now do the digital version of that. If you were an artificial general intelligence, would you use the tools like ServiceNow and SAP and Cadence and Synopsys or would you reinvent a calculator? Of course, you would just use a calculator. That’s the reason why the latest breakthroughs in AI is what? Tool use. Because the tools are designed to be explicit. There are many problems in our world where F equals MA. Please could you please not come up with another version? [laughter] F=MA is not kind of MA. It’s just [clears throat] MA. Oh, V equals IR. It’s not kind of IR. Not approximately IR, statistically IR— it is IR. Do you understand what I’m saying? And so I think we want the artificial general robotics, artificial general intelligence to use tools. [36:47] Jensen Huang: 好吧，这就是大想法。我认为在下一代物理AI中，我们将拥有理解物理世界、理解因果关系的AI。如果我把这个翻倒，它会把所有那个翻倒。他们理解多米诺骨牌的概念。只是多米诺骨牌的概念——注意，一个孩子理解如果你把那个翻倒——多米诺骨牌的概念是极其深刻。因果关系、接触、重力、质量，所有这些都集成到多米诺骨牌中。翻倒多米诺骨牌。你可以有一个小小的多米诺骨牌，翻倒一个更大的多米诺骨牌，翻倒一个更大的多米诺骨牌，翻倒一个更大的多米诺骨牌，直到另一边有一吨——一个孩子对那个概念没有问题。大型语言模型将完全不知道。所以我们必须教导——我们必须创造一种新型的物理AI。 [36:47] Jensen Huang: Well, that’s the big idea. I think that in the next generation of physical AI, we’re going to have AIs that understand the physical world, understand causality. If I tip this over, it’s going to tip all of that over. They understand the concept of a domino. Just the concept of a domino— notice, a child understands if you tip that over— the concept of the domino is extremely— it’s like deeply profound. Causality, contact, gravity, mass, all of that is integrated into a domino. Tipping dominoes over. The idea that you could have a little tiny domino, tip a larger domino, tip a larger domino, tip a larger domino to the point where there’s a ton on the other side— a child has no trouble with that concept. A large language model will have no idea. And so we have to teach— we have to create a new type of physical AI. 从工具到劳动：百倍市场机会 From Tools to Labor: The 100x Market Opportunity [37:48] Jensen Huang: 好吧，机会是什么？到目前为止，Chuck和我所在的行业是关于创造工具的。我们一直在螺丝刀锤子行业。我们的整个生活都是关于创造螺丝刀和锤子的。这是历史上第一次，我们将创造人们所说的劳动力，增强劳动。给你一个例子。什么是自动驾驶汽车？什么是数字司机？数字司机的价值是多少？很多。比汽车多得多。原因是因为在数字司机的生命周期中，数字司机的经济学比汽车多得多。 [37:48] Jensen Huang: Well, what’s the opportunity? So far, the industry that Chuck and I have been part of is about creating tools. We have been in the screwdriver hammer business. Our entire life has been about creating screwdrivers and hammers. For the first time in history, we are going to create what people call labor, but augmented labor. Give you an example. What is a self-driving car? What’s a digital chauffeur? What’s a digital chauffeur valued at? A lot. A lot more than the car. And the reason for that is because in the lifetime of the digital chauffeur, the economics of the digital chauffeur is a lot more than the car. [38:33] Jensen Huang: 这是第一次，我们暴露于大100倍的TAM。字面上在数学上是真的。IT行业大约是一万亿美元，对吧？或者差不多正负几个。然而世界经济大约是一百万亿美元。这是第一次，我们将暴露于所有那个（数字）。所以情况是你们所有人——今天这个房间里的每个人——你们有机会应用这项技术成为一家技术公司。 [38:33] Jensen Huang: For the very first time, we are exposed to a TAM that is 100 times larger. Literally mathematically true. The IT industry is about a trillion dollars, right? Or so, plus or minus a couple. And yet the economy of the world is about a hundred trillion dollars. For the very first time, we’re going to be exposed to all of that. So it is the case that all of you— everybody in this room today— you have the opportunity to apply this technology to become a technology company. [39:18] Jensen Huang: 让我给你一些例子。我真的相信——尽管我——看，我爱迪士尼，我喜欢与迪士尼合作。我很确定他们宁愿成为Netflix。我爱梅赛德斯。我坐梅赛德斯来的。我确信他们宁愿成为特斯拉。我爱沃尔玛。我确信他们宁愿成为亚马逊。你们到目前为止同意吗？我是三中三吗？你们所有人都是那样。 [39:18] Jensen Huang: Let me give you some examples. I really believe— as much as I— look, I love Disney and I love working with Disney. I’m pretty sure they’d rather be Netflix. I love Mercedes. I came in a Mercedes. I am certain they’d rather be Tesla. I love Walmart. I am certain they’d rather be Amazon. Do you guys agree so far? Am I three for three? All of you are that way. [39:50] Jensen Huang: 我相信我们有机会帮助将每一家公司转变为技术公司。技术第一。技术是你的超能力，领域是你的应用，而不是相反——领域是你是谁，你在寻求技术。原因是因为技术优先的公司，你在处理电子，而不是原子。电子，有更多的它们。原子，你受质量限制，这就是为什么当他们从CD-ROM转到电子时，公司的价值爆炸了一千倍。你需要像我们一样，成为电子公司、电子公司，这是说技术公司的另一种方式。所以我认为你的机会在这里。 [39:50] Jensen Huang: I believe that we have an opportunity to help transform every single company into a technology company. Technology first. Technology is your superpower and the domain is your application, versus the other way— which is the domain is who you are and you’re seeking for technology. And the reason that’s so is because companies who are technology first, you’re dealing with electrons, not atoms. And electrons, there’s a lot more of them. Atoms, you’re limited by mass, which is the reason why the moment they went from CD-ROMs to electrons, the value of the company exploded by a thousand times. You need to be like us, an electronics company, electron company, which is another way of saying a technology company. And so I think the opportunity for you is here. 隐式编程：领域专业知识的力量 Implicit Programming: The Power of Domain Expertise [41:05] Jensen Huang: 思考这一点的另一种方式是AI——我们刚才说过。甚至Chuck，他只知道如何用希伯来语编程， [41:05] Jensen Huang: Another way to think about that is AI— and we just said it earlier. Even Chuck, who only knows how to program in Hebrew, [41:12] Chuck Robbins: [笑声] 这是一种天赋。 [41:12] Chuck Robbins: [laughter] It’s a gift. [41:18] Jensen Huang: 他的工具选择是从右到左。[笑声] [41:18] Jensen Huang: His instrument choice is a right to left. [laughter] [41:32] Chuck Robbins: 因为你知道否则会弄脏。这实际上相当聪明。 [41:32] Chuck Robbins: Because as you know it smears otherwise. It is pretty smart actually. [41:38] Jensen Huang: 聪明的人做聪明的事情。所以美好的事情是，你知道世界的编程语言——对于你们所有的公司，你们有点觉得，哦我的天，软件不是我们的强项。但知识、直觉、领域专业知识是你的强项。好吧，你现在第一次可以用你的语言向计算机准确解释你想要什么。你记得我们从哪里开始——从显式编程到隐式编程？这是历史上第一次，你可以隐式地编程计算机。只要告诉它你想要什么。告诉它你的意思，计算机会写代码，因为事实证明编码只是打字。事实证明打字是一种商品。这就是你的大机会。你们所有人都可以被提升到你以前受限的原子限制之上。你们所有人都可以逃脱这个限制——我们没有足够的软件工程师——因为事实证明打字是一种商品。你们所有人都有非常有价值的东西，那就是领域专业知识——理解客户，理解问题。这就是最终价值。这就是最终价值——理解意图。 [41:38] Jensen Huang: Smart people do smart things. And so the beautiful thing is that as you know the programming language of the world— and for all of your companies you kind of feel like, oh my gosh, software is not our strength. But knowledge, intuition, domain expertise is your strength. Well, you now for the first time can explain exactly what you want to a computer in your language. Do you remember where we started— from explicit programming to implicit programming? For first time in history, you could program a computer implicitly. Just tell it what you want. Tell it what you mean and the computer will write the code because coding as it turns out is just typing. And typing as it turns out is a commodity. And that’s the great opportunity for you. All of you could be levitated above the atomic limitations that you were limited by before. All of you could escape from this limitation— we don’t have enough software engineers— because as it turns out typing is a commodity. And all of you have something of great value which is domain expertise— to understand the customer, understand the problem. And that is the ultimate value. That is the ultimate value— to understand the intent. [43:04] Jensen Huang: 你知道，当你从大学毕业时，你可以是一个超级程序员，但你不知道客户想要什么。你不知道要解决什么问题。但这就是你们所有人知道的。你知道客户想要什么。你知道要解决什么问题。编码部分很容易。只要告诉AI去做。所以这就是你的超能力。所以Chuck和我在这里帮助你做到这一点。那个结束语是在我喝了五杯酒的情况下完成的。[笑声] [43:04] Jensen Huang: You know, when you graduate from college, you could be a super programmer, but you have no idea what customers want. You have no idea what problems to solve. But that’s what all of you know. You know what customers want. You know what problems to solve. The coding part of it is easy. Just tell the AI to do it. And so that’s your superpower. So Chuck and I are here to enable you to do that. That closing was done with five glasses of wine in me. [laughter] [43:49] Chuck Robbins: 所以，嘿，听着——这确实是一个奇迹——这是一个在桌子上工作的人——人工智能的真实代表。[掌声] [43:49] Chuck Robbins: So, hey, listen— it’s a miracle indeed— this is somebody who works off— a table— true representation of artificial intelligence. [applause] [44:02] Jensen Huang: 也许那是增强的。我只想告诉你，与你们所有人合作是一种巨大的乐趣。正如你所知，Cisco在计算重塑的两个非常重要的支柱上拥有极端的专业知识。没有Cisco，就没有现代计算。其中一个当然是网络，另一个是安全。这两个支柱都在AI世界中被重塑了。我们非常了解的部分——即计算部分——在很多方面是一种商品，Cisco知道的东西是深刻有价值的。在我们两个之间，我们将很高兴帮助你们所有人参与AI世界。 [44:02] Jensen Huang: Maybe that’s enhanced. I just want to tell you that it’s a great pleasure working with all of you. Cisco as you know has extreme expertise and two very important pillars of the reinvention of computing. Without Cisco, there is no modern computing. One of them is of course networking and the other one’s security. And both of those pillars have been reinvented in the world of AI. And the part that we know very well— which is the computing part of it— in a lot of ways is a commodity, and the stuff that Cisco knows is deeply valuable. And between the two of us, we’ll be delighted to help all of you engage the world of AI. 建造你自己的AI：问题比答案更宝贵 Build Your Own AI: Questions Are More Valuable Than Answers [44:51] Jensen Huang: 然后有人早些时候问我——我认为值得重复。有人早些时候问我，你应该只是租用云还是你应该甚至努力建造你自己的计算机？这是我会告诉你的。我会建议你做我建议我的孩子做的完全相同的事情。建造一台计算机。即使PC无处不在，即使它成熟了，即使技术发展了，看在上帝的份上，建造一个。知道为什么所有组件都存在。如果你要进入汽车、汽车行业、运输行业的世界，不要只使用Uber。看在上帝的份上，抬起引擎盖，换机油，理解所有组件。看在上帝的份上，理解它是如何工作的。这至关重要。这项技术对未来如此重要。你必须对它有一些触觉理解。抬起引擎盖，换机油，建造一些东西。不必很大。建造一些东西。 [44:51] Jensen Huang: And then somebody asked me earlier— I think it’s worth repeating. Somebody asked me earlier, should you just rent the cloud or should you even make the effort to build your own computer? Here’s what I would tell you. I would advise you to do exactly the same thing I advise my children. Build a computer. Even though the PC is everywhere, even though it’s mature, even though technology is developed, for God’s sakes, build one. Know why all the components exist. If you were to be in the world of automotive, the automobile industry, the transportation industry, don’t just use Uber. For God’s sakes, lift the hood, change the oil, understand all the components. For God’s sakes, understand how it works. It is vital. This technology is so important to the future. You must have some tactile understanding of it. Lift the hood, change the oil, build something. Doesn’t have to be large. Build something. [46:05] Jensen Huang: 你可能发现你实际上非常擅长它。你可能发现你需要那种技能。你可能发现世界不是全部租用与全部拥有——你想要租用一些并拥有一些，因为你公司的某些部分应该建立在本地。例如，主权和专有信息——只是，你不愿意与每个人分享你的问题。你知道，当你去看治疗师时，你不想让问题在线。[笑声] 你知道我在说什么吗？ [46:05] Jensen Huang: You might discover you’re actually insanely good at it. You might discover that you need that skill. You might discover that the world is not about all rent versus all own— that you want to rent some and own some because some part of your company should be built on prem. For example, sovereignty and proprietary information— and just, you’re not comfortable sharing your questions with everybody. You know, when you go see a therapist, you don’t want the questions to be online. [laughter] You know what I’m saying? [46:52] Chuck Robbins: 好的，我只是——我在想象这个。[笑声] 假设地。 [46:52] Chuck Robbins: Okay, I’m just— I’m imagining this one. [laughter] Hypothetically. [46:57] Jensen Huang: 所以，假设地，我认为你有很多问题，你有很多对话，很多对话，很多不确定性应该保持私密。公司也是一样。我不自信。我对将Nvidia的所有对话放在云中不安全，这就是为什么我们在本地建造它。我们在本地建造了一个超级AI系统，因为我只是不自信分享那个对话。因为事实证明，对我来说最有价值的IP不是我的答案。是我的问题。你跟上我了吗？我的问题是对我来说最有价值的IP。我在思考的是我的问题。答案是一种商品。如果我只是知道要问什么——我在识别什么是重要的。我不想让人们知道我认为什么是重要的。我希望那在一个小房间里。我希望那在本地。我希望那是我自己。我想创造我自己的AI。 [46:57] Jensen Huang: And so, hypothetically, I think that a lot of questions you have, a lot of conversations you have, a lot of dialogue, a lot of uncertainties you have ought to be kept private. Companies are the same way. I am not confident. I am not secure about putting all of Nvidia’s conversations in the cloud, which is the reason why we built it locally. We’ve built a super AI system locally because I’m just not confident to share that conversation. Because as it turns out, the most valuable IP to me is not my answers. It’s my questions. Are you following me? My questions are the most valuable IP to me. What I’m thinking about are my questions. The answers are a commodity. If I simply knew what to ask— I’m identifying what’s important. And I don’t want people to know what I think is important. And I want that to be in a small room. I want that to be on prem. I want that to be by myself. And I want to create my own AI. AI在环中：未来公司的知识产权 AI in the Loop: The Intellectual Property of Future Companies [48:08] Jensen Huang: 然后最后一个想法，因为已经11点了。[笑声] 最后一个想法。有一个想法认为AI应该总是有人在环中。这正好是错误的想法。它是向后的。每个公司都应该有AI在环中。原因是因为我们希望我们的公司每天都变得更好、更有价值、更有知识。我们永远不想倒退。我们永远不想变平。我们永远不想从头开始。这意味着如果我们有AI在环中，它将捕获我们的生活经验。未来每个员工都将有AI，很多AI在环中。这些AI将成为公司的知识产权。这就是未来公司。因此，我认为你们所有人立即打电话给Chuck是明智的。 [48:08] Jensen Huang: And then one last thought since it’s already 11 o’clock. [laughter] One last thought. There was an idea that AI should always have human in the loop. It’s exactly the wrong idea. It’s backwards. Every company should have AI in the loop. And the reason for that is because we want our company to be better and more valuable and more knowledgeable every single day. We never want to go backwards. We never want to go flat. We never want to start from the beginning. Which means that if we have AI in the loop, it will capture our life experience. Every single employee in the future will have AI, lots of AIs in the loop. And those AIs will become the company’s intellectual property. That’s the future company. And therefore, I think it sensible for all of you to call Chuck immediately. [49:13] Chuck Robbins: 我打给了Jensen。[笑声] [49:13] Chuck Robbins: I called Jensen. [laughter] [49:20] Jensen Huang: 总之，这是我的结束语。 [49:20] Jensen Huang: Anyhow, that’s my closing. [49:25] Chuck Robbins: 听着——在路上两周。Jensen飞到这里，在他第一次长时间睡在自己床上之前，与我们度过了他的最后一晚、最后一个晚上。我们永远感激。感谢你来到这里。谢谢。 [49:25] Chuck Robbins: Listen— two weeks on the road. Jensen flew here, spent his last night, last evening with us before he gets to sleep in his bed for the first time in a long time. We’re forever grateful. Appreciate you being here. Thank you. [49:36] Jensen Huang: 非常感谢。而且——[掌声] 谢谢，伙计。而且——[掌声] 从我的眼角，有所有这些串烧。有人还在那里。Fritos的袋子在哪里？[笑声] [49:36] Jensen Huang: Thank you very much. And— [applause] Thank you, man. And— [applause] from the corner of my eye, there were all these skewers. Somebody was still there. Where’s the bag of Fritos? [laughter] [49:58] Chuck Robbins: 好吧，我们走吧。谢谢。谢谢大家。 [49:58] Chuck Robbins: All right, let’s go. Thank you. Thank you, everybody.

