Sam Altman 谈 ChatGPT 个性与通往超级智能之路

我们做过的对世界影响最大的事情就是设定 ChatGPT 的性格。新模型有什么不同？更聪明、更快、上下文更多。你脑子里最常见的想法是什么？我们的目标是什么？你的愿景是什么？几乎难以想象的繁荣。这是 Sam Altman。3 年前，他让我们踏上了发明超级智能的竞赛。但实际上，他在 20 年前就开始研究这个了，当时所有人都认为创造 AI 是不可能的。而且，在重重困难下，他和一群草根工程师做到了，改变了世界。现在，每周有超过 9 亿人使用 ChatGPT。在今天的节目中，我会问 Sam 一些他从未被问过的问题。好吧，我可能会搞砸。试试看。并让你罕见地一窥他对未来的愿景。所以你得有机器人。这样你才能领先一步，打造下一个大事件。太有趣了。谢谢你来上节目。谢谢你邀请我。是的，我太兴奋了。我看了你过去 20 年所有的采访。我记得这个。很酷，因为我觉得你在很多事情上都非常一致。其中之一就是你专注于建设者。很多人不知道。他们现在只看到你是 OpenAI 的 CEO，但不知道你 20 年前就对 AI 着迷了。你能告诉我你大学时最初接触 AI 的经历吗？是什么让你爱上了人工智能？

The thing we do that's had the most impact on the world is how we set the ChatGPT personality. What's different with the new model? Smarter, faster, more context. What's like the most common thought in your head? What are we aiming towards? Like what's your vision here? Almost unimaginable prosperity. This is Sam Altman. 3 years ago, he catapulted us into a race to invent superintelligence. But he actually started working on this 20 years ago when everyone thought creating AI was impossible. And against all odds, he and a team of scrappy engineers did it and changed the world. Now, over 900 million people use ChatGPT every week. In today's episode, I'm going to ask Sam questions he's never been asked before. Okay, I'm probably going to be really bad at this. Let's try it. And give us a rare look into his vision for the future. So you got to have robots. So you can get ahead of it and build the next big thing. That was so fun. Thanks for coming on the show. Thank you for having me. Yeah, I'm so excited. I've watched like all of your interviews for over the last 20 years. I remember this. It's been cool cuz I think there's been a lot of things that you've stayed very consistent on. Um one of them is your like focus on builders. And I think a lot of people don't know. They see you as like the CEO of OpenAI now, but they don't know that you were obsessed with AI like 20 years ago. Can you tell me about your initial venture to it in college and what made you love artificial intelligence?

第一，从技术角度看，它是有史以来最酷的东西。这个想法非常有趣：我们可以让计算机思考，为我们做事并提供帮助。我一直很喜欢我对人类历史上技术进步故事的理解：我们不断在工具之上发明工具，搭建起这个脚手架，让我们能做越来越多的事情。所以我认为这是一个美妙的想法和非常酷的技术。但这就引出了第二点，从技术和科学发现的角度来看，它可以让世界变得更好。如果我们能把这种工具交到人们手中，如果人们能用 AI 来建设、探索、创造，创建新公司、新艺术形式、新体验，以及他们能做的任何其他事情，我相信世界会因此变得更好。我也相信人们会因此获得成就感。在此之前，我的职业生涯在创业公司，我观察人们创办公司，我认为这对世界很好，对做这件事的人也很好。我现在认为我们将进入一个世界，你可以有一人创业公司或三人创业公司，等等。这将释放的人类潜力以及为我们所有人带来的好处，在没有这项技术的情况下是完全不可能的。我认为这将非常值得一看。

One, I think it is just from like a techno perspective the coolest thing ever. It's just such an interesting idea that we could make computers think and do stuff for us and help. And I have like loved this my understanding of the story of technological progress in human history where we keep inventing tools on top of tools on top of tools and build up this scaffolding that lets us do more and more. And that gets to the So I think it's just like a beautiful idea and very cool technology. But that gets to the second thing, which is in terms of technology and scientific discovery that can make the world better. If we can put this hand this tool in the hands of people and if people can use AI to build and explore and create and new companies and new kinds of art and new kinds of experiences for each other and whatever else they'll do with it, that is how I believe the world gets better. That's also how I believe people get fulfilled. So before this, my career was in startups and I used to watch people make startups and I thought it was like awesome for the world, awesome for the people doing it. I now think we're going to be in a world where you can have these one-person startups or three-person startups, whatever. And the amount of human potential that is going to unlock and the amount of good stuff that's going to make for all of us that was just totally impossible without this technology. I think it'll be like quite great to see.

你认为哪位历史人物会从拥有 AI 中受益最大？

Which historical figure do you think would have benefited the most from having AI?

达芬奇是我第一个想到的。一个非常聪明的思想家，对很多事情感兴趣，有着巨大的创造力，试图尽可能多地完成事情。

Da Vinci was the first one that came to mind. Someone like very bright thinker, interested in a lot of things that's like, you know, just huge creative energy trying to get as much done as possible.

这也很有趣，因为我觉得当 Transformer 最初出现时，它就像一个预测文本模型。那一个突破在发现方面让我们走了多远，真是疯狂。你考虑过这个吗？

It's interesting too because I feel like when the transformer initially came out, it's like a predictive text model. It's crazy how far that one breakthrough has gotten us in discovery. Do you think about that?

我一直在想。Ilya Sutskever 曾经说过一句非常简单的话，就像很多简单的话一样，它深深地印在了我的脑海里：预测非常接近智能。这个想法是，如果你能把关于世界、事物状态等的所有信息压缩成最小的表示，然后作为其中的一部分，预测接下来会发生什么，你就在某种程度上深入理解了它。当时，AI 领域的很多人对生成模型感到兴奋，但原因他们说不清楚。但我认为原因与预测非常接近智能有关，如果我们试图构建真正理解所有训练数据的系统，如果我们能让它们开始预测接下来会发生什么，那似乎是一个很好的步骤。当然，当你观察孩子开始理解世界时，我认为你可以观察到类似的现象。是的，我经常思考这个。看着 AI 变得更聪明让我充满希望，因为我想，好吧，如果我把所有这些信息都输入我的大脑，我也会得到同样的结果。

I all the time. Ilya Sutskever once said a very simple sentence that as many simple sentences said it really stuck in my mind, which is prediction is very close to intelligence. And the idea is that if you can compress all of the information about the world, the state of things, whatever, into its smallest representation, and then as part of that predict the thing that's going to happen next, you understand it in a sort of deep way. And this idea at the time, a lot of people in the AI field were quite excited about generative models for a reason they couldn't quite articulate. But I think the reason was something to do about this that prediction is very close to intelligence and if we're trying to build systems that really understand all of the data they're trained on, if we can get them to start predicting what comes next, that seems like a great step. And certainly when you watch like children begin to understand the world, I think you can observe a similar phenomenon. Yeah, I think about it a lot. Like watching AI get smarter makes me hopeful because I'm like, all right, if I just input all of that information to my brain, I would get the same outcome.

你是否认为某些人天生就更擅长物理、科学和数学？或者你相信如果给每个人相同的信息，他们会有相同的输出？

Do you think that certain people are just like more destined to be great at like physics or science and math? Or do you kind of believe that if you just gave everyone the same information, they would have the same output?

不，我不认为他们会，我很高兴他们不会。我很高兴我们拥有丰富的人类经验，人们有不同的兴趣、才能，当然还有不同的训练数据。但我认为，如果我们给每个人完全相同的训练数据，他们都有完全相同的想法、兴趣和其他一切，那将非常可悲。所以我庆幸这没有发生。

No, I don't think they wouldn't I'm happy they wouldn't. Like I'm happy we get like the rich tapestry of human experience and that people have different interests and talents and sort of you know, also different training data. But I think it'd be quite sad if we showed every person the exact same training data and they had all the exact same ideas and the exact same interests and anything else. So I'm grateful that doesn't happen.

