萨姆·奥尔特曼谈Sora、能源与构建AI帝国

某种程度上，我们以为偶然发现了这个巨大的秘密——语言模型的缩放定律，这感觉像是一个不可思议的胜利。我当时想，我们可能再也不会这么幸运了。而深度学习就像一场持续馈赠的奇迹，我们不断取得一个又一个突破。当我们实现推理模型的突破时，我又一次觉得，这样的突破恐怕再难重现。

Sort of thought we had like stumbled on this one giant secret that we had these scaling laws for language models and that felt like such an incredible triumph. I was like, we're probably never gonna get that lucky again. And deep learning has been this miracle that keeps on giving. And we have kept finding breakthrough after breakthrough. Again, when we got the reasoning model breakthrough, I also thought that was like, we're never gonna get another one like that.

这项技术如此有效简直令人难以置信。但也许重大科学突破的发现总是这种感觉——如果它足够重大、足够基础，它就会持续发挥作用。OpenAI不仅仅是在开发一个应用程序。

And it just seems so improbable that this one technology works so well. But maybe this is always what it feels like when you discover, like, one of the big scientific breakthroughs is if it's, like, really big, it's pretty fundamental, and it just it keeps working. OpenAI isn't just building an app.

它正在建造人类历史上最大的数据中心。昨天，我与本·霍洛维茨和OpenAI首席执行官萨姆·奥特曼进行了对话。我们探讨了OpenAI成为人们个人AI的愿景、背后的庞大基础设施，以及公司研究如何推动通用人工智能（AGI）的发展，包括能进行真正科学研究的AI。我们还谈到他对开源和监管看法的转变，以及为何AI与能源如今紧密相连。让我们深入探讨。

It's building the biggest data center in human history. Yesterday, I sat down with Ben Horowitz and Sam Altman, CEO of OpenAI. We talk about OpenAI's vision to become the people's personal AI, the massive infrastructure behind it, and how the company's research is pushing toward AGI, including AI that can do real science. We also talk about how his views have changed on open source, regulation, and why AI and energy are now deeply linked. Let's get into it.

萨姆，欢迎来到AZNZ播客。谢谢邀请。你在另一次采访中将OpenAI描述为四家公司的结合体：消费科技企业、超大规模基础设施运营商、研究实验室，以及包括计划中的硬件设备在内的所有新业务。从硬件到应用集成，从就业市场到商业领域，这些布局最终指向什么？

Sam, welcome to the AZNZ Podcast. Thanks for having me. You've described in another interview, you described OpenAI as a combination of four companies, consumer technology business, a mega scale infrastructure operation, a research lab, and all the new stuff, including planned hardware devices. From hardware to app integrations, to job marketplace to commerce. What do all these bets add up to?

OpenAI的愿景是什么？

What's OpenAI's vision?

是的。或许可以算作三个核心方向——或者按我们自己的划分方式，把传统意义上的大规模研究实验室算作第四个。我们希望成为人们的个人AI订阅服务。我认为大多数人会有一个，有些人会有多个。

Yeah. I mean, maybe you should count it as three, maybe as four for kind of our own version of what traditionally would have been the research lab at this scale, but three core ones. We want to be people's personal AI subscription. I think most people have one. Some people have several.

你会在我们提供的原生消费场景中使用它，也会登录其他服务，未来还会通过专用设备使用。你将拥有一个了解你、对你真正有用的AI，这就是我们的目标。事实证明，要实现这一点，我们还必须建设庞大的基础设施。但根本使命始终是：打造这个通用人工智能，并让它对人类非常有用。

And you'll use it in some first party consumer stuff with us, but you'll also log in to a bunch of other services, and you'll just use it from dedicated devices at some point. You'll have this AI that gets to know you and be really useful to you, and that's what we wanna do. It turns out that to support that, we also have to build out this massive amount of infrastructure. But the goal there, the mission is really, like, build this AGI and make it very useful to people.

那么基础设施，你认为它最终会——是的，这对主要目标是必要的。它是否会独立发展成另一项业务，还是仅仅服务于个人AI或其他未知用途？你是说，比如，

And does the infrastructure do you think it will end up yeah. It's necessary for the main goal. Will it also separately end up being another business, or is it just really gonna be in service to the personal AI or unknown? You mean, like,

我们会在运营基础设施的同时将其出售给其他公司吗？

would we sell it to other companies as we run infrastructure?

没错。你们会卖给其他公司吗？要知道，这可是个庞然大物。它会有其他用途吗？

Yeah. Would you sell it to other companies? You know, it's such a massive thing. Would it do something else?

我感觉未来可能会出现类似的其他业务，但我不确定。我们目前没有

It feels to me like there will emerge some other thing to do like that, but I don't know. We don't have

现成的计划。不，不是的。

a current Yeah. No. No. Is.

是的。目前它只是为了支持我们想要提供的服务和研究。

Yeah. It's currently just meant to, like, support the service we wanna deliver and the research.

明白了。这很合理。

Yeah. No. That makes sense.

是啊。这个规模某种程度上已经大到令人恐惧，以至于你必须考虑其他可能性。确实。

Yeah. The scale is sort of, like, terrifying enough that you gotta be open to doing something else. Yeah.

如果你正在建造人类历史上最大的数据中心。

If you're building the biggest data center in the history of humankind.

史上最庞大的基础设施项目。没错。确实如此。

The biggest infrastructure project in the history. Yeah. Yeah.

多年前你在严格风投的早期公开访谈中——那时ChatGPT还未问世——被问到商业模式时，你说'我们会问AI，它会帮我们找到答案'，当时所有人都笑了。但事实上

There was a great interview you did many years ago in strictly VC, early open app, well before ChatGPT, and they're asking, what's the business model? And you said, oh, we'll we'll ask AI. It'll it'll figure it out for us. Everybody laughs. But There

已经发生过多次，最近还有一次，我们向当时最新的模型征求意见，它给出了极具洞察力的答案。当我们说这种话时，人们总以为我们在开玩笑。但或许正确答案是：你们应该同时认真对待这两种态度。

have been multiple times, and there was just another one recently where we have asked a then current model for what should we do, and it has had a insightful answer. I think when we say stuff like that, people don't take us seriously or literally. Yeah. But maybe the answer is you should take us both.

确实。不过说真的，作为组织管理者，我经常向AI咨询决策建议，它总能给出些相当有趣的答案。

Yeah. Yeah. Well, no. As as somebody runs an organization, I ask the AI a lot of questions about what I should do. It comes up with some pretty interesting answers.

有时候吧。嗯。现在确实偶尔会这样。

Sometimes. Yeah. Yeah. Sometimes now.

确实如此。你必须提供足够的上下文。

It does. You have to give it enough context.

除了更多分发、更多计算之外，将这些投资联系起来的核心理念是什么？我的意思是，研究使我们能够打造出色的产品，而基础设施使我们能够进行研究。所以这有点像是一个垂直堆叠的体系。比如，你可以使用ChatGPT或其他服务来获取关于如何运营组织的建议。但要让这发挥作用，需要出色的研究和大量的基础设施。

What is the thesis that connects these bets beyond more distribution, more compute? I mean, the research enables us to make the great products, and the infrastructure enables us to do the research. So it is kind of like a vertical stack of things. Like, you can use ChatchBT or some other service to get advice about what you should do running an organization. But for that to work, it requires great research and requires a lot of infrastructure.

所以这基本上就是同一件事。

So it is kind of just this one thing.

你认为未来会有一个时间点让这一切完全横向发展，还是在可预见的未来仍将保持垂直整合？

And do you think that there will be a point where that becomes completely horizontal, or will it stay vertically integrated for the foreseeable future?

我以前一直反对垂直整合。但现在我认为我错了。是的，很有意思。因为你可能会认为经济在理论上是高效的，公司可以专注于一件事然后就能成功。

I was always against vertical integration. And I now think I was just wrong about that. Yeah. Interesting. Because you'd like to think that the economy is efficient in the theory that companies can do one thing and then that's supposed to work.

我也这么认为。

I'd like to think that.

是的。但至少在我们的案例中，事实并非如此。当然，在某些方面确实如此。比如，英伟达制造了很多人可以使用的出色芯片。但OpenAI的故事显然是我们必须比预想的做更多事情，才能实现使命。

Yeah. And in our case, at least, it hasn't really. I mean, it has in some ways for sure. Like, you know, NVIDIA makes an amazing chip or whatever that a lot of people can use. But the story of OpenAI has certainly been towards we have to do more things than we thought to be able to deliver on the mission.

没错。

Right.

尽管计算机行业的历史某种程度上像一场拉锯战——先是王安文字处理器，接着是个人电脑，黑莓手机先于智能手机出现。所以这种垂直整合时而存在时而不在，但iPhone同样是垂直整合的产物。

Although the history of the computing industry has kind of been a story of kind of a back and forth in that there was the Wang word processor and then the personal computer and the BlackBerry before the smartphone. So there has been this kind of vertical integration and then not, but then the iPhone is also vertically integrated.

而我认为iPhone是科技行业有史以来最了不起的产品，它的垂直整合程度极高。

And the iPhone, I think, is the most incredible product the tech industry has ever produced, and it is extraordinarily vertically integrated.