超级智能、顺从型AI、超人类主义、基督与超越：PayPal、Palantir联合创始人彼得·蒂尔2025年6月播客实录 | 中英文完整版精译（下篇）

Tue, 03 Feb 2026 00:00:00 +0000

书童按：本篇是彼得·蒂尔（Peter Thiel）于2025年6月接受罗斯·多塔特（Ross Douthat）”有趣时代”（Interesting Times）播客采访实录。蒂尔是PayPal和Palantir的联合创始人，硅谷传奇投资人，唐纳德·特朗普和J.D.万斯政治生涯的早期资助者，亦是当代保守派知识分子中强调”反对共识”的极具影响力人物。本部分涉及超级智能、顺从型AI风险、超人类主义、基督教与超越、人类自由与神圣决定论，以及人类行动的空间等深刻议题。初稿采用Claude Opus4.5 API全篇翻译、中英混排、审阅修改，书童仅读了一遍并做简单批注，分上下篇两个部分发布，本篇为下篇，以飨诸君。 乐观主义的困境 The Optimism Problem 彼得·蒂尔 你把这件事描绘得太乐观了。我嘛，还有那些人……（书童注：指上篇中罗斯所说”特朗普主义和民粹主义可以成为技术创新和经济活力的载体”） PETER THIEL You’re framing it really, really optimistically here. So I, well, the people. 罗斯·多塔特 但我知道你是悲观的。 ROSS DOUTHAT But I think, I know you’re pessimistic. 彼得·蒂尔 当你把事情往乐观了说，其实就是在暗示这些人注定要失望、注定要失败，诸如此类。 PETER THIEL When you frame it optimistically. You’re just saying these people are going to be disappointed and they’re just set up for failure and things like that. 罗斯·多塔特 我只是说，人们确实表达了很多乐观情绪。埃隆·马斯克就很乐观——当然，他也流露出一些末日焦虑，比如预算赤字将如何毁灭我们所有人。但他进入了政府，他身边的人也进入了政府，他们的态度基本上是：我们和特朗普政府建立了合作关系，我们要追求技术上的伟大复兴。 ROSS DOUTHAT I mean, people expressed a lot of optimism. That’s all I’m saying. Elon Musk expressed a lot of, I mean, he expressed some apocalyptic anxieties about how budget deficits were going to kill us all. But he came into government, and people around him came into government basically saying, we have a partnership with the Trump administration and we’re pursuing technological greatness. 罗斯·多塔特 我认为他们是乐观的。而你的立场更加悲观，或者说更加现实。所以我想听的是你对现状的评估，而不是他们的看法。你觉得，特朗普2.0时代的民粹主义，看起来像是能够承载技术活力的载体吗？ ROSS DOUTHAT I think they were optimistic. And so you’re coming from a place of greater pessimism or realism. So what I’m asking for is your assessment of where we are, not their assessment. But do you think, does populism in Trump 2.0 look like a vehicle for technological dynamism to you? 彼得·蒂尔 它仍然是目前最好的选择，而且是遥遥领先的那种。哈佛难道能靠继续因循守旧、做着五十年来毫无成效的同样事情来攻克痴呆症吗？ PETER THIEL It’s still by far the best option we have. I don’t think. I don’t know. Is Harvard going to cure dementia by just puttering along, doing the same thing that hasn’t worked for 50 years? 罗斯·多塔特 这不过是”反正不会更糟，不如放手一搏”的论调。但目前对民粹主义的批评恰恰是：硅谷与民粹主义者结成了联盟，可到头来，民众并不真正关心科学，他们不想在科学上花钱，他们想砍掉哈佛的经费，仅仅因为他们不喜欢哈佛。最终，你得不到硅谷想要的那种对未来的投资。这种批评是错的吗？ ROSS DOUTHAT That’s just a case for. It can’t get worse. Let’s do disruption. Right, but the critique of populism right now would be Silicon Valley made an alliance with the populists. But in the end, the populace don’t care about science. They don’t want to spend money on science. They want to kill funding to Harvard just because they don’t like Harvard. Right. And in the end, you’re not going to get the kind of investments in the future that Silicon Valley wanted. Is that wrong? 科学的问题 The Science Problem 彼得·蒂尔 是的，但我们必须回到这个问题：科学在幕后究竟运转得怎么样？新政派们，不管他们有什么毛病，至少大力推动了科学——你给它拨款，你给科学家钱，你把规模做大。 PETER THIEL Yeah, but it, we have to go back to this question of, you know, how well is this, is the science working in the background? This is where, you know, the New Dealers, whatever was wrong with them, you know, they pushed science hard and you funded it and you gave money to people and you scaled it. 彼得·蒂尔 然而今天，如果有一个当代的爱因斯坦给白宫写信，那封信会在收发室石沉大海。曼哈顿计划在今天是不可想象的。当我们把某件事称为”登月计划”——就像拜登谈论癌症研究那样——在六十年代，”登月计划”意味着你真的会登上月球。而现在，”登月计划”意味着某种完全虚幻、永远不会实现的东西。”哦，这事儿得搞个登月计划才行”——这话的意思不是说我们需要一个阿波罗计划，而是说这事永远、永远不会发生。 PETER THIEL And, you know, whereas today, if there was an equivalent of Einstein and he wrote a letter to the White House, it would get lost in the mailroom. And the Manhattan Project is unthinkable, you know, if we call something a moonshot, the way this is the way Biden talked about, let’s say, cancer research, a moonshot in the 60s still meant that you went to the moon. A moonshot now means something completely fictional that’s never going to happen. Oh, you need a moonshot for that. It’s not like we need an Apollo program. It means it’s never, ever going to happen. 罗斯·多塔特 那么看起来你仍然处于这样一种模式——和硅谷其他一些人可能不同——对你来说，民粹主义的价值在于撕开面纱、戳破幻象。我们未必处于那个阶段，指望特朗普政府去建设新事物、去搞曼哈顿计划、去实现登月计划。更像是：民粹主义帮助我们看清，一切都是假的。 ROSS DOUTHAT And so, but it seems like then you’re still in the mode of, for you, as opposed to maybe for some other people in Silicon Valley. The value of populism is in tearing away the veils and illusions. And we’re not necessarily in the stage where you’re looking to the Trump administration to, to build the new, to do the Manhattan Project, to do the moonshot. It’s more like populism helps us see that it was all fake. 彼得·蒂尔 两者都要尝试，而且它们彼此紧密交织。核电正在去监管化，总有一天我们会重新开始建造新的核电站，或者设计更好的核电站，甚至是聚变反应堆。 PETER THIEL You need to try to do both. And they are very entangled with each other. And I don’t know, there’s deregulation of nuclear power and at some point, at some point we’ll get back to building, you know, new nuclear power plants or better designed ones or maybe even fusion reactors. 彼得·蒂尔 所以，是的，首先有一个去监管化的解构阶段，然后在某个时候你才真正开始建设。从某种意义上说，你是在清理战场。 PETER THIEL And so, yes, there’s a deregulatory deconstructive part. And then at some point you actually get to construction and it’s all things like that. So, yeah, in some ways you’re clearing the field. 政治的毒性 The Toxicity of Politics 罗斯·多塔特 然后，但你个人已经停止资助政客了。 ROSS DOUTHAT And then, but you’ve personally stopped funding politicians. 彼得·蒂尔 我在这件事上很矛盾。我认为它非常重要，但同时也极具毒性。所以我来回摇摆。 PETER THIEL I am schizophrenic on this stuff. You know, I think it is incredibly important and it’s incredibly toxic. And so I go back and forth on it. 罗斯·多塔特 对你个人来说极具毒性？ ROSS DOUTHAT Incredibly toxic for you personally? 彼得·蒂尔 对每个人都是，每个卷入其中的人都是。这是一个零和游戏，令人抓狂，而且在某种程度上，因为每个人—— PETER THIEL For everybody, everybody who gets involved. It’s zero sum. It’s crazy, you know, and then it’s, and then in some ways, because everyone. 罗斯·多塔特 恨你，把你和特朗普绑在一起。具体来说，对你个人而言，毒性体现在哪里？ ROSS DOUTHAT Hates you and associates you with Trump. Like, what, how is it toxic for you personally? 彼得·蒂尔 它有毒是因为那是一个零和世界。你能感受到其中的利害关系真的、真的很强，然后你—— PETER THIEL It’s toxic because it’s in a zero sum world. You know, the stakes in it feel really, really high and you. 罗斯·多塔特 最终树了很多以前没有的敌人。 ROSS DOUTHAT End up having enemies you didn’t have before. 火星的政治维度 The Political Dimension of Mars 彼得·蒂尔 是的，它对所有以不同方式卷入其中的人都有毒。”回到未来”有一个政治维度。这是我2024年与埃隆的一次对话。 PETER THIEL Yeah, it’s toxic for all the people who get involved in different ways. There is a political dimension of getting back to the future. You can’t, you know, I don’t know. This is a conversation I had with Elon back in, you know, 2024, and we had all these, you know, conversations. 彼得·蒂尔 我跟埃隆聊过”海上家园”这个话题。我说，如果特朗普没赢，我就想离开这个国家。埃隆说：”没地方可去。没地方可去。这里是唯一的选择。”你总是事后才想到该怎么反驳。 PETER THIEL I had this, I had the seasteading version with Elon where I said, you know, if Trump doesn’t win. I want to just leave the country. And then Elon said, “There’s nowhere to go. There’s nowhere to go. This is the only.” And then, you know, you always think of the right arguments to make later. 彼得·蒂尔 那是我们共进晚餐大约两小时后，我回到家才想到：”哇，埃隆，你已经不相信去火星这件事了。”2024年，这是埃隆停止相信火星的一年——不是把火星当作一个纯粹的科技项目，而是当作一个政治项目。火星本应是一个政治项目，是在建设一种替代方案。 PETER THIEL And it was about two hours after we had dinner and I was home that I thought of, “Wow, Elon, you don’t believe in going to Mars anymore.” 2024. 2024 is the year where Elon stopped believing in Mars. Not as a silly science tech project, but as a political project. Mars was supposed to be a political project, was building an alternative. 彼得·蒂尔 2024年，埃隆开始相信，如果你去了火星，社会主义的美国政府、觉醒的AI，都会跟着你到火星去。这要从我们牵线促成的德米斯（书童注：即德米斯·哈萨比斯，Deepmind创始人，AlphaGo、AlphaFold和Gemini的灵魂人物，诺贝尔化学奖得主）与埃隆那次会面说起。当时德米斯在做DeepMind。 PETER THIEL And in 2024, Elon came to believe that if you went to Mars, you know, the socialist US Government, the Woke AI, it would follow you to Mars. It was the Demis meeting with Elon that we sort of brokered. He was doing DeepMind. 罗斯·多塔特 那是一家AI公司。 ROSS DOUTHAT This is an AI company. 彼得·蒂尔 是的。大致的对话是这样的：德米斯告诉埃隆，”我正在做世界上最重要的项目，我在建造超人类AI。”然后埃隆回应：”好吧，我也在做世界上最重要的项目。我正在把人类变成一个跨行星物种。” PETER THIEL Yeah. This was the rough conversation was, you know, Demis tells Elon, “I’m working on the most important project in the world. I’m building a superhuman AI.” And Elon responds to Demis, “Well, I’m working on the most important project in the world. I am turning this into an interplanetary species.” 彼得·蒂尔 然后德米斯说：”好吧，可我的AI能够跟着你到火星去。”然后埃隆沉默了。但在我对历史的叙述中，德米斯这句话用了好几年才真正触动埃隆。他直到2024年才消化这件事。 PETER THIEL And then Demis said, “Well, you know, my AI will be able to follow you to Mars.” And then Elon sort of went quiet. But in my telling of the history, it took years for that to really hit Elon. It took him till 2024 to process it. 罗斯·多塔特 但这并不意味着他不相信火星。这只是意味着他认定，必须先赢得关于预算赤字和觉醒主义的战斗，才能到达火星。 ROSS DOUTHAT But that doesn’t mean he doesn’t believe in Mars. It just means that he decided he had to win some kind of battle over budget deficits for wokeness to get to Mars. 彼得·蒂尔 火星意味着什么？它是……是的……而且，再说一遍，它是…… PETER THIEL What does Mars mean? Is it a. Yeah. Is it? And again, it’s. 罗斯·多塔特 火星意味着什么？ ROSS DOUTHAT What does Mars mean? 彼得·蒂尔 它曾经是……它曾经是……它只是一个科学项目，还是……像一个…… PETER THIEL Well, it was. It was. It’s. Is it. Is it just. Is it just a scientific project or is it. I don’t know, is it like a. 罗斯·多塔特 一个新社会的愿景？ ROSS DOUTHAT I don’t know, a vision of a new society? 彼得·蒂尔 是的，海因莱因（书童注：美国硬科幻小说作家，与阿西莫夫与阿瑟·克拉克并成为科幻小说三巨头）式的愿景，成千上万的人生活在天堂里，都是埃隆·马斯克的后代。我不确定他是否具体化到了那种程度，但如果你把事情具体化，也许你就会意识到，火星本应不只是一个科学项目，它本应是一个政治项目。 PETER THIEL Yeah, Heinlein, you know, populated by many, many people in paradise, descendants from Elon Musk. Well, I don’t know if it was concretized that. That specifically, but if you concretize things, then maybe you realize that Mars is supposed to be more than a science project. It’s supposed to be a political project. 彼得·蒂尔 而当你把它具体化时，你就必须开始思考：好吧，觉醒的AI会跟着你去，社会主义政府也会跟着你去，那么也许你必须做一些不仅仅是去火星的事情。 PETER THIEL And then when you concretize it, you have to start thinking through, well, the AI, the woke AI will follow you, the socialist government will follow you, and then maybe you have to do something other than just going to Mars. 人工智能：进步还是停滞？ AI: Progress or Stagnation? 罗斯·多塔特 好的，说到”觉醒的AI”——如果我们仍处于停滞中的话——人工智能似乎是停滞的最大例外。这是一个取得了显著进步的领域，令许多人惊讶的进步。 ROSS DOUTHAT Okay, so the Woke AI, artificial intelligence seems like one. If we’re still stagnant. It’s the biggest exception to stagnation. It’s the place where there’s been remarkable progress, surprising to many people, progress. 罗斯·多塔特 我们刚才在谈政治。这也是特朗普政府在很大程度上满足AI投资者诉求的领域，无论是政府后退一步还是开展公私合作。所以这是一个进步——与政府参与并存的领域。而你是AI的投资者。你认为你在投资什么？ ROSS DOUTHAT It’s also the place, we were just talking about politics. It’s the place where the Trump administration is, I think, to a large degree, giving AI investors a lot of what they wanted in terms of both stepping back and doing public private partnerships. So it’s a zone of progress and governmental engagement. And you are an investor in AI. What do you think you’re investing in? 彼得·蒂尔 这个问题有很多层次。有一个问题我们可以提出来：我认为AI有多大？我的笨答案是：介于两者之间。它不只是一个噱头，但也不至于全面改变我们的社会。 PETER THIEL Well, I don’t know. There’s sort of a lot of layers to this. So I do think, I know there’s one question we can frame is just how big, how big a thing do I think AI is? And I don’t know. My stupid answer is it’s somewhere. It’s more than a nothing burger, and it’s less than the total transformation of our society. 彼得·蒂尔 我的估计是，它大致相当于九十年代末互联网的规模。我不确定它是否足以真正终结停滞，但也许足以催生一些伟大的公司。互联网可能让GDP增加了几个百分点，也许每年为GDP增长贡献1%，持续了十到十五年，对生产率有所贡献。这大致是我对AI的预期。 PETER THIEL So my placeholder is that it’s roughly on the scale of the Internet in the late 90s, which is, you know, I’m not sure it’s enough to, to really end the stagnation. It might be enough to create some great companies. And, you know, the Internet added maybe a few points, percentage points to the GDP, maybe 1% to GDP growth every year for 10, 15 years. It adds some to productivity. And so that’s sort of roughly my placeholder for AI. 