这让我想起你在之前的一次采访中说过的一个有趣的事情：历史上从未有过这么多人都与同一个思想对话。每周有 9 亿 ChatGPT 用户。这太了不起了。这如何影响你塑造 ChatGPT 的性格？

It brings me to an interesting thing that you said in a previous interview, which is that there's never been another time in history where this many people have all talked to like one mind. Like there're 900 million ChatGPT users every week. It's extraordinary. How does that influence how you shape the personality of ChatGPT?

我们尝试过很多不同的方法。这很难做对。人们想要不同的性格。同一个人在不同日子想要不同的性格。不同时间跨度的人可能更喜欢非常不同的性格。比如，如果你只考虑今天它让你感觉如何，你可能想要一个告诉你你有多棒的模型。但如果你考虑的是如何最大化你长期的满足感和成就感，你可能想要一个更经常反驳你的模型。几乎没有人想去设置滑块，比如：我希望 ChatGPT 这样表现，我希望它这么有趣，对我这么好，反驳我这么多。我们在生活中对朋友也不会这样做。但我们确实会被不同的人吸引，或者在不同时间想要不同的人，或者希望生活中的人在不同环境、不同日子、不同背景下以不同方式支持我们，并且我们期望他们能理解这一点。

We've gone through many different approaches to this. It's so hard to get this right. People want different personalities. The same person wants different personalities on different days. People on different time horizons might prefer a very different personality. Like you might want a model if you're just thinking about how it makes you feel today, you might want a model that tells you how great you are. And if you want if you're thinking about like what's going to maximize your fulfillment and kind of accomplishment over a longer period of time, you might want a model that pushes back on you way more. Almost no one wants to go like set sliders about like, here's how I want ChatGPT to behave. I want it to be like this funny and I want it to be like this nice to me and push back on me this much. And we don't do that for friends in our lives either. But we do gravitate towards different people or different people at different times or want different um want the people in our lives to support us in different ways in different environments and different days and different contexts and we expect them to understand that.

而现在，ChatGPT 并不是这样工作的，大多数人也不期望它这样工作，但我觉得我们应该朝这个方向努力。不过我认为它确实有一点建模。比如，我感觉我的 ChatGPT 非常充满活力和乐观，而当我使用未登录的账户时，它就不是这样。

And right now, ChatGPT doesn't work that way and most people don't expect it to work that way, but that is what I think we should shoot for. I think it does kind of model a little bit though. Like I feel like my ChatGPT is like very energized and optimistic in a way that when I use the account logged out, it's not.

它确实有一点这样。它确实在一定程度上了解你。这正是我们一直追求的目标。我们正在推动这种记忆和理解能力越来越强。但以前我们没有这个功能。以前我们有那些滑块，让我们告诉它想要什么样的个性。

It definitely does a little bit of this. Like it definitely kind of gets to know you somewhat. And this is what we've been going for. We're pushing towards this kind of memory and understanding more and more. But we used to not have that. We used to have these sliders and tell us what the personalities want kind of thing.

是的。然后你推出了 40，这对你来说是一个有趣的时刻，因为对于不了解背景的人来说，它非常讨人喜欢，但在某些方面是好的。我记得在一次采访中，你谈到有些人给你发邮件说，这是我生命中唯一支持我的聊天。

Yeah. And then you had 40, which was an interesting moment for you because for people that don't know the context, it was very agreeable, but in a good way in some ways. And I remember in an interview you talked about how some people actually emailed you saying this is the only supportive chat in my life.

我经常想起那些邮件。是的，你如何处理这个问题？因为我认为很酷的一点是，你有机会通过正确的方式重塑一个人的大脑，使其更积极、更有职业道德，但这伴随着巨大的责任。这一点上的责任非常重大。你知道，我们经常谈论人工智能的风险和好处，我们衡量大的风险，比如生物风险或网络安全风险。但可能我们做的、至少历史上对世界影响最大的事情，就是我们如何设定 ChatGPT 的个性。

I still think about those emails a lot. Yeah, how do you navigate that? Because I think what's really cool is you have this opportunity to rewire someone's brain for positivity and work ethic if you do it right, but there's a lot of responsibility. The responsibility on this point is huge. You know, we talk a lot about the risks with AI and the benefits, and we measure the big risk like bio risk or cyber security risk. But probably the thing we do that at least historically has had the most impact on the world is how we set the ChatGPT personality.

是的。它应该有多鼓励？应该有多少严厉的爱？它应该多大程度上为你定制，而不是为我定制，或者根本不定制？它的行为应该有多可理解？比如，你最终需要看到那些滑块吗？历史上，不仅是我们，整个领域都没有像对待“不要制造新病原体”这类事情那样，以同样的严谨性、科学关注度和风险理解来对待这个问题。但这对世界的影响是巨大的。我认为它产生了巨大的积极影响。显然，40 也带来了一些负面影响。我还没有听到任何人以一种让我觉得“这就是答案”的方式来谈论这个问题。这就是世界应该如何看待默认个性的力量或这些模型中个性的限制。但这显然是一个大问题，而且会越来越大。你现在是怎么想的？

Yes. How encouraging should it be? How much tough love should it be? How much should it customize to you versus me versus not do that? How understandable should what it's doing be? Like how much do you need to see those sliders after all? And we have historically, not just us, the whole field, not treated this with the same amount of rigor and scientific focus and sort of risk understanding that we have on things like let's not make a new pathogen. But the impact this has had on the world is huge. I think it's had tremendous positive impact. Obviously with 40, it had some negative impact, too. I still have not heard anyone talk about this in a way where I'm like, this is the answer. This is how the world should think about the power of the default personality or the limits of personality in these models. But it is clearly a huge issue and going to get bigger. How are you thinking about it right now?

我请教了一小部分我认为在各方面都非常有智慧的人，你知道，来自伟大精神传统的人，比如杰出的临床心理学家，那些我认为真正理解人们如何互动、什么激励人、什么让人满足的人，让他们尝试为 ChatGPT 编写不同的指令手册。大致是：如何表现才能最大化人们的满足感、个人成长、成就感和生活乐趣。我想拿到这些手册，然后尝试让 ChatGPT 与它们的组合对齐，看看会发生什么。

I have asked a small number of people that I think are really wise in different ways, you know, people from great spiritual traditions, like great clinical psychologists, people who I just think really understand how people interact with each other and what motivates someone, what fulfills them, to try to write different instruction manuals for ChatGPT. About here is how to behave to maximize people's fulfillment and personal growth and sort of accomplishment and enjoyment of life. And I want to get those and I want to try aligning ChatGPT to the combination of those and see what happens.

它是否也需要根据文化而改变？

Would it have to change by culture, too?

我认为很大程度上会，但人类有一些普遍的东西似乎更关乎生物学而非文化。比如什么？有一本有趣的书，我可能会把书名说错一点。也许不会。我想它叫《人类普遍性》。是一些人类学家研究了他们能找到的每一种人类文化，寻找所有特征。如果某个特征哪怕在一个文化中不存在，他们就把它去掉，因为他们说那不是真正的普遍性，那只是某种文化现象。有一些特征在我看来并不明显会存在于每一种文化中，比如重视旅行，但它确实存在。还有很多特征我觉得有道理，会在每一种文化中都受到重视。

I think it largely will, but there are some universal things about people that seem more about biology than culture. Like what? There's this interesting book. I'm going to get the title slightly wrong. Maybe I won't. I think it's called Human Universals. But it was some anthropologists that went through every human culture they could find and looked for traits. And if something didn't exist in even one culture, they took it out because they said it's not really a universal. That's some sort of cultural thing. And there were some things that weren't obvious to me that would exist in every culture, like valuing travel, but it still did. And there were a lot of things that made sense to me that would value every in every culture.

