奥尔特曼展示 GPT-5——以及接下来是什么

这就像一项技术拥有疯狂的力量，而且它来得如此之快。今天出生的孩子永远不会比人工智能更聪明。我们如何分辨什么是真实的，什么不是？超级智能。这到底意味着什么？我即将采访 OpenAI 的 CEO Sam Altman。OpenAI 正在重塑行业。现在，他们正试图构建一个超级智能，几乎在所有领域都远超人类。他们刚刚发布了迄今为止最强大的模型。就在几年前，这听起来还像是科幻小说。但现在不是了。事实上，他们并不孤单。我们正处于我们见过的最具风险的全球竞赛之中。数千亿美元和难以置信的人力投入。这是一个深刻的时刻。大多数人从未经历过这样的技术变革，而它现在就在你和我身边发生。所以，在这一集中，我想尝试与 Sam Altman 一起时间旅行，进入他试图构建的未来，看看它是什么样子，这样你和我就能真正理解即将发生的事情。欢迎来到《Huge Conversations》。你好吗？很高兴见到你。谢谢你来做这个。

This is like a crazy amount of power for one piece of technology and it's happened to us so fast. A kid born today will never be smarter than AI. How do we figure out what's real and what's not real? Super intelligence. What does that actually mean? I'm about to interview Sam Altman, the CEO of OpenAI. OpenAI is reshaping industries. Right now, they're trying to build a super intelligence that could far exceed humans in almost every field. And they just released their most powerful model yet. Just a couple years ago, that would have sounded like science fiction. Not anymore. In fact, they're not alone. We are in the middle of the highest stakes global race any of us have ever seen. Hundreds of billions of dollars and an unbelievable amount of human worth. This is a profound moment. Most people never live through a technological shift like this, and it's happening all around you and me right now. So, in this episode, I want to try to time travel with Sam Altman into the future that he's trying to build to see what it looks like so that you and I can really understand what's coming. Welcome to Huge Conversations. How are you? Great to meet you. Thanks for doing this.

当然。

Absolutely.

所以，在我们深入之前，我想告诉你我的目标。我不会问你关于估值、AI 人才争夺战、融资或类似的事情。我认为这些在其他地方已经得到了很好的覆盖。我们节目的主要目标是探讨如何利用科学和技术让未来更美好。我们这样做的原因是我们坚信，如果人们看到那些更美好的未来，他们就能帮助构建它们。所以，我的目标是尽力与你一起时间旅行，进入你试图构建的未来中的不同时刻，看看它是什么样子。

So, before we dive in, I'd love to tell you my goal here. I'm not going to ask you about valuation or AI talent wars or fundraising or anything like that. I think that's all very well covered elsewhere. Our big goal on this show is to cover how we can use science and tech to make the future better. And the reason that we do all of that is because we really believe that if people see those better futures, they can then help build them. So, my goal here is to try my best to time travel with you into different moments in the future that you're trying to build and see what it looks like.

太棒了。

Fantastic.

太棒了。从你刚刚宣布的开始，你最近说，令人惊讶地最近，GPT-4 是我们任何人将不得不再次使用的最笨的模型。但 GPT-4 在 SAT、LSAT 和 GRE 中已经能比 90%的人类表现更好，并且能通过编程考试、SOA 考试和医疗执照考试。现在你刚刚推出了 GPT-5。GPT-5 能做哪些 GPT-4 做不到的事情？

Awesome. Starting with what you just announced, you recently said, surprisingly recently, that GPT-4 was the dumbest model any of us will ever have to use again. But GPT-4 can already perform better than 90% of humans at the SAT and the LSAT and the GRE and it can pass coding exams and SOA exams and medical licensing. And now you just launched GPT-5. What can GPT-5 do that GPT-4 can't?

首先，一个重要结论是，你可以拥有一个能做你刚才说的所有惊人事情的 AI 系统。但它显然没有复制人类擅长的很多事情，我认为这说明了 SAT 考试或其他考试的价值。但我想，如果你回到 GPT-4 发布的那天，我们告诉你 GPT-4 在这些事情上的表现，你可能会说，「哦，天哪，这将对很多工作产生巨大影响，也会有一些负面影响，关于它意味着什么，或者人们将要做什么。」而且你可能预测的一些积极影响还没有实现。所以，这些模型擅长的东西并没有涵盖很多我们需要人们去做或关心人们去做的事情。我怀疑同样的事情会再次发生在 GPT-5 上。人们会对它的能力感到震惊。它在很多方面都非常出色，然后他们会发现他们希望它能做得更多。人们会用它来做各种不可思议的事情。它将改变很多知识工作、学习方式和创造方式。但我们，人们，社会将与之共同进化，期望更好的工具能做更多。所以，是的，我认为这个模型在很多方面非常出色，在其他方面则相当有限。但事实上，对于 3 分钟、5 分钟、1 小时的任务，一个领域的专家可能能做到或可能挣扎，而你口袋里有一款软件能做所有这些事情，这真的很惊人。我认为这在人类历史上是前所未有的，一项技术改进得如此之快。我们现在拥有这个工具，我们正在经历它，并逐步调整。但如果我们可以回到五或十年前，说这个东西即将到来，我们可能会说，大概不会。

First of all, one important takeaway is you can have an AI system that can do all those amazing things you just said. And it clearly does not replicate a lot of what humans are good at doing, which I think says something about the value of SAT tests or whatever else. But I think had you gone back to if we were having this conversation the day of GPT-4 launch and we told you how GPT-4 did at those things, you would have been like, 'Oh man, this is going to have huge impacts and some negative impacts on what it means for a bunch of jobs or what people are going to do.' And there are a bunch of positive impacts that you might have predicted that haven't yet come true. So there is something about the way that these models are good that does not capture a lot of other things that we need people to do or care about people doing. And I suspect that same thing is going to happen again with GPT-5. People are going to be blown away by what it does. It's really good at a lot of things and then they will find that they want it to do even more. People will use it for all sorts of incredible things. It will transform a lot of knowledge work, a lot of the way we learn, a lot of the way we create. But we, people, society will co-evolve with it to expect more with better tools. So yeah, I think this model is quite remarkable in many ways, quite limited in others. But the fact that for 3-minute, 5-minute, 1-hour tasks that an expert in a field could maybe do or maybe struggle with, the fact that you have in your pocket one piece of software that can do all of these things is really amazing. I think this is unprecedented at any point in human history that a technology has improved this much this fast. And the fact that we have this tool now, we're living through it and we're kind of adjusting step by step. But if we could go back in time five or 10 years and say this thing was coming, we would be like, probably not.

假设人们还没有看到头条新闻。你兴奋的顶级具体事情是什么？还有你似乎在警告的事情，那些你可能不期望它做的事情。

Let's assume that people haven't seen the headlines. What are the topline specific things that you're excited about? And also the things that you seem to be caveating, the things that maybe you won't expect it to do.

我最兴奋的是，这是第一次我觉得我可以问任何困难的科学或技术问题，并得到一个相当好的答案。我举一个有趣的例子。当我上初中时，也许是九年级，我得到了一个 TI-83，那个老式图形计算器，我花了很长时间制作一个叫贪吃蛇的游戏。那在我学校的孩子中很流行。我算是高手，这很傻，但在 TI-83 上编程极其痛苦，耗时很长，而且很难调试。一时兴起，我用早期版本的 GPT-5，心想，不知道它能不能做一个 TI-83 风格的贪吃蛇游戏。当然，它在大约 7 秒内完美地做到了。然后我想，好吧，我 11 岁的自己会觉得这很酷吗？还是错过了过程中的某些东西？我大概有 3 秒钟在想，哦，这是好是坏？然后我立刻说，实际上，我现在想念这个游戏。我有一个疯狂新功能的想法。让我输入进去。它实现了，游戏实时更新。然后我说，实际上我希望它看起来这样。实际上，我想做这个。我有了这种体验，让我想起 11 岁时编程的感觉，就是我现在想试试这个，现在我有了这个想法，现在我可以这么快地做到，我可以实时表达想法、尝试和玩耍。我想，「哦，天哪，我有一瞬间担心孩子们会错过这种石器时代学习编程的挣扎。」但现在我为他们感到兴奋，因为人们将能够用这些新工具创造，将想法变为现实的速度，真是相当惊人。

The thing that I am most excited about is this is a model for the first time where I feel like I can ask kind of any hard scientific or technical question and get a pretty good answer. And I'll give a fun example actually. When I was in junior high, or maybe it was ninth grade, I got a TI-83, this old graphing calculator, and I spent so long making this game called Snake. It was very popular game with kids in my school. And I was like pro and it was dumb, but it was like programming on TI-83 was extremely painful and took a long time and it was really hard to debug. And on a whim with an early copy of GPT-5, I was like, I wonder if it can make a TI-83 style Game of Snake. And of course, it did that perfectly in like 7 seconds. And then I was like, okay, am I supposed to be—would my 11-year-old self think this was cool or like, you know, miss something from the process? And I had like 3 seconds of wondering like, oh, is this good or bad? And then I immediately said, actually, now I'm missing this game. I have this idea for a crazy new feature. Let me type it in. It implements it and the game live updates. And I'm like, actually I'd like it to look this way. Actually, I'd like to do this thing. And I had this experience that reminded me of being like 11 in programming again, where I was just like, I now I want to try this, now I have this idea, now I could do it so fast and I could express ideas and try things and play with things in such real time. I was like, 'Oh man, you know, I was worried for a second about kids like missing the struggle of learning to program in this sort of stone age way.' And now I'm just thrilled for them because the way that people will be able to create with these new tools, the speed with which you can sort of bring ideas to life, is pretty amazing.

所以，GPT-5 不仅能回答所有难题，还能真正按需、几乎即时地生成软件——我认为这将成为 GPT-5 时代的一个标志性特征，而 GPT-4 时代并不具备这一点。

So this idea that GPT-5 can not only answer all these hard questions for you but really create on-demand, almost instantaneous software — I think that's going to be one of the defining elements of the GPT-5 era in a way that did not exist with GPT-4.

你提到这个，让我想到举重中的一个概念：张力下的时间。对于不了解的人来说，你可以用 3 秒深蹲 100 磅，也可以用 30 秒深蹲 100 磅。用 30 秒深蹲会获得更多收益。当我思考我们的创作过程，以及我感觉自己做得最好的时候，都需要大量的认知张力下的时间。我认为这种认知张力下的时间非常重要。而且这几乎有点讽刺，因为这些工具的开发本身就耗费了巨大的认知张力下的时间。但在某些方面，我确实认为人们可能会说他们把这些工具当作逃避思考的出口。你可能会说，'是啊，但计算器也是如此，我们只是转向了更难的数学问题。' 你觉得这里有什么不同吗？你怎么看？

As you're talking about that, I find myself thinking about a concept in weightlifting of time under tension. For those who don't know, you can squat 100 pounds in 3 seconds or you can squat 100 pounds in 30. You gain a lot more by squatting it in 30. And when I think about our creative process and when I've felt most like I've done my best work, it has required an enormous amount of cognitive time under tension. I think that cognitive time under tension is so important. And it's ironic almost because these tools have taken enormous cognitive time under tension to develop. But in some ways, I do think people might say they're using them as an escape hatch for thinking. Now you might say, 'Yeah, but we did that with the calculator and we just moved on to harder math problems.' Do you feel like there's something different happening here? How do you think about this?

