从免费演示到超级助手：OpenAI 的旅程

ChatGPT 最初是完全免费的，原因是它原本只是一个演示，我们打算一个月后就关掉它。

ChatGPT originally was entirely free and the reason for that was that it was intended to be a demo and we were going to wind it down after a month.

后来我们意识到这个演示火了，人们很喜欢它，它实际上成了一个产品。但我们认识到，作为一个产品，你不能每次容量不够就把它下线。所以我们推出订阅，只是为了调节需求。这是一种在不得不拒绝用户时体面地拒绝他们的方式。

We then realized that the demo went viral and people loved it and it was actually a product. But we realized to be a product, you can't take the product down every time you're at capacity. So we shipped subscriptions simply because it could shape the demand. It was a way of gracefully turning users away when we had to turn someone away.

你们现在有 9 亿周活跃用户，这个增长太惊人了。下一个十亿用户将从哪里来？

You guys are at 900 million weekly active users now, and that growth has been incredible. The next billion users, where are they going to come from?

现在全球大约有 10% 的人在使用我们。还有 90% 的潜力，对吧？机会还多得很。

We've got about 10% of the world coming to us now. 90% left to go, right? There's so much more opportunity.

Nick，非常高兴你能来。

Well, Nick, so excited to have you here.

谢谢邀请，Peru。你从德国到美国布朗大学的经历很精彩。

Thank you for having me, Peru. You've had quite the journey from Germany to the US for Brown.

最近还在 Instacart 30 分钟送杂货，现在却把 AGI 带给数十亿人。我敢肯定这从一开始就是计划好的。

Most recently at Instacart delivering groceries in 30 minutes to now delivering AGI to billions. I'm sure that was the plan all along.

没错，显然是精心策划的。

Yeah, clearly total master plan.

跟我们讲讲你的经历吧。你是怎么加入 OpenAI 的？我知道这是个有趣的故事。你在 OpenAI 的三年半过得怎么样？

Well, tell us about your journey. How did you get to OpenAI? I know it's a fun story. And your three and a half years or so at OpenAI, how have they gone?

我所有工作决策的唯一主线完全基于人。所以我不敢说加入 OpenAI 或预测 ChatGPT 等有什么功劳。但我联系了一位我非常钦佩的人，Joanne，我在 Dropbox 认识她，她当时在这里工作。我请她帮我从 DALL-E 2 的候补名单中移除，她告诉我如果我想移除，就得参加面试。于是我上钩了，过程中完全被技术问题吸引，然后就成了现在这样。

The only through line in any employment decision has been entirely people based. So I don't claim any credit for joining OpenAI or predicting ChatGPT or anything like it. But I hit up someone I admire a lot, who I got to know at Dropbox, Joanne, who worked here at the time. I asked her to get off the DALL-E 2 waitlist, and she told me I had an interview if I wanted to get off the waitlist. So I took the bait and got totally nerd-sniped in the process, and here I am.

就是这样。DALL-E 2 的候补名单果然能吸引人。真是个绝妙的招聘工具。

There you go. The DALL-E 2 waitlist will get you. It's a great recruiting tool.

我们或许应该多用候补名单。

We should do more waitlists probably.

我们现在正处于 ChatGPT 的大超级周期。我猜月活用户超过 10 亿，周活跃用户 9 亿，这是从三年半前的零增长起来的。如果我想象 Nick Turley 的仪表盘，上面可能有用户数、付费订阅者、日活跃用户、留存率、参与度等等，大概有 15 项指标，也许全部都有。你的北极星是什么？你怎么运营？你在优化什么？Nick 每天看仪表盘上的什么？

Well, you know, the big super cycle we're in is ChatGPT. Now, I assume over a billion users on the monthly side, 900 million weekly active users that recently reported up from zero three and a half years ago. If I imagine what the dashboard of Nick Turley looks like, it could have users, paying subscribers, daily active users, retention, engagement. I mean, there's like 15 things, maybe all of them. What is your north star? How do you operate? What are you optimizing for? What is Nick looking at in his daily dashboard?

这很有趣，对吧？因为这是一个非常年轻的产品。正如你所说，才三年半。这类问题会随着你的成长和进化而变化，你会问自己：我们到底在建造什么？直到今天，我们想构建一个超级助手，真正帮助人们实现目标。最终，我们关心的是产品是否做到了这一点。它是否真的在帮你完成你使用产品的目的？这对不同的人来说差异很大。有些人想变得更健康，有些人想创业、学习新知识、报税等等。成功的真正衡量标准是我们是否在帮你做到这些。显然，我们特别关注周活跃用户，因为我们想知道你是否会回来使用产品。我们看留存率。但我们会综合看各种数据，因为确实没有单一指标可以优化。

It's funny, right? Because it's such a young product. It's been, to your point, three and a half years. And this kind of question changes as you evolve and grow up and ask yourself, what are we really building here? To this day, we want to build a super assistant that can actually help people achieve their goals. Ultimately, the thing we care about is whether our product is doing that. Is it actually helping you do the thing you're coming to the product to do? And it's so different for different people. Some people are trying to get healthy, others are trying to start a company, learn a new topic, do their taxes. There are all these different things. The true measure of success is whether we're helping you do that. Obviously, we look at WAU in particular because we want to know if you're coming back to the product. We look at retention. But we look at all kinds of stuff in aggregate because there really isn't one single thing you can optimize for.

如果你要给这些指标分配 100 分，按当前的重要性，你会怎么分配这 100 分？

If you were to allocate 100 units of points to these metrics, which metric would you distribute the 100 units across in order of importance for you right this second?