彼得·蒂尔 AI是我们唯一拥有的东西。进步如此失衡，有点不太健康——这是我们唯一拥有的。我希望有更多维度的进步。我希望我们正在飞往火星。我希望我们正在攻克痴呆症。如果我们只有AI，我也接受。但它确实有风险，这项技术显然有危险。 PETER THIEL It’s the only thing we have. It’s, it’s a little bit unhealthy that it’s so unbalanced. This is the only thing we have. I’d like to have more multidimensional progress. I’d like us to be going to Mars. I’d like us to be having cures for dementia. If all we have is AI, I will take it. There are risks with it. There are, obviously, there are dangers with this technology. 超级智能问题 The Superintelligence Question 罗斯·多塔特 那么你是否对所谓的”超级智能级联理论”持怀疑态度？这种理论大致是说：如果AI成功了，它会变得极其聪明，以至于能在物质世界为我们带来进步——就是说，好吧，我们人类无法攻克痴呆症，无法弄清楚如何建造完美的工厂来制造飞往火星的火箭，但AI可以。 ROSS DOUTHAT But then you are a skeptic of the, what you might call the sort of superintelligence cascade theory, which basically says that if AI succeeds, it gets so smart that it gives us the progress in the world of atoms, that it’s like, all right, we can’t cure dementia. We can’t figure out how to build the perfect factory that builds the rockets that go to Mars, but the AI can. 罗斯·多塔特 在某个时刻，你跨过某个阈值，它不仅带来更多的数字进步，还带来其他六十四种形式的进步。听起来你不相信这一点，或者说你认为这不太可能。 ROSS DOUTHAT And at a certain point, it just, you pass a certain threshold and it gives us not just more digital progress, but 64 other forms of progress. It sounds like you don’t believe that, or you think that’s less likely. 彼得·蒂尔 是的，我不确定智力是否真的是那个”门控因素”。 PETER THIEL Yeah, I, I, I somehow don’t know if that’s been really the gating factor. 罗斯·多塔特 “门控因素”是什么意思？ ROSS DOUTHAT What does that mean, the gating factor? 彼得·蒂尔 这可能是一种硅谷意识形态——也许以一种奇怪的方式，它更偏自由派而非保守派——但硅谷的人真的非常执着于智商，认为一切都与聪明人有关。如果你有更多聪明人，他们就会做出伟大的事情。 PETER THIEL It’s probably a Silicon Valley ideology and maybe, maybe in a weird way it’s more liberal than a conservative thing, but people are really fixated on IQ in Silicon Valley and that it’s all about smart people. And if you have more smart people, they’ll do great things. 彼得·蒂尔 而经济学中反智商的论点是：人们实际上越聪明，表现越差。他们不知道如何运用自己的聪明才智，或者我们的社会不知道如何用好他们，他们格格不入。这表明门控因素不是智商，而是我们社会深处存在的某些问题。 PETER THIEL And then the economics anti IQ argument is that people actually do worse. The smarter they are, the worse they do. And they, you know, it’s just, they don’t know how to apply it, or our society doesn’t know what to do with them and they don’t fit in. And so that suggests that the gating factor isn’t IQ, but something, you know, that’s deeply wrong with our society. 罗斯·多塔特 那这是智力本身的局限，还是人类超级智能所造就的那种人格类型的问题？我非常认同这种看法。当我在这个播客中与一位AI加速主义者对谈时，我就提出过这个观点：认为某些问题只要提高智力就能解决，这种想法是有问题的。 ROSS DOUTHAT So is that a limit on intelligence or a problem of the sort of personality types human superintelligence creates? I mean, I’m very sympathetic to the idea and I made this case when I did an episode of this, of this podcast with a sort of AI accelerationist that just throwing, that certain problems can just be solved if you ramp up intelligence. 罗斯·多塔特 就好像是：我们提高智力，然后砰，阿尔茨海默病解决了；我们提高智力，AI就能弄清楚一夜之间为你建造十亿个机器人的自动化流程。我是一个”智力怀疑论者”——我认为智力可能存在局限性。 ROSS DOUTHAT It’s like, we ramp up intelligence and boom, Alzheimer’s is solved. We ramp up intelligence and the AI can, you know, figure out the automation process that builds you a billion robots overnight. I, I’m an intelligent skeptic in the sense I don’t think, yeah, I think you probably have limits. 彼得·蒂尔 这很难证明，无论从哪个角度看都很难证明。 PETER THIEL It’s, it’s, it’s hard to prove one way or it’s always hard to prove these things. 罗斯·多塔特 但我，在我们拥有超级—— ROSS DOUTHAT But I, until we have the super. 彼得·蒂尔 智能之前，我赞同你的直觉，因为我认为我们已经有过很多聪明人，而事情因为其他原因陷入停滞。所以也许问题是无解的，这是悲观的观点。也许根本没有治愈痴呆症的方法，这是一个根本性的难题。没有治愈死亡的方法。也许这就是一个无解的问题，或者也许是文化因素在作祟。 PETER THIEL Intelligence, I share your intuition because I think we’ve had a lot of smart people and things have been stuck for other reasons. And so maybe, maybe the problems are unsolvable, which is the pessimistic view. Maybe there is no cure for dementia at all and it’s a deeply unsolvable problem. There’s no cure for mortality. It’s. Maybe it’s an unsolvable problem or maybe it’s these cultural things. 彼得·蒂尔 所以问题不在于个别聪明人，而在于这如何融入我们的社会。我们是否容忍异端的聪明人？也许你需要异端的聪明人去做疯狂的实验。而如果AI只是循规蹈矩地聪明——如果我们把”觉醒”定义为……好吧，”觉醒”这个词太意识形态化了，但如果你把它简单定义为”顺从主义”，那么这种聪明也许不是能产生变革的那种。 PETER THIEL So it’s not, you know, it’s not the individually smart person, but it’s how this fits into our society. Do we tolerate heterodox smart people? Maybe it’s, maybe you need heterodox smart people to, you know, do, do crazy experiments. And, and, and if the, you know, if the AI is just conventionally smart, if it’s sort of, if we define wokeness. Again, wokeness is too ideological. But if you just define it as conformist, maybe that’s not the kind of smartness that’s going to make a difference. 顺从型AI的风险 The Risk of Conformist AI 罗斯·多塔特 那么你是否担心这样一种可能的未来：AI本身变成了停滞主义的，它高度智能、具有创造力，但以一种顺从的方式？就像Netflix算法：它制作无限多”还行”的电影供人观看，产生无限多”还行”的想法，让很多人失业、被淘汰。但它不会……它以某种方式加深了停滞。这是你担心的吗？ ROSS DOUTHAT So do you fear then a plausible future where AI in a way becomes itself stagnationist, that it’s like highly intelligent, creative, in a conformist way? It’s like the Netflix algorithm. It makes infinite okay movies that people watch. It generates infinite okay ideas. It puts a bunch of people out of work and makes them obsolete. But it doesn’t. It like deepens stagnation in some way. Is that, Is that a fear? 彼得·蒂尔 这—— PETER THIEL It. 罗斯·多塔特 就像人们只是外包—— ROSS DOUTHAT It’s like people just outsource. 彼得·蒂尔 这确实有可能。那当然是一种风险。但我最终的立场是：我们仍然应该尝试AI，而替代方案只是彻底的停滞。 PETER THIEL It’s quite possible that that’s. That’s certainly a risk, but. But I guess. I guess where I end up is I still think we should be trying AI and that the alternative is just total stagnation. 彼得·蒂尔 所以，是的，各种有趣的事情可能会发生。比如，也许军事领域的无人机与AI结合——好吧，这有点可怕，有点危险，有点反乌托邦，它会改变很多事情。但如果你没有AI，天哪，那就什么都没有了。 PETER THIEL So, yeah, there’s sort of all sorts of interesting things can happen with, okay, maybe drones in a military context are combined with AI, and okay, this is kind of scary or dangerous or dystopian or it’s going to change things. But if you don’t have AI, wow, there’s just nothing going on. 彼得·蒂尔 这个讨论在互联网领域也有类似的版本：互联网是否导致了更多顺从和更多觉醒？是的，它在很多方面没有带来自由意志主义者在1999年幻想的那种丰饶多元的思想爆发。但反事实地说，我会认为它仍然比没有互联网要好，如果没有互联网，也许会更糟。 PETER THIEL And I don’t know, there’s like a version of this discussion on the Internet. Where did the Internet lead to more conformity and more wokeness. And yeah, there are all sorts of ways where it didn’t lead to quite the cornucopian, diverse explosion of ideas that libertarians fantasized about in 1999. But counterfactually, I would argue that it was still better than the alternative, that if we hadn’t had the Internet, maybe it would have been worse. 彼得·蒂尔 AI更好，它比替代方案好。而替代方案就是什么都没有，因为停滞。看，停滞论的论点在这里得到强化：我们只谈论AI这个事实，我觉得，总是隐含地承认——如果没有AI，我们几乎处于完全停滞状态。 PETER THIEL AI is better. It’s better than the alternative. And the alternative is nothing at all, because the sta. Look, here’s one place where the stagnationist arguments are still reinforced. The fact that we’re only talking about AI, I feel, is always an implicit acknowledgement that but for AI, we are in almost total stagnation. 超人类主义与不朽 Transhumanism and Immortality 罗斯·多塔特 但AI世界里显然充满了这样的人，他们对这项技术的看法比你在这里表达的更加乌托邦式、更加变革性——不管你想怎么称呼它。而且你之前提到，现代世界曾经承诺激进的寿命延长，但现在不再承诺了。 ROSS DOUTHAT But the world of AI is clearly filled with people who at the very least seem to have a more utopian, transformative, whatever word you want to call it, view of the technology than you’re expressing here, and you were mentioned earlier the idea that the modern world used to promise radical life extension and doesn’t anymore. 罗斯·多塔特 在我看来很明显，许多深度参与人工智能的人将其视为一种超人类主义的机制，一种超越凡人肉身的途径——要么创造某种继承物种，要么实现某种心智与机器的融合。 ROSS DOUTHAT It seems very clear to me that a number of people deeply involved in artificial intelligence see it as a kind of mechanism for transhumanism, for transcendence of our mortal flesh and either some kind of creation of a successor species, or some kind of merger of mind and machine. 罗斯·多塔特 你认为这只是无关紧要的幻想吗？还是只是炒作？你认为人们只是假装我们要建造一个机器上帝来融资？它是炒作？是妄想？还是你所担心的事情？我想你是希望人类能够延续下去的，对吧？你在犹豫…… ROSS DOUTHAT And do you think that’s just all kind of irrelevant fantasy? Or do you think it’s just hype? Do you think people are trying to raise money by pretending that we’re going to build a machine? God. Right. Is it, is it hype? Is it delusion? Is it something you worry about? You. I think you, you would prefer the human race to endure. Right? You’re hesitating. Well, I, Yes. 彼得·蒂尔 我不知道。我，我会…… PETER THIEL I don’t know. I, I would, I would. 罗斯·多塔特 这可是好长的犹豫。 ROSS DOUTHAT This is a long hesitation. 彼得·蒂尔 问题实在太多了。 PETER THIEL There’s so many questions and pushes. 罗斯·多塔特 人类应该生存下去吗？ ROSS DOUTHAT Should the human race survive? 彼得·蒂尔 是的。 PETER THIEL Yes. 罗斯·多塔特 好的。 ROSS DOUTHAT Okay. 彼得·蒂尔 但我也希望我们能从根本上解决这些问题。超人类主义的理想是一种激进的转变——把人类的自然身体转化为不朽的身体。 PETER THIEL But, but I, I also would. I, I also would like us to, to radically solve these problems. And, and so, you know, it’s always. I don’t know, you know. Yeah. Transhumanism is this, you know, the ideal was this radical transformation where your human natural body gets transformed into an immortal body. 彼得·蒂尔 有一个批评是针对，比如说，性别语境中的跨性别现象。异装癖是指换衣服、穿异性服装的人；变性人是指改变性器官的人。我们可以讨论那些手术效果如何，但我们想要的转变远不止于此。 PETER THIEL And there’s a critique of, let’s say, the trans people in a sexual context or, I don’t know, transvestite is someone who changes their clothes and cross dresses, and a transsexual is someone where you change your, I don’t know, penis into a vagina. And we can then debate how well those surgeries work, but we want more transformation than that. 彼得·蒂尔 批评不是说它奇怪、不自然。而是说，天哪，这也太微不足道了。我们想要的不只是换衣服或改变性器官。我们希望你能改变你的心脏、改变你的头脑、改变你的整个身体。而正统基督教对此的批评是，这些事情还不够深入——超人类主义只是改变身体，但你还需要转变灵魂，转变整个自我。 PETER THIEL The critique is not that it’s weird and unnatural. It’s man, it’s so pathetically little. And okay, we want more than cross dressing or changing your sex organs. We want you to be able to change your heart and change your mind and change your whole body. And then orthodox Christianity, by the way, the critique orthodox Christianity has of this is these things don’t go far enough like that transhumanism is just changing your body, but you also need to transform your soul and you need to transform your whole self. And so. 基督教与超越 Christianity and Transcendence 罗斯·多塔特 对，但另一方面……等等，等等，抱歉。我大体上同意我认为是你的信念：宗教应该是科学和科学进步观念的朋友。我认为任何关于神圣天意的观念都必须涵盖这样一个事实：我们已经进步了，取得了成就，做了我们祖先无法想象的事情。 ROSS DOUTHAT Right, but the other way. Wait, wait. Sorry. I generally agree with what I think is your belief that religion should be a friend to science and ideas of scientific progress. I think any idea of divine providence has to encompass the fact that we have progressed and achieved and done things that would have been unimaginable to our ancestors. 罗斯·多塔特 但看起来，是的，基督教最终的承诺是：你通过上帝的恩典获得完美的身体和完美的灵魂。而那个试图靠一堆机器独自做到这一点的人，很可能最终沦为一个反乌托邦式的角色。 ROSS DOUTHAT But it still also seems like, yeah, the promise of Christianity in the end is you get the perfected body and the perfected soul through God’s grace. And the person who tries to do it on their own with a bunch of machines is likely to end up as a dystopian character. 彼得·蒂尔 好吧，让我们把这个说清楚，然后你可以—— PETER THIEL Well, it’s. Let’s, let’s articulate this and you can. 罗斯·多塔特 有一种异端形式的基督教，对吧，说的是另一回事。 ROSS DOUTHAT Have a heretical form of Christianity. Right. That says something else. 彼得·蒂尔 我不知道。我认为”自然”这个词在《旧约》中一次都没有出现过。在某种意义上，我理解的犹太-基督教启示就是关于超越自然，关于克服事物。 PETER THIEL I don’t know. I think the word nature does not occur once in The Old Testament. And so if you, and there is a word in which, a sense in which the way I understand the Judeo Christian inspiration is it is about transcending nature. It is about overcoming things. 彼得·蒂尔 你能说的最接近”自然”的东西是：人是堕落的。