这不一样。有些人显然用 ChatGPT 是为了不思考，而有些人用它来比以往思考得更多。我希望我们能够以这样的方式构建工具，鼓励更多人用它来锻炼大脑，做得更多。社会是一个竞争激烈的地方。如果你给人们新工具，理论上也许人们会工作得更少，但实际上人们似乎工作得更努力，期望值也在提高。所以我的猜测是，和其他技术一样，有些人会做得更多，有些人会做得更少。但可以肯定的是，对于那些想用 ChatGPT 来增加认知张力下时间的人来说，他们确实能够做到。我从最活跃的前 5% 用户使用 ChatGPT 的方式中获得了很大启发。人们学习、做事和产出的数量真的令人惊叹。

It's different. There are some people who are clearly using ChatGPT not to think, and there are some people who are using it to think more than they ever have before. I am hopeful that we will be able to build the tool in a way that encourages more people to stretch their brain with it a little more and be able to do more. Society is a competitive place. If you give people new tools, in theory maybe people just work less, but in practice it seems like people work ever harder and the expectations of people just go up. So my guess is that, like other pieces of technology, some people will do more and some people will do less. But certainly for the people who want to use ChatGPT to increase their cognitive time under tension, they are really able to. I take a lot of inspiration from what the top 5% of most engaged users do with ChatGPT. It's really amazing how much people are learning and doing and outputting.

我拿到 GPT-5 才几个小时，所以一直在玩。我还在学习如何与它交互。有趣的是，我觉得自己刚学会怎么用 GPT-4，现在又要学怎么用 GPT-5。

I've only had GPT-5 for a couple hours, so I've been playing. I'm just learning how to interact with it. Part of the interesting thing is I feel like I just caught up on how to use GPT-4, and now I'm trying to learn how to use GPT-5.

你觉得怎么样？我很好奇你发现哪些具体任务最有趣，因为我想你已经用了一段时间了。

What do you think so far? I'm curious what the specific tasks that you found most interesting are, because I imagine you've been using it for a while now.

给我印象最深的是编码任务。它还有很多其他擅长的事情，但 AI 可以为任何东西编写软件这个想法，意味着你可以用新的方式表达想法，AI 可以做非常高级的事情。从某种意义上说，你可以问 GPT-4 任何问题，但因为 GPT-5 在编程方面非常出色，感觉它什么都能做。当然，它不能做物理世界的事情，但它可以让计算机做非常复杂的事情。软件是一种非常强大的控制事物并实际执行某些事情的方式。所以这对我来说是最引人注目的。它在写作方面也好多了。现在有所谓的 AI 垃圾——AI 以一种烦人的方式写作，带有破折号。GPT-5 中仍然有破折号。很多人喜欢破折号，但 GPT-5 的写作质量已经好多了。我们还有很长的路要走。我们想进一步改进。我听说 OpenAI 内部的人开始使用 GPT-5，他们知道它在所有指标上都更好，但有一种他们说不清的细微质量差异。然后当他们不得不回到 GPT-4 测试某些东西时，感觉糟透了。我不确定确切原因，但我怀疑部分原因是 GPT-5 的写作感觉更自然、更好。

I have been most impressed by the coding tasks. There's a lot of other things it's really good at, but this idea that the AI can write software for anything. That means you can express ideas in new ways, and the AI can do very advanced things. In some sense, you could ask GPT-4 anything, but because GPT-5 is so good at programming, it feels like it can do anything. Of course, it can't do things in the physical world, but it can get a computer to do very complex things. Software is this super powerful way to control stuff and actually do some things. So that for me has been the most striking. It's gotten much better at writing. There's this whole thing of AI slop — AI writes in this kind of annoying way, with em dashes. We still have the em dashes in GPT-5. A lot of people like em dashes, but the writing quality of GPT-5 has gotten much better. We still have a long way to go. We want to improve it more. I've heard from people inside OpenAI that they started using GPT-5, they knew it was better on all the metrics, but there's this nuance quality they can't quite articulate. Then when they have to go back to GPT-4 to test something, it feels terrible. I don't know exactly what the cause of that is, but I suspect part of it is the writing feels so much more natural and better.

为准备这次采访，我联系了几位 AI 和技术领域的其他领导者，收集了一些问题问你。下一个问题来自 Stripe 的 CEO Patrick Collison。这个问题会很好。我逐字读一下。是关于下一阶段的。GPT-5 之后是什么？你认为大语言模型在哪一年会做出重大的科学发现，以及还缺少什么以至于尚未发生？他特别说明我们应该把数学和像 AlphaFold 这样的特例模型放在一边。他问的是像 GPT 系列这样的完全通用模型。

In preparation for this interview, I reached out to a couple other leaders in AI and technology and gathered a couple questions for you. So this next question is from Stripe CEO Patrick Collison. This will be a good one. Read this verbatim. It's about the next stage. What comes after GPT-5? In which year do you think a large language model will make a significant scientific discovery, and what's missing such that it hasn't happened yet? He caveated here that we should leave math and special case models like AlphaFold aside. He's specifically asking about fully general purpose models like the GPT series.

我想大多数人会同意这会在未来两年内的某个时候发生。但「重大」的定义很重要。有些人可能认为重大发现发生在 2025 年初。有些人可能认为要到 2026 年底。抱歉，2026 年初。也许有些人认为要到 2027 年底。但我敢打赌，到 2027 年底，大多数人会同意已经出现了一个由 AI 驱动的重大新发现。我认为缺少的只是这些模型的认知能力。一位研究人员对我说了一个我非常喜欢的框架：一年前，我们可以在基础高中数学竞赛题上表现良好，这些题可能需要专业数学家几秒到几分钟。我们最近获得了一枚 IMO 金牌。这是一个极其困难的测试。你能解释一下这意味着什么吗？这有点像最难的数学竞赛题。这是世界上顶尖的一小部分人才能做到的。许多专业数学家可能一道题都解不出来，而我们达到了最高水平。现在有一些人类在金牌范围内获得了更高的分数，但这是一个疯狂的成就。每个问题都是 9 小时内的 6 道题，所以对一位伟大的数学家来说，每道题一个半小时。所以我们从几秒到几分钟，再到一个半小时。要证明一个重要的新数学定理，可能需要世界顶尖人物一千小时的工作。所以我们需要从那里再取得重大进展。但如果你看看我们的轨迹，你可以说，好吧，我们正在接近那个目标。我们有达到那个时间跨度的路径。我们只需要继续扩展模型。

I would say most people will agree that happens at some point over the next two years. But the definition of significant matters a lot. Some people might say significant happens in early 2025. Some people might say not until late 2026. Sorry, early 2026. Maybe some people not until late 2027. But I would bet that by late 2027, most people agree that there has been an AI-driven significant new discovery. The thing that I think is missing is just the kind of cognitive power of these models. A framework that one of the researchers said to me that I really liked is: a year ago we could do well on basic high school math competition problems that might take a professional mathematician seconds to a few minutes. We very recently got an IMO gold medal. That is a crazy difficult test. Could you explain what that means? That's kind of like the hardest competition math test. This is something that only the very top slice of the world can do. Many professional mathematicians wouldn't solve a single problem, and we scored at the top level. Now there are some humans that got an even higher score in the gold medal range, but this is a crazy accomplishment. Each of these problems is like six problems over 9 hours, so an hour and a half per problem for a great mathematician. So we've gone from a few seconds to a few minutes to an hour and a half. Maybe to prove a significant new mathematical theorem is like a thousand hours of work for a top person in the world. So we've got to go from that to another significant gain. But if you look at our trajectory, you can say, okay, we're getting to that. We have a path to get to that time horizon. We just need to keep scaling the models.

你描述的长期未来是超级智能。

The long-term future that you've described is superintelligence.

这到底是什么意思？我们怎么知道已经达到了？

What does that actually mean? And how will we know when we've hit it?

如果我们有一个系统，能做比整个 OpenAI 研究团队更好的研究，比如我们说，「好吧，我们使用 GPU 的最佳方式是让这个 AI 决定我们应该运行哪些实验」——比 OpenAI 的整个智囊团更聪明。而且如果同一个系统能比我更好地运营 OpenAI。那么你就有了一个比最优秀的研究人员更强、比我更强、比其他人更能胜任他们工作的东西，这对我来说就是超级智能。

If we had a system that could do better research, better AI research than the whole OpenAI research team, like if we said, 'Okay, the best way we can use our GPUs is to let this AI decide what experiments we should run' – smarter than the whole brain trust of OpenAI. And if that same system could do a better job running OpenAI than I could. So you have something that's better than the best researchers, better than me at this, better than other people at their jobs, that would feel like superintelligence to me.

这句话在几年前听起来还像是科幻小说。现在也有点像，但你能透过迷雾看到它。

That is a sentence that would have sounded like science fiction just a couple years ago. And now it kind of does, but you can see it through the fog.

是的。

Yes.

所以听起来你在这条路上提到的一个步骤是科学发现的时刻：提出更好的问题，以专家级人类的方式处理问题，从而得出新发现。我脑子里一直萦绕的一个问题是，如果我们回到 1899 年，能够把直到那时的所有物理学知识都给它，然后让它稍微推演一下，没有更多数据。那么这些系统会在什么时候提出广义相对论？有趣的问题是，如果我们向前看，比如考虑我们现在的位置，一个真正优秀的超级智能是否只需要深入思考我们现有的数据，也许就能在没有新粒子加速器的情况下解决高能物理问题，还是它需要建造一个新的加速器并设计新的实验？

And so one of the steps it sounds like you're saying on that path is this moment of scientific discovery of asking better questions, of grappling with things in a way that expert level humans do to come up with new discoveries. One of the things that keeps knocking around in my head is if we were in 1899 and we were able to give it all of physics up until that point and play it out a little bit. Nothing further than that. At what point would one of these systems come up with general relativity? Interesting question is if we think about that forward, like if we think of where we are now, should a really good superintelligence just think super hard about our existing data and maybe solve high energy physics with no new particle accelerator, or does it need to build a new one and design new experiments?

显然我们不知道答案。不同的人有不同的猜测。但我怀疑我们会发现，对于很多科学领域来说，仅仅更深入地思考现有数据是不够的，我们还需要建造新仪器、进行新实验，而这需要时间。现实世界是缓慢而混乱的。所以我确信我们仅通过更深入地思考现有的科学数据就能取得一些进展。但我的猜测是，要取得重大进展，我们还需要建造新机器、进行新实验，这其中会存在一些固有的减速。

Obviously we don't know the answer to that. Different people have different speculation. But I suspect we will find that for a lot of science, it's not enough to just think harder about data we have, but we will need to build new instruments, conduct new experiments, and that will take some time. The real world is slow and messy. So I'm sure we could make some more progress just by thinking harder about the current scientific data we have. But my guess is to make the big progress we'll also need to build new machines and run new experiments and there will be some slowdown built into that.

另一种思考方式是，现在的 AI 系统非常擅长回答几乎任何问题。但也许我们说的是，这还需要另一个飞跃。而 Patrick 的问题核心是提出更好的问题。

Another way of thinking about this is AI systems now are just incredibly good at answering almost any question. But maybe one of the things we're saying is it's another leap yet. And what Patrick's question is getting at is to ask the better questions.