好问题。我非常看重长期留存，我会把所有分数都放在那里。因为我为我们拥有的留存数据感到自豪。但最终，持久价值的标志是人们在三个月后是否回来，这意味着你真正解决了他们的问题。我认为像收入这样的东西是随之而来的，而不是直接去追求它们。我们在这些方面做了很多原则性的决策，取得了很大成功。一个很好的例子是 GPT-4 曾经在付费墙后面，因为我们无法为所有人提供服务，后来 GPT-4o 在推理能力上取得了突破，我们就免费提供了。结果这完全带来了正向收入和留存，因为它让更多人用上了这项技术。我认为当你这样决策并专注于客户时，最终会得到优秀的产品，收入自然也会随之而来。

It's a good question. I care a lot about long-term retention and I would put all my points there. Because I'm really proud of the retention stats we have. But ultimately, the sign of durable value is whether people are coming back in three months, because that means you're really solving their problems. I think things like revenue, they follow from that. Versus trying to go after those things directly. We've had a lot of success making very principled decisions on this stuff. One good example is GPT-4 used to be behind a paywall because we couldn't serve it to everyone, and then we had GPT-4o which was a total breakthrough in our ability to inference it, so we just gave it away for free. That ended up being totally revenue positive and retention positive because it provided access to the tech. I think when you make your decisions that way and focus on the customer, you end up with a great product and revenue obviously follows too.

太棒了。这从数据中就能看出来。我昨天发了一张第三方数据的图表。ChatGPT 的留存曲线是微笑曲线。看看，就是这样。这种情况非常罕见。你认为为什么会出现微笑曲线？你在 ChatGPT 中看到了什么，让那些可能停用了几周或几个月的人又回来了？

Phenomenal. Well, it shows up in the numbers. I posted this chart yesterday on the data we have from a third party. The retention curves for ChatGPT are smiling. Look at that. Just like that. And that is a very rare occurrence. Why do you think these smile curves exist? What are you seeing in ChatGPT that has people who have maybe turned off for a couple of weeks or months coming back?

没有单一原因。构建一个高留存的产品需要很多很多小细节，并且系统地让它变得更好。我想说的是，对于 AI，尤其是 ChatGPT，我发现人们需要一些时间才能真正理解他们生活中哪些部分可以委托给 AI。对许多用户来说，这是一个长达数月的过程，他们才能理解这个东西如何帮助他们，以及有哪些不同的方式可以将 ChatGPT 融入生活。但回想我们取得的一些突破和杠杆，比如搜索和个性化，它们帮助解决了用户问题，因为搜索为你提供了更多日常价值。过去，ChatGPT 是一个偏工作的产品。我们看到周末使用量下降，暑假期间很多人不上班时使用量也下降。而现在，我们是移动优先。

Look, there isn't one single thing. The way you build a retentive product is lots and lots of little things and really trying to make it better systematically. I will say that with AI and in particular ChatGPT, I found that it takes people some time to really understand all the parts of their life they can delegate. I think for many users, it's a multi-month process for them to understand how this thing can help them and what are all the different ways they could plug ChatGPT into their life. But when I think about some of the breakthroughs and levers we've had, things like search and personalization have helped solve those user problems because search provides way more daily value to you. It used to be that ChatGPT was a pretty worky product. We'd see usage go down on the weekend, go down during the summer months when a lot of people were off from work. Today, we're mobile first.

绝大多数是移动端，我们看到了所有这些个人用例。我认为搜索是一项重大投资，让我们走到了今天，而个性化让聊天对你来说更加相关，对吧？因为它会随着时间了解你。你也了解它。这两件事极大地改变了人们回访产品的方式。但还有很多事情要做。

The vast majority is mobile and we see all these personal use cases. I think search was a big investment that got us there, and personalization makes chat so much more relevant for you, right? Because it gets to know you over time. You get to know it. Those are two things that have materially moved the way people come back to the product. But there's lots more to do.

是的。

Yeah.

正如提到的，尽管我们显然非常自豪，但我并没有满足于我们的留存数据。

And as mentioned, I'm not resting on our retention stats, even though we're obviously very proud.

很好，很好，很好。

Nice, nice, nice.

另一件我关于 ChatGPT 看错的事是两年半前。我当时想，看看谁会赢得这场消费级 AI 竞赛。通常这些消费市场是赢家通吃或赢家拿走大部分。看看搜索：谷歌拥有近 90% 以上的市场份额，市值 3.5 到 4 万亿美元。移动端，苹果也一样。社交，Meta 也一样。我当时想，AI 方面：Meta 拥有所有分发渠道，谷歌也拥有所有分发渠道，他们有 34 亿用户。他们推出 AI 就像按个开关一样简单。但我错了。事情并没有那样发展。ChatGPT 现在有 9 亿周活跃用户，增长令人难以置信。显然分发是不够的，对吧？

The other thing I got wrong about ChatGPT was two and a half years ago. I thought, let's look at who's going to win this consumer AI race. Typically these consumer markets are winner-take-most or winner-take-all. Look at search: Google has near 90% plus market share, three and a half, four trillion market cap. Mobile, same thing with Apple. Social, same thing with Meta. I thought, well, AI: Meta has all the distribution, Google's got all the distribution, they've got 3.4 billion users. It would be a flick of a switch for them to roll out their AI. But I was wrong. That's not what happened. ChatGPT turns out you guys are at 900 million weekly active users now, and that growth has been incredible. Clearly distribution was not enough, right?

所以对于分发也是同样的问题：我们达到这个规模用了哪些杠杆？是模型质量、产品质量、功能、体验，还是像记忆和个性化或搜索这样的产品改进？同样的问题：你认为是什么推动了过去的增长和成功？

So the same question for distribution: what are the levers for us that have gotten us to the scale? Is it model quality, product quality, features, the experience or product improvements like memory and personalization or search? Same question: what would you say drove historical growth and success?