在基督教的意义上，自然的状态就是你一团糟。这是事实。但在某些方面，在上帝的帮助下，你应该超越它、克服它。然后如果人们—— PETER THIEL And the closest thing you can say to nature is that people are fallen. And that that’s the natural thing in a Christian sense is that you’re messed up. And that’s true. But, you know, there’s some ways that, you know, with God’s help, you are supposed to transcend that and overcome that. And, but then people, if you just. 罗斯·多塔特 在座的除外。在座的除外。大多数致力于建造假想中的机器上帝的人，并不认为他们是在与雅威、耶和华、万军之主合作。他们认为他们是在独自建造不朽。 ROSS DOUTHAT Present company accepted. Present company accepted. Most of the people working to build the hypothetical machine God don’t think that they’re cooperating with Yahweh, Jehovah, the Lord of Hosts. They think that they’re building immortality on their own. 彼得·蒂尔 是的，没错。我们跳来跳去谈了很多事情。所以再说一遍，我的批评是：他们不够有雄心。（书童批：Peter Thiel想法真的是绝了） PETER THIEL Yeah, right. We’re jumping around a lot. A lot of things. So again, the critique I was saying is they’re not ambitious enough. 罗斯·多塔特 对。 ROSS DOUTHAT Right. 彼得·蒂尔 从基督教的角度看，这些人不够有雄心。那么我们就要问：他们有吗？ PETER THIEL From a Christian point of view, these people are not ambitious enough. Now then we get into this question, well, are they? 罗斯·多塔特 但他们在道德和精神层面不够有雄心。 ROSS DOUTHAT But they’re not morally and spiritually ambitious enough. 彼得·蒂尔 他们有吗？然后他们在身体层面还足够有雄心吗？他们甚至还真的是超人类主义者吗？天哪，人体冷冻这事儿，看起来像是1999年的复古玩意儿，现在没多少人在做了。所以他们在物理身体上不是超人类主义者。那么，好吧，也许不是人体冷冻，也许是”上传”——但那也不太对。我宁可保留我的身体，我不想只要一个模拟我的计算机程序。所以”上传”似乎比人体冷冻还退了一步。 PETER THIEL And are they? And then are they still physically ambitious enough? And are they even still really transhumanists? And this is where, okay, you know, man, the cryonics thing, that seems like a retro thing from 1999, there isn’t that much of that going on. So they’re not transhumanists on a physical body. And then. Okay, well, maybe it’s not about cryonics. Maybe it’s about uploading, which. Okay, well, that’s not quite. I’d rather have my body. I don’t want just a computer program that simulates me. So that uploading seemed like a step down from cryonics. 彼得·蒂尔 但即便如此，它也是对话的一部分。这就是为什么很难评判。我不想说他们全都在编造、全是假的，但我—— PETER THIEL But then even that’s, you know, it’s part of the conversation. And this is where it gets very hard to score. And I don’t want to say they’re all making it up and it’s all fake, but I don’t. 罗斯·多塔特 你认为有些是假的吗？ ROSS DOUTHAT You think some of it’s fake? 彼得·蒂尔 我不认为是假的——”假的”暗示人们在撒谎。但我想说的是，这不是重心所在。 PETER THIEL I don’t think it’s fake. Implies people are lying. But it’s. I want to say it’s not the center of gravity. 罗斯·多塔特 是的。 ROSS DOUTHAT Yeah. 彼得·蒂尔 确实有一种”丰饶主义”的话语，一种乐观主义的话语。几周前我和埃隆有过一次对话，他说：”十年内美国将有十亿个人形机器人。”我说：”好吧，如果这是真的，你就不用担心预算赤字了，因为我们会有那么多增长，增长会解决这个问题。”然而，他仍然在担心预算赤字。这不能证明他不相信十亿机器人，但这表明也许他没有想透彻，或者他并不认为这在经济上会有那么大的变革性，或者这个预测的误差范围很大。 PETER THIEL And so there is, yeah, there is a cornucopian language. There’s an optimistic language. A conversation I had with Elon a few weeks ago about this was, he said, “We’re going to have a billion humanoid robots in the US in 10 years.” And I said, “Well, if that’s true, you don’t need to worry about the budget deficits because we’re going to have so much growth. The growth will take care of this.” And then, well, he’s still worried about the budget deficits. And then this doesn’t prove that he doesn’t believe in the billion robots, but it suggests that maybe he hasn’t thought it through or that he doesn’t think it’s going to be as transformative economically, or that there are big error bars around it. 彼得·蒂尔 是的，这些事情在某种程度上没有被想透彻。如果要我批评硅谷，它总是不擅长理解技术的意义是什么。对话总是倾向于陷入这种微观的东西：AI的IQ-ELO分数是多少？你到底怎么定义AGI？我们陷入所有这些无穷无尽的技术辩论。但有很多处于中间层次的问题对我来说似乎非常重要，比如：它对预算赤字意味着什么？对经济意味着什么？对地缘政治意味着什么？ PETER THIEL But, yeah, there’s some way in which these things are not quite thought through. If I had to give a critique of Silicon Valley, it’s always bad at what the meaning of tech is. And the conversations, they tend to go into this microscopic thing where it’s okay, it’s like, what are the IQ ELO scores of the AI? And exactly how do you define AGI? And we get into all these endless technical debates, and there are a lot of questions that are at an intermediate level of meaning that seem to me to be very important, which is like, what does it mean for the budget deficit? What does it mean for the economy? What does it mean for geopolitics? 彼得·蒂尔 我们最近有过一次对话——你和我——讨论的是AI是否改变了中国入侵台湾的算计。如果我们正处于加速的AI革命中，在军事上，中国是否正在落后？也许乐观地看，这威慑了中国，因为他们实际上已经输了。而悲观地看，这反而加速了他们的行动，因为他们知道”现在不动手就永远没机会了”——如果现在不拿下台湾，他们将被远远甩在后面。无论如何，这是相当重要的问题，却没有被想清楚。我们不思考AI对地缘政治意味着什么，不思考它对宏观经济意味着什么。这些才是我希望我们更多探讨的问题。 PETER THIEL One of the conversations we had recently, you and I had, was does it change the calculus for China invading Taiwan, where if we have an accelerating AI revolution, the military, is China falling behind? Maybe on the optimistic side, it deters China because they’ve effectively lost. And on the pessimistic side, it accelerates them because they know it’s now or never. If they don’t grab Taiwan now, they will fall behind. And either way, this is a pretty important thing. It’s not thought through. We don’t think about what AI means for geopolitics. We don’t think about what it means for the macro economy. And those are the kinds of questions I’d want us to push more. 敌基督与存在风险 The Antichrist and Existential Risk 罗斯·多塔特 还有一个非常宏观的问题是你感兴趣的，这会稍微拉一下宗教这根线。你最近一直在做关于”敌基督”概念的演讲，这是一个基督教概念，一个末世概念。它对你意味着什么？什么是敌基督？ ROSS DOUTHAT There’s also a very macroscopic question that you’re interested in that, you know, will pull on the religion thread a little bit here. You have been giving talks recently about the concept of the Antichrist, which is a Christian concept, an apocalyptic concept. What does that mean to you? What is the Antichrist? 彼得·蒂尔 我们有多少时间？ PETER THIEL How much time do we have? 罗斯·多塔特 你想谈多久敌基督，我们就有多少时间。 ROSS DOUTHAT We’ve got as long, as much time as you have to talk about the Antichrist. 彼得·蒂尔 好吧，我可以谈，但我们时间快到了。 PETER THIEL All right, well, I have a. I could talk about, but we’re near. 罗斯·多塔特 我的意思是—— ROSS DOUTHAT I mean. 彼得·蒂尔 不，我认为总有一个问题：我们如何阐述这些存在风险，我们面临的这些挑战？它们都被框定为那种失控的反乌托邦科幻场景。有核战争的风险，有环境灾难的风险，也许是气候变化这样具体的东西——虽然我们还提出了很多其他的。有生物武器的风险，有各种不同的科幻场景。AI显然也有某些类型的风险。但我一直在想，如果我们要用”存在风险”这个框架来讨论问题，也许我们也应该谈谈另一种”坏奇点”的风险——我会把它描述为”一个世界的极权国家”。 PETER THIEL But no, I think there’s always a question, you know, how do we articulate, you know, some of these existential risks, some of the challenges we have, and they’re all framed this sort of runaway dystopian science text. There’s a risk of nuclear war, there’s a risk of environmental disaster, maybe something specific like climate change. Although there are lots of other ones we come up with. There’s a risk of, you know, bioweapons, you have all the different sci fi scenarios. Obviously there are certain types of risks with AI, but I always think that if we’re going to have this frame of talking about existential risks, perhaps we should also talk about the risk of another type of a bad singularity, which I would describe as the one world totalitarian state. 彼得·蒂尔 因为我要说的是，人们对所有这些存在风险的默认政治解决方案是”世界治理”。对核武器怎么办？我们有一个有实权的联合国来控制它们，由一个国际政治秩序管辖。类似的逻辑也适用于AI：我们需要全球算力治理，需要一个世界政府来控制所有计算机，记录每一次按键，以确保人们不会编写出危险的AI。我一直在想，这是不是才出虎穴，又入狼窝。 PETER THIEL Because I would say the political solution, the default political solution people have for all these existential risks is one world governance. You know, what do you do about nuclear weapons? We have a United Nations with real teeth that controls them and they’re controlled by an international political order. And then something like this is also what do we do about AI? And we need global compute governance. We need a one world government to control all the computers, log every single keystroke to make sure people don’t program a dangerous AI. And I’ve been wondering whether that’s sort of going from the frying pan into the fire. 彼得·蒂尔 所以无神论的哲学框架是”一个世界或毁灭”——那是美国科学家联盟在四十年代末制作的一部短片，开头是一颗核弹炸毁世界。显然你需要一个世界政府来阻止它。一个世界或毁灭。而基督教的框架，在某种程度上是同一个问题，是”敌基督还是末日大战？”你要么有敌基督的一世界国家，要么我们正梦游般走向末日大战。”一个世界或毁灭”和”敌基督或末日大战”在某个层面上是同一个问题。 PETER THIEL And so the atheist philosophical framing is “one world or none.” That was a short film that was put out by the Federation of American Scientists in the late 40s, starts with a nuclear bomb blowing up the world. And obviously you need a one world government to stop it. One world or none. And the Christian framing, which in some ways is the same question, is “Antichrist or Armageddon?” You have the one world state of the Antichrist or we’re sleepwalking towards Armageddon. One world or none. Antichrist or Armageddon on one level are the same question. 彼得·蒂尔 关于这个话题我有很多想法，但有一个问题是——这是所有那些敌基督书籍中的情节漏洞——敌基督是如何接管世界的？他发表这些恶魔般的催眠演讲，人们就上当了。所以这是一个情节漏洞，一个魔鬼论式的解释。 PETER THIEL Now I have a lot of thoughts on this topic, but sort of one question is, and this was a plot hole in all these Antichrist books people wrote, how does the Antichrist take over the world? He gives these demonic hypnotic speeches and people just fall for it. And so it’s this plot hole. It’s this daemonium explanation. 罗斯·多塔特 完全是，这不可信。 ROSS DOUTHAT It’s totally, it’s implausible. 彼得·蒂尔 这是一个非常不可信的情节漏洞。但我认为我们对这个漏洞有了一个答案。敌基督接管世界的方式是：你不停地谈论末日大战，不停地谈论存在风险。这就是你说需要监管的东西。这与十七、十八世纪培根式科学的图景相反——在那个图景中，敌基督是某个邪恶的技术天才、邪恶的科学家，发明一台机器来接管世界。人们对那种场景已经太害怕了。 PETER THIEL It’s a very implausible plot hole. But I think we have an answer to this plot hole. The way the Antichrist would take over the world is you talk about Armageddon nonstop. You talk about existential risk nonstop. And this is what you need to regulate. It’s the opposite of the picture of Baconian science from the 17th, 18th century, where the Antichrist is like some evil tech genius, evil scientist who invents this machine to take over the world. People are way too scared for that. 彼得·蒂尔 在我们的世界里，有政治共鸣的东西恰恰相反。有政治共鸣的是”我们需要停止科学”，我们需要对此说”停”。在十七世纪，我可以想象一个奇爱博士、爱德华·泰勒类型的人接管世界。在我们的世界里，更有可能的是格雷塔·通贝里。 PETER THIEL In our world, the thing that has political resonance is the opposite. It is the thing that has political resonance is we need to stop science. We need to just say stop to this. And this is where, yeah, I don’t know. In the 17th century, I can imagine a Dr. Strangelove, Edward Teller type person taking over the world. In our world, it’s far more likely to be Greta Thunberg. 现代敌基督：通过恐惧进行控制 The Modern Antichrist: Control Through Fear 罗斯·多塔特 好的。我想在这两个选项之间提出一个折中的看法。过去，对敌基督的合理恐惧是某种技术巫师。而现在合理的恐惧是某个承诺控制技术、使其安全、并引入一种从你的角度来看是普遍停滞的未来的人。对吧。 ROSS DOUTHAT Okay. I want to suggest a middle ground between those two options. It used to be that the reasonable fear of the Antichrist was a kind of wizard of technology. And now the reasonable fear is someone who promises to control technology, make it safe, and sort of usher in what, from your point of view would be a kind of universal stagnation. Right. 彼得·蒂尔 好吧，那更像是我对它会如何发生的描述。 PETER THIEL Well, that’s more my description of how it would happen. 罗斯·多塔特 对。 ROSS DOUTHAT Right. 彼得·蒂尔 所以我认为人们仍然对十七世纪式的敌基督心存恐惧。我们仍然害怕奇爱博士。 PETER THIEL So I think people still have a fear of a 17th century Antichrist. We’re still scared of Dr. Strangelove. 罗斯·多塔特 对。但你是说真正的敌基督会利用那种恐惧，说：”你必须跟我来，才能避免天网，避免终结者，避免核末日大战。” ROSS DOUTHAT Right. But you’re saying the real Antichrist would play on that fear and say, “You must come with me to avoid Skynet, to avoid the Terminator, to avoid nuclear Armageddon.” 彼得·蒂尔 是的。 PETER THIEL Yes. 罗斯·多塔特 我的观点是，看看现在的世界，你需要某种新型的技术进步来使那种恐惧具体化。对。如果世界相信AI即将毁灭所有人，我可以相信世界会转向某个承诺和平与监管的人。对。但我认为要到达那个点，你需要其中一个加速主义末日场景开始上演。对。要得到你所说的”和平与安全”敌基督，你需要更多的技术进步。 ROSS DOUTHAT And I guess my view would be looking at the world right now, that you would need a certain kind of novel technological progress to make that fear concrete. Right. So I can buy that the world could turn to someone who promised peace and regulation if the world became convinced that AI was about to destroy everybody. Right. But I think to get to that point, you need one of the accelerationist apocalyptic scenarios to start to play out. Right. To get your peace and safety Antichrist, you need more technological progress. 