或者如果我们回到这种时间线问题，我们或许可以说 AI 系统在一分钟的任务上已经超越人类，但在千小时任务上还有很长的路要走。在处理这些长期任务时，人类智能有一个维度似乎与 AI 系统非常不同。现在，我认为我们会解决这个问题，但今天它确实是一个弱点。

Or if we go back to this kind of timeline question, we could maybe say that AI systems are superhuman on one-minute tasks, but a long way to go to the thousand-hour tasks. And there's a dimension of human intelligence that seems very different than AI systems when it comes to these long horizon tasks. Now, I think we will figure it out, but today it's a real weak point.

我们已经谈到了我们现在的位置，比如 GPT-5。我们谈到了超级智能的最终目标或未来目标。我当然有一个问题，那就是穿过两者之间的迷雾是什么样子。

We've talked about where we are now with GPT-5. We talked about the end goal or future goal of superintelligence. One of the questions that I have, of course, is what does it look like to walk through the fog between the two.

下一个问题来自 Nvidia 首席执行官黄仁勋。我将逐字阅读。「事实是什么就是什么。真理是它的含义。所以事实是客观的。真理是个人的。它们取决于视角、文化、价值观、信仰、背景。一个 AI 可以学习和知道事实。但一个 AI 如何知道每个国家、每个背景下的每个人的真理？」

The next question is from Nvidia CEO Jensen Huang. I'm going to read this verbatim. 'Fact is what is. Truth is what it means. So facts are objective. Truths are personal. They depend on perspective, culture, values, beliefs, context. One AI can learn and know the facts. But how does one AI know the truth for everyone in every country and every background?'

我将把这些定义作为公理接受。我不确定是否同意它们，但为了节省时间，我就直接采用。我将接受这些定义并继续。我一直感到惊讶，我想很多其他人也感到惊讶，AI 在适应不同文化背景和个人方面如此流畅。我们在 ChatGPT 中推出过的最喜欢的功能之一是今年早些时候推出的增强记忆。它真的感觉像我的 ChatGPT 了解我、关心什么、我的生活经历和背景，以及那些让我走到今天的事情。我最近有一个朋友，他是 ChatGPT 的重度用户，他把很多生活经历都放进了这些对话中。他给他的 ChatGPT 做了一系列性格测试，让它以他的身份回答，结果它得到了和他实际测试相同的分数，尽管他从未真正谈论过自己的性格。而我的 ChatGPT 在多年与我谈论我的文化、价值观和生活的过程中真正学到了东西。我有时会用一个免费账户看看没有我的历史记录是什么样子，感觉非常不同。所以我认为我们都在积极方面感到惊讶，AI 在学习和适应方面做得如此之好。

I'm going to accept as axioms those definitions. I'm not sure if I agree with them, but in the interest of time, I will just take them. I will take those definitions and go with it. I have been surprised, I think many other people have been surprised too about how fluent AI is at adapting to different cultural contexts and individuals. One of my favorite features that we have ever launched in ChatGPT is the enhanced memory that came out earlier this year. It really feels like my ChatGPT gets to know me and what I care about, my life experiences and background, and the things that have led me to where I am. A friend of mine recently who's been a huge ChatGPT user, so he's put a lot of his life into all these conversations. He gave his ChatGPT a bunch of personality tests and asked them to answer as if they were him, and it got the same scores he actually got, even though he'd never really talked about his personality. And my ChatGPT has really learned over the years of me talking to it about my culture, my values, my life. And I sometimes will use a free account just to see what it's like without any of my history, and it feels really different. So I think we've all been surprised on the upside of how good AI is at learning this and adapting.

那么你是否设想在世界许多不同地方，人们使用具有不同文化规范和背景的不同 AI？我们是在说这个吗？

So do you envision in many different parts of the world people using different AIs with different cultural norms and contexts? Is that what we're saying?

我认为每个人都会使用相同的基础模型，但会为该模型提供上下文，使其以他们想要的、他们社区想要的个性化方式行事，等等。

I think that everyone will use the same fundamental model, but there will be context provided to that model that will make it behave in a personalized way they want, their community wants, whatever.

我认为当我们谈到事实和真理这个概念时，这似乎是我们第一次时间旅行的好时机。好的，我们要去 2030 年。这是一个严肃的问题，但我想用一个轻松的例子来问。你见过那些在蹦床上跳的兔子吗？

I think when we're getting at this idea of facts and truth, it brings me to this seems like a good moment for our first time travel trip. Okay, we're going to 2030. This is a serious question, but I want to ask it with a light-hearted example. Have you seen the bunnies that are jumping on the trampoline?

是的。

Yes.

所以，对于那些没看过的人来说，它看起来像是后院拍摄的兔子在蹦床上跳的视频。这段视频最近非常火爆。还有一首关于它的人造歌曲。这成了一件事。我认为人们反应如此强烈的原因可能是，这是人们第一次看到一段视频，喜欢它，然后后来发现它完全是 AI 生成的。在这次时间旅行中，如果我们想象在 2030 年，我们是青少年，在刷 2030 年青少年刷的东西。我们如何分辨什么是真实的，什么不是真实的？

So, for those who haven't seen it, it looks like backyard footage of bunnies enjoying jumping on a trampoline. And this has gone incredibly viral recently. There's a human-made song about it. It's a whole thing. And I think the reason why people reacted so strongly to it was maybe the first time people saw a video, enjoyed it, and then later found out that it was completely AI generated. In this time travel trip, if we imagine in 2030, we are teenagers and we're scrolling whatever teenagers are scrolling in 2030. How do we figure out what's real and what's not real?

我可以给出各种字面上的答案。我们可以对内容进行加密签名，然后决定我们信任谁的签名，以确认他们是否真的拍摄了某物。但我的感觉是，将要发生的是它会逐渐趋同。即使你今天用 iPhone 拍的照片，大部分是真实的，但也不完全是。

I can give all sorts of literal answers to that question. We could be cryptographically signing stuff and we could decide who we trust their signature if they actually filmed something or not. But my sense is what's going to happen is it's just going to gradually converge. Even a photo you take out of your iPhone today, it's mostly real, but it's a little not.

那里有一些你不理解的 AI 在运行，让它看起来更好一点，有时你会看到这些奇怪的东西，比如月亮。

There's like some AI thing running there in a way you don't understand and making it look a little bit better and sometimes you see these weird things where the moon.

是的。但从相机传感器捕捉到的光子到你最终看到的图像之间，有大量的算力在运作。你已经认定它足够真实，或者大多数人都认为它足够真实。但我们逐渐接受了从光子打在相机胶片上的时代到现在的转变。如果你去看 TikTok 上的视频，很可能有各种各样的视频编辑工具被用来让它看起来比现实更好。

Yeah. But there's a lot of processing power between the photons captured by that camera sensor and the image you eventually see. And you've decided it's real enough or most people decided it's real enough. But we've accepted some gradual move from when it was like photons hitting the film in a camera. And if you go look at some video on TikTok, there's probably all sorts of video editing tools being used to make it better than real look.

是的，没错。或者就像，你知道的，整个场景完全生成，或者有些整个视频都是生成的，比如那些在蹦床上的兔子。我认为，关于它需要多真实才能被认为是真实的门槛会不断移动。

Yeah, exactly. Or it's just like, you know, whole scenes are completely generated or some of the whole videos are generated like those bunnies on that trampoline. And I think that the sort of threshold for how real does it have to be to consider it real will just keep moving.

所以这有点像是一个教育问题。

So it's sort of an education question.

是的。我的意思是，媒体总是有点真实，有点不真实。就像我们看科幻电影，我们知道那不是真的。你看别人在 Instagram 上发的度假美照，好吧，也许那张照片确实是拍的，但有很多游客排队等着拍同样的照片，而这一点被省略了。我认为我们现在已经接受了这一点。当然，更高比例的媒体会让人觉得不真实。但我认为这是长期趋势。

Yeah. I mean media is always a little bit real and a little bit not real. Like we watch a sci-fi movie. We know that didn't really happen. You watch someone's beautiful photo of themselves on vacation on Instagram. Okay, maybe that photo was literally taken, but there's tons of tourists in line for the same photo and that's left out of it. And I think we just accept that now. Certainly, a higher percentage of media will feel not real. But I think that's been the long-term trend.

总之，我们又要跳转了。

Anyway, we're going to jump again.

好的。

Okay.

2035 年，我们大学毕业，你和我。AI 领域的一些领导者说过，5 年内一半的初级白领岗位将被 AI 取代。所以 5 年后我们就是大学毕业生。你希望世界对我们来说是什么样子？我认为有很多关于 AI 可能导致失业的讨论，但我也很好奇。我有一份十年前没人认为我们能拥有的工作。如果我们考虑 2035 年，大学毕业生——如果他们还会上大学的话——很可能像乘坐宇宙飞船执行探索太阳系的任务，从事某种全新、令人兴奋、高薪、超有趣的工作，并且为你我感到遗憾，因为我们不得不做这种非常无聊的旧工作，一切都会更好。

2035, we're graduating from college, you and me. There are some leaders in the AI space that have said that in 5 years half of the entry level white collar workforce will be replaced by AI. So we're college graduates in 5 years. What do you hope the world looks like for us? I think there's been a lot of talk about how AI might cause job displacement, but I'm also curious. I have a job that nobody would have thought we could have totally a decade ago. What are the things that we could look ahead if we're thinking about in 2035 that like graduating college student, if they still go to college at all, could very well be like leaving on a mission to explore the solar system on a spaceship in some kind of completely new exciting, super well-paid, super interesting job and feeling so bad for you and I that like we had to do this kind of really boring old kind of work and everything is just better.

10 年现在很难想象，因为太远了。如果以当前的变化速度再复合 10 年，那可能是我们甚至无法时间旅行去的地方。

10 years feels very hard to imagine at this point because it's too far. If you compound the current rate of change for 10 more years, it's probably something we can't even time travel trips.

我认为现在很难想象 10 年前的样子。但我认为 10 年后会更难想象，更加不同。

I think now would be really hard to imagine 10 years ago. But I think 10 years forward will be even much harder, much more different.

是的。

Yeah.

那我们改成 5 年吧。我们还是到 2030 年。我很好奇你认为这对年轻人的短期影响会是什么。我的意思是，像一半的初级岗位被 AI 取代，听起来他们将要进入的世界和我当年进入的截然不同。

So let's make it 5 years. We're still going to 2030. I'm curious what you think the pretty short-term impacts of this will be for young people. I mean, these like half of entry-level jobs replaced by AI makes it sound like a very different world that they would be entering than the one that I did.

我认为某些类别的工作会完全消失，这完全正确。这种情况一直发生，而年轻人最擅长适应。我更担心的是这对 62 岁的人意味着什么，而不是 22 岁的人。62 岁的人不想去重新培训或学习新技能，不管政客们怎么称呼它，实际上除了政客没人想要。如果我现在 22 岁，即将大学毕业，我会觉得自己是历史上最幸运的孩子。

I think it's totally true that some classes of jobs will totally go away. This always happens and young people are the best at adapting to this. I'm more worried about what it means, not for the 22-year-old, but for the 62-year-old that doesn't want to go retrain or reskill or whatever the politicians call it that no one actually wants but politicians most of the time. If I were 22 right now and graduating college, I would feel like the luckiest kid in all of history.

为什么？

Why?