我们现在大约有全球 10% 的用户。还有 90% 有待拓展，对吧？所以还有更多机会去触达更多人，向他们介绍 AI 能带来的好处。但当我回顾过去时，我这么说是因为下一个十亿用户可能在如何互动、触达和提供价值方面非常不同。但回顾过去，大致是三分之一、三分之一、三分之一的比例，分布在经典的消除摩擦类工作上。比如，从纯粹影响来看，我们最大的时刻之一是移除认证墙。Sam 会说“我告诉过你”，因为我认为那是他从第一天起就给出的反馈：你不应该需要登录 ChatGPT。但就是这类事情，你为任何产品都会做，而且确实重要。有些东西永远不会变，对吧？然后另外大约三分之一是我所说的核心产品投资，这些通常是我们研究和产品团队共同完成的事情。搜索和个性化就是很好的例子，我们合作不仅解决了 UI/UX 演进，还解决了如何将这些变化后训练到模型中。正是我们合作的时刻。另一个最近的例子是我们有这些写作块，当你询问关于用模型写作的查询时会出现，精心打磨这些体验非常重要，我们的用户很喜欢。然后另外三分之一的增长纯粹是模型改进，比如从 GPT-3.5 到 GPT-4 的阶跃变化，再到从付费墙后的 GPT-4 到 o1 突然对所有人开放。但其中很多也是不那么引人注目的迭代，不值得一个命名发布。我对我们刚刚在 5.3、5.4 等版本上做的更新感到非常兴奋。因为那时我们收集了大量用户反馈，并系统地加以解决，这显然也体现在我们的留存数据中。所以大致是三分之一、三分之一、三分之一，分布在经典的消除摩擦和访问、核心产品投资，以及纯粹的模型改进。

We've got about 10% of the world coming to us now. 90% left to go, right? So there's so much more opportunity to reach more people and introduce them to the way that AI can benefit them. But when I look backwards, I only say that because the next billion users might be very different in terms of how you engage and reach and provide value. But when I look backward, it's been roughly sort of one third, one third, one third between classic friction removal type of work. Like one of our biggest moments when you look at pure impact was removing the authentication wall. And Sam will say 'I told you so' because I think that was his feedback from day one: you shouldn't have to log into ChatGPT. But it's stuff like that that you do for any product and it does matter. Some things never change, right? And then another third or so are what I would call core product investments, and they're really typically things that we've done together between research and product. So search and personalization are really good examples of that, where we came together and we figured out not just UI/UX evolution but also how to post-train these changes into the model. It was really the moments when we came together. Another recent example is we have these writing blocks that render when you ask about queries where you're trying to write with the model, and putting really good craft into those experiences really matters, and our users love it. And then another third of the growth has been just model improvements, like step changes from GPT-3.5 back then to GPT-4, then going from GPT-4 behind a paywall to o1 suddenly available to everyone. But a lot of it is also the iteration that isn't splashy, that doesn't warrant a named release. I'm really excited about the updates we just made with 5.3, 5.4, etc. Because that is when we take a lot of user feedback and we methodically address it, and obviously that shows up in our retention as well. So sort of one third, one third, one third between classic friction removal and access, core product investments, and then pure model improvements.

所以我一直想问你的问题是：我们如何获得下一个十亿用户？请谈谈这个。从外界的战争迷雾来看，似乎有很多变数。如果今天我是一个消费者，要在市场上挑选我的超级助手，你会看到几个不错的选择：Claude 就在那里，过去几周他们势头很好。Gemini 有超级分发，Uber 也有分发，而我们目前是领先的产品，至少在用户数量上。下一个十亿用户将从哪里来？

And so the question that I've really been waiting to ask you is: how do we get the next billion? And talk about that a little bit. There's a lot of it, it seems like at least from the outside fog of war. If I was a consumer today in the market to pick my super assistant, you would have a couple of great options: Claude out there, they're having some great traction last couple of weeks. Gemini mega distribution, Uber distribution, and us by the leading product today, at least in user numbers. The next billion users, where are they going to come from?

首先，把这个目标放在背景中。我们最终关心两件事。显然，触达更多人非常重要。这是我们使命的直接体现：我们能让更多人了解 AI 的好处，就越好。但我们也非常兴奋能深入下去。这意味着让今天在 ChatGPT 中找到价值的同一批用户，在他们的世界中获得更有意义的价值，比如真正帮助他们实现目标，而不仅仅是回答问题，对吧？所以我会谈谈我们如何获得更大规模，但我觉得重要的是要记住，这项技术演进的方式是，我们将很快超越纯聊天机器人。令人兴奋。我认为在规模方面，让我惊讶的是有多少人在 ChatGPT 当前的形式中找到了价值，因为我认为对大多数人来说，委派并不是一种天生的技能。而 ChatGPT 是一个强力工具：你使用它，它不会告诉你它是做什么的，你基本上得自己发现。你必须使用它，然后你会了解到某个很酷的提示词，然后也许你在 Twitter 上了解到另一个，或者在 Instagram 上了解到另一个。但产品就像一个原始设备。我认为在触达下一批用户时，我们真正需要做好的一件事是让产品有更多的可供性。因为我认为对大多数人来说，他们非常非常忙。世界上每个人都有智能受限的问题——更多智能可以帮助解决的问题——但你需要向人们阐明这一点。是的。我仍然觉得我们有点太像计算机终端了，它需要感觉更像软件，或者一个软件操作系统，对吧？所以这是一件事。

First of all, just to contextualize that goal. We care about two things at the end of the day. Obviously, reaching more people is really important. It's the direct manifestation of our mission to the world: the more people we can introduce the benefits of AI, the better that is. But we're also really excited to go deeper. And that means taking the same billion users that find value in ChatGPT today and actually providing more meaningful value in their world, like actually helping them achieve their goals, not just answering questions, right? So I'll talk about how we get to more scale, but I think it's important to remember that the way this technology is evolving is we're going to go beyond pure chatbots pretty fast. Exciting. I think on scale, it's shocked me how many people have found value in ChatGPT as it works today, because I don't think delegation is a natural skill for most people. And ChatGPT is a power tool: you come to it, it doesn't tell you what it's for, you kind of have to discover it on your own. You have to use it, and then you'll learn about this prompt that was really cool, and then maybe you're on Twitter and you learn about another one, or you're on Instagram and you learn another. But the product is like a raw appliance. And I think one thing we really need to nail as we reach the next set of users is a product that has a bit more of an affordance. Because I think for most people, they're very, very busy. And everyone in the world has intelligence-constrained problems—problems that more intelligence could help with—but you need to frame that to people. Yeah. And I still feel like we're a little bit too much like a computer terminal, and it needs to feel more like software, or an operating system of software, right? So that's one thing.