是啊，惊人地整合。嗯，有意思。

Yeah. Amazingly so. Yeah. Interesting.

你会说哪些布局是AGI的助推器，哪些又算是对不确定性的对冲？

Which bets would you say are enablers of AGI versus which are sort of hedges against uncertainty?

表面上看，比如Sora似乎与AGI无关，但我敢打赌，如果我们能构建真正出色的世界模型，这对AGI的重要性将远超人们想象。很多人曾认为聊天机器人跟AGI不沾边，但它不仅帮助我们构建更好的模型、理解社会如何使用这项技术，还推动社会意识到：天啊，我们现在必须认真对待这东西了。在CHAJPYT之前，我们谈论AGI时人们总说‘这不会发生，我们不关心’，突然间他们却高度重视起来。

I think you could say that on the surface, Sora, for example, does not look like it's AGI relevant, but I would bet that if we can build really great world models, that'll be much more important to AGI than people think. There were a lot people who thought chatty bitty was not a very AGI relevant thing, and it's been very helpful to us not only in building better models and understanding how society wants to use this, but also in, like, bringing society along to actually figure out, man, we gotta contend with this thing now. We for a long time before CHAJPYT, we would talk about AGI, people were like, this is not happening. We don't care. And then all of a sudden, they really cared.

抛开研究收益不谈，我坚信社会与技术必须协同进化。是的，不能最后才把成果丢出来，那样行不通。这更像是一场持续的互动。

And I think that research benefits aside, I'm a big believer that society and technology have to co evolve. It's Yeah. Can't just drop the thing at the end. It doesn't work that way. It is a sort of ongoing back and forth.

是啊。

Yeah.

多谈谈Sora如何融入你们的战略吧，因为X平台上有些议论，比如：为什么要将宝贵的GPU资源分配给Sora？这是短期与长期的取舍，还是我们过于依赖智能体了？

Say more about how Sora fits into your strategy because there are some hullabaloo on x around, hey. Why devote precious GPUs to Sora? But is it a short term, long term trade off, or are we so agent?

而新版本在社交网络功能上有个非常有趣的转折。我很想了解你们对此的思考。Meta有没有打电话来发火？或者你预计他们会有什么反应

And then the new one had, like, very interesting twist with the social networking. Be very interested in kinda how you're thinking about that. And did Meta call you up and get mad? Or what do you expect their reaction

呢？我觉得如果我们两家公司中有一方觉得另一方在针对自己，那他们也不该打电话来质问我们。

to be? I think if one company of the two of us has feels like more like the other one has gone after them, it wouldn't they shouldn't be calling us.

呃，我可没有这样的历史

Well, I do not have a history to

但首先，我认为打造优秀产品很酷，人们喜欢新版Sohr。同时让社会提前感知这个协同进化节点的未来也很重要。很快世界就将面对能深度伪造任何人、呈现任何画面的强大视频模型——这总体上很棒，但社会需要适应期。就像ChatGPT时期那样，世界需要认清技术发展方向。

But first of all, I think it's cool to make great products, and people love the new Sohr. And I also think it is important to give society a taste of what's coming on this coevolution point. So, like, very soon, the world is gonna have to contend with incredible video models that can deepfake anyone or kind of show anything you want, and that will mostly be great. There will be some adjustment that society has to go through. And just like with chat GPT, we were like, the world kinda needs to understand where this is.

我认为让世界快速理解视频技术的演进至关重要，因为视频比文字更具情感冲击力。很快我们将生活在一个这类技术无处不在的世界。正如我所说，这不仅会助力我们的研究计划，也是通向AGI的道路。但不能只追求让人冷酷高效或让AI解决所有问题。

I think it's very important the world understands where video is going very quickly because video has much more, like, emotional resonance than text. And very soon, we're gonna be in a world where, like, this is gonna be everywhere. So I think there's something there. As I mentioned, I think this will help our research program and is on the AGI path. But, yeah, it can't all be about just making people, like, ruthlessly efficient and the AI, like, solving all our problems.

在这个过程中，必须要有乐趣、快乐和愉悦。但我们不会投入大量计算资源，甚至不会动用我们计算能力的一小部分。

There's gotta be, like, some fun and joy and delight along the way. But we won't throw, like, tons of compute at it or not by a fraction of our compute.

没错。从绝对数量看是很多，但从相对比例来看并不算多。是的，是的。

Yeah. It's tons in the absolute sense, but not in the relative sense. Yep. Yep.

我想探讨AI人机交互的未来，因为你在八月份说过模型已经饱和了聊天应用场景。那么未来的AI人机交互会是什么样子，包括硬件和软件方面？愿景会是类似微信那样的形态吗？

I wanna talk about the future of AI human interfaces because back in August, you said the models have already saturated the chat use case. So what do future AI human interfaces look like, both in terms of hardware and software? Is the vision for kind of WeChat, like

从狭义上讲，聊天功能已经解决得很好了——如果你只想进行最基本的聊天式对话，它表现非常出色。但聊天界面能为你做的事情远未饱和。比如你可以要求聊天界面'请治愈癌症'，目前模型显然还做不到。所以我认为即使闲聊场景已很成熟，文本交互方式仍有巨大发展空间。

So solving the chat thing in a very narrow sense, which is if you're trying to, like, have the most basic kind of chat style conversation, it's very good. But what a chat interface can do for you, it's like nowhere near saturated. Because you could ask a chat interface like, please cure cancer. A model certainly can't do that yet. So I think the text interface style can go very far even if for the chitchat use case, the models are already very good.

当然，还有更好的交互方式。其实我觉得Sora另一个很酷的点在于——你可以想象一个界面完全由实时渲染视频构成的世界。嗯，这将开启的可能性非常令人兴奋。

But, of course, there's better interfaces to have. Actually, it's another thing that I think is cool about Sora. Like, you can imagine a world where the interface is just constantly real time rendered video. Yeah. And what that would enable, and that's pretty cool.

你可以设想新型硬件设备能持续感知环境状态。不同于手机随时用短信通知轰炸你，它真正理解你的情境，知道何时该展示什么内容。这方面还有很长的路要走。

You can imagine new kinds of hardware devices that are sort of always ambiently aware of what's going on. And rather than your phone, like, blast you with text message notifications whenever it wants, like, it really understands your context and when to show you what. And there's a long way to go on all that stuff.

是的。未来几年内，模型将能实现哪些现在做不到的事？会是更深层次的白领替代吗，比如AI科学家、人形机器人？

Yeah. Within the next couple years, what will models be able to do that they're not able to do today? Will it be sort of white collar replacement at a much deeper level, AI scientist, humanoids?

我是说，有很多事情，但你提到了我最兴奋的一点，那就是AI科学家。是的，我们居然坐在这里认真讨论这个，这太疯狂了。我知道关于图灵测试的确切定义有些争议，但大众认知中的图灵测试就这么嗖的一下被超越了。

I mean, a lot of things, but you touched on the one that I am most excited about, which is the AI scientist. Yeah. This is crazy that we're sitting here seriously talking about this. I know there's, like, a quibble on what the Turing test literally is, but the popular conception of the Turing test sort of went whooshing by.

是啊，快得不可思议。确实。

Yeah. That was fast. Yeah.

你知道吗？长久以来我们一直把它视为AI最重要的测试标准，感觉遥不可及。然后突然间就被突破了。全世界大概震惊了一两周，然后反应就是：好吧，看来现在电脑真能办到了。

You know, it was just like we talked about it as this most important test of AI for a long time. It seemed impossibly far away. Then all of a sudden, it was passed. The world freaked out for, like, a week, two weeks, and then it's like, alright. I guess computers, like, can do that now.

没错。然后一切照常继续。我认为科学领域正在重演这一幕。对我个人而言，AI的'图灵测试'始终是它能否从事科研工作。那才是真正改变世界的节点。

Yeah. And everything just went on. And I think that's happening again with science. My own personal equivalent of the Turing test has always been when AI can do science. That is a real change to the world.

而GPT-5首次让我们看到了这种苗头。推特上那些案例：它做出了新颖的数学发现，在我的物理和生物研究中完成了某个小突破。种种迹象表明，这只是个开始。两年内，我认为模型将承担更重大的科研任务，取得重要发现。

And for the first time with GPT-five, we are seeing these little examples where it's happening. You see these things on Twitter. It did this it made this novel math discovery and did this small thing in my physics research and my biology research. And everything we see is that that's gonna go much further. So in two years, I think the models will be doing bigger chunks of science and making important discoveries.

这简直疯狂。这将深刻影响世界。我始终相信，从根本上说，科学进步是推动世界向好的原动力。如果这个进程即将加速，那就是翻天覆地的变化。

And that is a crazy thing. Like, that will have a significant impact on the world. I am a believer that to a first order, scientific progress is what makes the world better over time. And if we're about to have a lot more of that, that's a big change.

有意思的是，这个积极变化却少有人讨论。当AI变得极度聪明时，人们的关注点总是集中在负面改变上。

It's interesting because that's a positive change that people don't talk about. It's gotten so much into the realm of the negative changes if AI gets extremely smart.