我之前谈到我对人工智能非常兴奋，以及我认为它能给世界带来什么。但越来越多的人向我们提出一个担忧：好吧，假设你是对的。假设你通过这项技术赋予每个人巨大的能动性，人们与这项技术合作共同为世界创造巨大的繁荣。比如人们仍然会工作，因为他们想为了乐趣而工作，但没有人必须工作，每个人都会拥有美好的生活。越来越多的人说，你谈论繁荣的权利，但奋斗呢？对逆境的需求呢？克服挑战、学习、不依赖一切被照顾好的需求呢？这有多重要？还有几件类似的事情，似乎对我们如何进化也很重要。

I was talking earlier about how I'm very excited about AI and what I think it can do for the world. But increasingly one of the concerns we're hearing from people is, okay, let's say you're right. Let's say you do give with this technology. You give everybody on Earth a ton of agency and people working with this technology collectively make huge prosperity for the world. Like people will still work because they want to work for fun, but nobody will have to and everybody will have this great life. Increasingly people are saying, well, you talk about a right to prosperity, but what about the struggle? What about the need for adversity? What about the need to overcome challenges and learn and not have everything taken care of and how important that is? And there are a few things like that that seem important to how we evolve as well.

同意。我也觉得这有点虚假的等同，因为如果你看看任何重大的技术革命，工作总量从未真正减少过。只是工作发生了转移。我们曾被承诺四小时工作周之类的。我们被承诺更少的压力、更多的幸福和更多的富足。也许如果我们仍然满足于一百年或五百年前的生活质量，实际上我们不必那么努力工作。我们可以得到那种生活。但这是最有价值的吗？

Agree. I also feel like that's a little bit of a false equivalency that's being made because I don't think if you look at any big technological revolution, there's never really been an overall decrease in jobs. It's just that the jobs have shifted. We were promised four-hour workweeks or whatever. And we were promised less stress and more happiness and more abundance. And maybe if we were still content with the quality of life from a hundred or five hundred years ago, actually wouldn't have to work that hard. We could get that. But is it the most valuable?

而且越来越高。标准不断提高。更重要的是，无论新世界是什么样子，我们都想取得成就，我们想竞争，我们想对彼此有用。我们想推动新事物，发现新前沿，发明新产品和服务，创造新东西，你知道吗？

And more and more. The bar keeps going up. And more than that, we want to accomplish and we want to compete and we want to be useful to each other whatever the new world looks like. And we want to push on new things and discover new frontiers and invent new products and services and make new stuff, you know?

完全同意。我曾经看到过一件事，几十年前一位音乐制作人说音乐已经变得如此之好，他真的不明白为什么还需要创作更多的音乐。太疯狂了。

Totally. I saw something once where some music producer said decades ago that music had gotten so good, he really didn't see why there was ever going to be a need to create any more music. Crazy.

事情不是那样的。是的，当然。

Just don't work that way. Yeah, of course.

所以，幸运或不幸，取决于你的看法，人们仍然会努力工作。人们仍然会有压力。人们仍然会不快乐。人们仍然会努力创造，试图以对他们有意义的方式克服逆境，并从中找到满足感和成长。也许它看起来完全不像我们今天所拥有的挣扎或工作，但我敢打赌，它的精神会非常相似。

So, fortunately or unfortunately, depends on your take, people are still going to be working hard. People are still going to be stressed. People are still going to be unhappy. People are still going to be striving to create and trying to overcome adversity in whatever way is meaningful to them and through that find fulfillment and growth. And maybe it looks nothing like the kind of struggles that we have today or the kind of work we have today, but I bet the spirit of it will be very similar.

是的，我认为这是一个有趣的观点，因为如果你看看人工智能的总体民调，它在美国并不积极。

Yeah, I think that's an interesting point because if you look at how AI is polling in general, it is like not polling positively in America.

但我对此非常兴奋，和 ChatGPT 聊天时感觉自己像个在糖果店里的孩子，因为它打开了所有探索的新大门。我想很多创始人朋友也有同感。但当我看到新闻里经常说‘50%的工作将被淘汰’时，你觉得这种说法为什么流行起来？实际上会发生什么？

And yet I'm so excited about it and I feel like a kid in a candy store when I chat with ChatGPT because it opens up all these new doors to explore. And I think a lot of my founder friends feel similarly. But then I look at how it's often covered in the news and it will be like 50% of jobs are going to be wiped out. Why do you think that narrative has taken off and what do you think is actually going to happen?

有很多想法。我认为人们总是喜欢末日场景。新闻就报道这类事情。人们似乎喜欢阅读和谈论未来会有多糟糕。坏消息比好消息传播得更快。对于任何新技术和这种程度的变化，谨慎是有理由的。说到人类深层的进化特征，我们似乎进化出思考坏事、谈论坏事的倾向，这大概有助于我们防御它。可能有一种重要的社会共同谨慎。我知道一些 AI CEO 说‘50%的工作将消失’。更不用说某人说‘我的公司将消除 50%的工作，成为人类历史上最有价值的公司，这将是多么美妙，但你们中的 50%将失去工作’是多么缺乏同理心。我不认为这是正确的思考方式。工作会消失。每一次技术革命工作都会消失。工作会改变。就在昨天，有人对我说了一句让我印象深刻的话：‘我可以使用新模型 GPT-5.5 和 Codex 在一小时内完成两年前需要几周的工作。’我想我在那个世界里会轻松很多。

A lot of thoughts. I think people do kind of always like doomsday scenarios. The news covers that kind of thing. People seem to love reading about and talking about how horrible the future is going to be. Bad stuff travels better than good stuff. With any new technology and this degree of change, there is reason for caution. Speaking of deep human evolutionary traits, we seem to evolve to think about the bad, talk about the bad, and that probably helps us defend against it. There's probably an important societal shared caution there. I know some AI CEOs are saying things like 50% of jobs are going to go away. To say nothing of how tone deaf it is for someone to say, 'My company's going to eliminate 50% of jobs and be the most valuable company in human history, and how wonderful that's going to be, but 50% of you are going to lose your jobs.' I don't think that's the right way to think about it. Jobs will go away. Jobs have gone away with every technological revolution. Jobs will change. Someone said to me just yesterday that really stuck with me: 'I can use the new model GPT-5.5 in Codex to accomplish in an hour what would have taken me weeks two years ago.' And I thought I would have been much less busy in that world.

是啊，现在你比以往任何时候都忙。

Yeah, and now you're doing more than ever.

我一生中从未这么忙过。我半夜醒来做更多工作。简直想让它停下来。

Never been busier in my life. I'm waking up in the middle of the night to do more work. It's like make it stop, please.

太多了。有了新工具，我认为我们会以新的方式创造。我毫不怀疑经济会发生巨大变化，工作也会发生巨大变化。我认为谨慎是必要的，关于新社会契约、新经济体系的严格辩论也是必要的。但我不认为我们会无所事事地过着没有意义、没有工作的生活。它只是会不同。我还认为科学突破真的即将到来，非常令人兴奋。

It's too much. With new tools, I think we will create in new kinds of ways. I have no doubt that the economy is going to change a lot and jobs are going to change a lot. And I think caution is warranted and rigorous debate about new social contracts, new economic systems are warranted as well. But I don't think it's like we're all going to sit around in a life without meaning and without work. It's just going to be different. I also think that scientific breakthroughs are really coming and super exciting.

我想和你深入探讨这个问题。我有很多问题。但首先想到的一个是回到它就像一个预测模型的想法。我认为这里有两种情况。一种是如果你给一个人足够的时间和所有这些信息，他们会做出同样的突破吗？第二种是它有点像第 37 手，在围棋比赛中 AI 走出了人类从未想过的一步。我们走的是哪条路？

I want to dive into that with you. I have so many questions here. But one of the first ones that came to mind is going back to the idea that it's like a predictive model. I think there are two scenarios playing out here. One is if you gave a human enough time and all this information, would they develop the same breakthrough? And two, is it kind of like move 37, which was in the game of Go when AI came up with a move that humans never would have done. Which path are we on?