另一个触及同样限制的问题是开始变得主动。在一个很多人忙得没时间把问题委托给 AI，或者不知道从何下手的时代，我认为能够主动帮助你非常重要。但我认为这些都是基于当前技术可以做的产品演进。而让我特别兴奋的是将我们的下一代技术或推理模型产品化，因为事实是，当你审视推理和聊天机器人时，它只对一小部分人有用。它只对那些试图充分利用 ChatGPT 的人有用。但我从根本上相信推理是变革性的。如果你能想出一种方式将推理产品化，让它以人们甚至不知道的方式为他们工作，那看起来就像是模型在为你执行长期任务。这并不意味着你接触到这个概念，只是意味着它让你受益。所以还有很多工作要做，产品必须进化才能与这种规模相关。

Another thing that gets at the same constraint is beginning to be proactive. In a world where a lot of folks are too busy to delegate their problems to AI or don't quite know where to start, I think being able to help you proactively is really important as well. But I think all these are product evolutions that we could make on top of the current tech. And the thing that gets me particularly excited is productizing our next generation tech or reasoning models because the truth is when you look at reasoning and chat today, it's relevant to a very small group of people. It's relevant for the people who are trying to get the most out of ChatGPT. But I fundamentally believe that reasoning is transformative. And if you can figure out how to productize reasoning in a way that works on people's behalf without them even knowing, and that looks very much like the model doing long horizon tasks on your behalf. It doesn't mean you encounter the concept. It just means it's benefiting you. So there's so much work to do and the product certainly has to evolve to be relevant for this kind of scale.

是的。我期待已久的一件事，Brad 两年前打了个赌：ChatGPT 什么时候能帮我采取行动？ChatGPT 能帮我更主动吗？我认为他的赌注去年年底到期了。所以我们很好奇。什么时候会实现？我帮你梳理一下，因为二十年前用 Google 搜索引擎，你会得到 10 个蓝色链接。你可能要花一个小时才能找到答案。现在用 ChatGPT 可以立刻得到答案。感觉下一步就是行动。

Yeah. One of the things that I've been hoping for a while and Brad made a bet two years ago: when can ChatGPT help me take actions? Can ChatGPT help me be more proactive? And I think his bet expired end of last year. So we're very curious. When is that coming? I'll frame that for you because with search engines in Google two decades ago, you'd have gotten the 10 blue links. You could have spent an hour getting the answer. You can now get the answer instantly with ChatGPT. And it feels like the next step is actions.

100%。感觉下一步就是行动。Pulse 是一个很好的主动式产品。我有一个每周运行的 Pulse，但我真正想要的是：嘿，Nick 提到了某件事，就找到我，确保我知道 Nick 提到了这个，或者嘿，我关心的某件 XYZ 事情发生了。什么时候能变得主动？会是什么形式？

100%. And it feels like the next step is actions. Pulse is a great proactive product. I have a Pulse that runs weekly, but what I would really like is: hey, Nick spoke about something, just find me and make sure I know that Nick spoke about this, or hey, this XYZ thing happened that I cared about a lot. When is one of that going to get proactive? What is the modality going to look like?

是的。我认为有两个概念。一是聊天机器人应该做事而不仅仅是回答，二是聊天机器人应该主动。我认为当你把两者结合起来，就会感觉它像一个超级助手。因为这些是相互叠加的。在采取行动方面，严格来说，聊天机器人今天就能做事。只是行动空间非常有限，对吧？它可以搜索网络，这意味着它可以像人类一样使用搜索工具或浏览器。它可以制作图片。它可以做所有这些事情，对吧？但它没有人类用电脑时那样的行动空间。而这正是我们想要构建的。时机在这些赌注中至关重要，对吧？我也不假装自己很擅长把握时机。看看我们过去的尝试，比如 ChatGPT 智能体，它就有类似的能力。只是稍微早了一点。模型还不够好，无法达到真正的逃逸速度。问题是，如果没有逃逸速度，用户就不会学会信任它。他们甚至不会尝试。所以，当你看人们在原始版本的 ChatGPT 智能体上做的事情时，都是那些碰巧能用的东西，比如把文件服务器迁移到云端之类的。有用但非常小众。随着这些东西变得更好，我们只需要让它达到一个点，让人们尝试用它来解决生活中真正有意义的问题，因为那样我们就可以开始爬山了。这就是 ChatGPT 的魔力所在：ChatGPT 在发布时就已经足够好，让人们真正尝试用例，即使最初并不成功。比如聊天机器人最初写作很糟糕，编程也很差，但人们尝试了，并从中获得了足够的价值，这样我们就可以把这些用例变得更好。我确实认为我们即将达到通用智能体的那个点：它足够好，让你至少获得部分成功，而因为获得了部分成功，你会得到非常好的任务反馈，然后魔法就开始了。一旦你有一组可以爬山的用例，我们就可以让它们变得很棒。所以关于任务，我认为我们接近了，但我认为即使是 OpenAI 内部的人也很难准确预测什么时候会变得很好。我们已经兴奋了一段时间。关于主动性，Pulse 是一个非常棒的第一步，因为我们想要构建一种形式，不是你提示模型，而是模型提示你。原因我之前说过：人们很难委托任务，也很难弄清楚自己的问题是什么。如果 AI 理解了你的目标和感兴趣的事情，然后开始主动为你做事呢？Pulse 能提供的价值有限，因为它没有连接到你的生活，也不能采取行动。所以它只是为你生成信息，人们喜欢这样。我也喜欢。我的 Pulse 也在运行。但我认为魔法在行动和主动性结合时才开始，因为那时它可以开始推测性地检测，嘿，你刚刚降落在目的地。我来帮你叫辆出租车。或者如果你在工作，它会说，嘿，我主动运行了这个分析，因为我看到你的指标下降了。所以我认为这些东西真的会叠加，我们需要掌握多个构建模块才能真正实现我们希望的转型和形式。