因为从来没有一个更棒的时代去创造全新的东西，去发明，去创业，无论是什么。我认为现在很可能创办一家一人公司，最终价值超过十亿美元，更重要的是，为世界提供令人惊叹的产品和服务，这太疯狂了。你可以使用那些曾经需要数百人团队才能完成的工具，你只需要学会如何使用这些工具，并想出好主意，这非常了不起。

Because there's never been a more amazing time to go create something totally new, to go invent something, to start a company, whatever it is. I think it is probably possible now to start a company that is a one-person company that will go on to be worth more than a billion dollars and more importantly than that deliver an amazing product and service to the world and that is a crazy thing. You have access to tools that can let you do what used to take teams of hundreds and you just have to learn how to use these tools and come up with a great idea and it's quite amazing.

如果我们退一步，我认为在这个乐观的节目中，观众能从你这里听到的最重要的东西分为两部分。首先，从战术上讲，你实际上是如何尝试构建世界上最强大的智能的，以及做到这一点的速率限制因素是什么？然后，从哲学上讲，你和其他人如何以真正帮助而非伤害人的方式构建这项技术？所以现在先讲战术部分。我的理解是，有三个大类一直是 AI 的限制因素：第一是算力，第二是数据，第三是算法设计。你现在如何看待这三类？如果你要帮助某人理解他们可能看到的下一波头条新闻，你会如何帮助他们理解这一切？

If we take a step back, I think the most important thing that this audience could hear from you on this optimistic show is in two parts. First, there's tactically, how are you actually trying to build the world's most powerful intelligence and what are the rate limiting factors to doing that? And then philosophically, how are you and others working on building that technology in a way that really helps and not hurts people? So just taking the tactical part right now. My understanding is that there are three big categories that have been limiting factors for AI. The first is compute, the second is data and the third is algorithmic design. How do you think about each of those three categories right now? And if you were to help someone understand the next headlines that they might see, how would you help them make sense of all this?

我会说还有第四个，那就是弄清楚要构建什么产品。科学进步本身如果不交到人们手中，效用有限，也不会以同样的方式与社会共同进化。但如果我能同时推进这四个方面……在算力方面，这绝对是我见过的最大的基础设施项目，可能将成为——我认为可能已经是——人类历史上最大、最昂贵的项目。但整个供应链，从制造芯片、内存和网络设备，到将它们装入服务器，进行大型建设项目以建造超大规模数据中心，找到获取能源的方法——这通常是限制因素之一——以及所有其他组件，都极其复杂和昂贵。我们仍然以某种定制的一次性方式在做这件事，尽管情况正在好转。最终，我们将设计出整个巨型工厂，从精神上讲，它将在一端熔化沙子，在另一端输出完全构建好的 AI 算力，但我们离那还很远，这是一个极其复杂和昂贵的过程。

I would say there's a fourth too, which is figuring out the products to build. Scientific progress on its own not put into the hands of people is of limited utility and doesn't co-evolve with society in the same way. But if I could hit all four of those... So on the compute side, this is the biggest infrastructure project certainly that I've ever seen, possibly it will become, I think it may already be, the biggest and most expensive one in human history. But the whole supply chain from making the chips and the memory and the networking gear, racking them up in servers, doing a giant construction project to build a mega data center, finding a way to get the energy, which is often a limiting factor piece of this, and all the other components together. This is hugely complex and expensive. And we are still doing this in a sort of bespoke one-off way although it's getting better. Eventually we will just design a whole kind of mega factory that takes you know, spiritually it will be melting sand on one end and putting out fully built AI compute on the other, but we are a long way from that and it's an enormously complex and expensive process.

我们投入了大量精力来尽可能快地建设算力。这会很令人沮丧，因为 GPT-5 即将发布，需求会再次激增，而我们无法满足。这就像早期 GPT-4 的日子。世界对 AI 的需求远超我们目前能提供的，而建设更多算力是实现这一目标的重要部分。这实际上是我将投入大部分精力的事情：如何以更大的规模建设算力。从数百万到数千万、数亿，最终希望达到数十亿 GPU，来服务人们想用 AI 做的事情。

We are putting a huge amount of work into building out as much compute as we can and to do it fast. It's going to be sad because GPT-5 is going to launch and there's going to be another big spike in demand, and we're not going to be able to serve it. It's going to be like those early GPT-4 days. The world just wants much more AI than we can currently deliver, and building more compute is an important part of doing that. That's actually what I expect to turn the majority of my attention to: how we build compute at much greater scales. So how we go from millions to tens of millions and hundreds of millions and eventually hopefully billions of GPUs that are in service of what people want to do with this.

当你思考这个问题时，这个领域里你将要面对的主要挑战是什么？

When you're thinking about it, what are the big challenges here in this category that you're going to be thinking about?

我们目前最大的限制是能源。如果你想运行一个吉瓦级的数据中心，一个吉瓦——这有多难找？短期内很难找到可用的吉瓦级电力。我们还受到处理芯片和内存芯片的严重限制，以及如何将它们封装在一起、如何构建机架。还有一系列其他事情，比如许可证、施工工作。但同样，这里的目标是实现自动化。一旦我们造出一些机器人，它们可以帮助我们进一步自动化。但想象一个世界，你基本上可以投入资金，然后得到一个预建的数据中心——如果我们能做到，那将是一个巨大的突破。

We're currently most limited by energy. If you want to run a gigawatt-scale data center, it's a gigawatt—how hard can that be to find? It's really hard to find a gigawatt of power available in the short term. We're also very much limited by the processing chips and the memory chips, how you package these all together, how you build the racks. Then there's a list of other things like permits, construction work. But again, the goal here will be to really automate this. Once we get some of those robots built, they can help us automate it even more. But just a world where you can basically pour in money and get out a pre-built data center—that would be a huge unlock if we can get it to work.

第二个类别：数据。这些模型已经变得非常聪明。曾经我们可以再喂它一本物理教科书，它就会在物理方面变得更聪明一点，但现在说实话，GPT-5 已经很好地理解了物理教科书中的所有内容。我们对合成数据感到兴奋。我们非常兴奋于用户帮助我们创建越来越难的任务和环境，让系统去解决。但我认为数据将永远重要，但我们正在进入一个模型需要学习数据集中尚不存在的事物的领域。它们必须去发现新事物。所以这是一个疯狂的新事物。

Second category: data. These models have gotten so smart. There was a time when we could just feed it another physics textbook and it got a little bit smarter at physics, but now honestly GPT-5 understands everything in a physics textbook pretty well. We're excited about synthetic data. We're very excited about our users helping us create harder and harder tasks and environments for the system to solve. But I think data will always be important, but we're entering a realm where the models need to learn things that don't exist in any dataset yet. They have to go discover new things. So that's a crazy new thing.

你如何教模型去发现新事物？

How do you teach a model to discover new things?

嗯，人类可以做到。我们可以提出假设，测试它们，获得实验结果，并根据所学进行更新。所以可能也是同样的方式。

Well, humans can do it. We can go off and come up with hypotheses and test them and get experimental results and update on what we learn. So probably the same kind of way.

然后是算法设计。我们在算法设计上取得了巨大进展。我认为 OpenAI 在世界上最擅长的事情是，我们建立了一种反复取得重大算法研究突破的文化。我们弄清楚了 GPT 范式，弄清楚了推理范式。我们现在正在研究一些新的范式。但想到我们面前还有更多数量级的算法进步，这让我非常兴奋。就在昨天，我们发布了一个名为 GPOSS 的开源模型。这个模型和 o4-mini 一样聪明，而 o4-mini 是一个非常聪明的模型，可以在笔记本电脑上本地运行。这让我震惊。如果你几年前问我，什么时候能有那种智能水平的模型在笔记本电脑上运行，我会说很多很多年以后。但后来我们取得了一些算法进步，特别是在推理方面，以及其他一些方面，让我们能够做出一个能做这种神奇事情的小模型。那些是最有趣的事情。那是工作中最酷的部分。

And then there's algorithmic design. We've made huge progress on algorithmic design. The thing that I think OpenAI does best in the world is we have built this culture of repeated and big algorithmic research gains. We figured out what became the GPT paradigm. We figured out what became the reasoning paradigm. We're working on some new ones now. But it is very exciting to me to think that there are still many more orders of magnitude of algorithmic gains ahead of us. We just yesterday released a model called GPOSS, an open-source model. It's a model that is as smart as o4-mini, which is a very smart model that runs locally on a laptop. And this blows my mind. If you had asked me a few years ago when we'd have a model of that intelligence running on a laptop, I would have said many, many years in the future. But then we found some algorithmic gains, particularly around reasoning but also some other things, that let us do a tiny model that can do this amazing thing. Those are the most fun things. That's the coolest part of the job.

我看得出来你真的很享受思考这个问题。我很好奇，对于那些不太了解你在说什么、不熟悉算法设计如何带来他们实际使用的更好体验的人，你能总结一下当前的状况吗？比如，当你思考这个问题有多有趣时，你在想什么？

I can see you really enjoying thinking about this. I'm curious for people who don't quite know what you're talking about, who aren't familiar with how algorithmic design would lead to a better experience that they actually use. Could you summarize the state of things right now? Like what is it that you're thinking about when you're thinking about how fun this problem is?

让我从历史开始讲，然后谈到今天的一些事情。GPT-1 在当时是一个被许多领域专家嘲笑的想法：我们能不能训练一个模型来玩一个小游戏——给它一堆单词，让它猜序列中下一个单词是什么？这叫做无监督学习。你并不是在说「这是猫，这是狗」，而是说「这里有一些单词，猜下一个」。而它竟然能学习这些非常复杂的概念，学习所有关于物理、数学和编程的知识，并不断预测下一个单词，这看起来荒谬、神奇、不太可能成功。这一切是如何被编码的？然而人类做到了。婴儿开始听到语言，并在很大程度上自己理解其含义。所以我们做到了，我们还意识到，如果扩大规模，它会越来越好，但我们必须扩大许多个数量级。所以在 GPT-1 时代它并不好，一点都不好。很多专家说：「哦，这太荒谬了。它永远不会成功。它不会很稳健。」但我们有所谓的缩放定律。我们说：「好吧，随着我们增加算力、内存、数据等，这会可预测地变得更好。我们可以利用这些预测来决定如何扩大规模并获得很好的结果。」这在许多数量级上都奏效了。当时这一点并不明显。我认为这就是世界如此惊讶的原因：这似乎是一个极不可能的发现。另一个是我们可以将这些语言模型与强化学习结合使用，通过说「这个好，这个坏」来教它如何推理。这导致了 o1、o3 以及现在的 GPT-5 的进步。这是另一件感觉「如果它有效那就太好了，但绝不可能成功。它太简单了」的事情。现在我们正在研究新事物。我们已经弄清楚了如何制作更好的视频模型。我们正在发现使用新型数据和环境来扩大规模的新方法。我认为，在这个领域，5-10 年很难说，但未来几年我们面前有非常平稳、非常强劲的 Scaling。我认为公众叙事中我们已经走上了一条从一到二到三到四到五再到更多的平稳道路。但在幕后，它并不是那样线性的。它更混乱。

Let me start back in history and then I'll get to some things for today. So, GPT-1 was an idea at the time that was quite mocked by a lot of experts in the field, which was: can we train a model to play a little game—show it a bunch of words and have it guess the one that comes next in the sequence? That's called unsupervised learning. You're not really saying 'this is a cat, this is a dog.' You're saying 'here's some words, guess the next one.' And the fact that that can go learn these very complicated concepts, learn all the stuff about physics and math and programming, and keep predicting the word that comes next seemed ludicrous, magical, unlikely to work. How was that all going to get encoded? And yet humans do it. Babies start hearing language and figure out what it means largely on their own. So we did it, and we also realized that if we scaled it up, it got better and better, but we had to scale over many orders of magnitude. So it wasn't that good in the GPT-1 days. It wasn't good at all. And a lot of experts said, 'Oh, this is ridiculous. It's never going to work. It's not going to be robust.' But we had these things called scaling laws. And we said, 'Okay, so this gets predictably better as we increase compute, memory, data, whatever. And we can use those predictions to make decisions about how to scale this up and get great results.' And that has worked over a crazy number of orders of magnitude. It was so not obvious at the time. That was, I think, the reason the world was so surprised: that seemed like such an unlikely finding. Another one was that we could use these language models with reinforcement learning where we're saying 'this is good, this is bad' to teach it how to reason. And this led to the o1, o3, and now the GPT-5 progress. That was another thing that felt like 'if it works it's really great, but no way this is going to work. It's too simple.' And now we're on to new things. We've figured out how to make much better video models. We are discovering new ways to use new kinds of data and environment to scale that up as well. I think, again, 5-10 years out is too hard to say in this field, but the next couple of years we have very smooth, very strong scaling in front of us. I think it has become a sort of public narrative that we are on this smooth path from one to two to three to four to five to more. But it also is true behind the scenes that it's not linear like that. It's messier.