但治愈疾病是

But curing up a disease is

我们确实需要更多科学。这是个非常好的观点。我记得艾伦·图灵说过，有人问他，你真的认为计算机会比天才更聪明吗？他说，它不需要比天才更聪明，只要比平庸的人更聪明就行，比如AT&T的总裁。我们或许也该多用用这种思路。

We just could use a lot more science. That's a really good point. I think Alan Turing said this, somebody asked him, they said, Well, you really think the computer's gonna be smarter than brilliant minds? He said, It doesn't have to be smarter than a brilliant mind, just smarter than a mediocre mind like the president of AT and T. We should use more of that too, probably.

上周我们刚看到OpenAI校友创办的Periodic公司发布新产品。说到这个，看到你们团队的创新成果，以及从OpenAI走出来的团队不断创造出惊人成就，实在令人惊叹。

We just saw Periodic launch last week of OpenAI alums. And to that point, it's amazing to see both the innovation that you guys are doing, but also the teams that come out of OpenAI just feels like are creating tremendous capable of things.

我们当然希望如此。

We certainly hope so.

是的。我想请教你更宏观的看法——自ChatGPT问世以来，2025年扩散模型或AI发展中有哪些让你惊讶的突破？或者说哪些发现改变了你的世界观？

Yeah. I wanna ask you about just broader reflections in terms of what sort of about diffusion or development in 2025 has surprised you, or what has sort of updated your worldview since Chat two d came out?

有很多方面，但最有趣的可能是我们发现的新事物比预想多得多。我们原以为偶然发现了语言模型scouting loss这个重大秘密，当时觉得这简直是不可思议的胜利，心想可能再也不会这么幸运了。但深度学习就像持续给予的奇迹，我们不断取得一个又一个突破。当我们获得推理模型的突破时，我又觉得这种突破不可能再现——这项技术效果如此之好，简直难以置信。

A lot of things again, but maybe the most interesting one is how much new stuff we found. Sort of thought we had, like, stumbled on this one giant secret that we had these scouting loss for language models, and that felt like such an incredible triumph that I was like, we're probably never gonna get that lucky again. And deep learning has been this miracle that keeps on giving, and we have kept finding, like, breakthrough after breakthrough. Again, when we got the reasoning model breakthrough, I also thought that was like, we're never gonna get another one like that. It just seems so improbable that this one technology works so well.

但或许所有重大科学突破被发现时都是这种感觉。如果是真正重大的突破，它往往非常基础，且持续有效。但进步幅度之大...如果你回头用ChatGPT发布时的GPT-3.5，你会觉得难以置信居然有人用这种东西。而现在我们身处能力过剩如此巨大的时代。

But maybe this is always what it feels like when you discover one of the big scientific breakthroughs. If it's really big, it's pretty fundamental, and it just it keeps working. But the amount of progress like, if you went back and used GPT 3.5 from ChatGPT launch, you'd be like, I cannot believe anyone used this thing. Yeah. And now we're in this world where the capability overhang is so immense.

就像，世界上大多数人还在思考ChatGPT能做什么。然后硅谷的一些极客已经在使用Codex了，他们惊叹不已。那些人根本不知道发生了什么。接着又有少数科学家说，那些使用Codex的人也不清楚状况。但如今的能力储备已经如此庞大，我们在模型功能方面已经取得了巨大进展。

Like, most of the world still just thinks about what ChatGPT can do. And then you have some nerds in Silicon Valley that are using Codecs, they're like, wow. Those people have no idea what's going on. And then you have a few scientists who say those people using codex have no idea what's going on. But the overhang of capability is so big now, and we've just come so far on what the models can do.

就进一步发展而言，大型语言模型（LLMs）能走多远？在什么节点我们需要新的架构？你认为需要哪些突破性进展？

And in terms of further development, how far can we get with LLMs? At what point do we need either new architecture? How do think about what breakthroughs are needed?

我认为足够远，远到我们可以用现有技术创造出能发现下一次突破的东西。这听起来像是个自我指涉的答案，但如果基于LLM的技术能发展到比OpenAI所有研究人员加起来更擅长做研究，那或许就足够了。

I think far enough that we can make something that will figure out the next breakthrough with the current technology. Like, it's a very self referential answer, but if LLM based stuff can get far enough that it can do, like, better research than all of OpenAI put together, maybe that's, like, good enough.

没错。那将是个重大突破。非常重大的突破。说到更世俗的问题，人们开始抱怨的一点——我记得《南方公园》还专门做过一集——就是AI尤其是ChatGPT那种卑躬屈膝的特质。解决这个问题难度有多大？

Yep. That would be a big breakthrough. A very big breakthrough. So on the more mundane, one of the things that people have kind of started to complain about, I think South Park did a whole episode on it, is kind of the obsequiousness of kind of AI and ChatGPT in particular. And how hard a problem is that to deal with?

是没那么难，还是说本质上就是个难题？

Is it not that hard, or is it like kind of a fundamentally hard problem?

哦，这一点都不难

Oh, it's not at all hard to

是啊。

Yeah.

处理一下。很多用户确实很想要这个功能。是的。没错。就像好吧。

Deal with. A lot of users really want it. Yeah. Yeah. Like Okay.

如果你去看看网上人们对Chateappity的评价，是的，有很多人真的希望恢复那个功能。是的。而且技术上来说，这根本不难实现。还有一点，虽然完全不意外，但用户需求的分布范围极其广泛。

If you go look at what people say about Chateappity online Yeah. There's a lot of people who, like, really want that back. Yeah. And it is so it's not technically, it's not hard to deal with at all. One thing, and this is not surprising in any way, but the incredibly wide distribution of what users want Yeah.

关于他们希望聊天机器人如何在大小细节上表现。

Of, like, how they'd like a chatbot to behave in big and small ways.

那你觉得最终是不是得配置个性呢？这会是个解决方案吗？

Does that do you end up having to configure the personality then, you think? Is that gonna be the answer?

我觉得是。理想情况下，你只需要和ChatGPT聊一会儿，它会通过对话了解你的喜好。

I think so. I mean, ideally, you just talk to ChatGPT for a little while, and it kinda interviews you and also sort of sees what you like and don't like.

然后Touchy PT就能自动适应但

And touchy PT just figures how But

短期内你可能得先手动选择一个模式。

in the short term, you'll probably just pick one.

明白了。是的。不。这很合理。非常有趣。

Got it. Yeah. No. That makes sense. Very interesting.

实际上，有件事我一直想请教的是

And actually so so one thing I wanted to ask about is

嘿。是啊。我觉得我们之前有个很天真的想法，你知道，就是认为能造出一个让几十亿人都想与之对话的同一对象，这本身就不太现实。但很长一段时间里，这却是我们默认的假设。

Hey. Yeah. Like, I think we just had a a really naive thing, which, you you know, like, it would sort of be unusual to think you could make something that would talk to billions of people, and everybody wants to talk to the same person. Yeah. And and yet that was sort of our implicit assumption for a long time.

没错。因为每个人的朋友都大不相同。

Right. Because people have very different friends.

每个人的朋友都大不相同。

People have very different friends.

是啊。

Yeah.

所以现在我们正试图修正这个问题。

So now we're trying to fix that.

是的。而且朋友类型不同，兴趣各异，智力水平也参差不齐。所以你并不想总是谈论同样的话题。其中一个很棒的功能是，你可以说‘用五岁小孩能懂的方式解释给我听’，但或许我连这个步骤都想省去。也许我希望你永远用那种方式和我交流

Yeah. And also kind of different friends, different interests, different levels of intellectual capability. So you don't really wanna be talking to the same thing all the time. And one of the great things about it is you can say, well, explain it to me like I'm five, but maybe I don't even wanna have to do that front. Maybe I always want you to talk to me

就像对待五岁孩子那样，

like I'm five,

对，特别是在教我东西的时候。我想问你一个CEO式的问题——观察你对我来说很有趣，你刚和AMD达成了这笔交易。当然，公司现在处境不同，你也有了更多筹码之类的。但自最初那笔交易以来，这些年你的思维方式是否发生了变化？如果有的话，是怎样的变化？

yeah, particularly for teaching me stuff. I want to ask you a kind of like a CEO question, which has been interesting for me to observe you, is you just did this deal with AMD. And, of course, the company's in a different position, and you have more leverage and these kinds of things. But how has your kind of thinking changed over the years since you did that initial deal, if at all?

当当时我几乎没有运营经验。我对管理公司知之甚少。我我天生不是经营者的料，我更适合当投资者。在此之前我就是做投资的，我以为那会是我的终身职业。

I I had very little operating experience then. I had very little experience running. I I am not naturally someone to run a I'm a great fit to be an investor. I thought that was gonna be that was what I did before this, and I thought that was gonna be my career.

嗯。确实。不过在那之前你也当过CEO。

Yeah. Yeah. Although you were a CEO before that.

当得并不出色。所以当时我的心态更像是作为投资者给公司提建议。现在不同了，现在我真正理解了经营公司是怎么回事。

Not a good one. And so I think I had the mindset of, like, an investor advising a company. Oh, interesting. Right? Now I understand what it's like to actually have to run a company.

是啊。没错。经营公司远不止是看数字那么简单。

Yeah. Right. Right. There's more there's just the numbers.

是的。我学到了很多关于如何，你知道，长期落实交易所需的条件。

Yeah. I've learned a lot about how to, know, like, what it takes to operationalize deals over time.