Yeah. So there's two concepts I think. There's chat should be doing stuff rather than just answering, and then there's chat should be being proactive. I think when you put them together, you start feeling like it feels like a super assistant. Because I think these things compound. On the action taking piece, strictly speaking, chat can do stuff today. The action space is just very limited, right? It can search the web, which means it can use a search tool or browser in the same way that a human would. It can make images. It can do all these things, right? But it doesn't have the same action space that a human with a computer would have. And that is what we aim to build. Timing is everything on these bets, right? And I don't pretend to be great at timing either. You look at past attempts that we've made like the ChatGPT agent for example, which kind of has capabilities like this. It was just slightly too early. The models weren't quite good enough to hit real escape velocity. And the problem is if you don't have escape velocity, users don't learn to trust it. They don't even try. So when you look at a lot of things people were doing in the original version of ChatGPT agent, it was the things that happened to work like migrating your file server into the cloud or something like that. Useful stuff but very niche. As this stuff gets better, we just have to get it to a point where people try to use it for real meaningful problems in their life because then we can start hill climbing. This has been the magic of ChatGPT where ChatGPT upon launch was good enough to get real attempts at use cases even if they didn't initially work. Like chat was a pretty bad writer originally, it was a bad software engineer, but people tried and got enough value out of it that we could take those use cases and make them great. And I do think we're about to get to that point with general purpose agents where it works well enough that you get at least partial credit, and because you're getting partial credit, you get really good tasks back and then the magic begins. Once you have a set of use cases that you can climb the hill on, we can make them awesome. So on task, I think we're close, but I think even people inside of OpenAI would have had a hard time predicting exactly when this gets good. We've been excited about it for a while. On proactivity, Pulse was a really great first step because what we wanted to build was a form factor where you're not prompting the model, the model's prompting you. For the reasons that I described earlier, which is, it's so hard for people to delegate and to figure out what their problems are. What if the AI understood your goals and the things you're interested in and just could start being proactive on your behalf? Pulse is limited in the value it can provide for you because it's not connected to your life and it can't take action. So, it's producing information for you and people love that. I love that. I've got mine running, too. But I think the magic begins when you have actions and proactivity because then it can begin speculatively actually detecting, hey, you just landed where you were supposed to go. I'm going to call a cab for you. Or if you're at work, it's like hey, I proactively ran this analysis because I saw your metrics dropped. So I think these things really compound and we need to nail multiple of the building blocks to really achieve the transformation and the form factor that we hope for.

你回答这些问题时，我现在又多了 15 个问题。所以我希望你有 15 分钟。但好吧，一个一个来。我们从你提到的行动和任务开始。关于时机，很难预测。但你认为任务或智能体有没有某种形态或顺序，比如，嘿，无论何时实现，这类事情可能会先出现？

As you were answering those questions, I now have 15 more questions for you. So I hope you have 15 more minutes. But okay, one by one. We'll start with what you said on actions and tasks. Got it on timing, tough to predict. But is there a shape or ordinality of tasks or agents that you think, hey, this is the kind of thing that's likely to come first whenever it does?

我的意思是，已经先出现的是领域特定智能体，对吧？如果你看看代码领域正在发生的事情，我们已经完全实现了。

I mean, the thing that's already come first is the domain specific agents, right? If you look at what's happening in code, we're fully there.

你知道吗，这很不可思议，但我们有太多工程师从来不打开他们的 IDE。而对我来说，作为一个曾经写代码但后来变得非常忙的人，它让我重新回到了这个领域。所以 Codex 和类似的产品显然已经具备了逃逸速度，人们完全在用它们做各种智能体式工作。如果你只是把人们正在做的事情做得更好，你基本上就能达到目标。我不会惊讶于看到这种情况发生在其他形式的量化知识工作上，因为它恰好具备代码的那些属性：可测试，你能知道它是否有效，而且它非常有利于强化学习。但领域特定的智能体已经能用了。我认为每个人都在努力的方向是通用智能体，能处理任何事情。

You know, it's mind-bending, but we've got so many engineers who don't open their IDE ever. And for me, as someone who used to code and then unfortunately got very busy, it's brought me back in the game. So Codex and products like it clearly have escape velocity, where people are absolutely using it for all kinds of agentic work. And if you just take what people are doing and make it work even better, you kind of get all the way there. I won't be surprised if you see this happen for other forms of quantitative knowledge work, just because it happens to have the properties that code has: it's testable, you know if it worked or not, and it's very RL-friendly. But the domain-specific ones already work. I think the thing everyone's working for is general-purpose agents that just work for anything.

是的。我认为这就是为什么你需要赢得消费者，因为很难训练人们去接受“哦，它能行”。比如 Deep Research 就是一个消费产品，它真的是我们推出的第一个智能体式产品。但我认为消费者想要的是：我可以随便问它任何事，它就能完成需要做的事情，而无需任何重新训练。我们会达到那一步，只是时间问题。至少一个心理上的目标是订机票、订餐厅、购物，所有这些事情。消费者问题太多了，而这些正是你会发起的那种任务，对吧？

Yeah. And I think that's why you need to win a consumer, because it's very hard to train people into like, okay, it can work. Like Deep Research was a consumer product, and it was really our first agentic thing out there. But I think what consumers want is: I can just ask it anything, and it'll do what needs to be done without any sort of retraining. And we'll get there, just a matter of time. At least a psychological goal is flight bookings, restaurant bookings, shopping, all this stuff. There are so many consumer problems, and those are just the type of things that you would kick off, right?

对。

Yeah.

一旦你有了生产力，有些你甚至不认为是智能体式任务的事情也会出现。比如你想健身。你不会认为那是可以委托的任务，除非你有教练，那样你才会，但大多数人没有，对吧？但如果 AI 知道这一点，它完全可以开始在后台为你长期工作，给你制定健身计划。好，我实际上已经帮你报名了这个项目。你可以想象，如果它与你的长期利益一致，它会非常有帮助。

The minute you have productivity, there are things you don't even think of as agentic tasks. Like you're trying to get in shape. You don't think of that as a task you would delegate, unless you have a trainer, in which case you do, but most people don't, right? But if the AI knew that, it could totally start working in the background for you over very long periods of time, and getting you, here's your fitness plan. Okay, I actually signed you up for this thing. You could imagine it being quite helpful if it's aligned with your long-term interest.