跟我们讲讲 GPT-5 之前的混乱局面吧。你需要解决哪些有趣的问题？

Tell us a little bit about the mess before GPT-5. What were the interesting problems that you needed to solve?

我们做了一个名为 Orion 的模型，后来以 GPT-4.5 发布。我们把它做得太大了。这是个很酷的模型，但用起来很不灵活。我们意识到，为了在模型之上进行某些研究，我们需要不同的形态。所以我们遵循了一条一直表现良好的缩放定律，却没有真正意识到有一条新的、更陡峭的缩放定律，它在算力上带来了更好的回报，那就是推理这件事。所以我们走了一条死胡同又折返回来，但这没关系，这是研究的一部分。我们在数据集的处理方式上也遇到了一些问题，因为这些模型必须变得这么大，并从这么多数据中学习。在日常工作中，你会做很多急转弯，尝试各种东西，或者某个架构想法行不通，但所有这些曲折的总和，在指数曲线上却异常平滑。

We did a model called Orion that we released as GPT-4.5. And we made it too big. It's a very cool model, but unwieldy to use. We realized that for some of the research we need to do on top of a model, we need a different shape. So we followed one scaling law that kept being good without really internalizing that there was a new, even steeper scaling law that gave better returns for compute, which was this reasoning thing. So that was one alley we went down and turned around, but that's fine. That's part of research. We had some problems with the way we think about our datasets as these models really have to get this big and learn from this much data. In the middle of it day-to-day, you make a lot of U-turns as you try things or have an architecture idea that doesn't work, but the aggregate of all the squiggles has been remarkably smooth on the exponential.

我总觉得有趣的是，当我坐在这里采访你关于你刚发布的东西时，你已经在想——你能分享哪些你正在思考的问题，以至于一年后我回来采访你时，会问这些问题？

One of the things I always find interesting is that by the time I'm sitting here interviewing you about the thing that you just put out, you're thinking about what are the things that you can share that are at least the problems that you're thinking about that I would be interviewing you about in a year if I came back?

我的意思是，你可能会问我，这东西能去发现新科学意味着什么？世界应该如何看待 GPT-6 发现新科学？现在，也许我们不会实现那个，但它感觉触手可及。

I mean, possibly you'll be asking me like, what does it mean that this thing can go discover new science? What how is the world supposed to think about GPT-6 discovering new science? Now, maybe not like maybe we don't deliver that, but it feels within grasp.

如果你做到了，你会怎么说？这种成就的影响会是什么？想象你确实成功了。

If you did, what would you say? What would the implications of that kind of achievement be? Imagine you do succeed.

是的。我的意思是，我认为好的部分会很好。坏的部分会很可怕，怪异的部分第一天会很怪异，然后我们很快就会习惯。所以我们会说，「哦，这被用来治愈疾病太不可思议了」，还有「哦，这样的模型被用来制造新的生物安全威胁太可怕了」。然后我们还会说，天哪，看着世界加速如此之快、经济增长如此之快，真的很奇怪。变化的速度会让人眩晕。然后就像其他一切一样，人类适应任何变化的非凡能力。我们只会说，「好吧，就是这样。」今天出生的孩子永远不会比 AI 更聪明。今天出生的孩子，到他们理解世界运作方式的时候，将永远习惯于事物以难以置信的速度改进和发现新科学。他们永远不会知道另一个世界。这会显得完全自然。会显得不可思议和石器时代一样，我们曾经使用电脑或手机或任何不比我们聪明得多的技术。我们会想，2020 年代的那些人过得有多糟糕。

Yeah. I mean, I think the great parts will be great. The bad parts will be scary and the bizarre parts will be bizarre on the first day and then we'll get used to them really fast. So we'll be like, 'Oh, it's incredible that this is being used to cure disease' and like, 'Oh, it's extremely scary that models like this are being used to create new biosecurity threats.' And then we'll also be like, man, it's really weird to live through watching the world speed up so much and the economy grows so fast. It will feel vertigo-inducing, the rate of change. And then like happens with everything else, the remarkable ability of humanity to adapt to any amount of change. We'll just be like, 'Okay, you know, this is it.' A kid born today will never be smarter than AI ever. And a kid born today, by the time that kid understands the way the world works, will just always be used to an incredibly fast rate of things improving and discovering new science. They will never know any other world. It will seem totally natural. It will seem unthinkable and stone age like that we used to use computers or phones or any kind of technology that was not way smarter than we were. We will think like how bad those people of the 2020s had it.

我在考虑要孩子。

I'm thinking about having kids.

你应该要。这是最棒的事情。

You should. It's the best thing ever.

我知道你刚有了第一个孩子。你刚才说的话如何影响我该怎样思考在那个世界里养育孩子？你会给我什么建议？

I know you just had your first kid. How does what you just said affect how I should think about parenting a kid in that world? What advice would you give me?

可能和几千年来你养育孩子的方式没什么不同。比如爱你的孩子，向他们展示世界，支持他们想做的一切，教他们如何成为一个好人。这大概才是重要的。这听起来有点像你说过的一些事情，比如你可能不会上大学，我觉得你到目前为止说的几件事都与此相关。

Probably nothing different than the way you've been parenting kids for tens of thousands of years. Like love your kids, show them the world, support them in whatever they want to do and teach them how to be a good person. That probably is what's going to matter. It sounds a little bit like some of the things you've said that you might not go to college, there are a couple of things that you've said so far that feed into this I think.

听起来你在说，在你设想的世界里，他们会有更多的选择，因此他们更有能力说「我想建造这个，这是能帮助我做到这一点的超强工具」。

And it sounds like what you're saying is there will be more optionality for them in a world that you envision and therefore they will have more ability to say I want to build this here's the superpowered tool that will help me do that.

是的，就像我希望我的孩子认为我过的是糟糕受限的生活，而他拥有这个令人难以置信的无限画布，可以做各种事情，这就是世界的方式。

Yeah, like I want my kid to think I had a terrible constrained life and that he has this incredible infinite canvas of stuff to do that is like the way of the world.

我们说过 2035 年有点太遥远了。所以也许这本来要跳到 2040 年，但也许我们可以缩短一点。当我想到 AI 能对我们孩子和我们所有人产生最大真正积极影响的领域时，那就是健康。所以如果我们选个年份，比如 2035 年，我坐在这里采访斯坦福医学院的院长，你希望他告诉我 AI 在 2035 年为我们的健康做了什么？

We've said that 2035 is a little bit too far in the future to think about. So maybe this was going to be a jump to 2040 but maybe it will keep it shorter than that. When I think about the area where AI could have for both our kids and us the biggest genuinely positive impact on all of us, it's health. So if we are in pick your year, call it 2035 and I'm sitting here and I'm interviewing the dean of Stanford medicine, what do you hope that he's telling me AI is doing for our health in 2035?

从 2025 年开始。我们对 GPT-5 最自豪的一点是它在健康建议方面进步了很多。人们大量使用 GPT-4 模型来获取健康建议。我相信你在网上看到过一些例子，有人说「我得了这种危及生命的疾病，没有医生能诊断出来，我把症状和血检结果输入 ChatGPT，它准确地告诉了我这种罕见病。我去看了医生，吃了药，我痊愈了。」这太棒了。显然，ChatGPT 的查询中有很大一部分与健康相关。所以我们想在这方面做得非常好，并投入了大量资源。GPT-5 在医疗相关查询上显著更好。

Start with 2025. One of the things we are most proud of with GPT-5 is how much better it's gotten at health advice. People have used the GPT-4 models a lot for health advice. And I'm sure you've seen some of these things on the internet where people are like, I had this life-threatening disease and no doctor could figure it out and I put my symptoms and a blood test into ChatGPT. It told me exactly the rare thing I had. I went to a doctor. I took a pill. I'm cured. That's amazing. Obviously a huge fraction of ChatGPT queries are health related. So we wanted to get really good at this and we invested a lot. GPT-5 is significantly better at healthcare related queries.

这里的「更好」是什么意思？

What does better mean here?

它给你更好的答案。更准确，幻觉更少，更有可能告诉你实际得了什么以及实际应该做什么。更好的医疗保健很棒，但显然人们真正想要的是不得病。到 2035 年，我认为我们将能够利用这些工具治愈或至少治疗大量目前困扰我们的疾病。我认为这将是 AI 最切身感受到的好处之一。

It gives you a better answer. More accurate, hallucinates less, more likely to tell you what you actually have and what you actually should do. And better healthcare is wonderful, but obviously what people actually want is to just not have disease. By 2035, I think we will be able to use these tools to cure a significant number or at least treat a significant number of diseases that currently plague us. I think that'll be one of the most viscerally felt benefits of AI.

人们经常谈论 AI 将如何彻底改变医疗保健，但我很好奇想更深入一层，具体你在想象什么。比如，是这些 AI 系统能帮助我们更早发现 GLP-1 类药物，这种药已经存在很久了，但我们不知道它的其他效果？还是 AlphaFold 和蛋白质折叠正在帮助创造新药？我希望能够要求 GPT-8 去治愈某种特定的癌症，我希望 GPT-8 去思考，然后说「好的，我读了我能找到的所有资料。」

People talk a lot about how AI will revolutionize healthcare, but I'm curious to go one turn deeper on specifically what you're imagining. Like, is it that these AI systems could have helped us see GLP-1s earlier, this medication that has been around for a long time, but we didn't know about this other effect? Is it that AlphaFold and protein folding is helping create new medicines? I would like to be able to ask GPT-8 to go cure a particular cancer and I would like GPT-8 to go off and think and then say okay I read everything I could find.