嗯，它们可能没那么不同。我笑了，因为我记得当我们有第一批 GPT 模型时，有很多听起来很聪明的科学家或 AI 专家说：‘下一个词预测永远不会发展出新知识。它做不到。它是基于它所看到的数据建模的。它无法发现任何新东西。’他们听起来很聪明。他们有各种花哨的解释说明为什么会这样。然后到了 5.4，稍微在 5.3，模型开始以小的方式为人类集体知识贡献新知识。比如什么？证明未证明的数学定理，一些小的新物理学，诸如此类。我预计这会继续。从某种意义上说，第 37 手已经是这样的例子。所以这个想法——我们可以训练一个模型仅仅基于它已经看到的东西预测下一个词，然后用这种能力去发现根本不存在的新事物——表面上并不那么明显。事实上，你会说人们被证明是错误的，即它不应该这样做。但这些模型通过下一个词预测的过程真正学习的是推理。理解如何理解它们看到的所有数据并完成接下来会发生什么，即使那是它们以前没见过的东西。这种推理过程可以应用于你从未见过的事物。这真的很了不起。人类也这样做，对吧？人类可以学习所有已知的物理学，然后继续运行他们的预测模型，通过应用他们不仅从事实中，而且从物理学训练中发展出的底层思维过程所获得的推理能力，去发现新的物理学。我认为这些模型也在做同样的事情。那么，人类如果有更多时间和更多脑力能做到吗？可能可以。我实际上会说可以。

Well, they might not be that different. I was smiling because I was remembering when we had the first GPT models, there were all these really smart-sounding scientists or AI experts that would say, 'Next-token prediction will never develop new knowledge. It can't. It's modeled off of the data it's been shown. It can't figure out anything new.' And they sounded so smart. They had all these fancy explanations for why this was going to be the case. And then with really with 5.4, a little bit with 5.3, was the first time where models started contributing in small ways new knowledge to humanity's collective knowledge. Like what? Proving unproven mathematical theorems, some smallish new pieces of physics, things like that. I expect this to keep going. In some sense, move 37 was already an example of this. And so this idea that we can train a model to just predict the next token based on things it has already seen and then use that ability to go discover fundamentally new things that didn't exist anywhere is not so obvious on its face. In fact, you would say what people turned out to be wrong about, which is it shouldn't do that. But really what these models are learning to do through this process of next-token prediction is to reason. To understand how to make sense of all of the data they have seen and complete what comes next even if it's something they haven't seen before. This reasoning process can be applied to things that you have not seen before. And this is really quite remarkable. People do this too, right? People can go study all of the known physics and then keep running their predictive model or whatever and by applying that reasoning ability they have learned through not just the facts, but the underlying thinking process that they developed during their physics training go discover new physics. And I think that's what these models are doing, too. Now, could people do it with more time and more brainpower? Probably yes. I would say yes, actually.

好的。

Okay.

但制造一个更快、更大的模型比想办法给人类更大的大脑要容易得多。所以我个人很兴奋我们有这些新的外部工具，可以请它们去深入思考一个我们可能更难自己思考的问题。当你看到这些模型在几秒钟内阅读数十万页并综合所有内容时，就像，也许如果我们有一个更大的大脑，我们也能做到，但以我们目前的大脑大小，我们做不到。

But it is much easier to go make a faster, bigger model than it is to figure out how to give people much bigger brains. So I for one am thrilled we have these new kinds of external tools that we can ask to go think really hard about a problem that maybe would be harder for us to think about ourselves. When you see these models read hundreds of thousands of pages in a few seconds and synthesize across all of them, it's like, maybe if we had a bigger brain, we could do that, but we cannot with our current size brains.

不过有趣的是，它有点像生物大脑，这让我好奇。你认为自然界中还有其他东西我们可以模仿来实现技术突破吗？比如飞机是基于鸟类的。还有其他例子吗？

It is interesting though that it kind of is similar to biological brains and it made me curious. Are there other things in nature that you think we could copy for tech breakthroughs? Like the airplane is based on the bird. Are there other examples of that?

一位伟大的科学家曾经说过，没有替代方案。那是我们唯一想出的办法。显然这不完全正确，但神经网络，人工神经网络显然受到大脑中神经元连接方式的启发。我当然不认为字面上说自然是我们发现新科学的唯一灵感来源，但天哪，它是一个很好的起点。

A great scientist once said that there is no alternative. That that's the only thing that we figured out how to do. Now, obviously that's not quite true, but neural networks, artificial neural networks were clearly inspired by the way neurons in a brain connect. I certainly don't think it's literally true that nature is our only source of inspiration for discovering new science, but man, is it a good place to start looking.

你现在有没有想实现的东西？这周对你新的科学模型来说很重要。你怎么看待科学突破，你在关注什么，接下来是什么？我看到有个澳大利亚人治好了他狗的癌症。

Is there anything that you're thinking about now that you want to implement? This is a big week for your new science model. How are you thinking about science breakthroughs, what you're focusing on there, what comes next? I saw that an Australian guy helped his dog's cancer be cured.

那是个具体的事。我昨晚刚和 YC 拜访的一位公司创始人聊过，他也在想类似的事，但把它规模化。针对人类癌症的定制 mRNA 疫苗。这看起来非常令人兴奋。

That is a specific thing. I was just talking to a founder of a company I got to visit YC last night, who was thinking about a similar thing, but making that scaled. Custom mRNA vaccines for cancer in people. That seems tremendously exciting.

为什么我们还没做到？

Why haven't we done it yet?

据我所知，原因很多，但一个大问题是 FDA 没有很好地准备好考虑如何做到这一点，尽管在快速改进。

As I understand it, there are many reasons, but one of the big ones is the FDA is not well set up to think about how we're going to do that, although getting better fast.

这很有趣，因为当我想到个性化医疗时，它必须是下一个前沿，因为我们都有不同的 DNA 和风险结果。如果你得了癌症，公司或实验室可以为你制作只针对你癌症的个性化疫苗，而且很可能有效，这听起来像是我们都应该要求的显而易见的事情。

It's interesting because when I think about personalized medicine, it has to be the next frontier because all of us have such different DNA and risk outcomes. The idea that if you get cancer, a company or lab can make you a personalized vaccine just for your cancer, and it's very likely to be effective, sounds like an obvious thing we should all demand.

你现在用 ChatGPT 来管理健康吗？

Do you use ChatGPT now for your health?

是的。我可能用得过度了。以前他们叫网络疑病症，我不知道 ChatGPT 版本叫什么，但任何轻微症状我都会陷入 ChatGPT 的兔子洞。像其他人一样，我把血检结果放进去。我很高兴我这么做，但有时它会说‘哦，这个有点异常’，我就会想‘我该做点什么吗？’

I do. I am probably an overuser of it. I think they used to call them cyberchondriacs. I don't know what they call the ChatGPT version of this, but any mild symptom I get, I will go down a ChatGPT rabbit hole. Like everybody else, I put my blood test in there. I'm happy I do, but sometimes it'll really be like, 'Oh, this is slightly off,' and I'm like, 'Should I do something about this?'

它最近帮到你了吗？怎么帮的？

It helped you recently? How did it help you?

我跑步导致应力性骨折，医生出城了。我做了核磁共振，把结果放进去，它帮我读了核磁共振。显然你得核实，但它是准确的。这让我震惊。我们刚推出 ChatGPT 时，也有点这样，但效果不好。人们说‘人们永远不会用 ChatGPT 寻求医疗建议，它不够好，永远不够好。即使它好，大家也更愿意和医生谈。’人们当然还是想和医生谈，但用 ChatGPT 问医疗问题并得到非常有帮助信息的人数相当惊人。

I had a stress fracture from running and my doctor went out of town. I had an MRI and I put it in and it read the MRI for me. Obviously, you got to check, but it was accurate. It blew my mind. When we first launched ChatGPT and there was a little bit of this, it was not very good. People said, 'People will never use ChatGPT for medical advice. It's just not nearly good enough. It's never going to be good enough. And even if it were good, everybody would rather talk to a doctor.' People still definitely want to talk to their doctor, but the amount of ChatGPT usage of people asking medical questions and getting really helpful information is quite extraordinary.