你这是在给 Ozanic 制造竞争啊。我们得小心进入哪些业务，但希望我们能帮上忙。

You're going to give Ozanic a run for the money. We got to be careful what businesses we get into, but hopefully we can help.

那太好了。等不及了。等不及了。你提到的第二件事是主动用户，这可能要求我们超越聊天机器人。有什么模态的例子能让 ChatGPT 超越聊天机器人？

That'll be great. Cannot wait. Cannot wait. The second thing you said was proactive users, and that might require us to go beyond chatbots. What's an example of a modality that might take ChatGPT beyond a chatbot?

所以聊天永远是我的心头好。这是我们成长的方式。它是一个重要的模态，应该保留。我认为对我来说，与其说是聊天，不如说是自然语言——你可以用非常自然的方式向机器表达自己，无论是文字、语音，还是模型渲染的结构化 UI，这都非常强大，而且会一直存在。但我认为会改变的是：聊天是表达意图的好方式，是与机器沟通的好方法，但它不是一个好的输出。在很多情况下，你希望得到的是一个制品：这是你的旅行计划，这是分析结果，这是我为你交付的成果。我刚刚帮你赚了五块钱。这才是我希望我的 AI 为我做的事情，对吧？

So chat will always be close to my heart. It's the way we grew up. And it's an important modality to stay. I think it's less about chat and more about natural language to me, where the fact that you can express yourself to the machine in ways that are very natural to you, whether that's text, whether that's voice, whether that is structured UI that is rendered by the model, that is just very powerful and that's here to stay. But I think the thing that'll change is that chat is a great way of expressing your intent. It's a good way of communicating with the machine, but it's not a great output. In many cases, what you want back is an artifact: here's your plan for your trip, here is the analysis, here is an outcome that I delivered for you. I just made you five bucks. This is what I want my AI doing for me, right?

是的，完全同意。我的意思是，这才是人们关心的，对吧？我认为聊天会一直存在，作为你澄清意图和启动任务的方式，但我不认为它一定是最终的交付物。我认为这就是我们可以演进的方向。所以希望这是一个非常优雅的过渡，因为我很幸运，也来之不易，每周有十亿人来找你，做他们喜欢的事情。

Yeah, totally. I mean, this is what people care about, right? And I think chat will always be there as the way that you sort of disambiguate your intent and you kick off the task, but I don't think it's necessarily the final deliverable. And I think that's the way in which we can evolve. So hopefully that's a very graceful transition, because I'm very lucky and it's hard earned, to have a billion people coming to you weekly for a thing that they love.

是的。但我认为这是一个很好的起点，因为我们有太多未满足的意图，人们显然想做一些事情，聊天已经足够有帮助，但它可以更有帮助得多。我认为这就是我们演进的方向。

Yeah. But I think it's a great jumping off point because we have so much unsatisfied intent from people where they're clearly trying to do something, and chat is helpful enough, but it could be so much more helpful. And I think that's where we evolve.

是的。你一定掌握着大量这样的数据，人们来聊天并尝试，正如你所说，三年前他们至少还在尝试。

Yeah. And you must be sitting on so much of this data where people are showing up to chat and attempting, as you said, three years ago they were at least making the attempt.

是的。所以你至少有一个频率直方图，比如，嘿，这是人们想用我们实现的所有事情。我们确实有非常棒的分类器自动运行。它完全保护隐私，但能让我们了解人们有哪些用例。这很重要，因为当你发布一个新模型、做一次模型更新时，你想知道哪些用例变好了，哪些用例变差了。除非你的系统有非常好的分析能力，否则这并不总是容易搞清楚的。但我的很多学习实际上是定性的，我会习惯性地联系一组相当随机的用户，去了解他们在做什么。我从未在这样一个产品上工作过，三年半后你还在不断学习，因为通常到那个时候你已经知道你的产品能实现哪些用例了。但我们的技术如此不寻常，我总能了解到一些疯狂的事情，我原本不知道是可能的。

Yeah. So you might have at least the frequency histogram of like, hey, here are all the things that people want to achieve with us. We do have really awesome classifiers that run automatically. It's fully privacy preserving but gives us a sense of what use cases people have. And it's important, because when you make a new model, you make a model update, you want to know what use cases just got better, what use cases got worse. And that's not always trivial to figure out unless you have really good analytics on the system. But so much of my learning is actually qualitative, where I will just, I have a habit of reaching out to a fairly random set of users to just figure out what they're doing. And I've never worked on a product where three and a half years later you're still learning every time, because usually by that time you know what the use cases are that your product can deliver on. But our tech is so unusual in the fact that I keep learning about something crazy I didn't know was possible.

哇，太棒了。但基本上，十亿用户，我猜其中一小部分是重度用户，他们从 200 美元的订阅中获得了数千甚至数万美元的价值。绝大多数是中间用户，还有一些轻度用户，他们开始把 ChatGPT 当作搜索工具，或者让我教 AI，或者帮我做作业。你的关注点是什么，也许在这些群体中：重度用户、轻度用户和早期用户，或者随便你怎么划分？我们对这三个群体的关注点分别是什么？

Wow, that's awesome. But basically, a billion users, I suspect a small fraction of them are power users who are getting maybe thousands, maybe tens of thousands of value on their $200 subscription. The vast majority is middle of the pack, and then a few casual users who are starting to use ChatGPT as search, or teach me about AI, or help me with my homework. What is your focus, maybe in those constituents: power users, casual users, and early users, or however you frame it? What is our focus for each of those three factions?

是的。首先，我对我们的整个用户群都负有责任。

Yeah. Well, first of all, I feel accountable to our entire user base.

事实上，我们的非用户也是如此，因为像 ChatGPT 这样的产品可能对所有人类产生真正的外部性。

In fact, our non-users too because your products like ChatGPT can have real externalities on all humans.