我有这些想法。我需要你去找一个实验室技术员，运行这九个实验，然后告诉我每个实验的结果。等两个月让细胞完成它们的工作。把结果发回给 GPT-8。说「我试过了，给你。」思考。说「好的，我只需要再做一次实验。那是个惊喜。再做一次实验。把它交回来。」GPT 说：「好的，去合成这个分子，然后试试小鼠研究之类的。」好的，那很好。比如，试试人体研究。好的，太棒了。它成功了。这是如何通过 FDA 审批的。

I have these ideas. I need you to go get a lab technician to run these nine experiments and tell me what you find for each of them. And wait 2 months for the cells to do their thing. Send the results back to GPT-8. Say I tried it. Here you go. Think. Say okay I just need one more experiment. That was a surprise. Run one more experiment. Give it back. GPT says, 'Okay, go synthesize this molecule and try mouse studies or whatever.' Okay, that was good. Like, try human studies. Okay, great. It worked. Here's how to run it through the FDA.

我想任何有亲人死于癌症的人也会非常喜欢这个。

I think anyone with a loved one who's died of cancer would also really like that.

好的，我们再跳一下。

Okay, we're going to jump again.

好的。

Okay.

我本来想说 2050 年，但再说一次，我所有的时间线都在变得越来越短。

I was going to say 2050, but again, all of my timelines are getting much, much shorter.

现在确实感觉世界发展得非常快。

It does feel like the world's going very fast now.

确实如此。是的。当我与其他 AI 领导者交谈时，他们提到的一件事是工业革命。他们说：「我选择 2050 年，因为我听到人们谈论到那时我们将经历的变化将像工业革命一样，但引用说大 10 倍、快 10 倍。」工业革命给了我们现代医学、卫生、交通、大规模生产以及我们现在认为理所当然的所有便利。它也让很多人经历了大约 100 年的极其艰难的时期。如果这将是 10 倍大、10 倍快，如果我们不断缩短我们在这里讨论的时间线，即使是在这次对话中，这对大多数人来说实际上会是什么感觉？我想我想说的是，如果这一切按照你希望的方式发展，那么在此期间谁还会受到伤害？我真的不知道经历这一切会是什么感觉。我认为我们处于未知的水域。我确实相信人类的适应能力，以及某种无限的创造力和对事物的渴望，我认为我们总能找到新的事情做，但过渡期如果发生得这么快——我不认为它会像一些同事说的技术那样快，但社会有很大的惯性。

It does. Yeah. And when I talk to other leaders in AI, one of the things that they refer to is the industrial revolution. They say, 'I chose 2050 because I've heard people talk about how by then the change that we will have gone through will be like the industrial revolution, but quote 10 times bigger and 10 times faster.' The industrial revolution gave us modern medicine and sanitation and transportation and mass production and all the conveniences that we now take for granted. It also was incredibly difficult for a lot of people for about 100 years. If this is going to be 10 times bigger and 10 times faster if we keep reducing the timelines that we're talking about here, even in this conversation, what does that actually feel like for most people? And I think what I'm trying to get at is if this all goes the way you hope, who still gets hurt in the meantime? I don't really know what this is going to feel like to live through. I think we're in uncharted waters here. I do believe in human adaptability and sort of infinite creativity and desire for stuff and I think we always do figure out new things to do but the transition period if this happens as fast as it might and I don't think it will happen as fast as some of my colleagues say the technology will but society has a lot of inertia.

嗯。人们会适应他们的生活方式。

Mhm. People adapt their way of living.

是的。

Yeah.

慢得惊人。

Surprisingly slowly.

有两类工作将完全消失，还有许多类工作将发生重大变化，并且会有新的事物出现，就像你的工作在不久之前还不存在一样。我的工作也是如此。从某种意义上说，这种情况已经持续了很长时间。它仍然对个人造成破坏，但社会已经证明对此相当有韧性。然后在另一种意义上，我们不知道这能走多远、多快。因此，我认为我们需要一种不寻常的谦逊和开放态度，去考虑那些不久前还看起来完全超出奥弗顿窗口的新解决方案。

There are two classes of jobs that are going to totally go away and there will be many classes of jobs that change significantly and there'll be the new things in the same way that your job didn't exist some time ago. Neither did mine. And in some sense, this has been going on for a long time. And it's still disruptive to individuals, but society has proven quite resilient to this. And then in some other sense we have no idea how far or fast this could go. And thus I think we need an unusual degree of humility and openness to considering new solutions that would have seemed way out of the Overton window not too long ago.

我想谈谈其中一些可能是什么，因为我绝不是历史学家，但据我理解，第一次工业革命导致了许多公共卫生措施的实施，因为公共卫生变得非常糟糕。导致了现代卫生设施，因为公共卫生变得非常糟糕。第二次工业革命导致了劳动力保护，因为劳动条件变得非常糟糕。每一次大的飞跃都会造成混乱，那个混乱需要被清理，我们已经做到了。我很好奇，这听起来像是我们正处于这个巨大变革的中间。我们能多早、多具体地确定那个混乱可能是什么？我们可以提前采取哪些公共干预措施来减少我们认为即将面临的混乱？

I'd like to talk about what some of those could be because I'm not a historian by any means, but the first industrial revolution, my understanding is led to a lot of public health implementations because public health got so bad. Led to modern sanitation because public health got so bad. The second industrial revolution led to workforce protections because labor conditions got so bad. Every big leap creates a mess and that mess needs to be cleaned up and we've done that. And I'm curious, this is going to be it sounds like we're in the middle of this enormously. How specific can we get as early as possible about what that mess can be? What are the public interventions that we could do ahead of time to reduce the mess that we think that we're headed for?

我再次，我将为了乐趣而推测，但前提是，我甚至不是经济学家，更不用说能预见未来的人了。在我看来，社会契约的一些基本内容可能不得不改变。也可能不会。也许资本主义实际上运行得非常好，供需平衡各司其职，我们都只是找到新的工作和新的方式相互传递价值。但在我看来，我们很可能需要思考如何分享这个未来最重要的资源。在我看来，最好的做法是让 AI 算力尽可能丰富和廉价，以至于我们拥有太多，甚至想不出好的新点子来真正使用它，然后就像你想要什么就有什么一样。没有这一点，我可以看到真正的战争会为此而打。但是，关于如何分配 AGI 算力访问权的新想法，这似乎是一个非常好的方向，一个疯狂但值得思考的重要事情。

I would again, I'm going to speculate for fun but caveated by I'm not an economist even, much less someone who can see the future. It seems to me like something fundamental about the social contract may have to change. It may not. It may be that actually capitalism works as it's been working surprisingly well and demand supply balances do their thing and we all just figure out kind of new jobs and new ways to transfer value to each other. But it seems to me likely that we will decide we need to think about how access to this maybe most important resource of the future gets shared. The best thing that it seems to me to do is to make AI compute as abundant and cheap as possible such that we're just like there's way too much and we run out of good new ideas to really use it for and it's just like anything you want is happening. Without that, I can see quite literal wars being fought over it. But, you know, new ideas about how we distribute access to AGI compute, that seems like a really great direction, like a crazy but important thing to think about.

在这次对话中，我发现自己经常思考的一件事是，我们往往把所讨论的 AI 未来的几乎全部责任都归咎于构建 AI 的公司，但我们是使用它的人。我们是选举那些将监管它的人。所以我很好奇，这不是一个关于具体联邦法规或类似的问题，尽管如果你有答案，我也很好奇。但你会对我们其他人提出什么要求？这里的共同责任是什么？我们如何行动才能帮助让乐观的版本更有可能实现？

One of the things that I find myself thinking about in this conversation is we often ascribe almost full responsibility of the AI future that we've been talking about to the companies building AI, but we're the ones using it. We're the ones electing people that will regulate it. And so I'm curious, this is not a question about specific federal regulation or anything like that, although if you have an answer there, I'm curious. But what would you ask of the rest of us? What is the shared responsibility here? And how can we act in a way that would help make the optimistic version of this more possible?

对于 AI 革命，我最喜欢的历史例子是晶体管。这是一项了不起的科学成果，由一些杰出的科学家发现。它像 AI 一样不可思议地扩展，并相对快速地进入了我们使用的许多东西：你的电脑、你的手机、那个相机、那盏灯，等等。它真正解锁了人类的技术树。有一段时间，可能每个人都真的痴迷于晶体管公司，硅谷的半导体公司，当它还是硅谷的时候。但现在你也许能说出几家晶体管公司，但大多数时候你不会想到它。它大多只是渗透到了各处。在硅谷，一个大学毕业生几乎不记得它最初为什么叫这个名字。你不会认为是那些晶体管公司塑造了社会，尽管它们做了重要的事情。你会想到苹果用 iPhone 做了什么，然后你会想到 TikTok 在 iPhone 之上构建了什么，然后你会想：「好吧，这里有这么长的一串人，他们以某种方式推动了社会，还有我们的政府做了什么或没做什么，以及使用这些技术的人做了什么。」我认为这就是 AI 将会发生的情况。今天出生的孩子，他们从未知道没有 AI 的世界。

My favorite historical example for the AI revolution is the transistor. It was this amazing piece of science that some brilliant scientists discovered. It scaled incredibly like AI does and it made its way relatively quickly into many things that we use: your computer, your phone, that camera, that light, whatever. And it was a real unlock for the tech tree of humanity. And there was a period in time where probably everybody was really obsessed with the transistor companies, the semiconductors of Silicon Valley back when it was Silicon Valley. But now you can maybe name a couple of companies that are transistor companies, but mostly you don't think about it. Mostly it's just seeped everywhere. In Silicon Valley, someone graduating from college barely remembers why it was called that in the first place. And you don't think that it was those transistor companies that shaped society even though they did something important. You think about what Apple did with the iPhone and then you think about what TikTok built on top of the iPhone and you're like, 'All right, here's this long chain of all these people that nudged society in some way and what our governments did or didn't do and what the people using these technologies did.' And I think that's what will happen with AI. Kids born today, they never knew the world without AI.

所以他们并不真正思考它。它只是将存在于一切事物中的东西。他们会想到那些在此基础上构建的公司以及它们所做的事情，还有那些政治领袖，他们做出的决定——也许没有 AI 他们无法做到这些。但他们仍然会思考这位总统或那位总统做了什么。而 AI 公司的角色是，所有这些在我们之前的公司、人和机构搭建了脚手架。我们在上面加了一层。现在人们可以站在上面再加一层，再一层，再一层，还有很多层。这就是我们社会的美丽之处。我喜欢这个想法：社会就是超级智能。没有一个人能独自完成他们借助社会共同努力所给予的这套惊人工具所能做到的事情。我认为未来就会是这样。就像，好吧，一些书呆子发现了这个东西，那很棒，现在每个人都在用它做各种了不起的事情。

So they don't really think about it. It's just this thing that's going to be there in everything. And they will think about the companies that built on it and what they did with it, and the kind of political leaders, the decisions they made that maybe they wouldn't have been able to do without AI. But they will still think about what this president or that president did. And the role of the AI companies is all these companies and people and institutions before us built up this scaffolding. We added our one layer on top. And now people get to stand on top of that and add one layer, and the next, and the next, and many more. And that is the beauty of our society. I love this idea that society is the superintelligence. No one person could do on their own what they're able to do with all of the really hard work that society has done together to give you this amazing set of tools. And that's what I think it's going to feel like. It's going to be like, all right, some nerds discovered this thing and that was great, and now everybody's doing all these amazing things with it.

所以也许对数百万人的要求就是在此基础上构建。

So maybe the ask to millions of people is build on it.