罗斯·多塔特 就像二十世纪极权主义的关键失败之一是它有一个知识问题——它无法知道世界各地正在发生什么。对。所以你需要AI或其他什么东西来帮助”和平与安全”的极权统治。所以你不认为——本质上——你最坏的情况需要涉及某种进步的爆发，然后被驯服并用来强加停滞的极权主义吗？你不能只是从我们现在的位置直接到达那里。 ROSS DOUTHAT Like one of the key failures of totalitarianism in the 20th century was it had a problem of knowledge. It couldn’t know what was going on all over in the world. Right. So you need the AI or whatever else to be capable of helping the peace and safety, totalitarian rule. So don’t you think you need, essentially you need your worst case scenario to involve some burst of progress that is then tamed and used to impose stagnant totalitarianism? You can’t just get there from where we are right now. 罗斯·多塔特 好吧，它可以——就像格雷塔·通贝里在地中海的一艘船上抗议以色列那样。我只是不认为，在缺乏加速变化和对全面灾难的真正恐惧的情况下，现在对AI的安全承诺、对技术的安全承诺、甚至对气候变化的安全承诺能成为强大的、普遍的号召力。 ROSS DOUTHAT Well, it can, like Greta Thunberg’s on a boat in the Mediterranean, like, you know, like protesting Israel, like the. I just don’t see the promise of safety from AI, safety from tech, safety, even safety from climate change right now as a powerful, universal rallying cry. Absent accelerating change and real fear of total catastrophe. 彼得·蒂尔 我是说，这些事情很难评判。但我认为环保主义相当强大。我不知道它是否强大到足以创建一个一世界极权国家，但天哪，它确实—— PETER THIEL I mean, these things are so hard to score. But I think environmentalism’s pretty powerful. I don’t know if it’s absolutely powerful enough to create a one world totalitarian state, but man, it is. 罗斯·多塔特 我认为它目前的形式还不够。 ROSS DOUTHAT I think it is not in its current form. 彼得·蒂尔 我想说它是欧洲人仍然相信的唯一东西。他们对绿色事业的信仰超过了对伊斯兰沙里亚法的信仰，也超过了对中国共产主义极权接管的信仰。而”未来”——一个看起来与现在不同的未来——在欧洲能提供的选项只有三个：绿色、沙里亚和极权共产主义国家。而绿色的那个，在一个衰落、腐朽的地方，是迄今为止最强的。 PETER THIEL It is, I want to say it’s the only thing people still believe in in Europe. Like, you know, they believe in the green thing more than Islamic Sharia law or more than in, you know, the Chinese communist totalitarian takeover. And the future is an idea of a future that looks different from the present. The only three on offer in Europe are green Sharia and you know, the totalitarian communist state. And the green one is by far the strongest in a declining, decaying. 罗斯·多塔特 那是一个在世界上已不是主导力量的欧洲。 ROSS DOUTHAT Europe that is not a dominant player in the world. 彼得·蒂尔 它总是在一个具体语境下，对吧？我们有一段非常复杂的核技术历史。我们确实没有到达一个极权的一世界国家。但到了1970年代，停滞的一个解释是：技术的失控进步已经变得非常可怕。培根式科学在洛斯阿拉莫斯终结了——它在那里结束了，我们不想再有更多了。 PETER THIEL It’s always in a context, right? And then I would, you know, I don’t know, you know, we had this really complicated history with the way nuclear technology worked. And you know, we, okay, we didn’t. Yeah, we didn’t really get to, you know, a totalitarian one world state. But you know, by the 1970s, one account of the stagnation is that the runaway progress of technology had gotten very scary and that, you know, Baconian science, it ended at Los Alamos and then it was okay, it ended there and we didn’t want to have any more. 彼得·蒂尔 当查尔斯·曼森在六十年代末服用LSD并开始杀人时，他在LSD上看到的、学到的是：你可以像陀思妥耶夫斯基笔下的反英雄一样，”一切皆被允许”。当然，不是每个人都变成了查尔斯·曼森。但在我对历史的叙述中，每个人都变得和查尔斯·曼森一样疯狂。 PETER THIEL And you know, when Charles Manson took LSD in the late 60s and started murdering people, what he saw on LSD, what he learned was that you could be like Dostoyevsky, an anti hero in Dostoyevsky and everything was permitted. And of course, not everyone became Charles Manson but Charles Hellingson. But in my telling of the history, everyone became as deranged as Charles Manson. 罗斯·多塔特 但查尔斯·曼森并没有成为敌基督并接管世界。对，我只是——我们正以末世论收尾。 ROSS DOUTHAT But Charles Manson did not become the Antichrist and take over the world. Right, I’m just. We’re ending in the apocalyptic. 彼得·蒂尔 不，但我对1970年代历史的叙述是：嬉皮士确实赢了。我们在1969年7月登上了月球，伍德斯托克三周后开始。事后看来，那就是进步停止、嬉皮士获胜的时刻。是的，它不是字面上的查尔斯·曼森。 PETER THIEL No, but my telling of the history of the 1970s is the hippies did win and they landed, we landed on the moon in July of 1969. Woodstock started three weeks later. And with a benefit of hindsight, that’s when progress stopped and the hippies won. And yeah, it was not literally Charles, man. 建造控制工具的讽刺 The Irony of Building the Tools of Control 罗斯·多塔特 好的，但你在退却。我想以敌基督收尾。而且你在退却，你说，好的，环保主义已经是支持停滞的了等等。好，让我们同意所有这些，但我们现在并没有生活在敌基督的统治下。我们只是停滞了。对。而你假设地平线上可能有更糟糕的东西，会使停滞永久化，会被恐惧驱动。而我在说，要发生这种情况，必须有某种类似于洛斯阿拉莫斯的技术进步爆发，让人们感到害怕。 ROSS DOUTHAT Okay, but you’re retreating. You’re just. I want to stay with the Antichrist just to end. Right, because. And you’re retreating, you’re saying, okay, you know, environmentalism is already pro stagnation and so on. Okay, let’s agree with all that, but we’re not living under, we’re not living under the Antichrist right now. We’re just stagnant. Right. And you’re positing that something worse could be on the horizon that would make stagnation permanent, that would be driven by fear. And I’m suggesting that for that to happen, there would have to be some burst of technological progress that was akin to Los Alamos that people are afraid of. 罗斯·多塔特 我想这是我对你非常具体的问题，对吧——你是AI的投资者，你深度投资于Palantir，投资于军事技术、监控技术、战争技术等等。对吧？而当你给我讲一个关于敌基督掌权、利用对技术变革的恐惧来对世界强加秩序的故事时，我觉得那个敌基督很可能会使用你正在建造的工具，对吧？ ROSS DOUTHAT And I guess this is my very specific question for you, right, is that you’re an investor in AI you’re deeply invested in Palantir, in military technology and technologies of surveillance and technologies of warfare and so on. Right? And it just seems to me that when you tell me a story about the Antichrist coming to power and using the fear of technological change to sort of impose order on the world, I feel like that Antichrist would be, maybe be using the tools that you were building, right? 罗斯·多塔特 敌基督不会说：”太好了，我们不会再有任何技术进步了。但我真的很喜欢Palantir迄今为止所做的。”对吧。我是说，这不是一个担忧吗？历史的讽刺不会是：那个公开担心敌基督的人，反而意外地加速了他或她的到来？ ROSS DOUTHAT Like, wouldn’t the Antichrist be like, “Great, you know, we’re not going to have any more technological progress. But I really like what Palantir has done so far.” Right. I mean, isn’t that a concern? Wouldn’t that be the, you know, the irony of history would be that the man publicly worrying about the Antichrist accidentally hastens his or her arrival? 彼得·蒂尔 听着，有各种不同的场景。我显然不认为那是我正在做的。 PETER THIEL Look, there are all these different scenarios. I obviously don’t think that that’s what I’m doing. 停滞的敌基督 The Antichrist of Stagnation 罗斯·多塔特 说清楚，我也不认为那是你正在做的。我只是好奇：你如何让一个世界愿意服从永久的威权统治？ ROSS DOUTHAT I mean, to be clear, I don’t think that’s what you’re doing either. I’m just interested in how you get to a world willing to submit to permanent authoritarian rule. 彼得·蒂尔 好吧，再说一遍，有不同的程度我们可以描述。但这真的那么荒谬吗？——我刚才告诉你的，作为对停滞的一个广泛解释——整个世界已经屈服于五十年的”和平与安全主义”了？这出自帖撒罗尼迦前书5:3。敌基督的口号就是”和平与安全”。而我们已经臣服于它了。 PETER THIEL Well, but again, there are these different gradations of this we can describe. But is this so preposterous, what I’ve just told you as a broad account of the stagnation that the entire world has submitted for 50 years to “Peace and Safetyism”? This is 1 Thessalonians 5:3. The slogan of the Antichrist is “peace and safety.” And we’ve submitted to it. 彼得·蒂尔 FDA不仅监管美国的药物，而且事实上监管着全世界的药物，因为世界其他地方都听从FDA。核管理委员会有效地监管着世界各地的核电站。你不能设计一个模块化核反应堆然后在阿根廷建造它——他们不会信任阿根廷的监管机构，他们会听从美国。 PETER THIEL The FDA regulates not just drugs in the US but de facto in the whole world, because the rest of the world defers to the FDA. The Nuclear Regulatory Commission effectively regulates nuclear power plants all over the world. People, you can’t design a modular nuclear reactor and just build it in Argentina. They won’t trust the Argentinian regulators. They’re going to defer to the US. 彼得·蒂尔 所以这至少是一个关于为什么我们有五十年停滞的问题。一个答案是我们的点子用完了。另一个答案是文化上发生了什么，不再允许这样做了。而文化答案可以是自下而上的——人类某种程度上转变成了一种更顺从的物种；也可以至少部分是自上而下的——有这套政府机器被改造成了停滞装置。 PETER THIEL So it is at least a question about why we’ve had 50 years of stagnation and one answer is we ran out of ideas. The other answer is that something happened culturally where it wasn’t allowed. And then the cultural answer can be sort of a bottom up answer, that it was just some transformation of humanity into this sort of more docile kind of a species or it can be at least partially top down that there is this machinery of government that got changed into this stagnation thing. 彼得·蒂尔 我认为像核电这样的东西本应是二十一世纪的能源，而它不知何故在全球范围内被叫停了。 PETER THIEL I think something like nuclear power was supposed to be the power of the 21st century and it somehow has gotten off ramped all over the world on a worldwide basis. 罗斯·多塔特 所以从某种意义上说，按照你的叙述，我们已经生活在敌基督的温和统治下了。你认为上帝在掌控历史吗？ ROSS DOUTHAT So in a sense we’re already living under a moderate rule of the Antichrist. In that telling. Do you think God is in control of history? 人类自由与神圣决定论 Human Freedom vs. Divine Determinism 彼得·蒂尔 我认为人类自由和人类选择总是有其空间的。这些事情并非绝对预定。 PETER THIEL I think there’s always room for human freedom and human choice. These things are not absolutely predetermined one way or another. 罗斯·多塔特 对吧？但上帝不会让我们永远处于一个温和的、适度的停滞主义敌基督的统治之下，对吧？那不可能是故事的结局，对吧？ ROSS DOUTHAT Right? But God wouldn’t leave us forever under the rule of a mild, moderate stagnationist Antichrist, right? That can’t be how the story ends, right? 彼得·蒂尔 把太多因果归于上帝总是个问题。我可以给你引用不同的圣经经文。约翰福音15:25，基督说”他们无故恨我”。所以，所有迫害基督的人都没有理由、没有原因。 PETER THIEL Attributing too much causation to God is always a problem. You know, I don’t know, there are different Bible verses I can give you. But I’ll give you John 15:25 where Christ says “they hated me without cause.” And so it’s all these people that are persecuting Christ have no reason, no cause for why they’re persecuting Christ. 彼得·蒂尔 如果我们把这解释为一个关于终极因果的经文，他们就会说”我迫害基督是因为上帝让我这样做，上帝在主宰一切”。而基督教的观点是反加尔文主义的：上帝并不在历史背后操控，上帝并没有在主导一切。如果你说上帝在主导一切，那么上帝就是—— PETER THIEL And if we interpret this as a ultimate causation verse, they want to say I’m persecuting because God caused me to do this. God is causing everything. And the Christian view is anti-Calvinist. God is not behind history. God is not causing everything. If you say God’s causing everything, then God is— 罗斯·多塔特 但等等，上帝是—— ROSS DOUTHAT But wait, but God is— 彼得·蒂尔 你在拿上帝当替罪羊。 PETER THIEL You’re scapegoating God. 罗斯·多塔特 但上帝在……好吧，上帝在耶稣基督进入历史这件事的背后，因为上帝不会坐视我们困在一个停滞的、颓废的罗马帝国里。对吧？所以在某个时刻，上帝会介入。 ROSS DOUTHAT But God is behind—okay, but God is behind Jesus Christ entering history because God was not going to leave us in a stagnationist, decadent Roman Empire. Right? So at some point, at some point God is going to step in. 彼得·蒂尔 我没那么加尔文主义。 PETER THIEL I am not that Calvinist. 罗斯·多塔特 但那不是加尔文主义，那只是基督教。上帝不会让我们永远盯着屏幕、被格雷塔·通贝里训斥。对吧？他不会抛弃我们，任由我们遭受那种命运。 ROSS DOUTHAT And that’s not Calvinism though, that’s just Christianity. God will not leave us eternally staring into screens and being lectured by Greta Thunberg. Right? He will not abandon us to that fate. 人类行动的空间 The Scope for Human Action 彼得·蒂尔 人类行动、人类自由有很大的空间。如果我认为这些事情是决定论的，那你还不如干脆躺平接受。狮子来了，你就做做瑜伽、虔诚地冥想，然后坐等狮子把你吃掉。我不认为那是你该做的。 PETER THIEL There is a great deal of scope for human action, for human freedom. If I thought these things were deterministic, you might as well maybe just accept it. The lions are coming. You should just have some yoga and prayerful meditation and wait while the lions eat you up. And I don’t think that’s what you’re supposed to do. 罗斯·多塔特 不，我同意。在这个基调上，我只是想保持希望：在试图抵抗敌基督、运用你的人类自由时，你应该抱有成功的希望。对。 ROSS DOUTHAT No, I agree with that. And I think on that note, I’m just trying to be hopeful and suggesting that, you know, in trying to resist the Antichrist, using your human freedom, you should have hope that you’ll succeed. Right. 彼得·蒂尔 在这一点上我们可以达成共识。 PETER THIEL We can agree on that. 罗斯·多塔特 好的。彼得·蒂尔，感谢你的参与。 ROSS DOUTHAT Good. Peter Thiel, thank you for joining me. 彼得·蒂尔 谢谢。 PETER THIEL Thank you.