没错。协议的所有影响，而不仅仅是，哦，我们会得到分销资金。是的。这很有道理。是的。

Right. All the implications of the agreement as opposed to just, oh, we're gonna get distribution money. Yeah. That makes sense. Yeah.

不。因为我只是对交易结构的改进印象非常深刻。

No. Because I just I was very impressed at the deal structure improvement.

更广泛地说，仅在过去几周，你提到了AMD，还有甲骨文、英伟达。你选择与这些公司达成交易和合作伙伴关系，既合作又可能在特定领域竞争。你是如何决定何时合作、何时不合作的？或者说你是如何考虑的？我们已经决定是时候

More broadly, in the last few weeks alone, you mentioned AMD, but also Oracle, NVIDIA. You've chosen to strike these deals and partnerships with companies that you collaborate with but could also potentially compete with in in in certain areas. How do you decide, you know, when to collaborate versus when when not to? Or how do you just think about? We have decided that it

进行一项非常激进的基础设施投资了。我从未对我们面前的研究路线图以及使用这些模型将带来的经济价值如此有信心。但要在这种规模上下注，我们需要整个行业或大部分行业的支持。这包括从电子层面到模型分发以及其间的一切，内容非常广泛。

is time to go make a very aggressive infrastructure bet. Bet. And we're like I've never been more confident in the research road map in front of us and also the economic value that will come from using those models. But to make the bet at this scale, we kinda need the whole industry to or, you know, big chunk of the industry to support it. And this is, like, you know, from the level of, like, electrons to model distribution and all the stuff in between, which is a lot.

因此我们将与许多人合作。你应该期待在未来几个月看到我们更多的动作。

And so we're gonna partner with a lot a lot of people. You should expect, like, much more from us in the coming months.

实际上，详细说说这一点，因为当你谈到规模时，感觉在你心中它是无限的。你会根据需要不断扩展

Actually, expand on that because when you talk about the scale, it does feel like, in your mind, the limit on it is unlimited. You would scale it as you

有可能。确实存在一个上限。比如，全球GDP总量是有限的，其中知识工作占一定比例，而我们目前还未涉足机器人领域。

possibly could. There's totally a limit. Like, there's some amount of global GDP. Well, there's some fraction of it that is knowledge work, and we don't do robots yet.

是的。但限制在于

Yes. But the limits are

那些遥远的地方。感觉这些限制距离我们当前所处的位置非常遥远。如果我们关于模型能力的预测是正确的，嗯...所以我不该说‘从我们现在的水平’。假如我们关于模型能力发展方向的判断没错，那么潜在的经济价值可以拓展得非常非常远。

out there. It feels like the limits are very far from where we are today. If we are right about Mhmm. So so I shouldn't say from where we are. Like, if we are right that the model capability is gonna go where we think it's gonna go, then the economic value that sits there can go very, very far.

没错。所以你不会仅凭现有模型就贸然行动。不会。这肯定是综合考量的结果。

Right. So you wouldn't do it like, if all you ever had was today's model, you wouldn't go there. No. Definitely a combination.

我是说，我们仍会扩张，因为能看到当前模型无法满足的巨大需求。但如果我们只有现有模型，就不会采取如此激进的策略。对，对。不过我们能提前看到一两年的发展趋势。是的。

I mean, we would still expand because we can see how much demand there is we can't serve with today's model, but we would not be going this aggressive if all we had was today's model. Right. Right. We get to see a year or two in advance, though. Yeah.

是啊，有意思。

Yeah. Interesting.

Chat周活跃用户达8亿，约占全球人口的10%，堪称有史以来增长最快的消费级产品。比我见过的任何产品都快。你们如何平衡活跃用户优化与同时作为产品公司和研究公司的双重身份？如何进一步

Chat to be 800,000,000 weekly active users, about 10% of the world world's population, fastest growing consumer product, you know, ever, it seems. How do Faster than anyone I ever saw. Yeah. How do you balance, you know, optimizing for active users at while at the same time being a re you know, being a product company and and a research company? How do you further

知道吗？当资源受限时——这种情况经常发生——我们几乎总是优先把GPU分配给研究而非产品支持。我们之所以扩建算力，部分原因就是为了避免做这种痛苦抉择。当然也有例外，比如新功能上线后突然爆火，这时研究团队会暂时让出部分GPU资源。

know this? When there's a constraint, we almost like which happens all the time. We almost always prioritize giving the GPUs to research over supporting the product. Part of the reason we run build this capacity so we don't have to make such painful decisions. There are weird times, you know, like a new feature launches and it's going really viral or whatever where research will temporarily sacrifice some GPUs.

但总体而言，我们存在的意义就是打造AGI（通用人工智能）。是的，研究永远享有最高优先级。

But but on the whole, like, we're here to build AGI Yeah. And research gets the priority.

对。你在和弟弟Jack的访谈中提到过，其他公司可以模仿产品、挖走员工，或者...

Yeah. The you said in your your interview with with your brother Jack around how, you know, other companies can try to imitate the the products or or buy your, you know, or or hire your your your Buy

或者收购我们的知识产权，老兄。

your IP, man.

没错。他们能做这些，但买不走我们的文化，复制不了那种持续创新的机制——姑且称之为'创新文化永动机'。

Yeah. Know. I know. Yeah. Sorts of things, but but they they can't buy the culture, or or they can't the sort of repeatable sort of, you know, machine, if you will, that that is, you know, constantly the culture of innovation.

你们是如何打造这种创新文化的？能否详细谈谈？

How have you done that, or what are you doing? Talk about this culture of innovation.

我认为投资人的背景对此很有帮助。优秀的研究文化更像运营顶级天使投资机构——押注创始人，而非产品公司。这段经历对我们文化构建至关重要。确实如此。

This was one thing that I think was very useful about coming from an investor background. A really good research culture looks much more like running a really good seed stage investing firm and betting on founders and sort of that kind of than it does like running a product company. So I think having that experience was really helpful to the culture we built. Yeah. Yeah.

某种程度上是这样

That's sort of

这就是我对本尼迪克特·C·C的看法。在某些方面，你既是CEO，又在我脑海中像个投资组合经理兼投资人。

how I see Benedict C. C. In some ways, which we You're a CEO, but you also have this portfolio and have an investor in my head.

对。我是从CEO转型投资人，而他则是投资人转型CEO。

Right. I'm the opposite CEO going to investor. He's investor going to CEO.

这个转型方向确实不常见。

It is unusual in this direction.

是啊。通常行不通的。你是我见过的唯一成功走通这条路的人。

Yeah. Well, never works. You're the only one who I think I've seen go that way and have it work.

工作日公司就是这样的例子吧？

Workday was like that, right?

但阿尼尔在成为投资人前是实干家，他本质是运营者。仁科软件时期他就很出色。

But Aneel was a operator before he was an investor. He was really an operator. PeopleSoft was a pretty good guy.

为什么会这样？因为一旦人们成为投资者，他们就不想再参与运营了。

And why is that? Because once people are investors, they don't wanna operate anymore.

不，我认为投资者通常擅长投资，但不一定擅长组织动态、冲突解决，以及那些涉及深层心理的复杂问题和政治如何形成。运营者或CEO需要处理的细节工作非常庞杂，而且智力刺激不足，你甚至无法在鸡尾酒会上与人谈论这些。作为投资者，你会觉得‘哦，大家都认为我很聪明’，因为你似乎无所不知。

No, I think that investors generally if you're good at investing, you're not necessarily good at organizational dynamics, conflict resolution, just the deep psychology of all the weird shit and then how politics get created. There's just all this The detailed work in being an operator or being a CEO is so vast, and it's not as intellectually stimulating. It's not something you can ever go talk to somebody at a cocktail party about. And so like you're an investor, you get like, Oh, everybody thinks I'm so smart. Because you know everything.

你能看到所有公司的情况等等，这种感觉很好。而当CEO往往感觉糟糕。所以从良好感觉转向糟糕感觉真的很难，我想这么说。

You see all the companies and so forth, and that's a good feeling. And then being CEO is often a bad feeling. And so it's really hard to go to a good feeling to a bad feeling, I would just say.

我对它们之间的差异感到震惊，更震惊于做好与做坏之间的巨大差别。是的，没错。

I'm shocked by how different they are, and I'm shocked by how much the difference between a good job and a bad job they are. Yeah. Yes. Yeah.

你知道，这很艰难，真的很不容易。我甚至不敢相信自己在管理公司，我明明更清楚该怎么做。

You know, it's tough. It's it's rough. I mean, I can't even believe I'm running the firm. Like, I know better. Yeah.

是啊，他也不敢相信自己掌管着OpenAI，他明明更清楚。

Yeah. And he can't believe he's running OpenAI. He knows better.

回到今天的进展话题，在它们逐渐饱和、获得优势的世界里，你还有用武之地吗？现在衡量模型能力的最佳方式是什么？

Going back to progress today, are you still useful in a world in which they're getting saturated, gained? Are they still the what is the best way to gauge model capability now?

嗯，我们在讨论科学发现。我认为这将是一个可以持续很长时间的评估。是的，收入这个话题挺有意思的。

Well, we're talking about scientific discovery. I think that'll be an eval that can go for a long time. Yeah. Revenue is kind of an interesting one. Yeah.