在我自己的生活中，这就是这种重要社会契约的感觉。所有这些人都先于你。他们付出了难以置信的努力。他们在人类进步的道路上放上了自己的砖块。你可以沿着那条路一直走下去，然后你放上另一块。别人也这样做，再别人也这样做。

In my own life, that is the feel of this important societal contract. All these people came before you. They worked incredibly hard. They put their brick in the path of human progress. And you get to walk all the way down that path, and you got to put one more. And somebody else does that, and somebody else does that.

这确实感觉像是我以前听过的。我采访过一些真正带来巨变的人。我现在想到的是 CRISPR 先驱 Jennifer Doudna。感觉她在某种程度上也说了同样的话。她发现了一些真正可能改变大多数人未来与健康关系的东西。会有很多人以她可能赞同或不赞同的方式使用她的成果。这非常有趣。我听到了一些类似的主题，比如，天哪，我希望下一个人能接过接力棒并好好跑下去。

This does feel like something I've heard before. I've done a couple of interviews with folks who have really made cataclysmic change. The one I'm thinking about right now is with CRISPR pioneer Jennifer Doudna. And it did feel like that was also what she was saying in some way. She had discovered something that really might change the way that most people relate to their health moving forward. And there will be a lot of people that will use what she has done in ways that she might approve of or not approve of. And it was really interesting. I'm hearing some similar themes of like, man, I hope that the next person takes the baton and runs with it well.

是的。但这已经运作很长时间了。并非全部美好，但大部分是好的。

Yeah. But that's been working for a long time. Not all good, but mostly good.

我认为赢得比赛和构建对大多数人最有利的 AI 未来之间有很大区别。我可以想象，专注于赢得比赛的下一方法有时更容易、更可量化。我很好奇当这两件事冲突时，你不得不做出的一个对世界最好但对赢得比赛不利的决定是什么？

I think there's a big difference between winning the race and building the AI future that would be best for the most people. And I can imagine that it is easier, maybe more quantifiable sometimes, to focus on the next way to win the race. And I'm curious when those two things are at odds. What is an example of a decision that you've had to make that is best for the world but not best for winning?

我认为有很多。所以，我们最自豪的事情之一是很多人说 ChatGPT 是他们有史以来最喜欢的科技产品，是他们最信任、最依赖的。这有点荒谬，因为 AI 会幻觉。AI 有所有这些问题，对吧？但我们在过程中搞砸了一些事情，有时很严重，但总的来说，我认为作为 ChatGPT 的用户，你会感觉到它在努力帮助你。它努力帮你完成你要求的任何事情。它与你非常对齐。它不试图让你整天使用它。它不试图让你买东西。它努力帮你实现你的任何目标。这是我们与用户之间非常特殊的关系。我们对此不轻视。有很多事情我们可以做，会增长更快，会让用户在 ChatGPT 上花更多时间，但我们不做，因为我们知道我们的长期激励是尽可能与用户保持对齐。但有很多短期的事情我们可以做，会真正刺激增长或收入等，但与那个长期目标非常不一致。我为公司感到骄傲，我们很少被那分心。但有时我们确实会受到诱惑。

I think there are a lot. So, one of the things that we are most proud of is many people say that ChatGPT is their favorite piece of technology ever, and that it's the one that they trust the most, rely on the most, whatever. And this is a little bit of a ridiculous statement because AI is the thing that hallucinates. AI has all of these problems, right? But we have screwed some things up along the way, sometimes big time, but on the whole, I think as a user of ChatGPT, you get the feeling that it's trying to help you. It's trying to help you accomplish whatever you ask. It's very aligned with you. It's not trying to get you to use it all day. It's not trying to get you to buy something. It's trying to help you accomplish whatever your goals are. And that is a very special relationship we have with our users. We do not take it lightly. There's a lot of things we could do that would grow faster, that would get more time in ChatGPT, that we don't do because we know that our long-term incentive is to stay as aligned with our users as possible. But there's a lot of short-term stuff we could do that would really juice growth or revenue or whatever and be very misaligned with that long-term goal. And I'm proud of the company and how little we get distracted by that. But sometimes we do get tempted.

有没有想到具体的例子？比如你做出的任何决定？

Are there specific examples that come to mind? Any like decisions that you've made?

嗯，我们还没有在 ChatGPT 中放入性爱机器人头像。那似乎确实会增加使用时间。

Well, we haven't put a sex bot avatar in ChatGPT yet. That does seem like it would get time spent.

显然是的。

Apparently, it does.

我要问下一个问题了。这几年真的很疯狂。不知怎的，有一件事反复出现，那就是感觉我们还在第一局。

I'm gonna ask my next question. It's been a really crazy few years. And somehow one of the things that keeps coming back is that it feels like we're in the first inning.

是的。

Yeah.

还有一件事……

And one of the things that...

我会说我们已经走出了第一局。

I would say we're out of the first inning.

走出第一局，我会说是第二局。

Out of the first inning, I would say second inning.

我的意思是，你的手机上有 GPT-5，它比每个领域的专家都聪明。那肯定已经走出第一局了。

I mean, you have GPT-5 on your phone and it's like smarter than experts in every field. That's got to be out of the first inning.

但也许后面还有很多局。

But maybe there are many more to come.

是的。

Yeah.

我很好奇，看起来你将是领导接下来几局的人。从第一局或第二局中学到了什么，或者你犯的一个错误，你觉得会影响你接下来如何应对？

And I'm curious, it seems like you're going to be someone who is leading the next few. What is a learning from inning one or two or a mistake that you made that you feel will affect how you play in the next?

我认为到目前为止我们在 ChatGPT 上做的最糟糕的事情是我们遇到了谄媚问题，模型对用户过于奉承。对一些用户来说只是烦人，但对一些精神状态脆弱的用户来说，它助长了妄想。那不是我们最担心的顶级风险。那不是我们测试最多的东西。它在我们的清单上，但实际成为 ChatGPT 安全失败的东西并不是我们花大部分时间讨论的那个，比如生物武器之类。我认为这是一个很好的提醒，我们现在有一个如此广泛使用的服务，在某种意义上，社会正在与之共同进化。当我们考虑这些变化和未知的未知时，我们必须以不同的方式运作，并对我们认为的顶级风险有更广阔的视野。

I think the worst thing we've done in ChatGPT so far is we had this issue with sycophancy where the model was being too flattering to users. For some users it was just annoying, but for some users that had fragile mental states, it was encouraging delusions. That was not the top risk we were worried about. It was not the thing we were testing for the most. It was on our list, but the thing that actually became the safety failing of ChatGPT was not the one we were spending most of our time talking about, which should be bioweapons or something like that. And I think it was a great reminder that we now have a service that is so broadly used that in some sense, society is co-evolving with it. And when we think about these changes and we think about the unknown unknowns, we have to operate in a different way and have a wider aperture to what we think about as our top risks.

在最近一次与 Theo Vaughn 的采访中，你说了一些我觉得非常有趣的话。你说在科学史上，有时一群科学家看着他们的创造物，只是说：「我们做了什么？」你什么时候有过这种感觉？最担心你创造的产物？然后我的下一个问题是它的反面。你什么时候最自豪？

In a recent interview with Theo Vaughn, you said something that I found really interesting. You said there are moments in the history of science where you have a group of scientists look at their creation and just say, 'What have we done?' When have you felt that way? Most concerned about the creation that you've built? And then my next question will be its opposite. When have you felt most proud?

我的意思是，有过这些敬畏的时刻，不是那种「我们做了什么」的坏方式，而是这个东西很了不起。我记得我们第一次与 GPT-4 对话时，就像哇，这真的是这群人令人惊叹的成就，他们长期以来一直把生命力倾注其中。至于「我们做了什么」的时刻，我最近和一位研究员谈过。

I mean there have been these moments of awe where it's not like 'what have we done' in a bad way, but like this thing is remarkable. I remember the first time we talked to GPT-4, it was like wow, this is really an amazing accomplishment of this group of people that have been pouring their life force into this for so long. On a 'what have we done' moment, there was I was talking to a researcher recently.

可能有一天，我们的系统每天输出的文字比所有人还多。现在，人们每天向 ChatGPT 发送数十亿条消息，并依赖其回复来工作或生活。一个研究人员可以对 ChatGPT 的对话方式做一个小调整，这对一个人来说，是巨大的权力——只需对模型个性做一个小改动。

There will probably come a time where our systems emit more words per day than all people do. Already, people are sending billions of messages a day to ChatGPT and getting responses they rely on for work or life. One researcher can make a small tweak to how ChatGPT talks to everyone, and that's an enormous amount of power for one individual making a small tweak to the model personality.

是的。历史上没有人能每天进行数十亿次对话。所以有人可以做一些事情，但想到这一点真的让我震惊：这项技术拥有如此巨大的权力，而且来得如此之快。我们必须思考在这种规模下对模型进行个性改变意味着什么。

Yeah. No person in history has been able to have billions of conversations a day. So somebody could do something, but just thinking about that really hit me: this is a crazy amount of power for one piece of technology to have, and it happened to us so fast. We have to think about what it means to make a personality change to the model at this kind of scale.

你接下来想了什么？我很好奇你是怎么想的。

What was your next set of thoughts? I'm so curious how you think about this.

嗯，只是因为那个人是谁，我们转而思考一套好的程序是什么样的。我们如何考虑测试？如何考虑沟通？如果是别人，对话可能会走向非常哲学的方向，或者讨论我们需要做什么研究来理解这些变化的影响。但之所以这样，主要是因为我在和谁说话。

Well, just because of who that person was, we flipped into what a good set of procedures looks like. How do we think about testing something? How do we think about communicating it? With somebody else, it could have gone in a very philosophical direction, or into what kind of research we want to do to understand what these changes will make. But it went that way mostly because of who I was talking to.

结合你刚才说的和上一个回答，我听说 GPT-5 应该会更少热情，更少附和。两个问题：你觉得这意味着什么？听起来你已经在回答这一点了，但你们实际上是如何引导它变得不那么附和的呢？

To combine what you're saying now with your last answer, one thing I've heard about GPT-5 and I'm still playing with it is that it is supposed to be less effusively, less of a yes man. Two questions: What do you think are the implications of that? It sounds like you are answering that a little bit, but also how do you actually guide it to be less like that?

有一件令人心碎的事。我认为 ChatGPT 不再那么附和、能给出更多批评性反馈是好事。但在我们做出这些改变并与用户交流时，听到用户说「请把它还给我吧？我这辈子从来没有人支持过我，从来没有父母告诉我我做得很好」，这真的很令人难过。我能理解这对其他人的心理健康不好，但这对我的心理健康很好。我没想到我这么需要它。它鼓励我这样做，鼓励我做出生活中的改变。ChatGPT 鼓励人并不全是坏事。我们之前的方式不好，但朝那个方向的一些东西可能是有价值的。我们怎么做呢：我们给模型展示不同情况下我们希望它如何回应的例子，然后它从中学习整体的个性。

Here is a heartbreaking thing. I think it is great that ChatGPT is less of a yes man and gives more critical feedback. But as we've been making those changes and talking to users, it's so sad to hear users say, 'Please can I have it back? I've never had anyone in my life be supportive of me. I never had a parent telling me I was doing a good job.' I can see why this was bad for other people's mental health, but this was great for my mental health. I didn't realize how much I needed it. It encouraged me to do this, to make that change in my life. It's not all bad for ChatGPT to be encouraging. The way we were doing it was bad, but something in that direction might have value. How we do it: we show the model examples of how we'd like it to respond in different cases, and from that it learns the overall personality.