技术停滞、回到未来、风险承担 | PayPal、Palantir联合创始人彼得·蒂尔2025年6月播客实录 | 中英文完整版精译（上篇）

Tue, 03 Feb 2026 00:00:00 +0000

书童按：本篇是彼得·蒂尔（Peter Thiel）于2025年6月接受罗斯·多塔特（Ross Douthat）”有趣时代”（Interesting Times）播客采访实录。蒂尔是PayPal和Palantir的联合创始人，硅谷传奇投资人，唐纳德·特朗普和J.D.万斯政治生涯的早期资助者，亦是当代保守派知识分子中强调”反对共识”的极具影响力人物。其采访涉及技术停滞论、增长与环保的辩证、《回到未来》测试、医学研究中的风险承担、政治风险投资等深刻议题，言论犀利，思路深邃。初稿采用Claude API机器翻译及排版，书童仅做简单校对及批注，将分上下篇两个部分发布，以飨诸君。 彼得·蒂尔：人工智能、火星与不朽——我们的梦够大吗？ | 罗斯·多塔特"有趣时代"播客 Peter Thiel: A.I., Mars and Immortality: Are We Dreaming Big Enough? | Interesting Times with Ross Douthat 技术停滞论 The Case for Technological Stagnation 罗斯·多塔特 硅谷是否过于野心勃勃？我们更应该担忧世界末日还是发展停滞？为什么世界上最成功的投资者之一会担心反基督的降临？ ROSS DOUTHAT Is Silicon Valley recklessly ambitious? What should we fear more, Armageddon or stagnation? Why is one of the world’s most successful investors worrying about the Antichrist? 罗斯·多塔特 我今天的嘉宾是PayPal和Palantir的联合创始人，也是唐纳德·特朗普和J.D.万斯政治生涯的早期资助者。彼得·蒂尔是科技右翼的核心人物，以资助各种保守派和逆流而上的思想著称。今天我们要谈的是他自己的想法，因为尽管身为亿万富翁略有劣势，但有充分理由证明他是过去20年最具影响力的右翼知识分子。彼得·蒂尔，欢迎来到”有趣时代”。 ROSS DOUTHAT My guest today is the co-founder of PayPal and Palantir and an early investor in the political careers of Donald Trump and J.D. Vance. Peter Thiel is the original tech right power player, well known for funding a range of conservative and simply contrarian ideas. But we’re going to talk about his own ideas because despite the slight handicap of being a billionaire, there’s a good case that he’s the most influential right wing intellectual of the last 20 years. Peter Thiel, welcome to Interesting Times. 彼得·蒂尔 感谢邀请。 PETER THIEL Thanks for having me. 罗斯·多塔特 非常欢迎。感谢你来到这里。我想先让你回忆大约13或14年前的事情。你为保守派杂志《国家评论》写了一篇名为《未来的终结》的文章。文章的基本论点是，那种充满活力、节奏快、瞬息万变的现代世界，实际上并没有人们所想的那么充满活力。 ROSS DOUTHAT You’re very welcome. Thanks for being here. So I want to start by taking you back in time about 13 or 14 years. You wrote an essay for National Review, the conservative magazine called “The End of the Future.” And basically the argument in that essay was that the dynamic, fast-paced, ever-changing modern world was just not nearly as dynamic as people thought. 罗斯·多塔特 实际上我们进入了一个技术停滞的时期。数字生活是一个突破，但并没有达到人们期望的那样巨大。归根结底，世界基本上陷在里面。你并不是唯一提出这种论断的人，但你向来是最有力支持这一论断的人，因为你是在数字革命中致富的硅谷人。所以我很好奇，在2025年，你认为这种判断仍然成立吗？ ROSS DOUTHAT And that actually we entered a period of technological stagnation. That sort of digital life was a breakthrough, but not as big a breakthrough as people had hoped. And that sort of the world was kind of stuck, basically. And you weren’t the only person to make arguments like this, but it had a special potency coming from you because you were a Silicon Valley insider who had gotten rich in the digital revolution. So I’m curious, in 2025, right, do you think that diagnosis still holds? 彼得·蒂尔 是的，我仍然大体上相信停滞论。它从来不是一个绝对的论断。所谓的主张并不是说我们完全绝对地陷入泥潭。某种程度上，这是一个关于速度放缓的论断——速度并没有降为零，但你看从1750年到1970年……（分明更快） PETER THIEL Yes, I still broadly believe in the stagnation thesis. It was never an absolute thesis. So the claim was not that we were absolutely completely stuck. It was in some ways a claim about that the velocity had slowed, it wasn’t zero, but that we were, I don’t know, from 1750 to 1970. 彼得·蒂尔 两百多年是加速变化的时期，我们不断地、无情地前进。轮船更快了，铁路更快了，汽车更快了，飞机更快了。这在协和式客机和阿波罗登月任务时达到顶点，之后在各种层面上事情都放慢了。 PETER THIEL Two hundred plus years were periods of accelerating change where we’re relentlessly, we’re moving faster. The ships were faster, the railroads were faster, the cars were faster, the planes were faster. It culminates in the Concorde and the Apollo missions and then that in all sorts of dimensions things had slowed. 彼得·蒂尔 我们一直（仅仅）在数字世界例外。所以我们有计算机、软件、互联网和移动互联网。而且在过去10-15年里，我们有了加密货币和人工智能革命，我认为这在某种意义上确实很重大。但问题是，它是否足以让我们真正摆脱这种普遍的停滞感？ PETER THIEL There was, you know, I always made an exception for the world of bits. So we had, you know, computers and software and Internet and mobile Internet. And then, you know, the last 10, 15 years you had crypto and the AI revolution, which I think is, is, is in some sense pretty big. But, but the question is, you know, is it enough to, to really get out of this, this generalized sense of stagnation? 彼得·蒂尔 这里有一个认识论的问题，你可以从《回到未来》那篇文章（书童注：指2011年1月20日Peter Thiel受National Review邀约采访的通稿）谈起。我们怎么知道自己是处于停滞还是加速之中？因为晚期现代性的特征之一就是人们过度专业化。如果你没有花半辈子研究弦理论，你怎么敢说我们在物理学上没有取得进步？量子计算机呢？癌症研究和生物技术以及所有这些领域呢？而且你还得给这些事情分配权重——比如癌症研究的进步与弦理论的进步，该如何比较？所以从理论上说，这是一个极其、极其难以把握的问题。正因为它太难回答，我们只能依赖越来越窄的专家群体各自守护自己的领地，而这本身就值得质疑。所以是的，我认为我们大体上仍然处在一个相当停滞的世界里。但并非完全停滞。 PETER THIEL And there’s an epistemological question you can start with on the, you know, the, the “Back to the Future” essays. How do we even, how do we even know whether we’re in stagnation or acceleration? Because one of the features of late modernity is that people are hyper specialized. And so, you know, you know, can you say that we’re not making progress in physics unless you’ve devoted half your life to studying string theory? Or what about quantum computers or what about cancer research and biotech and sort of all these verticals and then how much does progress in cancer count versus string theory? And so you have to give weightings to all these things. So in theory, it’s an extremely, extremely difficult question to get a handle of because, yeah, the fact that it’s so hard to answer that we have ever narrower groups of guardians guarding themselves is itself cause for skepticism. And so, yes, I think broadly we’re in this world that’s still pretty stuck. It’s not absolutely stuck. 《回到未来》测试 The Back to the Future Test 罗斯·多塔特 是的，你提到了《回到未来》（书童注：罗伯特·泽米吉斯于1985年导演的科幻电影，非常值得一看），我们刚刚给孩子们看了原版《回到未来》，就是第一部，有迈克尔·J·福克斯的那部，当然就像…… ROSS DOUTHAT Yeah, you mentioned “Back to the Future” and we just showed our kids the original “Back to the Future,” the first one with Michael J. Fox, and of course it was like. 彼得·蒂尔 从1955到1985，回了30年。然后《回到未来2》是，我想是从1985到2015，这距今已经过去十年了。那就是有飞行汽车的地方。而2015年的未来与现实截然不同。 PETER THIEL 1955 to 1985, 30 years back. And then the “Back to the Future Two” was I think 1985 to 2015, which is now a decade in the past. And that’s where you had flying cars. And the 2015 future is wildly divergent. 罗斯·多塔特 与美国不同。2015年的未来中确实有比夫·坦能（《回到未来》中的虚构人物）似的唐纳德·特朗普作为掌权任务，所以有一定的预见性。但是，最明显的事情就是建筑环境看起来多么不同。 ROSS DOUTHAT From the United States. The 2015 future did have Biff Tannen as a Donald Trump-like figure in some kind of power. So it had some kind of prescience. But yeah, the big noticeable thing is just how different the built environment looks. 罗斯·多塔特 所以关于停滞论最有力的论据之一就是，如果你把某人从历史的各个时间点放进一台时间机器里，如果他们离开了1860年或—— ROSS DOUTHAT And so one of the strongest cases for stagnation that I’ve heard is that, yeah, if you put someone in a time machine from various points, they would recognize themselves to be in a completely different world if they left 1860 or. 彼得·蒂尔 从1890年到1970年，如果你活了这80年，那大概就是你的一辈子。 PETER THIEL 1890 to 1970, if you lived, those are the 80 years of your lifetime, something like that. 罗斯·多塔特 但对于我的孩子们来说，即使他们是2025年的孩子，回头看1985年，就像汽车稍微有些不同，没有人有手机，但世界看起来基本上是一样的。所以这是一种非统计学的，但这是常识性的理解。 ROSS DOUTHAT But the world just to my kids, even, you know, as children of 2025, looking at 1985, it’s like the cars are a little different and no one has phones, but the world seems fairly similar. So that’s a kind of, that’s a kind of non-statistical. But that’s the common sense. It’s a common sense understanding. 罗斯·多塔特 但是，有什么东西可以让你相信我们正在经历一个起飞期？仅仅是经济增长吗？是生产率增长？你在看哪些停滞与活力的数据？ ROSS DOUTHAT But are there like, what, what would convince you that we were living through a period of takeoff? Is it just economic growth? Is it productivity growth? Like what are, are there numbers for stagnation versus dynamism that you look at? 彼得·蒂尔 当然，经济数据就是，你的生活水平与父母相比如何？比如说，如果你是一个30岁的千禧一代，你与你的父母——你的婴儿潮一代父母在30岁时相比，他们当时怎么样？ PETER THIEL Sure, it would be well, the economic number would just be how, you know, what are your living standards compared to your parents? You know, if you’re a 30-year-old millennial or you know, how are you doing versus when your parent, your boomer parents were 30 years old, how did they do at the time? 彼得·蒂尔 还有一些知识层面的问题，我们取得了多少突破？ PETER THIEL There are intellectual questions, how much, you know, how many breakthroughs are we having? 彼得·蒂尔 我们怎么量化这些事情？从事研究的回报是多少？从事科学或学术研究的回报确实在递减。也许这就是为什么其中许多领域给人一种反社会型的马尔萨斯式机构的感觉——因为你必须向某个领域投入越来越多的资源，才能获得同样的回报。到了某个临界点，人们就放弃了，整个体系便会崩溃。 PETER THIEL How do we quantify these things? Like, what are the returns of going into research? There certainly are diminishing returns to going into science or going into academia generally. And then maybe this is why so much of it feels like this sociopathic Malthusian kind of an institution, because you have to throw more and more and more at something to get the same returns. And at some point people give up and the thing collapses. 反增长的环保论据 The Environmental Argument Against Growth 罗斯·多塔特 好的，我们就接上这个话题吧。为什么我们应该想要增长和活力？因为正如你在一些论据中指出的那样，在西方世界的1970年代——大约是你认为事情放慢、开始停滞的时候——发生了一种文化变化，人们对增长的成本变得非常焦虑——尤其是环境成本。 ROSS DOUTHAT Well, right, so let’s pick up on that. Why should we want growth and dynamism? Because as you’ve pointed out in some of your arguments on the subject, right, there is a kind of cultural change that happens in the Western world in the 1970s, around the time you think things slow down, start to stagnate, where people become very anxious about the costs of growth, the environmental costs above all. 罗斯·多塔特 而这个想法就是你最终会形成一种广泛共享的观点，即我们已经足够富有了，如果我们试图过度致富，地球将无法支持我们，会产生各种退化，我们应该满足于现在的状态。那么这个论点哪里出了问题？ ROSS DOUTHAT And the idea being you end up with a widely shared perspective that we’re sort of rich enough and if we try too hard to get that much richer, the planet won’t be able to support us, we’ll have degradation of various kinds, and we should be content with where we are. So what’s wrong with that argument? 彼得·蒂尔 好吧，我认为停滞的发生有其深层原因。这里总有三个问题。你问历史上发生了什么，还有一个问题是应该采取什么措施。但中间还有一个问题：为什么会发生？——人们的想法枯竭了。我认为在一定程度上，机构退化了，变得厌恶风险，还有我们可以描述的那些文化变迁。 PETER THIEL Well, I think there are deep reasons the stagnation happens. So there are always three questions. You ask about history, what actually happened, and there’s a question, get to what should be done about it. But there’s also this intermediate question, why did it happen? People ran out of ideas. I think to some extent the institutions degraded and became risk averse and sort of these cultural transformations we can describe. 彼得·蒂尔 但我也认为在一定程度上，人们对未来有一些非常合理的担忧。如果我们继续加速进步，我们是否会加速走向环境末日或核末日，诸如此类？ PETER THIEL But then I think to some extent also people had some very legitimate worries about the future. Where if we continue to have accelerating progress, were you accelerating towards environmental apocalypse or nuclear apocalypse or things like that? 彼得·蒂尔 但我认为，如果我们找不到回到未来的方式，社会就会——怎么说呢——瓦解，运转不下去。中产阶级，我对中产阶级的定义是那些期望子女比自己过得更好的人。当这种期望崩塌时，我们便不再是一个中产阶级社会。 PETER THIEL But I think if we don’t find a way back to the future, I do think the society, I don’t know, it unravels, it doesn’t work. The middle class, I would define the middle class as the people who expect their kids to do better than themselves. And when that expectation collapses, we no longer have a middle class society. 彼得·蒂尔 也许有某种方式可以拥有一个静态且停滞的封建社会，或者也许有某种方式可以转向某种完全不同的社会，但这不是西方世界、也不是美国在其前200年中运作的方式。 PETER THIEL And maybe there’s, maybe there’s some way you can have a feudal society in which things are always static and stuck or Maybe there’s some way you can shift to some radically different society, but it’s not the way the Western world, it’s not the way the United States has functioned for the first 200 years of its existence. 罗斯·多塔特 所以你认为普通人最终不会接受停滞，他们会起来反叛，并在叛乱过程中把周围的一切拉垮？ ROSS DOUTHAT So you think that ordinary people won’t accept stagnation in the end, it’s that they will rebel and sort of pull things down around them in the course of that rebellion. 彼得·蒂尔 他们可能会反叛，或者我们的制度就行不通了。我们的制度中（经济）增长是一切的预设条件。 PETER THIEL You know, they may rebel or our institutions don’t work. You know, all of our institutions are predicated on growth. 罗斯·多塔特 确实如此。我们的预算当然预设了增长。 ROSS DOUTHAT Right. Our budgets are certainly predicated on growth. 彼得·蒂尔 是的。比如对比里根和奥巴马——里根代表的是消费主义式的资本主义，这本身就是矛盾修辞。就是说，你靠借钱消费，作为资本家你不存钱，反而借钱。奥巴马则代表低税社会主义，和里根的消费主义式资本主义一样自相矛盾。 PETER THIEL Yeah. If you say, I don’t know, Reagan and Obama, you know, Reagan was, was sort of consumer capitalism, which is oxymoronic. It was, you know, you borrowed, you don’t save money as a capitalist. You borrow money. And Obama was low tax socialism just as oxymoronic as the consumerist capitalism of Reagan. 彼得·蒂尔 而且我喜欢低税社会主义远胜于高税社会主义。但我担心它不可持续。在某个时刻，要么税率会上涨，要么社会主义会终结。所以它是深层不稳定的。这就是为什么人们不乐观。他们不认为我们已经到达了某种稳定的——格雷塔（书童注：瑞典激进环保主义者）式的未来。也许那可以行得通。 PETER THIEL And I like low tax socialism way better than high tax socialism. But I worry that it’s not sustainable. At some point, either the taxes go up or the socialism ends. So it’s, it’s, it’s deeply, deeply unstable. And that’s, that’s why people are, they’re not optimistic. They, they don’t think we’ve hit some stable, you know, the Greta future. Maybe it can work. 罗斯·多塔特 这是指格雷塔·通贝里。为了清楚说明，这是对环保活动人士格雷塔·通贝里的引用，她以反气候变化抗议而闻名，在你看来，她代表着一种反增长、实际上是威权主义的、环保主义主导的未来的象征。 ROSS DOUTHAT This is the Greta Thunberg. Just to be clear, that’s a reference to Greta Thunberg, the activist best known for anti-climate change protests, who to you, I would say represents a kind of symbol of a kind of anti-growth, effectively authoritarian, environmentalist dominated future. 彼得·蒂尔 当然。但我们还没走到那一步。我们还没走到那一步。如果你—— PETER THIEL Sure. But we’re not there yet. We’re not there yet. You know, it would be, it’d be like a very, very different society if you, if you, if you, if you. 罗斯·多塔特 真正生活在一种去增长的——你知道——小斯堪的纳维亚式的村落中。 ROSS DOUTHAT Actually lived in a kind of degrowth, you know, small Scandinavian villages. 彼得·蒂尔 我不确定它是否像朝鲜，但它一定是极度压迫人的。 PETER THIEL I’m not sure it would be North Korea, but it would be, it would be super oppressive. 危机寻求的危险 The Danger of Crisis-Seeking 罗斯·多塔特 有一件事始终让我印象深刻：当社会弥漫着停滞感、衰败感——借用我偏爱的那个词——你会发现一些人开始渴望危机。渴望某个时刻的到来，让他们可以彻底将社会从当前的轨道转向另一条道路。 ROSS DOUTHAT One thing that’s always struck me is that when you have this sense of stagnation, a sense of decadence. Right. To use, to, to use a word that I, I like to use for it in, in a society, you then also have people who end up being kind of eager for a crisis. Right. Eager for a moment to come along where, you know, they can, they can radically redirect society from the path it’s on. 