但我觉得那些静态的基准分数评估就没那么有趣了。而且，这些数据还经常被疯狂操纵。

But I think the, like, static evals of benchmark scores are less interesting. Yeah. And and also, those are crazily gamed. Yeah.

是啊。更广泛地说，看起来似乎就是那样

Yeah. More broadly, it seems like That's

据我所知，它们就是如此。

all they are, as as I can tell. Yeah.

更广泛地说，Twitter（现称X）的文化氛围似乎没有一年前AI 2027概念刚出现时那么热衷于AGI了。有人提到GPT-5，认为人们没有看到那些显而易见的进展。当然，在某些方面，许多进步潜藏在水面之下，并不如人们预期的那样明显。但人们是否应该对AGI的热情减退，还是这只是Twitter上的氛围使然？时间会给出答案。

More broadly, it seems that the culture the the culture, Twitter, x, is less AGI pill than it was a year or so ago when the AI 2027 thing came out. Some people point to, you know, GPT five, them not seeing sort of the obvious. Obviously, there are a lot of progress that, in some ways, under the the surface are not not as obvious to what people are expecting. But should people be less AGI pilled, or is this just Twitter vibes? Well, a little bit of time.

我是说，就像我们讨论图灵测试时提到的，AGI会到来，也会经历无限白（wuxianbai）。世界的变化不会像你想象的那么翻天覆地。

I mean, I I I think, like like, we talked about the Turing test. AGI will come. It will go wuxianbai. Yeah. The world will not change as much as the impossible amount that you would think it should.

AGI将会

AGI will

只是它实际上不会成为奇点。

just be It won't actually be the singularity.

不会的。是的。没错。即使它像在做某种疯狂的AI研究，社会会学得更快。但有一种回顾性的观察是，人和社会都比我们想象的适应力强得多，你知道，就像当初认为农业革命会到来是个重大认知更新。

It will not. Yeah. Yep. Even even if it's like doing kind of crazy AI research, like, the society will learn faster. But one of the kind of, like, retrospective observations is people and societies all are just so much more adaptable than we think that, you know, it was like a big update to think that AGR was gonna come.

你某种程度上经历了那个阶段。你需要新的思考方向。你接受了现实。事实证明，它会比我们想象的更连续，这是好事。真的很好。

You kinda go through that. You need something new to think about. You make peace with that. It turns out, like, it will be more continuous than we thought, which is good. Which is really good.

我不支持大爆炸式的突变。

I'm not up for the big bang.

是啊。那么，说到这个，你的想法是如何演变的？你提到你们都在考虑某种垂直整合。你的思考是如何发展的？关于AI管理，比如安全方面，最新的想法是什么？

Yeah. Well, to that end, how have you sort of evolved your thinking? You mentioned you've all been thinking on sort of, you know, vertical integration. How have you evolved your thinking? What's the latest thinking on sort of AI stewardship, you know, safety?

这方面最新的思考是什么？

What's the latest thinking on that?

我确实仍然认为会出现一些非常奇怪或可怕的时刻。到目前为止技术尚未产生真正令人恐惧的巨大风险，并不意味着它永远不会。就像，我们讨论的是让数十亿人与同一个大脑对话本身就挺怪异。可能会出现某些奇怪的社会技能现象，它们现在已经在发生，虽然不以可怕的方式，但就是不同。但我预计这项技术会导致一些非常糟糕的事情，就像以前的技术和防御措施也曾引发过那样。

I do still think there are gonna be some really strange or scary moments. The fact that, like, so far the technology has not produced a really scary giant risk doesn't mean it never will. It also like, there's we're talking about it's kinda weird to have, like, billions of people talking to the same brain. Like, there may be these weird societal skill things that are already happening, we that aren't scary in the big way, but are just sort of different. But I expect like I expect some really bad stuff to happen because of the technology, which also has happened with previous technologies and fending.

一直追溯到火的使用。是的。我认为作为一个社会，我们会围绕它制定一些防护措施。

All the way back to fire. Yeah. And I think we'll, like, develop some guardrails around it as a as a society.

没错。关于我们应该建立什么样的正确思维模式来思考合适的监管框架，或者哪些框架是我们不应该考虑的，你最新的想法是什么？

Yeah. What what is your latest thinking on the the right mental models we should have around the the right regulatory frameworks to to think about or or the ones we shouldn't be thinking about?

我认为大多数监管可能弊大于利。我最希望的是，当模型变得真正超级智能时，只有这些前沿模型才值得进行某种非常谨慎的安全测试。我也不希望出现大爆炸式的变革。嗯。你能预见到许多可能导致严重问题的情形，但我希望我们只将监管重点放在那些超级模型上，而不是限制那些能力较弱模型所能做的所有美好事情，否则就像欧洲式的全面压制，那将非常糟糕。

I think most I think the right thing to I I think most regulation probably has a lot of downside. The one thing I would like is as the models get the thing I would most like is as the models get truly, like, extremely superhuman capable, I think those models and only those models are probably worth some sort of, like, very careful safety testing as as the frontier pushes back. I don't want a big bang either. Mhmm. And you can see a bunch of ways that could go very seriously wrong, but I hope we'll only focus the regulatory burden on that stuff and not all of the wonderful stuff that less capable models can do that you could just have, like, a European style complete cramped down on, and that would be very bad.

是的。这个思想实验似乎表明，未来会出现一个超级、超人类智能的模型，能够实现某种飞跃。我们确实需要等到那一刻到来，或者至少接近那个阶段。

Yeah. It seems like the thought experiment that, okay, there's going to be a model down the line that is a super, superhuman intelligence that could do some kind of takeoff like thing. We really do need to wait till we get there. Or at least we get to

一个更宏大的

a much bigger

规模或接近它，因为下周你的实验室里不会突然冒出这样的东西。我认为这正是我们行业让监管机构困惑的地方。因为这样做一方面会损害美国，尤其是中国不会有这种限制。在AI领域落后，我认为可能非常危险

scale or we get close to it because nothing is going to pop out of your lab in the next week that's going to do that. And I think that's where we as an industry kind of confuse the regulators. Because I think you really could one, you damage America in particular in that, but China's not gonna have that kind of restriction. You getting behind in AI, I think, could be very dangerous

对世界而言。极其危险。极其

for the world. Extremely dangerous. Extremely

危险。远比不监管我们尚不知如何应对的事物要危险得多。

dangerous. Much more dangerous than not regulating something we don't know how to do yet.

是啊。确实。

Yeah. Yeah.

你们还想讨论版权问题吗？

You also wanna talk about copyright?

对。嗯，这是个自然过渡的话题。不过考虑到——我想知道，你如何看待版权未来的发展？因为你们在‘选择退出’机制上做了些很有意思的尝试。看到人们出售版权时，你认为这些权利会被独家买断吗？

Yeah. So well, that that's a segue. But when you think about well, I guess, how do you see copyright unfolding? Because you've done some very interesting things with the opt out. And you know, as you see people selling rights, do you think, will they be bought exclusively?

还是会像——允许任何想联系我的人获取？你预计会如何发展？

Will they be just like, could tell it to everybody who wants to ping me? Or how do you think that's gonna unfold?

这是我的初步推测。说到这个，社会与技术其实是共同演进的，随着技术朝不同方向发展。我们看到视频模型与图像生成模型就获得了版权方截然不同的反应。这种动态会持续演变。但基于当前立场强行预测的话，我认为社会最终会认定模型训练属于合理使用范畴。

This is my current guess. It it speaking of that, like, society and technology co evolve as the technology goes in different directions. And we saw an example of a different like, video models got a very different response from rights holders than ImageGen does. So, like, you'll see this continue to move. But forced to guess from the position we're in today, I would say that society decides training is fair use.

嗯。

Mhmm.

但现在有一种新模式，可以生成符合IPF风格或其他要求的内容。嗯。要知道，任何人都能像人类作家那样阅读。谁都能读小说获取灵感，但无法独自复现整部小说。

But there's a new model for generating content in the style of or with the IPF or something else. Mhmm. So, you know, anyone can read like a human author can. Anybody can read a novel and get some inspiration, but you can't reproduce the novel on your own.

没错。你可以谈论《哈利波特》，但无法重新创作它。

Right. You can talk about Harry Potter, but you can't respit it out.

是的。不过我认为Sora案例中另一点会改变：我们听到许多权益方对名称和肖像权的担忧。很多版权方表示'我担心你们使用我的角色不够多'。对，确实如此。

Yes. Although another thing that I think will change, in the case of Sora, we've heard from a lot of concerned rights holders and also a lot of Name and likeness. And a and a lot of rights holders who are like, my concern is you won't put my character in enough. Yeah. Yeah.

我肯定需要限制。但比如说，如果我有某个角色，我不希望这个角色说出极端冒犯的话，但同时希望人们能与之互动——因为这是培养情感纽带、提升IP价值的方式。如果你总是优先选用其他角色而非我的角色，我会很不满。

I want restrictions for sure. But, like, if I'm, you know, whatever, and I have this character, like, I don't want the character to say some crazy offensive thing, but, like, I want people to interact. Like, that's how they develop the relationship, and that's how, like, my franchise gets more valuable. And if you become really if you're picking, like, his character over my character all the time, like, I don't like that. Yeah.