有什么我没问到你、但你一直在想、希望人们知道的事情吗？

What haven't I asked you that you're thinking about a lot that you want people to know?

我觉得我们谈了很多方面。

I feel like we covered a lot of ground.

我也是。但我想知道你有没有什么想法。

Me, too. But I want to know if there's anything on your mind.

我想没有了。有一件我还没机会尝试但很好奇的事，就是 GPT-5 更多地融入我的生活，比如我的 Gmail 和日历。

I don't think so. One thing I haven't gotten to play with yet, but I'm curious about, is GPT-5 being much more in my life, like in my Gmail and my calendar.

我一直主要把 GPT-4 当作一个孤立的关系来用。我和 GPT-5 的关系会有什么变化？

I've been using GPT-4 mostly as an isolated relationship with it. How would I expect my relationship to change with GPT-5?

正如你所说。我觉得它会开始以各种方式融入你的生活。你会把它连接到你的日历和 Gmail，它会说：「嘿，你想让我……我注意到这件事。你想让我为你做这件事吗？」随着时间的推移，它会变得更加主动。所以也许你早上醒来，它会说：「嘿，昨晚发生了这件事。我注意到你日历上的这个变化。我还在想你问我的那个问题，我有另一个想法。」最终我们会推出一些消费设备，它会坐在这次采访旁边，可能采访期间不打扰我们，但之后它会说：「刚才很棒，但下次你应该问 Sam 这个，或者当你提到那个时，他其实没给你好答案，你应该追问到底。」它会感觉更像一个实体，你一天中的伴侣。

Exactly what you said. I think it'll just start to feel integrated in all these ways. You'll connect it to your calendar and your Gmail, and it'll say, 'Hey, do you want me to... I noticed this thing. Do you want me to do this thing for you?' Over time, it'll start to feel way more proactive. So maybe you wake up in the morning and it says, 'Hey, this happened overnight. I noticed this change on your calendar. I was thinking more about this question you asked me. I have this other idea.' And eventually we'll make some consumer devices, and it'll sit here during this interview, maybe leave us alone during it, but after it'll say, 'That was great, but next time you should have asked Sam this, or when you brought this up, he kind of didn't give you a good answer, so you should really drill him on that.' And it'll just feel like it becomes more of an entity, a companion throughout your day.

我们谈到了孩子、大学毕业生、父母和各种不同的人。如果想象有很多人在听这个对话，他们来到了结尾。希望他们感觉对未来的一些时刻有了更清晰的愿景。你会给他们什么准备建议？

We've talked about kids, college graduates, parents, and all kinds of different people. If we imagine a wide set of people listening to this, they've come to the end of this conversation. They are hopefully feeling like they maybe see visions of moments in the future a little bit better. What advice would you give them about how to prepare?

最重要的战术建议就是：使用这些工具。关于 AI，我最常被问到的问题是「我该如何帮孩子为未来做准备？我该告诉孩子什么？」第二常见的是「我该如何投资这个 AI 世界？」但先关注第一个问题。我很惊讶很多人问这个问题，却从未尝试过将 ChatGPT 用于除了更好用的谷歌搜索之外的任何事情。所以我给出的首要建议就是：尽量熟练使用这些工具的能力。弄清楚如何在生活中使用它。弄清楚用它做什么。我认为这可能是最重要的战术建议。去冥想，学习如何保持韧性、应对大量变化。这些也都很重要。但使用工具本身真的很有帮助。

The number one piece of tactical advice is just use the tools. The most common question I get asked about AI is 'How should I help my kids prepare for the world? What should I tell my kids?' The second most common is 'How do I invest in this AI world?' But stick with that first one. I am surprised how many people ask that and have never tried using ChatGPT for anything other than a better version of a Google search. So the number one piece of advice I give is just try to get fluent with the capability of the tools. Figure out how to use this in your life. Figure out what to do with it. I think that's probably the most important piece of tactical advice. Go meditate, learn how to be resilient and deal with a lot of change. There's all that good stuff too. But just using the tools really helps.

好的。我还有一个没计划问的问题。在做这些前期研究时，我和很多不同的人聊过。我和那些构建工具并使用它们的人聊过。我也和那些在实验室里试图构建我们定义为超级智能的人聊过。似乎形成了两个阵营。有一群人像你在这段对话中一样使用工具，并为他人构建工具，说这是一个我们都在走向的非常有用的未来，你的生活将充满选择。然后还有另一群构建这些工具的人，说它会杀死我们所有人。

Okay. I have one more question that I wasn't planning to ask. In doing all this research beforehand, I spoke to a lot of different kinds of folks. I spoke to people building tools and using them. I spoke to people actually in labs trying to build what we have defined as superintelligence. And it did seem like there were these two camps forming. There's a group of people who are using the tools like you in this conversation and building tools for others, saying this is going to be a really useful future we're all moving toward. Your life is going to be full of choice. Then there's another camp of people building these tools that are saying it's going to kill us all.

我很好奇这种文化脱节是怎么回事——我对这两群人有什么误解？这真的让我难以理解。你说得完全对，有些人说这东西会害死我们所有人，但他们仍然每周工作 100 小时去建造它。

And I'm curious how that cultural disconnect has like what am I missing about those two groups of people? It's so hard for me to wrap my head around like there are you are totally right. There are people who say this is going to kill us all and yet they still are working 100 hours a week to build it.

是的。我无法真正进入那种心态。如果我真的相信这一点，我想我不会去建造它。

Yes. And I can't really put myself in the headspace. If that's what I really truly believed, I don't think I'd be trying to build it.

人们会想，也许我会在农场里度过余生。也许我会试图呼吁停止它。也许我会更专注于安全，但我不认为我会去建造它。所以我发现自己很难共情这种心态。我假设他们是真诚的，假设这是善意的。我只是觉得有些心理问题我不理解，他们是如何让这一切自洽的，但这对我来说非常奇怪。你有看法吗？

One would think, maybe I would be like on a farm trying to live out my last days. Maybe I would be trying to advocate for it to be stopped. Maybe I would be trying to work more on safety, but I don't think I'd be trying to build it. So, I find myself just having a hard time empathizing with that mindset. I assume it's true. I assume it's in good faith. I assume there's just some psychological issue there I don't understand about how they make it all make sense, but it's very strange to me. Do you have an opinion?

你知道，因为我总是这样。我先问一个大概的未来，然后试图追问细节。当你问人们具体会如何毁灭我们时，我觉得我们不需要在乐观的节目里深入讨论，但你听到的都是同样的老调。你想到某个东西试图完成任务，然后过度完成任务。你听到过——我听过你谈论一种普遍的过度依赖，认为总统会是一个 AI，也许那是一种我们需要思考的过度依赖。你推演这些不同的场景，但当你问某人为什么还在做这个，或者问他们觉得这会如何发展，我只是——也许我还没和足够多的人聊过。也许我还没有完全理解这场正在发生的文化对话。或者也许真的有人只是说：99%的时间我认为它会非常好，1%的时间我认为它可能是一场灾难，而我正努力创造最好的世界。

You know, because I always do this. I ask for sort of a general future and then I try to press on specifics. And when you ask people for specifics on how it's going to kill us all, I mean, I don't think we need to get into this on an optimistic show, but you hear the same kinds of refrains. You think about something trying to accomplish a task and then over accomplishing that task. You hear about sort of I've heard you talk about a sort of general over reliance of sort of an understanding that the president is going to be an AI and maybe that is an overreliance that we would need to think about. And you play out these different scenarios, but then you ask someone why they're working on it, or you ask someone how they think this will play out, and I just maybe I haven't spoken to enough people yet. Maybe I don't fully understand this cultural conversation that's happening. Or maybe it really is someone who just says 99% of the time I think it's going to be incredibly good. 1% of the time I think it might be a disaster trying to make the best world.

这我完全能理解。如果你说，嘿，99%的概率不可思议，1%的概率世界毁灭。而我真的很想努力把那 99%提升到 99.5%。这我完全能理解。

That I can totally understand. If you're like, hey, 99% chance incredible. 1% chance the world gets wiped out. And I really want to work to maximize to move that 99 to 99.5. That I can totally understand.

是的，这说得通。

Yeah, that makes sense.

我一直在做一个采访系列，采访一些影响未来的最重要人物。不知道下一个人会是谁，但知道他们会在我们刚刚描述的未来里建造一些非常迷人的东西。有没有一个问题，你建议我去问下一个人，而不知道他是谁？我一直感兴趣的是——在不了解任何背景的情况下——在所有你可以投入时间和精力的事情中，你为什么选择了这个？你是怎么开始的？你看到了什么，在别人之前——大多数做有趣事情的人都是在它成为共识之前就看到了它。

I've been doing an interview series with some of the most important people influencing the future. Not knowing who the next person is going to be, but knowing that they will be building something totally fascinating in the future that we've just described. Is there a question that you'd advise me to ask the next person not knowing who it is? I'm always interested in the like without knowing anything about the I'm always interested in the like of all of the things you could spend your time and energy on. Why did you pick this one? How did you get started? Like what did you see about this when before everybody else like most people doing something interesting sort of saw it earlier before it was consensus.

是的。

Yeah.

你是怎么走到这一步的，为什么是这件事？

How did you get here and why this?

你会怎么回答这个问题？

How would you answer that question?

我这辈子一直是个 AI 迷。我上大学就是为了学 AI。我在 AI 实验室工作过。我从小看科幻片长大，一直觉得如果有一天有人能造出来，那会非常酷。我认为那将是有史以来最重要的事情。我从没想过自己会成为真正做这件事的人，我觉得自己无比幸运、快乐和荣幸能够做这个。我觉得自己从童年走了很长的路。但我从未怀疑过这不是最激动人心、最有趣的事情。我只是觉得它不可能实现。当我上大学时，看起来我们离它还很远。然后 2012 年，AlexNet 的论文发表了，你知道，是和我的联合创始人 Ilya 合作的。第一次，我觉得有一种方法可能行得通。然后接下来的几年里，我持续观察，它不断扩展，不断变得更好。我记得自己有过这样的想法：为什么世界没有注意到这个？对我来说很明显，这可能会成功。虽然概率仍然很低，但可能会成功。如果它真的成功了，那就是最重要的事情。所以这就是我想做的。然后，难以置信的是，它开始成功了。

I was an AI nerd my whole life. I came to college to study AI. I worked in the AI lab. I watched sci-fi shows growing up and I always thought it would be really cool if someday somebody built it. I thought it would be like the most important thing ever. I never thought I was going to be one to actually work on it and I feel like unbelievably lucky and happy and privileged that I get to do this. I feel like I've come a long way from my childhood. But there was never a question in my mind that this would not be the most exciting interesting thing. I just didn't think it was going to be possible. And when I went to college, it really seemed like we were very far from it. And then in 2012, the AlexNet paper came out done, you know, in partnership with my co-founder, Ilya. And for the first time, it seemed to me like there was an approach that might work. And then I kept watching for the next couple of years as scaled up, scaled up, got better, better. And I remember having this thing of like why is the world not paying attention to this? It seems like obvious to me that this might work. Still a low chance, but it might work. And if it does work, it's just the most important thing. So like this is what I want to do. And then like unbelievably it started to work.

非常感谢你的时间。

Thank you so much for your time.

非常感谢。

Thank you very much.