是的。但当我思考我们的构建方式时，想象极端情况非常有用。一个极端是完全不关心 AI 的用户，他们生活忙碌，需要被说服我们提供的价值，因为这迫使你真正打磨界面，并以人们能够理解的方式展示模型中隐藏的能力。另一个有用的极端是我们的重度用户群体，因为重度用户教会我们什么是可能的。实际上，我们不可能独自完成所有产品发现，仅仅因为这项技术是如此经验性，而且你在发布后学到的东西非常多。所以为这两个极端构建都是有价值的。但我们的用户群体极其多样化，人们有这么多不同的用例，这就是为什么我喜欢看各种不同的细分，不仅仅是使用频率，还有你来找我们是为了什么用例。但 ChatGPT 的用户群体确实非常多样化。

Yeah. But when I think about the way we build, it's really useful to imagine the extremes. One extreme being a user who doesn't care about AI at all, who has a busy life, and needs to be convinced of the value that we can provide because that forces you to really nail the interface and expose the capabilities hidden in the model in a way that people can actually grok. And then the other useful extreme is our power user base, because power users are the users who teach us what's possible. It's actually impossible for us to do all the product discovery on our own simply because of how empirical this technology is and how much you actually learn post launch. So building for each of those extremes can be valuable. But our user base is incredibly diverse and people have so many different use cases, and this is why I like to look at all kinds of different segmentations, not just frequency but also what use cases you are coming to us for. But definitely huge variety in the ChatGPT user base.

是的。

Yeah.

我以 Mac OS 为例，它真的对完全不懂技术的人有效。它完全是神奇的，但如果你是重度用户，你有终端、有设置，几乎可以配置 Mac OS 中的任何东西，而且它做得非常漂亮，复杂性逐步展现。所以你可以与它互动并喜爱它的简单，但你也有所有旋钮，开发者喜欢它，对吧？所以我认为这有点像我们在 ChatGPT 中想要成为的灵感。这并不意味着我们总是能达到，但这意味着为重度用户构建极其重要。而且这不仅仅是一个我觉得美学上令人兴奋的特性。在 AI 中它也非常重要，因为正是重度用户向你展示了什么是可能的。他们实际上在做产品发现，因为对我们来说，用这样一项经验性技术，不可能独自完成所有产品发现。所以那种订阅 ChatGPT Pro、在 Codex 还没完全好用时就使用它、现在是最强大的酷工具代币倡导者并教我们什么是可能的用户，是社区中非常有价值的成员。这可能不会在你的周活跃用户中显示为一个数字，对吧？但这正是为什么没有一个单一的北极星指标，你真的需要非常认真地对待这些不同的细分。所以我喜欢为重度用户构建。

I look up to Mac OS for example as an example where it really works for people who don't understand technology at all. It's entirely magical, but if you are a power user, you've got terminal, you got settings, you can configure almost anything in Mac OS, and it's really beautifully done where the complexity is progressively disclosed. So you can interact with it and love the simplicity of it all, but you also have all the knobs and developers love it, right? And so I think this is kind of the inspiration for how we want to be in ChatGPT. That doesn't mean we always live up to it, but it means that building for power users is extremely important. And that's not just a property that I think is sort of aesthetically exciting. It's also really important in AI because it's the power users who show you what's possible. They are actually doing the product discovery because it would be impossible for us with such an empirical tech to do all the product discovery on our own. So the type of user who subscribes to ChatGPT Pro, who used Codex before it quite worked, who is now the strongest advocate of cool tools tokens and teaching us what's possible, that is an incredibly valuable member of the community. And it might not show up in your weekly active users as just one number, right? But this is exactly why there isn't a single north star, and you really need to take these different segments very seriously. So I love building for power users.

好的。所以我们非常关注整个用户群体。从重度用户那里学到很多。我可能还想说的是，现在的重度用户正在获得很多价值，几乎是太多的价值。

Okay. So we're very focused on the entire user base. Learn a lot from the power users. The other thing I might say is the power users right now are getting a lot of value, almost too much value.

没有这回事。

No such thing.

没有这回事。最常见的类比是 2015 年时代的 Uber 和 Lyft，对吧？你知道，这花了一段时间，但我知道你一直在思考这个问题。我知道你们在定价上考虑了很多。也许跟我们谈谈定价。现在定价相当简单。有没有一条路径，让那些获得巨大价值的用户以不同的方式为产品定价，满足他们的需求，另一方面呢？

No such thing. The analog that is most common is the Uber and Lyft of the 2015 era, right? And you know, it took a while, but I know you were thinking about it a lot. I know you guys are thinking about pricing quite a bit. Maybe tell us a little bit about pricing. Right now pricing is pretty simple. Is there a path for folks who are getting a lot of great value to price that product differently and meet them where they are, and on the other side?

我的意思是，定价——当技术变化如此之快时，定价不可能不发生重大演变，对吧？ChatGPT 最初完全免费，原因是它原本是一个演示，我们打算一个月后关闭它。然后我们意识到这个演示病毒式传播了，人们喜欢它，它实际上是一个产品。但我们意识到，作为一个产品，你不能每次容量不足时就下线。所以我们推出了订阅，仅仅因为它可以调节需求。这是一种优雅地拒绝用户的方式，当我们不得不拒绝某人时，感觉最公平、最平等的方式就是说：“嘿，如果你真的需要这个产品，付订阅费就能用。”然后我们想办法让产品稳定下来，我们面临选择：是保留订阅还是回到免费？我们意识到我们一直有更多无法扩展的技术。GPT-4 是第一个例子，因为我们有太多免费用户无法服务 GPT-4，所以我们把它放在了 Plus 计划后面。所以，我们偶然进入订阅的方式是试图解决用户问题。当时感觉这是提供最大技术访问权限的正确方式。从那以后，我们有了许多其他突破，包括测试时计算，你可以随意扩展智能。是的，差不多。整个行业花了一点时间才将其转化为产品价值，但现在我们到了这里，我们的重度用户想要使用越来越多的智能。有可能在当今时代，拥有无限计划就像拥有无限电力计划一样。这根本说不通，因为人们可能需要大量电力，并且从中获得巨大价值。你不能买无限电力是有原因的，对吧？所以，显然我想非常深思熟虑地发展我们的计划、SKU 和订阅，但如果考虑到我们取得的技术突破的规模和深刻性以及随之而来的产品突破，它没有变化的话，你会非常惊讶。