不断有人怀疑这项技术是否会产生影响，这对你来说很难吗？

Is it tough for you to constantly have people doubt that the technology is going to be impactful?

是的。它本不该再困扰我了，但还是让我烦得要命。

Yes. It shouldn't bother me anymore. It still annoys the hell out of me.

我也会烦。我觉得如果你看任何重大技术突破，比如在飞机发明之前，报纸说‘我们永远不会飞，还要一百年。’然后下周我们就上天了。那个例子在 OpenAI 早期我们经常谈论。莱特兄弟的《纽约时报》文章。我们一直谈论它，说 AI 也会这样。结果我们是对的。但老实说，早期这让我很烦，但当时还不明确，所以我认为批评者说‘这可能不会有很大影响’至少是智力上诚实的。现在看到有人说这真的没有价值，不会对世界产生影响，这不该烦我。但显然很荒谬，却如此烦人。如此智力上不诚实，如此烦人。而且，我觉得当你每天都在竞技场上，努力推动进步，你希望人们相信它。在我们的视频中，我经常试图向人们展示惊人的技术和未来的样子，因为我认为你需要看到它才能锁定并构建。最终，我觉得人们在努力做自己关心的事情时最满足。

It would bother me, too. I feel like if you look at any big tech breakthrough, like before we flew planes, the newspaper was like, 'We will never fly. It will be a hundred years.' And then the next week we're in the sky. That example in the early days of OpenAI we used to talk about all the time. The Wright brothers New York Times article. We used to talk about it all the time and we said AI is going to be like this. We turned out to be right. But honestly, it annoyed me a lot in the early days, but it was not super clear, so I thought it was at least intellectually honest of the critics to say maybe this isn't going to have a big impact. Now watching people say there's really no value in this, it's going to have no impact on the world, it shouldn't bother me. I mean, it's obviously ridiculous, but it's so annoying. It's so intellectually dishonest and so annoying. And also, I feel like when you're in the arena every day and you're trying so hard to push the ball forward, you want people to believe in it. With our videos, I often try to show people amazing technology and what the future can look like because I think you need to see it to then lock in and build. Ultimately, I feel like people are the most fulfilled when they're working hard on something that they care about.

如果你今天和一个 22 岁的人交谈，你会好奇了解他们对世界的感受的哪些方面？那会如何影响你构建的东西？

If you were talking to a 22-year-old today, what types of things would you be curious to know about how they're feeling about the world? And how would that inform what you build?

过去几周我一直在尝试做的一件事就是和人们坐在一起，使用最新模型和 Codex，了解它将如何影响他们的工作，他们对什么兴奋，对什么不兴奋，他们需要我们还没构建的东西。我主要和公司经营者或高级工程师做了这些，我真的应该和一些年轻人坐下来，说‘试试这个。’然后观察他们做什么，倾听他们的担忧。

A thing that I have been trying to do over the last couple of weeks is really sit with people using the latest model and using Codex to understand how it's going to impact their work, what they're excited about, what they're not excited about, what they need from us that we haven't already built. I've done this mostly with people running companies or senior engineers at companies and I really should go sit down with some young people and say, 'Try this out.' And watch what they do and listen to their concerns.

你有独特的视角，因为你给很多年轻创始人建议。几年前你在乔·罗根的播客上谈到缺乏 25 岁的创始人。从那以后有变化吗？

You have a unique perspective because you advise so many young founders. When you were on Joe Rogan's podcast a few years ago, you talked about how there was a lack of 25-year-old founders. Has that changed since then?

完全变了。

That's totally changed.

你觉得是什么改变了它？

What do you think changed it?

我认为有几件事同时发生。我不再能真正给创始人建议了，因为生活太忙，但我一直在想我需要找到某种方式重新做这件事，因为这项技术最重要的事情之一是它正在催生的创业精神。我在那方面感到脱节，我真的很不喜欢这样。我理智上理解它，但我想和那些两个创始人、一万块 GPU 的公司一起在战壕里。我最近见过几个这样的，但这提醒我，我得想办法再次接近初创公司。为什么当时没有年轻创始人而现在有：我认为有很多原因。美国教育系统经历了一个非常黑暗的时期，同时发生了新冠疫情。我们有点在打击这一整群人，告诉他们未来会很糟，资本主义很糟，公司很糟，野心很糟。这似乎已经纠正了。我们回来了。有一个蒂莫西·柴勒梅德的事情火了，他说他多么想赢一个奖，人们对此很兴奋。他们说‘再次在乎真酷。’这很好。它本不该不是那样。那种‘你不被允许’……

I think there were a few things happening at once. I don't really get to advise founders anymore because life got so busy, but I have been thinking that I need to find some way to do that again because one of the most important things about this technology is the entrepreneurship it's enabling. And I feel out of touch on that in a way I really don't like. I intellectually understand it, but I want to go be in the trenches with people building companies with two founders and 10,000 GPUs. I've met a few of these recently, but this is reminding me that I got to figure out some way to get closer to startups again. Why there weren't young founders then and why there are now: I think it was a lot of things. The US educational system went through a very dark period where COVID happened at the same time. We were kind of demotivating this whole set of people and telling them that the future is going to be bad, capitalism is bad, companies are bad, ambition is bad. That seems to have corrected. We're back. There was a Timothée Chalamet thing that went viral where he was saying how much he wanted to win an award and people were stoked about it. They were like, 'It's so cool to care again.' That's great. It should never have not been that. The sort of 'you weren't allowed to'...

要有雄心或者……是的，那真是一段非常奇怪的时期。然后我认为另一件事是，初创公司在充满活力和新事物时蓬勃发展，你知道，当技术格局发生变化时。这种情况发生在 iPhone 应用商店推出时，大概是 2008 年，也发生在 AWS 推出时，早了几年。然后很长一段时间都没有这样的变化，直到人工智能出现。所以，那就像一段荒芜期。仍然有成功的初创公司，但不像真正技术变革时那么多。七年前你在博客上说过：“我们该迎来下一次技术变革了。”然后你做到了，你就是那个实现它的人，这很酷。

Be ambitious or to like Yeah, it was really weird. Really weird time. And then I think another thing is startups thrive when there's dynamism and newness and the you know, there's a change in the technological landscape. And that happened when the iPhone App Store launched. Yep. In 2008, I guess, that happened when AWS launched a few years earlier and then it didn't happen for like a very long time until AI came along. So, there was just like a kind of period in the wilderness. There were still successful startups, but not as many as there can be when you know, there's a real technological shift. You said that on your blog seven years ago. You were like, "We're due for another technological shift." And you did and you're the man that made it, which is cool.

谢谢。

Thank you.

你现在如何看待专注？比如，在 AGI 方面你接下来想关注哪些领域？显然你们最近关闭了 Sora。哪些领域得到了最多的关注，为什么？

How do you think about focus now? Like what are the next areas that you want to focus on with AGI and obviously you guys like shut down Sora recently. What are the areas that get the most focus and why?