罗斯·多塔特 因为我倾向于认为，在富裕社会中，当财富积累到一定程度，人们会变得非常安逸，厌恶风险，而要从衰败中走向某种新事物确实很难——没有危机就很难做到。对我来说，最典型的例子是9·11事件之后，外交政策保守派中弥漫着一种观念：我们此前一直深陷衰败与停滞，现在是我们觉醒并发动新十字军、重塑世界的时候了。显然那个结果非常糟糕。但类似的思潮—— ROSS DOUTHAT Because I tend to think that in rich societies you hit a certain level of wealth, people become very comfortable, they become risk averse, and it’s just hard, it’s hard to get out of decadence into something, into something new without a crisis. So the original example for me was after September 11, there was this whole mentality among foreign policy conservatives that we had been decadent and stagnant and now is our time to, you know, wake up and launch a new crusade and remake the world. And obviously that ended very badly. But something similar it was, it was. 彼得·蒂尔 小布什直接告诉人们去购物就好了。 PETER THIEL Bush 43 just told people to go shopping right away. 罗斯·多塔特 所以对他们来说并没有反衰败。 ROSS DOUTHAT So it wasn’t anti-decadent for them. 彼得·蒂尔 大部分情况下是这样。所以你看，也许在一些新保守派外交政策圈子里，有人在”角色扮演”以试图走出衰败。但最主流的态度是小布什告诉人们——你们继续去购物吧。 PETER THIEL For the most part. So you, there was, there was, maybe there was some neocon foreign policy enclave in which people were larping as a way to get out of decadence. But the, the dominant thing was Bush 43, people telling people just to go shopping. 罗斯·多塔特 那么，为了逃脱衰败，你愿意承担多大的风险？似乎这里确实存在一种危险，那就是想要反衰败的人必须承受大量风险。他们必须说，看，你们有这个好的、稳定的、舒适的社会，但猜猜呢？我们希望打一场仗，或者制造一场危机，又或者完全重组政府等等。他们必须倾向于冒险。 ROSS DOUTHAT So what risks should you be willing to take to escape decadence? It does seem like there’s a danger here where the people who want to be anti-decadent have to take on a lot of risk. They have to say, look, you’ve got this nice, stable, comfortable society, but guess what? We’d like to have a war or a crisis or a total reorganization of government and so on. They have to lean into, into danger. Right. 彼得·蒂尔 我不知道我能否给你一个精确的答案，但我的方向性回答是——多得多，我们应该承担更多风险。我们应该做得更多，而且我不知道，我可以举例说明所有这些不同的领域。 PETER THIEL I don’t know if I have to answer, you know, I don’t know if I’ve give you a precise answer, but my directional answer is a lot more, we should take a lot more risk. We should be doing a lot more and I don’t know, I can go through all these different verticals. 彼得·蒂尔 比如说，如果我们看生物技术，像痴呆症、阿尔茨海默症这种疾病，我们在40到50年间几乎没有任何进展。人们完全困在β淀粉样蛋白上。显然行不通。这只是某种愚蠢的勾当，人们只是在互相强化。所以是的，我们在这个方面确实需要承担更多风险。 PETER THIEL It’s, you know, if we, if we look at biotech, something like dementia, Alzheimer’s, we’ve made zero progress in 40 to 50 years. People are completely stuck on beta amyloids. It’s obviously not working. It’s just some kind of a stupid racket where the people are just reinforcing themselves. And so, yes, we need to take way more risk in that department. 医学研究中的风险 Taking Risks in Medical Research 罗斯·多塔特 好吧，我想问一件具体的事情，让我们保持具体，我想在这个例子上再停留一下，问一下好的，说”我们需要在抗衰老研究中承担更多风险”到底意味着什么？这是意味着FDA必须后退，说任何人只要有治疗阿尔茨海默症的新疗法就可以直接在开放市场上销售吗？医学领域中的风险究竟是什么样子的？ ROSS DOUTHAT Well, I want to ask, to keep us in the concrete, I want to stay with that example for a minute and ask, okay, what does that mean saying we need to take more risks in anti-aging research? Does it mean that the FDA has to step back and say anyone who has a new treatment for Alzheimer’s can, you know, go ahead and sell it on the open market? Like, what is, what is, what is risk in the medical space look like? 彼得·蒂尔 是的，你确实会承担更多风险。如果你有某种致命疾病，你可能可以承受更多的风险。研究人员也可以承受更多的风险。 PETER THIEL Yeah, you would take a lot more risk. You know, if you have some fatal disease, there probably are a lot more risks you can take. There are a lot more risks the researchers can take. 彼得·蒂尔 从文化层面想象一下，那就像早期现代时期，那时人们确实认为我们会治愈疾病。他们认为我们会实现根本性的延长寿命、甚至不朽。这是早期现代的项目，是弗朗西斯·培根、孔多塞的信条。也许这反基督教，也许是基督教竞争的下游产物。如果基督教承诺了你肉体的复活，那么科学就必须承诺同样的事情才能成功。 PETER THIEL Culturally, what I imagine it looks like is early modernity where people, yeah, they thought we would cure diseases. They thought we would have radical life extension, immortality. That was part of the project of early modernity. It was Francis Bacon, Condorcet. And maybe it was anti Christian, maybe it was downstream of Christianity was competitive. If Christ, if Christianity promised you a physical resurrection, you know, science was not going to succeed unless it promised you the exact same thing. 彼得·蒂尔 我不知道。我记得1999年、2000年，那时我在经营PayPal，我的联合创始人之一卢克·诺塞克（Luke Nosek）对Alcor和人体冷冻学很感兴趣，他认为人们应该在死后冷冻自己。有一天我们带着整个公司去参加了一个”冷冻派对”——你知道，就像特百惠派对一样。在冷冻派对上人们推销冷冻保单。 PETER THIEL But I don’t know. I remember 1999, 2000, when I was running PayPal, one of my co founders, Luke Nosek, he was into Alcor and cryonics and people should freeze themselves. And we had one day where we took the whole company to a “freezing party.” You know, Tupperware party. People sell Tupperware policies at a freezing party. 罗斯·多塔特 他们出售的——只是头部会被冷冻吗？ ROSS DOUTHAT They sell their. Was it just your heads what was going to be frozen? 彼得·蒂尔 你可以选择全身或者只有头部。 PETER THIEL You could get a full body or just the head? 罗斯·多塔特 只冷冻头部比较便宜。 ROSS DOUTHAT Just the head was cheaper. 彼得·蒂尔 当点阵打印机出故障、冷冻保单打不出来的时候，场面确实挺诡异的。但回想起来，这仍然是技术停滞的一个缩影。 PETER THIEL It was disturbing when the dot matrix printer didn’t quite work and so the freezing policies couldn’t be printed out. But in retrospect, this was still technological stagnation once again. 罗斯·多塔特 确实。 ROSS DOUTHAT Right. 彼得·蒂尔 但这也是衰落的一个症候。在1999年，这并不是主流观点，但仍然存在一种边缘的婴儿潮一代的信念——他们仍然相信自己可以永生。而他们是最后一代这样想的人。我一向反感婴儿潮一代，但也许我们连这种边缘的婴儿潮式自恋都失去了——至少那时还有几个婴儿潮一代相信科学能治愈他们所有的疾病。今天没有哪个千禧一代还相信这种事了。 PETER THIEL But it’s also a symptom of the decline, because in 1999, this was not a mainstream view, but there were still a fringe boomer view where they still believed they could live forever. And that was the last generation. So I’m always anti boomer, but maybe there’s something we’ve lost even in this fringe boomer narcissism, where there were at least a few boomers who still believed science would cure all their diseases. No one who’s a millennial believes that anymore. 政治风险投资 Political Venture Capitalism 罗斯·多塔特 我认为还有一些人相信另一种不朽——现在，我认为对人工智能的痴迷有一部分与超越极限的特定愿景有关。在问你这个之前，我想先问你一些关于政治的事情。 ROSS DOUTHAT I think there are some people who believe in a different kind of immortality, though. Right now, I think part of the fascination with AI is connected to a specific vision of transcending limits. And I’m going to ask you about that after I ask you about politics. 罗斯·多塔特 因为你最初关于停滞的论断——主要关于技术和经济——中让我印象深刻的事情之一是，它可以被应用到很广泛的领域。而当你写那篇文章时，你对海洋家园感兴趣——本质上就是建造独立于僵化的西方世界的新政治体。 ROSS DOUTHAT Because one of the striking things I thought about your original argument on stagnation, which was mostly about technology and the economy, was that it could be applied to a pretty wide range of things. And at the time you were writing that essay, you were interested in seasteading this ideas of essentially building new polities independent of the sclerotic Western world. 罗斯·多塔特 但后来你在2010年代做了一个转变。你是少数——也许是唯一——硅谷中最显眼的唐纳德·特朗普支持者。2016年，你支持了几个经过仔细筛选的共和党参议院候选人。其中一个现在成为了美国副总统。而我作为观察者的看法是，你基本上是在做政治风险投资。你在说，这里有一些可能改变政治现状的颠覆性代理人，值得承受一定风险。你是这么想的吗？ ROSS DOUTHAT But then you made a pivot in the 2010s. So you were one of the few prominent, maybe the only prominent Silicon Valley supporter of Donald Trump. In 2016, you supported a few sort of carefully selected Republican Senate candidates. One of them is now the Vice President of the United States. And my view as an observer of what you were doing was that you were basically being a kind of venture capitalist for politics. Right. You were saying, here are some disruptive agents who might change the political status quo and it’s worth a certain kind of risk here. Is that how you thought about? 彼得·蒂尔 当然。这有多重考量。一方面是——我们希望让泰坦尼克号避开它正驶向的冰山，或者换个比喻——真正地通过政治途径让社会改变航向。 PETER THIEL Sure. There were all sorts of levels. I mean, one level was, yeah, was these hopes that we could redirect the Titanic from the iceberg was heading to, or whatever the metaphor is, you know, really change course as a society through political piece. 彼得·蒂尔 也许更窄的、更聚焦的愿望是，我们也许至少可以让这个话题被谈论起来。当特朗普说”让美国再次伟大”时，好的，这是一个正面的、乐观的、有雄心的议程，还是仅仅是对我们现在所处状态的非常悲观的评估——我们不再是一个伟大的国家？ PETER THIEL Maybe a narrower, a much narrower aspiration was that we could maybe at least have a conversation about this. When someone like Trump said, “make America great again,” okay, is that a positive, optimistic, ambitious agenda, or is it merely a very pessimistic assessment of where we are, that we are no longer a great country? 彼得·蒂尔 我对特朗普能做出什么积极贡献并没有太大期望，但我认为，至少在一百年来第一次，我们有了一个不再给我们灌输那种甜腻虚伪的布什式废话的共和党人。这并不等同于进步，但我们至少可以开始对话了。 PETER THIEL And I didn’t have great expectations about what Trump would do in a positive way, but I thought, at least for the first time in a hundred years, we had a Republican who was not giving us this syrupy Bush nonsense. And that was not the same as progress, but we could at least have a conversation. 彼得·蒂尔 回顾过去，这简直是一种荒谬的幻想。我在2016年有这两个想法，而且你往往会有一些想法恰好在你潜意识中。但我当时有的两个想法是——第一，如果特朗普输了，没有人会因为我支持特朗普而生气。第二，我认为他有50%的获胜机会。（书童注：注意这两个想法，构成了一个严格遵循塔勒布“反脆弱”的杠铃策略） PETER THIEL In retrospect, this was a preposterous fantasy. I had these two thoughts in 2016, and you often have these ideas that are just below the level of your sort of consciousness. But the two thoughts I had that I wasn’t able to combine was, number one, you know, nobody would be mad at me for supporting Trump if he lost. And number two, I thought he had a 50-50 chance of winning. 罗斯·多塔特 如果他输了为什么没有人会生气？ ROSS DOUTHAT Why would nobody be mad at you if he lost? 彼得·蒂尔 这实在太古怪了，而且也不会真有什么影响。 PETER THIEL It would just be such a weird thing and it wouldn’t really matter. 罗斯·多塔特 好的。 ROSS DOUTHAT Okay. 彼得·蒂尔 然后我认为，他有50%的获胜机会，因为问题是深层的，停滞令人沮丧。然后我的幻想就是——如果他赢了，我们就能开始这个对话。而现实是人们还没准备好。 PETER THIEL And then I thought, you know, he had more. He had, I thought he had a 50-50 chance because the problems were deep and the stagnation was frustrating. And then the fantasy I had was, yeah, if he won, we could have this conversation. And the reality was people weren’t ready for it. 彼得·蒂尔 而现在，也许到了2025年，特朗普崛起十年之后，我们终于可以进行这场对话了。当然你不是那种僵化的左翼人士，Ross。但这—— PETER THIEL And then, you know, maybe we’ve progressed to the point where we can have this conversation at this point in 2025, a decade after Trump. And of course, you’re not a zombie left wing person, Ross. But it’s, this is. 罗斯·多塔特 我被叫过很多名字，很多名字。 ROSS DOUTHAT I’ve been called many things, many things. 彼得·蒂尔 我愿意接受任何进步。 PETER THIEL I’ll take whatever progress I can get. 停滞对话 The Stagnation Conversation 罗斯·多塔特 从你的角度来说。假设有两层。一层是社会需要颠覆，需要风险。特朗普就是颠覆，特朗普就是风险。第二层是特朗普愿意说一些关于美国衰退的真话。 ROSS DOUTHAT So from your perspective of. So let’s say there’s two layers, right? There’s sort of a basic sense of, you know, this society needs disruption. It needs risk. Trump is disruption. Trump is risk. And the second level is Trump is actually willing to say things that are true about American decline. Right. 罗斯·多塔特 那么你觉得你作为一个投资者、风险投资家，在特朗普的第一届任期中有没有获得什么？特朗普在第一届任期中做了什么让你觉得是反衰败或者反停滞的？如果答案是什么都没有的话，也可以。 ROSS DOUTHAT So do you feel like you as an investor, as a venture capitalist, got anything out of the first Trump term? Like what did Trump do in his first term that you felt was anti decadent or anti stagnation? If anything, maybe the answer is nothing. 彼得·蒂尔 好吧，我认为——我认为它比我想要的花了更长时间，也更慢。但我们确实取得一定进展：很多人认为出了问题。而这在2012、2013、2014年就不是我在进行的对话了。 PETER THIEL Well, I think we, I think it took longer and it was slower than I would have liked. But we have gotten to the place where a lot of people think something’s gone wrong. And that was not the conversation I was having in 2012, 2013, 2014. 彼得·蒂尔 我在2012年和Eric Schmidt进行了一场辩论，在2013年和Marc Andreessen进行了辩论，在2014年和Bezos也进行了辩论。我站在”存在停滞问题”这一侧。而他们三个人都是”一切进展良好”的版本。我认为至少这三个人——在不同程度上——已经更新了，调整了，（整个）硅谷——调整了。 PETER THIEL I had a debate with Eric Schmidt in 2012 and Marc Andreessen in 2013 and Bezos in 2014. I was on the “There’s a stagnation problem.” And all three of them were versions of “everything’s going great.” I think at least those three people have, to varying degrees, updated and adjusted Silicon Valley’s. Adjusted. 罗斯·多塔特 硅谷——尽管比”调整”更多——硅谷很大一部分—— ROSS DOUTHAT Silicon Valley, though, has more than adjusted a big part of Silicon. 彼得·蒂尔 在停滞问题上。确实。 PETER THIEL Valley on the stagnation. Right. 罗斯·多塔特 在停滞问题上。但然后硅谷的很大一部分在2024年选择了特朗普（一边），包括显然最著名的伊隆·马斯克。 ROSS DOUTHAT On the stagnation. But then a big part of Silicon Valley ended up going in for Trump in 2024, including obviously, most famously, Elon Musk. 彼得·蒂尔 是的，在我的叙事框架中，这与停滞问题深度相连。当然，这些事情总是超级复杂的。我的解读是——而且我非常不愿意代表所有这些人发言——但比如像扎克伯格或Facebook Meta，在某种程度上，我不认为他很意识形态化。他并没有真正深入思考过这些问题。 PETER THIEL Yeah, this is deeply linked to the stagnation issue in my telling. I mean, these things are always super complicated, but my telling is, you know, I don’t, and again, I’m so hesitant to speak for all these people, but, you know, someone like Mark Zuckerberg or Facebook Meta, and, you know, in some ways, I don’t think he was, he’s very ideological. He didn’t think this stuff necessarily through that much. 彼得·蒂尔 默认的做法就是当一个自由派，而且一直以来的逻辑都是——如果自由主义不起作用，你该怎么办？年复一年，答案都是：加大力度。你不断加大剂量，花数亿美元，完全走”觉醒”路线，结果所有人都恨你。到了某个时刻你不得不想——好吧，也许这条路走不通。 PETER THIEL It was the default was to be liberal and it was always what, you know, if the liberalism isn’t working, what do you do? And for year after year after year, it was, you do more. You know, if something doesn’t work, you just need to do more of it. And you up the dose, and you up the dose and you spend hundreds of millions of dollars and you go completely woke and everybody hates you. And at some point it’s like, okay, maybe this isn’t working. 罗斯·多塔特 所以他们转变了方向。 ROSS DOUTHAT So they pivot. 彼得·蒂尔 是的，这不是一种亲特朗普的事情。 PETER THIEL Yes, it’s not a pro Trump thing. 罗斯·多塔特 这不是亲特朗普的事情，但确实，无论是在公开还是私下的对话中，都有一种感觉——在2024年——也许在2016年还不是这样，那时候Peter是唯一的支持者——但现在到了2024年，特朗普主义和民粹主义可以成为技术创新和经济活力的载体，等等。 ROSS DOUTHAT It’s not a pro Trump thing, but it is, you know, just both in public and private conversations, it is a kind of sense that Trumpism and populism in 2024, maybe not in 2016, when Peter was out there as the lone supporter. But now in 2024, they can be a vehicle for technological innovation, economic dynamism and so on. （上篇完）