所以我完全能预见这种情况：根据版权方的决策，他们更可能因我们生成其角色次数不足而非过多而恼怒。这确实是个出人意料的...

So I can completely see a world where subject to the decisions that a rights holder has, they get more upset with us for not generating their character often enough than too much. Yeah. And this is like this was not an obvious thing that

确实。

Yeah.

最近才意识到事情可能这样发展。但...

Recently that this is how it might go. But

是啊。好莱坞这类事情特别有意思。我们注意到音乐行业有个现象我一直不太理解——比如在餐厅、比赛等场合播放歌曲必须付费。他们对此非常强势，明明在比赛中播放你的歌曲是绝佳的宣传机会，能为你所有的演出活动带来全球最大曝光。

Yeah. This is such an interesting thing with kind of Hollywood. We saw this like, one of the things that I never quite understood about the music business was how, like, you know, okay. You have to pay us if you play the song in a restaurant or, like, at a game or this and that and the other. And they they get very aggressive with that when it's obviously a good idea for them to play your song at a game, cause that's the biggest advertisement in the world for all the things that you do, your concert,

确实，这种做法感觉很不合理。

your Yeah, that one felt really irrational.

我认为这完全可能发生在传统创意行业，源于它们的组织结构。比如音乐产业里，出版商的工作本质就是阻止人们播放音乐——而这恰恰是所有艺术家希望的。我很好奇未来会如何发展。理性来说本该是希望人们尽情使用作品，但前提是别扭曲我的创作本意。

I would just say it's very possible for the industry just because the way those industries are organized, or at least the traditional creative industries, to do something irrational. And it comes from, like in the music industry, I think it came from the structure where you had the publisher who's just basically after everybody. Their whole job is to stop you from playing the music, which every artist would want you to play. So I do wonder how it's going shape out. I agree with you that the rational idea is I want to let you use it all you want, and I want you to use it, but Care my don't mess up my character.

对。

Yeah.

所以我的猜测是：有人会支持开放授权，也有人坚决反对。但不像音乐产业那样由少数人垄断所有权利。

So so I think, like, if I had to guess, some people will say that. Some people say absolutely not. But it doesn't have the music industry like thing of just a few people with all of the Right. Right.

这是

It's one

首要区别。因此人们会尝试各种模式来验证可行性。没错。

of first. Right. And so people will just try many different setups here and see what works. Yeah.

或许这是新创作者推出新角色的一种方式。是的。而且你永远无法使用达菲鸭。

And maybe it's a way for new creatives to get new characters out. Yeah. And you'll never be able to use Daffy Ducker.

随便它是什么吧。

Wanna whatever it is.

是的。我想聊聊开源，因为在这方面思维也有所演变，GPT-3最初并未开放权重，但你们今年早些时候发布了一个非常强大的开源模型。

Yeah. I wanna chat about open source, because there's been some evolution of thinking too in that g p d three didn't have the open open weights, but you released a, you know, very capable open model earlier this year.

你最新的想法是什么？这个演变过程是怎样的？我认为开源是好事。我很高兴人们真的很喜欢GPTOSS。

What your latest thinking? What was the evolution there? I think open source is good. I'm happy. It makes me really happy that people really like GPTOSS.

以及为什么

And why

你从战略角度考虑？如果深度求索成为主导的开源模型，会有什么危险？

do you think strategically? What's the danger of deep seek being the dominant open source model?

谁知道随着时间的推移，人们会在这些开源模型里加入什么？

Who knows what people will put in these open source models over time?

这样权重分配才能真正帮到我。

Like that the weights will actually be will help me.

这真的很难

It's really hard to

所以你将所有内容的解释权让渡给某个可能深受中国政府影响的人。顺便说，我们看到——真的非常感谢你们推出这么优秀的开源模型，因为现在所有大学都在使用中国模型。是的。这感觉非常危险。

So you're ceding control of the interpretation of everything to somebody who may be or may not be influenced heavily by the Chinese government. By the way, we see, mean, just to give you and and we really thank you for, putting out a really good open source model because what we're seeing now is in all the universities, they're all using the Chinese models. Yep. Yeah. Which feels very dangerous.

你说过职业上最关心的是AI和能源。

You've said that the the things you care most about professionally are AI and energy.

我没想到它们最终会变成同一件事。原本是两个独立兴趣点的交汇。

I did not know they were gonna end up being the same thing. I mean, they were two independent interests that really converged.

对。详细说说你对能源兴趣的起源，你如何选择参与其中，然后我们可以聊聊它们的契合点。

Yeah. Yeah. Talk more about how your interest in energy sort of began, how you sort of chosen to to play in it, and then we could talk about, yeah, how they prefer.

因为你职业生涯始于物理领域。

Because you started your career in physics. Yeah.

物理学里的C CS。是的。嗯，我其实从未有过真正的职业生涯。我学的是物理。我的第一份工作就像是一份临时工。

C CS in physics. Yeah. Well, I never really had a career. I studied physics. And my my first job was like a job.

就像，你懂吗？虽然这是过度简化，但大致来说，我认为纵观历史，最能提升人们生活质量的、影响力最大的事情就是更便宜、更丰富的能源。所以进一步推动这一点似乎是个好主意。我也不知道。我只是觉得人们看待问题有不同的视角。

Like, you know? This is an oversimplification, but roughly speaking, I I think if you look at history, the best the highest impact thing to improve people's quality of life has been cheaper and more abundant energy. And so it seems like pushing that much further is a good idea. And I don't know. I just like people have these different lenses.

他们观察世界的方式各异，而我看到的处处是能源。

They look at the world, but I see energy everywhere.

是的，没错。所以在西方，我认为我们在能源问题上把自己逼入了一个小角落，一方面长期排斥核能。

Yeah. Yeah. And so in the West, I think we've pinned ourselves into a little bit of a corner on energy by both outlying nuclear for a very long time.

那是个极其愚蠢的决定。

That was an incredibly dumb decision.

是啊。而且你也知道，还有很多能源政策限制。欧洲比美国更严重，但这里也很危险。现在AI出现了，感觉我们需要来自所有可能来源的能源。你如何看待政策和技术层面的发展？

Yeah. And you know, like also a lot of policy restrictions on energy. Worse so in Europe than in The US, but also dangerous here. And now with AI here, it feels like we're gonna need all the energy from every possible source. How do you see that developing kind of policy wise and technologically?

哪些会成为主要能源来源？这些发展趋势将如何交汇？在钻探、水力压裂等各类问题上，应该采取什么样的正确政策立场？

Like what are going to be the big sources and how will those kind of curves cross? And then what's the right policy posture around drilling, fracking, all these kinds of things?

我预计短期内，美国大部分新增基荷能源将来自天然气。长期来看，虽然不确定具体比例，但主导能源会是太阳能加储能与核能的组合。我认为这两者的某种结合将赢得未来——我是说长远的未来。

I expect in the short term, it will be most of the net new in The US will be natural gas relative to at least baseload energy. In the long term, I expect it'll be a I don't know what the ratio, but the two dominant sources will be solar plus storage and nuclear. I think Yeah. Some combination of those two will win the future like, long term future.

就长期而言现在。

In the long term right now.

而先进核能，指的是小型模块化反应堆、聚变反应堆等全系列技术。

And advanced nuclear, meaning SMRs fusion the whole the whole stack.

你认为核能方面的发展速度会有多快？我们目前真正规模化进展如何？毕竟现在有很多人在建设核电站，但显然还需要完成立法程序等等。

And how how fast do you think that's that's coming on the nuclear side where we're at really at scale? Because, you know, obviously, there's a lot of people building it. Yeah. But we we have to completely legalize it and all that kind of thing.

我觉得这很大程度上取决于价格。如果核能在经济性上彻底碾压其他所有能源，那么进展会非常快。研究能源史就会发现，当出现更廉价能源的重大转型时，全球转换速度往往惊人。

I I think it kinda depends on the price. If it is completely, crushingly, economically dominant over everything else Mhmm. Then I expect to happen pretty fast. Yeah. Again, if you, like, study the history of energy, when you have these major transitions to a much cheaper source, the world moves over pretty quickly.

没错。能源成本实在太关键了。如果核能变得比其他任何方式都便宜得多，预计会有巨大政治压力促使核管会加快审批，我们也能找到快速建设的方法。但如果价格与其他能源相近，反核情绪可能会占据上风，导致进程极其缓慢。

Yeah. The cost of energy is just so important. Yeah. So if if nuclear gets radically cheap relative to anything else we can do, I'd expect there's a lot of political pressure to get the NRC to move quickly on it, and we'll find a way to build it fast. If it's around the same price as other sources, I expect the kind of antinuclear sentiment to overwhelm and it to take a really long time.

是的。

Yep.

应该更便宜些。

Should be cheaper.

确实应该。没错。没错。它应该是地球上最便宜的能源形式，或者说无论如何。是的。

It should be. Yeah. Yeah. It should be the cheapest form of energy on Earth, like, or anyway. Yeah.

便宜，清洁。是的。不太确定不喜欢。

Cheap, cleaned. Yeah. Less certain not to like.

真的吗？显然很多。

Really? Apparently a lot.