我认为现在对我们来说最重要的三件事是加速研究。我们之前稍微谈过这个，这涵盖了从 AI 研究到物理学、生物学等一切领域。加速研究是因为研究和科学理解对人类贡献巨大。第二是加速经济。我们讨论过自动化初创公司、大公司利用 AI 提高生产力。最终，你知道，比如建造太空殖民地之类的。第三是个人化的 AGI。ChatGPT 就是一个小小的预览。你知道，也许你可以输入你的医疗问题并获得一些建议。但你真正想要的——或者至少我真正想要的——是一个始终为我工作的 AGI，它了解我的全部背景、我的整个生活，花费算力让我的生活变得更好。这三个是最重要的关注点。它们在使能技术和平台方面惊人地相似，但这些是我认为社会将真正感受到价值的领域。

I think the three most important things for us now are accelerating research. We talked a little bit about this and this goes from like AI research to physics to biology, everything. But accelerate research because research and scientific understanding does so much for humanity. Second is accelerate the economy. Talked about this automated startups, big companies using AI to be more productive. And all of, you know, eventually like building the space colonies or whatever. And then third is the sort of like personal AGI. ChatGPT was like a little preview of this. You know, maybe you can type in your medical questions and get some advice. But you would really like or at least I would really like an AGI working for me with my whole context, my whole life all the time. Spending compute to like make my life better. Those are the three most important focuses. They're shockingly similar in terms of the enabling technology and platform, but those are the areas where I think society will really feel the value.

具体到科学突破，你们有基金会专注于阿尔茨海默病研究。你认为未来一年我们还能在哪些领域期待突破？

For scientific breakthroughs specifically, you guys have the foundation where you're focusing on Alzheimer's research. What other areas do you think we can expect breakthroughs like in the next year?

我预计数学领域的进展会非常惊人，或者类似的事情。

I would expect the progress in math to be astonishing or something like that.

比如在哪些方面？

Like in what way?

我们会发现极其重要的新数学，解决看似遥不可及的数学问题。就像历史上许多次一样，我预计如果我们发现新数学，它将指引新物理学、新密码学，谁知道还有什么实际应用。但我希望我们对自己要求更高，致力于一些更复杂、更困难的科学理解，这些理解对现实世界有更大影响。所以，我不认为我们能在明年治愈阿尔茨海默病，甚至不能真正治疗，但我希望我们能看到一些有希望的新方向，可以继续推进。

We'll just discover hugely important new math and solve math problems that seemed out of reach. And like many other times in history, I expect if we discover new math, it'll point the way to new physics and other new cryptography, who knows what, real-world applications. But I hope we hold ourselves to a higher bar and work on some of like the messier, more difficult scientific understanding that has more of a real-world impact. So, I don't think we'll get Alzheimer's cured in the next year, or even really treated in the next year, but I hope we can start to see like some new promising vectors that we can go push on.

是的，我记得马克·扎克伯格在一次采访中谈到，当他与 AI 领域的人交谈时，他们说：“我们将解决所有疾病。”但当他与医生交谈时，他们说：“这不会发生。”所以，这两个领域之间显然存在脱节。你怎么看？

Yeah, I remember Mark Zuckerberg in an interview talked about how when he talks to people that work in AI, they're like, "We are going to solve every disease." And then when he talks to doctors, they're like, "That is not going to happen." So, there's clearly a disconnect in the two fields. How do you think about it?

这需要的时间会比 AI 人士认为的要长，但比医生认为的要短。我喜欢这个说法。是的，我完全同意，而且我认为即使回顾几年前，现在 AI 能实现的突破类型，我们显然处于指数增长中。

It will take longer than the AI people think and shorter than the doctors think. Love that. Yeah, I totally agree and I think if you even look back to a few years ago, the types of breakthroughs that are now possible with AI, it's just like we're definitely on an exponential.

似乎更长的上下文窗口在这个指数增长中会非常重要。我们如何实现？是更多的算力吗？需要发生什么？

It seems like longer context windows is going to be super important in that exponential. How do we do that? Is it more compute? What has to happen?

我不认为它需要一个字面意义上的 10 亿或 1 万亿 token 的上下文窗口，尽管我假设我们也能做到。我认为你关心的是模型能够有效地理解你的整个生活、你的整个公司或你关心的所有事情。已经有了一些惊人的新方法，可以利用当前的上下文窗口，但真正找出重要的部分，或者在必要时使用工具去寻找不太重要的部分，从而更好地利用相同数量的上下文。所以，我认为这将继续发展，随着新模型以及我们在未来几个月内将添加到新模型中的东西。我不想说它会感觉像无限上下文，但它感觉像是，好吧，这个模型真的理解了很多东西。它脑子里装的东西比我能想到的多得多。

I don't think it needs to be like a literal 1 billion or 1 trillion token context window, although I assume we'll be able to do that, too. I think what you care about is that somehow the model can effectively understand your whole life or your whole company or all the things you care about. And there have been amazing new methods to use the current context windows, but really figure out the important bits, or to use tools to go off and find the less important bits when necessary, and make way better use of the same amount of context. So, I think that will keep going, and with the new model and the things we'll add to the new model in the coming months. I don't want to say it will feel like infinite context, but it feels like okay, this model really understands a lot. It has way more stuff in its head than I have in mind.

新模型有什么不同？你们改变了什么？

What's different with the new model? Like, what did you guys change?

更聪明、更快、更多上下文，而且我找不到合适的词。更可靠，这么说吧。我认为它更好地理解了我真正想要的东西。并且尝试几次，知道什么时候走在正确的轨道上，什么时候不是，最终给我正确的结果。所以，主观体验是，我让模型做某事时，它在更多时候做对了。

Smarter, faster, more context, and I don't have the right word for this. More reliability, let's say. Like, I think it does a better job of understanding what I actually want. And trying a few times, knowing when it's on track for that and not, and actually getting me the right thing. So, the subjective experience is way more of the time that I ask the model to do something, it does the right thing.

有趣。因为它基于训练来理解，你们更新了算法吗？

Interesting. Because it understands based on its training, like did you guys update algorithms or?

我们做了很多算法上的改变。它是一个更新、更好、更大的基础模型，采用了不同的架构，或者说架构上的改进，然后还有我们学到的关于后训练的一切，人们如何使用这些模型，以及如何将它们连接到世界、人们的系统、人们的上下文以提供帮助。

We have a lot of algorithmic change in it's a newer, better, bigger base model with a different architecture, and then or architectural improvements, and then all of the things we've learned about post-training, how people want to use these models, and how to kind of connect them to the world, people's systems, people's context to be helpful.

我在想这个。告诉我这个理解是否正确。似乎 AI 在三个方面变得更好：更好的算法、更多的数据，然后可能是更多的能量或更多的算力。是的。这些是我们能够推动的三个方面吗？

I'm thinking about it. Tell me if this is the right understanding. It kind of seems like AI gives better in three ways. It's better algorithms, more data, and then like maybe more energy or more compute. Yeah. Are these kind of the three things that we can push on?

基本上是的。更多数据是一个非常宽泛的类别。比如，我们是指字面上更多的训练数据，还是指我们将它连接到一个循环中，让它在你做某事并失败时持续学习？那很酷。但总的来说，我同意这三个类别。

Effectively, yes. There's more data is a very broad category. Like, is this you know Do we mean by that like literally just more training data, or do we mean like, you know, we're going to connect it in a loop that it can learn continuously as you're doing something and it's failing? That's cool. But yeah, broadly speaking, I agree those are the three categories.

哪一个最容易取得突破？

Which one's easiest one to have a breakthrough in?

我认为构建更多算力是最确定的。那里涉及的科学最少。它只需要大量的资金和复杂的供应链，但你可以直接去做。算法突破回报最高，但最难找到且最不确定，而更好的数据则介于两者之间。

I think building more compute is the most certain one. There's the least science there. It just takes a lot of money and a lot of complex supply chain, but you can just do it. Algorithmic breakthroughs are the highest payoff, but the hardest and most uncertain to find, and better data's in the middle.

更好的数据是否与递归学习有关，比如模型自我学习，还是有所不同？

Yeah, the better data, does it tie to recursive learning, like the model teaches itself, or is that different?

可以，这完全是一种方式。如果模型足够聪明，它可以去证明一个未证明的定理，那么在下一轮训练中，模型就能学到这个新证明。这是一个例子。

It can, yeah. I mean, that's totally one way to do it. If the model is really smart, it can go prove an unproven theorem, and now in the next training run, there's one more thing the model can learn. We have this new proof. That's an example.