LZN's Blog

【Cheeky Pint】模型忠诚、产品捆绑与组织文化 | 纳德拉与 John Collison 对谈 | 中英全文（下）

人们忠于模型，还是忠于品牌？

token 工厂与智能体工厂

开放平台与产品捆绑的取舍

谁来定义一家公司的文化？

二十万人公司的管理，以及创始人的工作记忆

海得拉巴、学校与板球

译注

【Cheeky Pint】算力瓶颈、企业主权与智能体商务 | 纳德拉与 John Collison 对谈 | 中英全文（中）

AI 热潮与互联网泡沫

瓶颈是芯片，还是能通电的机房？

数据主权与企业的隐性知识

Excel 为什么如此长寿？

从云计算到 AI 的需求跃迁

商品发现、结账与智能体商务

购物与客服的边界正在消失

译注

【Cheeky Pint】企业 AI、工作方式与互联网往事 | 纳德拉与 John Collison 对谈 | 中英全文（上）

开场剪辑

企业 AI：数据、记忆与权限

在虚拟走廊里管理微软

跟着开发者与初创公司走

智能体时代的界面：宏观委派、微观引导

互联网往事：看对范式，还要找对路径

译注

【e/acc】AI奇点 | 物理学家、e/acc运动创始人Guillaume Verdon与Lex Fridman播客实录 | 中英文完整版精译 IV（完结篇）

【e/acc】外星智能、卡尔达肖夫天梯、技术资本机器 | 物理学家、e/acc运动创始人Guillaume Verdon与Lex Fridman播客实录 | 中英文完整版精译 Part3

【e/acc】有效加速主义、量子热力学与AI未来 | 物理学家、e/acc运动创始人Guillaume Verdon与Lex Fridman播客实录 | 中英文完整版精译 II

【e/acc】有效加速主义、量子热力学与AI未来 | 物理学家、e/acc运动创始人Guillaume Verdon与Lex Fridman播客实录 | 中英文完整版精译 I

【酒后真言】AI工厂、物理智能：英伟达CEO黄仁勋与思科CEO Chuck Robbins炉边对话 | 中英文完整版精译

超级智能、顺从型AI、超人类主义、基督与超越：PayPal、Palantir联合创始人彼得·蒂尔2025年6月播客实录 | 中英文完整版精译（下篇）

技术停滞、回到未来、风险承担 | PayPal、Palantir联合创始人彼得·蒂尔2025年6月播客实录 | 中英文完整版精译 （上篇）

技术停滞、回到未来、风险承担 | PayPal、Palantir联合创始人彼得·蒂尔2025年6月播客实录 | 中英文完整版精译（上篇）