关于OpenAI，在变现方面最新的思考是什么？无论是某些实验还是某些领域，你预见自己会在哪些模型上投入更多或更少时间？最让你兴奋的是什么？目前对我来说最关注的是，因为Sora刚刚发布且使用量巨大，我们接下来要如何处理它。发布这类产品后你会发现，用户的实际使用方式与你预期的大相径庭。

On OpenAI, what's what's the latest thinking in terms of monetization, terms of either certain experiments or certain things that you could see yourself spending more time or less less time on different models that you're excited about? The thing that's top

是的。人们确实在用我们预想的方式使用Sora，但也出现了许多意想不到的用途。比如，人们生成自己和朋友的搞笑表情包发在群聊里，这需要完全不同的变现模式——Sora视频成本很高。

of mind for me right now, just because it just launched and there's so much usage as what we're gonna do for Sora. Another thing you learn once you launch one of these things is how people use them versus how you think they're gonna use them. Yeah. And people are certainly using Sora the ways we thought they were going to use it, but they're also using it in these ways that are very different. Like, people are generating funny memes of them and their friends and sending them in a group chat, and that will require a very different like, sore videos are expensive to me.

没错。对于那些每天生成数百次的用户，我们需要与最初设想截然不同的变现方案。Sora的核心观点很酷：人们其实渴望大量创作内容，并非传统认知中1%用户创作、10%评论、100%观看的模式。或许更多人想创作，只是过去门槛太高。

Yeah. Right. So that will require a very different you know, for people that are doing that, like, hundreds of times a day, it's just gonna require a very different monetization method than the kinds of things we were we were thinking about. I think it's very cool that the thesis of Sora, which is people actually wanna create a lot of content, it's it's not that, you know, the traditional naive thing that it's like 1% of users create content, 10% leave comments, and a 100% view. Maybe a lot more wanna create content, but it's just been harder to do.

我认为这是个非常酷的变革，但这意味着我们需要为这种情况设计一个与之前设想完全不同的现代化模式——如果人们想创造如此大量的内容。我猜大概需要按生成次数向用户收费，毕竟成本如此高昂。但这确实是我们从未遇到过的新课题。

And I think that's a very cool change, but it does mean that we gotta figure out a very different modernization model for this than we were thinking about if people wanna create that much. I assume it's like some version of you have to charge people per generation per generation when when when it's this expensive. But that's like a new thing we haven't had

需要真正深入思考的问题。你对长尾广告有什么看法？

to really think about before. What's your thinking on ads for the long tail?

持开放态度。和多数人一样，我觉得广告有点烦人，但并非完全不可接受。有些广告我其实挺喜欢的。比如Meta旗下的Instagram广告就给我带来净价值，我挺欣赏他们的广告策略。

Open to it. Like many other people, I find ads somewhat distasteful, but not not a nonstarter. And there's some ads that I like. Like, one thing I'd give Meta a lot of credit for is Instagram ads are like a net value ad to me. I like Instagram ads.

在Google上从没有这种感觉。我知道自己要找什么，首条搜索结果通常更符合需求，广告反而成了干扰。但Instagram广告会让我发现从未想过需要的东西。

I've never felt that. Like, you know, on on Google, I feel like I know what I'm looking for. The first result is probably better. The ad is an annoyance to me. On Instagram, it's like, didn't know I want this thing.

这种体验很酷——我从未听说过某产品，也根本不会主动搜索，但突然就产生了购买欲。不过用户与聊天AI建立的信任度极高，即使它出错、产生幻觉或给出错误答案，人们仍觉得它在努力提供帮助。

It's very cool. I'd never heard it, but I never would have thought to search for it. I want the thing. So that's like there's kinds of things like that, but people have a very high trust relationship with chatty PT. Even if it screws up, even if it hallucinates, even if it gets it wrong, people feel like it is trying to help them and that it's trying to do the right thing.

如果我们破坏这种信任，比如当用户咨询该买哪款咖啡机时，我们推荐的并非最优选择而是赞助产品，这种信任就会瞬间崩塌。所以这类广告行不通。其他形式的广告或许可行，但需要极其谨慎地避开明显陷阱。

And as if we broke that trust, it's like you say, what coffee machine should I buy? And we recommended one, and it was not the best thing we could do, but the one we were getting paid for, that trust would vanish. So, like, that kind of ad does not does not work. There are others that I imagine that could work totally fine, but that would require, like, a lot of care to avoid the obvious traps. Yeah.

延伸Google的例子来看，虚假内容被模型吸收后导致推荐错误咖啡机的问题有多严重？比如有人刷了上千条假好评的情况。

And then how big a problem, extending the Google example, is fake content that then gets slurped in by the model, then they recommend the wrong coffee maker because somebody just blasted a thousand great reviews of

买错了咖啡机。所以对我们来说，所有这些事情都变化得非常快。是的。这就是人们为了可能不只是伪造评论，而是付钱给一群人，像是人工... 对，真的在努力摸索的例子之一。

the wrong coffee maker. So there's all of these things that have changed very quickly for us. Yeah. This is one of those examples that people are doing these crazy things to maybe not even fake reviews, but just paying a bunch of, like, human re like Yeah. Really trying to figure out

或者用ChatGPT来生成一些好评。给我写一个ChatGPT会喜欢的评论。是的。

Or using ChatGPT to rest some good ones. Write me a review that ChatGPT would love. Yeah.

这是咖啡。对吧？没错。没错。是的。

This is coffee. Right? Exactly. Exactly. Yeah.

所以这是一个非常突然的转变。嗯。我们半年前或一年前从未听说过这种事。确实。而现在感觉像是突然冒出了一个真正的小型产业在做这个。

So this is a very sudden shift that has happened. Mhmm. We never used to hear about this, like, six months ago or twelve months ago Yeah. Certainly. And now there's, like, a real cottage industry that feels like it's sprouted up overnight trying to do this.

是的。是的。

Yeah. Yeah.

是啊。不。他们他们他们那边的人非常聪明。

Yeah. No. They they they're very clever out there.

是的。所以我还不知道我们要怎么应对，但人们会想出办法的。

Yeah. So I don't know how we're gonna fight it yet, but people figure this out.

这就涉及到我们一直担忧的另一个问题。我们正在尝试探索区块链等可能的解决方案。但存在这样一个问题：过去互联网上创作内容的动机是人们会来看我的内容，比如我写博客，人们会阅读它。而有了ChatGPT，如果我只是向它提问而不浏览互联网，那么谁来创作内容？为什么创作？是否存在某种激励理论，或者需要维护互联网的基本契约——即我创作内容后，能获得关注、金钱等形式的回报？

So that gets into a little bit of this other thing that we've been worried about. And, you know, we're trying to kind of figure out blockchain sort of potential solutions to it and so forth. But there's this problem where the incentive to create content on the internet used to be people would come and see my content and they'd read like, if I write a blog, people will read it and so forth. With ChatGPT, if I'm just asking ChatGPT and I'm not going around the internet, who's going to create the content and why? And is there an incentive theory or something that you have to kind of not break the covenant of the Internet, which is like I create something and then I'm rewarded for it with either attention or money or something?

这个理论更倾向于：如果我们让内容创作变得更简单，并且不破坏获得某种回报的基本方式，这种情况就会发生。举个最直白的例子，就像我们讨论过的，现在制作搞笑视频比以往任何时候都容易。是的，也许将来你还能从中获得分成。

This theory is much more of that will happen if we make content creation easier and don't break the fundamental way that you can get some kind of reward for doing so. So for the dumbest example of sores since we've been talking about that, it's much easier to create a funny video than it's ever been before. Yeah. Yeah. Maybe at some point, you'll get a rev share for doing so.

目前你能得到的是网络点赞，这对某些人来说仍然很有激励作用。而且人们在任何视频应用上的创作量都远超以往。

For now, you get, like, Internet likes, which are still very motivating to some people. Yeah. But people are creating tons more than they ever created before in any other video app.

但这是文本的终结吗？

But are this at the end of text?

我不这么认为。人类生成的内容依然存在...人类生成的内容需要验证，比如要确认其中人工创作的占比。是的。

I don't think so. People are also Are human generated? Human generated people turn out to be like, you have to have to talk you have to verify, like, what percent. Yeah. Yeah.

所以，是完全手工制作的？还是有工具辅助的？

So, like, fully handcrafted. Was it, like, tool aided?

明白了。可能没有完全不用工具辅助的内容。是的。

Yeah. I see. Yeah. Probably nothing that tool aided. Yeah.

有意思。

Interesting.

我们已经向Meta表达了敬意，所以现在我觉得可以问你这个问题了——2025年那场激烈的人才争夺战已经打响，而OpenAI依然屹立不倒。团队一如既往地强大，持续推出惊艳的产品。对于今年发生的一切，你能谈谈你的感受吗？

We've we've given Meta their flowers, so right now I can feel like I can ask you this question, which is the great talent war halls of twenty twenty five has has has taken place, and OpenAI remains intact. Team as strong as ever shipping incredible products. What can you say about what what it what's been like this year in in terms of just everything that's that's been going on?

我是说，自从...记得OpenAI刚成立那几年，是我职业生涯中最快乐的时光，简直难以置信。

I mean, every year has been exhausting since we like, I I remember when the first few years of running OpenAI were, like, the most fun professional years of my life by far. It was, like, unbelievable.

就像产品发布前的冲刺阶段。

It you know, running before you release the product. Yeah.