有趣。我们现在是否已经到了模型能大幅自我改进的阶段？

Interesting. Are we at that point where the model is improving itself a lot right now, or no?

这个问题很难准确界定。从某种意义上说，显然是肯定的。如果我们的工程师因为 Codex 效率提高了三倍，并且能用前一个模型更快地编写下一个模型的代码，这必须算进去。

It's so hard to frame that question properly. In some sense, clearly yes, right? If our engineers are three times as productive as they used to be because of Codex, and they can write the code for the next model faster using the previous model, you kind of got to count that.

完全同意。

Totally agree.

而在精神层面上，人们的意思是，我们是否只是按下一个按钮，说‘去造下一个模型，想出新的算法思路’？绝对不是。

And then in the spiritual sense, people mean of like, are we just pushing a button and saying, go make the next model and come up with these new algorithmic ideas? Definitely not.

我也认为在供应链方面，比如如何建造所有这些数据中心，机器人技术非常令人兴奋。你说过机器人技术是你的一个重点。能让我了解你的想法吗？你对机器人技术兴奋的是什么，路线图是什么？

I also think on the supply chain, like how do we build all these data centers, robotics is so exciting. You said that robotics is a big priority for you. Can you bring me into your mind, like what excites you about robotics, and then what's the road map?

我们生活在物理世界中，即使身处虚拟世界，也需要物理世界的巨大复杂性来支撑。我们需要制造芯片、建造数据中心、运营发电厂等等。

We live in the physical world, and even when we're in the virtual world, we need this massive complexity in the physical world to enable that. We need to make the chips and build the data centers, and run the power plants and whatever else.

所以，一个非常可悲的未来是，计算机能做这些不可思议的事情，但由于我们没有解决机器人问题，我们不得不作为 AGI 的执行器在物理世界中跑来跑去，AGI 会说‘请去移动这张桌子，做这个，做那个’。

So, a very sad future would be where computers can do these incredible things, but because we didn't figure out robots, we have to go run around the physical world as the actuators of the AGI that'll say, please go move this table and do this and do that.

那种情景。真的很糟糕。所以，你必须要有机器人。你认为哪种类型的机器人最好？

That scenario. Really bad. Really bad. So, you got to have robots. What type of robots do you think will be the best?

我并不特别关注某种特定形态。我想要的是自动化制造，以及能够说‘我们需要更多这种东西’的能力，并且拥有与 ChatGPT 相同通用性的机器人工厂，可以自我重新配置并制造更多这种东西。

I am not that focused on any particular morphology. I want automated manufacturing, and the ability to say, we need more of whatever this thing is, and with the same generality of ChatGPT, a factory of robots that can reconfigure itself and make more of that thing.

你认为你会亲自制造它们，还是合作？

Do you think you would ever physically manufacture them, or would you partner?

不知道。

Don't know.

除此之外，AI 硬件是你的优先事项吗？我知道 Johnny Ive 参与了。

Is AI hardware outside of that a priority to you? Like, I know Johnny Ive is involved.

是的。哦，你指的是消费级 AI 硬件。完全同意。我们之前谈到，你希望 AI 能了解你生活中的所有上下文。当前的硬件很棒。我认为 iPhone 是目前最伟大的消费硬件，它取得的成就令人难以置信。同意。但它并非为需要吸收你生活所有上下文的硬件而设计。你可以使用手机，也可以停止使用，把它放进口袋，但它有点像开或关。

Yeah. Oh, you mean consumer AI hardware. Totally. We were talking earlier about how you want an AI to have all the context in your life. And current hardware, which is amazing. I think the iPhone is currently the greatest piece of consumer hardware ever made by a lot, like incredible what that has done. Agree. But it was not meant for a world where you needed a piece of hardware that could absorb all of the context of your life. You can use the phone. You can stop using the phone, you can put it in your pocket, but it's kind of like on or off.

当我们不使用它时，比如这次非常有趣的对话，我希望以后我的个人 AGI 能参考它，但我的手机在口袋里，它不会理解。

And when we are not using it, like this has been a very interesting conversation. I would love this to be referenceable by my personal AGI later, but my phone is in my pocket, and it's not going to understand.

是的。你希望有一个设备，如果我愿意，它可以参与、理解并了解这次对话。

Yeah. And you would like a device that if I wanted to, can participate and understand and know about this conversation.

完全同意。是的，我认为获得外部见解也很有趣，比如我最近下载了播客《Acquired》每一集的文字记录。我喜欢那个播客。我试图逆向工程他们节目成功的原因。所以我下载了大约 400 份文字记录，放入 ChatGPT，让它分析他们的故事结构，结果令人惊叹。

Totally. Yeah, I think also it would be interesting to get outside insights, like I recently downloaded the transcripts of every episode of the podcast Acquired. I love that podcast. And I was trying to reverse engineer what makes their show successful. So, I downloaded like 400 transcripts from the show, put into ChatGPT, and had it analyze their story structure, and it was amazing.

我想你也能从自己的对话以及作为领导者处理事情的方式中获得类似的见解。

And I imagine that you could have similar insights of your own conversations and how you approach things as a leader.

是的。但我也知道，当人们看到始终开启的录音设备时，会感到不适。完全同意。你怎么看？

Yep. But I also know that when people see an always-on recorder, there's an ick with it. Totally. What do you think?

我最初想和 Johnny 谈的原因之一是，我在思考 AI 世界的硬件会是什么样子，以及我对那些过于侵入生活的技术（比如智能音箱）感到的不适。

One of the reasons I initially wanted to talk to Johnny is I was thinking about what hardware for the AI world is going to be, and the ick that I feel with technology that is just too present in my life, like even a smart speaker.

完全同意。我认为 Johnny 对如何设计一个能平衡所有这些因素的东西有深刻的见解，我相信他会做得很好。

Totally. I thought Johnny would have great insight about how to design something that held all of these things in tension, and I think he'll do great.

你认为关于你的方法，最大的误解是什么？

What do you think will be the biggest thing that is misunderstood about your approach?

我还不知道。好吧。肯定会有很多事情，我们可以回头再谈。

I don't know yet. Okay. Sure there will be many things we can come back and talk about there.

有趣。我也对 AI 在后台的使用很感兴趣，比如智能体。这意味着什么，你怎么看？

Interesting. I'm also very interested in the use of AI kind of in the background, like agents. What does that mean, and how do you think about it?

当团队第一次制作 Codex 应用时，我把它装在了我的电脑上。它有一个当时我们称之为 YOLO 模式的功能。我想我们后来给它起了一个更优雅的名字。但基本上你可以说，你可以在我的电脑后台运行并做事情，而且不必每次都问我。我当时想，我绝对不会打开那个功能。但几个小时后，我因为每一步都要获得许可而感到非常恼火，就打开了它。然后这个智能体就在我的电脑上到处运行，在后台做事。很快，我就不想合上电脑了，因为我不想停止工作。这个转变非常平滑，毫无波澜。我觉得自己这么做有点疯狂，有点不负责任，但我确实这么做了。后来我们找到了让它更负责任的方法。但我从认为自己不会舒服，变成了喜欢智能体在我的电脑上跑来跑去、做有用事情的想法。

When the team first made the Codex app, I put it on my computer. And it had this thing that at the time we called YOLO mode. I think we found a more polished name for it eventually. But you could basically say, you can just run in the background of my computer and do stuff. And you don't have to ask me every time. And I was like, I'm absolutely never going to turn that thing on. And I lasted a few hours, and I got kind of so annoyed by having to get permission every step, I just put it on. And there was this agent running all over my computer doing stuff in the background, and then pretty soon after that, I didn't want to close my computer because I didn't want to stop working. And the transition there was so smooth, so uneventful. I thought it was kind of crazy I was doing it, sort of irresponsible, but here I was. We've since figured out how to make it a more responsible thing to do. But I went from thinking I wasn't going to be comfortable to loving the idea that an agent was just running around my computer doing useful stuff.