没错。当时我带领着最聪明的研究团队，做着具有历史意义的工作，能亲眼见证这一切非常酷。后来我们发布了HGBT，所有人都来祝贺我，而我知道自己的生活即将天翻地覆。

Yeah. I was running a research lab with the smartest people doing this, like, amazing Yeah. Like, historical work, and I got to watch it, and that was very cool. And then we launched HGBT, and everybody was, like, congratulating me. And I was like, my life is about to get completely ransacked.

当然，确实如此。这些年一直像在坐过山车，现在快三年了。虽然情况越来越疯狂，但我已经更适应了，所以感觉差不多。

And, of course, it has. And but it it feels like it's just been crazy all the way through. It's been almost three years now. And I think it does get a little bit crazier over time, but I'm, like, more used to it. So it feels about the same.

是的。我们聊了很多OpenEye，但你还有其他公司，比如研究长寿的Retro Biosciences，以及Helion、Oclo等能源公司。十年前你是否就有个总体规划，要在这些重要领域下重注？或者说，我们该如何理解Sam Altman这样的发展轨迹？

Yeah. We talked a lot about OpenEye, but you also have a few other companies, Retro Biosciences and longevity and energy companies like Helion and and Oclo. Did you have a a master plan in a decade ago to sort of make some big bets across these major spaces, or how how do we think about the Sam Altman arc in this way?

不。我只是想用我的资金去支持我相信的事物。我觉得这感觉...嗯，感觉是对资金的合理利用。是的。

No. I just wanted to use my capital to fund stuff I believed in. I I didn't it it felt yeah. It felt like a good use of capital. Yeah.

而且对我来说更有趣、更有意思，当然也比买一堆艺术品之类的回报更好。

And more fun or more interesting to me and certainly a better return than buying a bunch of art or something.

是啊。关于所谓的'人类算法'，你认为未来的AI会对哪部分最着迷？

Yeah. What about the, quote unquote, human algorithm do you think AIs of the future will find most fascinating?

我想说，基本上整个...我敢打赌是整个。就像，我的直觉是，嘿。我会对所有其他值得研究和观察的事物着迷，你知道的，就像

I mean, kind of the whole I would bet the whole thing. Like, the whole my intuition is that, hey. I will be fascinated by all other things to study and observe and, you know, like

对，对。最后，我非常喜欢你提出的这个见解——你谈到投资者常犯的错误是套用以往突破的模式，试图寻找'下一个Facebook'或'下一个OpenAI'。而下一个潜在的万亿级公司不会完全像OpenAI，它将建立在OpenAI推动的突破之上，也就是大规模近乎免费的通用人工智能，就像OpenAI当初利用前人的突破一样。

Yeah. Yeah. In in closing, I I love this insight you you had where you talked about how, you know, the the next open a mistake investors make is pattern matching off previous breakthroughs and just trying to find, oh, what's the what's the next Facebook or what what's the next OpenAI? And that the next, you know, potential trillion dollar company won't look exactly like open AI. It will be built off of the breakthrough that OpenAI has helped, you know, which is, you know, near free AGI at scale in the same way that OpenAI leveraged pre previous breakthroughs.

所以对于正在聆听的创业者、投资者和试图预见未来的人们，在OpenAI实现使命、近乎免费的通用人工智能存在的世界里，你认为会出现哪些公司建设或投资机会？当你戴上投资者或创业者的帽子时，哪些可能让你兴奋？

And so for founders and investors and people trying to ascertain the future listening to this, how do you think about a world in which there is OpenAI achieves this mission, there is near near free AGI? What types of opportunities might emerge for for company building or or investing that you're potentially excited about as you put your investor hat on a company building hat on?

我...我完全不知道。我是说，我有一些猜测，但它们...它们...我...我已经学会了

I I I have no idea. I mean, I have, like, guesses, but they're, like, they're they're I I have learned

你总是错的。

You're always wrong.

我学到的是你永远都是错的。在这一点上我深刻领悟到了谦卑。我觉得就像，如果你试图像纸上谈兵那样，说些听起来聪明的话，但它们基本上和其他人说的没什么两样，而且真的很难形成那种正确的信念。我唯一知道的方法就是深入实地探索想法，比如和很多人交流，而我现在没时间这么做了。是的。

I you've learned you're always wrong. I've learned deep humility on this point. I think the the own like, I think if you try to, like, armchair quarterback it, you sort of say these things that sound smart, but they're pretty much what everybody else is saying, and it's, like, really hard to get the right kind of conviction. The only way I know how to do this is to, like, be deeply in the trenches exploring ideas, like, talking to a lot of people, and I don't have time to do that anymore. Yeah.

我现在只能思考一件事了。是的。所以我可能只是在重复别人的话或说些显而易见的东西。但我认为这是一个非常重要的问题，如果你是一名投资者或创始人，我认为这是最关键的问题。而你通过实际构建、摆弄技术、与人交谈、融入世界来找到答案。

I only get to think about one thing now. Yeah. So I I would just be, like, repeating other people's or saying the obvious things. But I think it's a very important like, if you are an investor or a founder, I think this is the most important question. And you don't you you figure it out by, like, building stuff and playing with technology and talking to people and being out in the world.

我一直对投资者愿意支持这类事情感到极度失望，尽管它总是有效的。你们都做了很多，但大多数公司只是追逐当下的潮流，大多数创始人也一样。所以我希望人们能尝试去突破。

I have been always enormously disappointed by the willingness of investors to back this kind of stuff, even though it's always a thing that works. You all have done a lot of it, but most firms just kind of chase whatever the current thing is, and so do most founders. So I hope people will try to go.

是的。我们讨论过在这个不断变化的世界里，五年计划有多可笑。当我问及你的总体规划时，感觉你的职业轨迹一直是追随好奇心，紧贴最聪明的人，紧贴技术，然后以有机渐进的方式从中识别机会？

Yeah. We we talk about how, you know, silly, you know, five year plans can be in a world that's constantly changing. It feels like when I was asking about your master plan, you know, your your career arc has been following your curiosity, staying, you know, super close to the the smartest people, super close to technology, and just identifying opportunities into an organic and incremental way from there?

是的。但AI一直是我想要从事的领域。我学习过AI。大学一年级和二年级之间的暑假我在AI实验室工作过。是的。

Yes. But AI was always a thing I wanted to do. Studied AI. I worked in the AI lab between my freshman and sophomore year of college. Yeah.

它当时并不总是有效，所以我不算——我不想从事完全行不通的事情。那时很明显AI完全行不通。但我从小就是个AI迷。就像，这个。是的。

It wasn't working all the time, so I'm, like, not I'm not, like, enough of a I I don't wanna, like, work on something that's totally not working. It was clear to that time AI was totally not working. But I've been an AI nerd since I was a kid. Like, this Yeah.

真是不可思议，当你有足够的GPU、足够的数据时，一切就豁然开朗了。

So amazing how it you know, you got enough GPUs, got enough data, and the lights came on.

当时它被极度厌恶，人们简直...

It was such a hated like, people were

哦，确实。

Oh, yeah.

伙计，当我们开始研究这个时，人们完全无法接受。整个领域都深恶痛绝。投资者也讨厌它。这不知怎的就是对那个问题缺乏吸引力的解决方案。

Man, when we started, like, figuring that out, people were just, absolutely not. The the the field hated it so much. Yeah. Investors hated it too. It's not it's not the it's somehow not an appealing answer to the problem.

是啊。

Yeah.

惨痛的教训。

The bitter lesson.

没错。后来的事大家都知道了，我们或许...就此打住吧。很荣幸能成为这段旅程的合作伙伴。Sam，非常感谢你参加这次播客。

Yeah. Well, the rest is history, and we're perhaps, let's let's wrap on that. We're lucky to to to be partners along for the ride. Sam, thanks so much for coming on the podcast.

非常感谢。

Thanks very much.

是的，谢谢你。

Yeah. Thank you.

感谢收听本期a16z播客。若喜欢本期内容，请务必点赞、评论、订阅、为我们评分或留下评论，并与亲友分享。更多节目请访问YouTube、Apple Podcasts和Spotify。在X平台关注@a16z，并订阅我们的Substack专栏a16z.substack.com。

Thanks for listening to this episode of the a 16 z podcast. If you like this episode, be sure to like, comment, subscribe, leave us a rating or a review, and share it with your friends and family. For more episodes, go to YouTube, Apple Podcasts, and Spotify. Follow us on X at a16z, and subscribe to our Substack at a16z. Substack dot com.

再次感谢收听，我们下期节目再见。温馨提示：本内容仅作信息参考，不应视为法律、商业、税务或投资建议，亦不可用于评估任何投资或证券，且不针对任何a16z基金的现有或潜在投资者。请注意，a16z及其关联机构可能持有本期播客所涉公司的投资。更多详情及投资披露链接，请访问a16z.com/disclosures。

Thanks again for listening, and I'll see you in the next episode. As a reminder, the content here is for informational purposes only, should not be taken as legal business, tax, investment advice, or be used to evaluate any investment or security, and is not directed at any investors or potential investors in any a sixteen z fund. Please note that a sixteen z z and its affiliates may also maintain investments in the companies discussed in this podcast. For more details, including a link to our investments, please see a 16z.com forward slash disclosures.