它为你做了什么？

What was it doing for you?

处理我的消息和邮件。最后，我尝试了类似‘看看我的电脑，找出你能做什么来帮助我’的功能。

Deal with my messages, deal with my email. Eventually, I tried something which is like, look around my computer and figure out what you can do to be useful to me.

哇。它做了什么吗？

Whoa. Did it do anything?

第一次尝试时没有，但它促使我做了这个小项目，制作了一个自动待办事项列表。就是这样。这很酷。自动完成待办事项列表是非常酷的事情。

That time I first tried it, no, but it led me to working on this little project of making this automatic to-do list. That's it. Which is sick. Like, auto-completing to-do lists is a very cool thing.

完全同意。它是内置在 ChatGPT 主页里，还是不同的东西？

Totally agree. Is it built into like ChatGPT homepage, or is it a different thing?

只是我写的一个小程序。

It's just a little program I made.

很酷。是的，因为我总是下载不同的待办事项应用，但从来坚持不下来，最后只能给自己发短信。这很酷。

That's cool. Yeah, cuz I feel like I always download different to-do list apps, and then they never stick, and I end up just texting myself. That's cool.

是的。

Yes.

你认为会有协同工作的智能体吗？比如，你会有一个像私人教练一样的智能体吗？还是说……我对此非常好奇。

Do you think that there will be agents that kind of work together? Like, will you have one agent that's like your personal trainer, will it just be like kind of the I wonder about this so much.

这就像是我最想得到答案的产品设计问题之一。人们会如何想要工作？我猜想人们会有一种不同智能体的概念模型，然后可能还有他们的个人助理、参谋长之类的角色，花很多时间协调它们。

This is like one of the product design questions I would most like an answer to. How people are going to want to work. I suspect that people will have a conceptual model of different agents, and then maybe their personal assistant, chief of staff, whatever you want to call it, that coordinates among them a lot of time.

是的。

Yeah.

好的，假设你和我穿越到未来，到了 2050 年，我知道那还很遥远。感觉连你都无法预测六个月后的事。但我很好奇你的梦想是什么，如果我们一起想象未来会是什么样子。我们的目标是什么？你的愿景是什么？

Okay, so let's say that you and I time travel into the future, and we go to 2050, which I know is a long way out. It feels like even you can't predict 6 months away. But I'm curious what your dream is, if we were to dream together what the future looks like. What are we aiming towards? What's your vision here?

天哪，那感觉太遥远了。几乎难以想象的繁荣似乎很可能出现。我所希望的，但也是我认为我们必须真正努力去实现的，是极端水平的人类能动性，人们可以做到和创造出超乎任何人想象的东西，并且我们避免了权力集中的趋势。至于世界实际的样子，我不知道，也许到时候会有太空殖民地。

Man, that feels so far away. Almost unimaginable prosperity seems likely. What I hope for, but what I think we have to really work for, is radical levels of human agency, where people can just do and create beyond anyone's imagination, and we avoided the kind of centralization of power tendencies. And then in terms of what the world actually looks like, I don't know, space colonies by then, maybe.

飞行汽车。

Flying cars.

是的，也许吧。飞行火车会很酷。

Yeah, maybe. Flying trains would be cool.

好吧，所以我希望它看起来像未来。

All right, so I hope it looks like the future.

我也是。是的，我希望它看起来有点像这样。

Me, too. Yeah, I hope it kind of looks like this.

为了结束这个视频，你听说过盲排吗？基本上，我会给你选项，但你不知道接下来是什么，你必须对它们进行分级排名。

To end this video, have you ever heard about blind ranking? Basically, I'll give you options, but you won't know what's coming next, and you have to tier list things.

好的。

Okay.

我想做技术突破。我必须把它们从 1 到 5 排名，而且我不知道接下来是什么。1 到 5，你认为哪个最重要？但你不知道接下来是什么，所以这是一个挑战。你得给自己留点余地。

I want to do tech breakthroughs. And I have to rank them 1 to 5, and I won't know what's coming. 1 through 5, what's the most important in your mind? But you won't know what's coming next, so it's a challenge. You got to give yourself some room here.

好吧，我可能在这方面会很差，但让我们试试。

Okay, I'm probably going to be really bad at this, but let's try it.

我相信你会很棒的。1 到 5，1 是最重要的。

I'm sure you're going to be amazing. 1 through 5, 1 is the most important.

是的。

Yes.

好的，我先给你一个开始。火。

Okay, I'm going to give you one to start us off. Fire.

唯一困难的是我会认为它们都是第一。

The only thing that's going to be hard is I'm going to think they're all ones.

你太棒了。第三。

You're awesome. Three.

好吧，这是个好答案。

Okay, that's a good answer.

印刷机。第四。

The printing press. Four.

好的。

Okay.

太空卫星。第五。

Satellites in space. Five.

我喜欢这个答案，很聪明。

I like that answer, that's smart.

人工智能。第一。

AI. One.

第一，好的。

One, okay.

自动驾驶汽车。所以，你剩下的唯一选项就是第二。

Self-driving cars. So, your only option left is like Two.

第二。

Two.

现在知道了所有选项，你会交换任何排名吗？

Would you swap any of them now knowing all the options?

我会排为人工智能、火、印刷机、卫星、自动驾驶汽车。

I would go AI, fire, printing press, satellites, self-driving cars.

好答案。为什么人工智能排在火前面？

Good answer. Why AI over fire?

火在人类历史上显然极其重要。从食物到蒸汽机，再到恶劣气候中的温暖，以及介于两者之间的许多其他东西。但我打赌，从现在往后看 100 年或 1000 年，它们都将是有史以来最伟大的通用技术中的两个，而人工智能在总量上会做得更多。但很难说。我想反过来讲，我不会反驳它们。

Fire was clearly extremely important in human history. From food to steam engines and warmth in difficult climates and way other stuff in between. But I would bet that viewed backwards 100 or 1,000 years from now, they will both be two of the great enabling general-purpose technologies of all time, and AI will have done more in total. But yeah, tough to say. I wanted to say it the other way, I wouldn't fight them.

好了，我最后一个问题。你脑子里最常见的想法是什么？比如，你每天想得最多的是什么？

All right, my last question for you. What's the most common thought in your head? Like, what do you think the most every day?

在这一点上，它一直是：成功的社会推广会是什么样子？不仅仅是技术，而是我们如何鼓励所有这些能动性和创业精神？我们如何思考社会契约必须是什么样子？生活在 GDP 下降但生活质量大幅提升的世界意味着什么？我们如何在供应链上足够积极，以建立我认为我们都需要的算力，以实现一个美好公平的未来，同时又不短期内破坏经济？诸如此类的事情。

At this point, it's been like what does the successful societal rollout of this look like? Not just the technology, but how do we encourage all of this agency and entrepreneurship? How do we think about what the social contract is going to have to look like? What does it mean to live in a world of declining GDP, even if quality of life is going way up? How do we go be aggressive enough on the supply chain to build out the compute power I think we all need for a good and fair future without breaking the economy in the short term? Those sorts of things.

嗯，你有一个有趣且令人兴奋的挑战，既要考虑现在，又要考虑 5 年、10 年、15 年后。我可能应该多考虑一下现在，但确实如此。

Well, you have an interesting, exciting challenge of having to think about the now, but then also think about 5, 10, 15 years. I probably should think about the now a little bit more, but yes.

嗯，非常感谢你来做客。你太棒了。

Well, thanks so much for coming on. You're awesome.

真的很享受这次。是的，这太棒了。好吧，太有趣了。老兄，你太棒了。

Really enjoyed this. Yeah, this is epic. All right, that was so fun. Dude, you're awesome.

谢谢你做这些。

Thank you for doing that.