I mean, pricing is — there's no world in which pricing doesn't significantly evolve when the technology is changing this quickly, right? ChatGPT originally was entirely free, and the reason for that was that it was intended to be a demo, and we were going to wind it down after a month. We then realized that the demo went viral and people loved the demo, and it was actually a product. But we realized to be a product, you can't take the product down every time you're at capacity. So we shipped subscriptions simply because it could shape the demand. It was a way of gracefully turning users away when we had to turn away someone, and it felt like the fairest and most equitable way of doing so is saying, 'Hey, if you really need this product, pay a subscription fee and you got it.' Then we figured out how to make the product stable, and we had the choice of do we keep the subscription thing or do we go back to free, and we realized we had consistently more tech that we couldn't scale. GPT-4 being the first example, because we had way too many free users to serve GPT-4, and we put it behind the Plus plan. And so, the way we stumbled into subscriptions was sort of accidental by trying to just solve for the user. And it felt like the right way at the time to provide maximal access to our tech. Since then, we've had so many other breakthroughs, including test-time compute where you can scale up intelligence as much as you want. Yeah, more or less. And it took us, in the entire industry, a little bit of time to turn that into product value, but we're here now where our power users want to use more and more intelligence. And it's possible that in the current era, having an unlimited plan is like having an unlimited electricity plan. It just doesn't make sense because people may need a lot of electricity and they're getting a lot of value out of that. There's a reason you can't buy that, right? So, obviously I want to be really thoughtful about the way that we evolve our plans and SKUs and subscriptions, but you would be incredibly surprised if it didn't change given the magnitude and profoundness of the technical breakthroughs that we've had and the product breakthroughs that follow.

是的。与此相关，我想你会为重度用户提供一些东西。

Yeah. And relatedly, I imagine you're going to have something for the power users.

嗯。

Mhm.

那另一方面呢？我们如何让普通用户参与进来，同时还能从中盈利？

What about the other side? How do we get the casual users into the wheel and still monetize them?

正如提到的，我们的商业模式会不断演变，北极星指标是“可访问性”。我们希望提供一种方案，让尽可能多的人能够使用我们最强大的工具。长期以来，这靠的是订阅制。但订阅制的缺点是，在很多市场，人们没有信用卡，或者不用信用卡订阅软件。我们对其他能最大化技术可及性的方式很感兴趣。我们的广告试点就是出于这个精神。我们把它看作一种工具，将 ChatGPT 以及我们最广泛的智能带给全球任何人。这体现了我们如何不断演进，找到最佳方式让需求与我们的供给相匹配。

As mentioned, our business model will evolve and the north star is access. We would like to provide an offering that maximizes the number of people who can access our most powerful tools. For the longest time, that has been subscriptions. Subscriptions have the downside that in many markets people don't have credit cards or don't use them to subscribe to software. We're interested in other ways that can maximize access to the tech. Our ads pilots are in that spirit. We view it as a tool to bring ChatGPT and our intelligence most broadly to anyone around the world. It's an example of how we constantly need to evolve and figure out the best way to bring demand in line with what we can offer.

有道理。广告这块一直很棘手，因为 Sam 历来对广告表示过犹豫，而且我们在做广告的同时还得维持高度的信任。那么，是什么改变了？

Makes sense. The ads piece has been tricky because Sam has historically expressed reluctance about ads, and we've got to maintain a lot of trust while delivering that. So, what changed?

我认为在我在 OpenAI 的历程中，我们多次讨论过这个问题。每次提到，我们都说如果要投放广告，必须非常慎重地考虑方式。我们从去年年底开始做的第一件事，就是让公司真正参与进来，讨论如何在 ChatGPT 中做广告。原则应该是什么？如何在获得广告好处（即让任何人，无论支付能力如何，都能使用我们最先进的技术）的同时，保留 ChatGPT 的神奇之处？我对最终确定的原则非常满意。在体验方面，我们还处于非常早期的阶段，但在原则方面，我感到非常自豪。ChatGPT 的答案保持独立性非常重要，尊重用户隐私也非常重要，而且从过去十年技术的发展中可以学到很多。我喜欢我们在真正开始之前就已经确立了原则。我们的试点还非常早期。有趣的是，我焦虑地查看我们的支持工单和数据，发现关于广告最常见的问题不是如何禁用或关闭广告，而是如何投放广告？因为整个生态系统都非常兴奋，想要成为这个故事的一部分，并找到与 ChatGPT 用户沟通的方式。未来还有很多工作要做，但我非常渴望把它做好。

I think we've talked about this several times in my history at OpenAI, and every time it came up, we said if we were to do ads, we'd have to be really thoughtful about the way we do it. The first thing we did, starting end of last year, was to really engage the company on how we should approach ads in ChatGPT. What should the principles be? How do you preserve the things that are magical about ChatGPT while getting the benefits of ads, which is our ability to bring our most advanced tech to anyone regardless of their ability to pay? I really love where we ended up on the principle side. On the experience side, we're very early, but on the principle side I feel really proud. It's very important that the answers of ChatGPT be independent, respecting user privacy is very important, and there's a lot to learn from how tech has evolved over the last decade. I like that the principles are out there before we've even really gotten started. We're very early with our pilots. It's kind of interesting: I was anxiously looking at our support inbounds and data, and the most common inquiry about ads is not how to disable or turn off ads, but how do I run an ad? Because the entire ecosystem is really excited to be part of the story and to figure out a way to talk to ChatGPT users. There's a lot more to come, but I'm very eager to get this right.

换个话题，Nick，我们之前聊过一点分销和合作伙伴关系。去年有几个大合作：苹果、Reliance 与 Gemini——那是两个庞大的用户群，覆盖大量印度用户和 iOS 用户。跟我们讲讲你如何看待 ChatGPT 通过合作触达用户群，特别是这两个合作。

Switching gears, Nick, something you and I have spoken about a little bit is distribution and partnerships. There were a couple of big partnerships last year: Apple, Reliance with Gemini — those are two big user bases, a lot of India, a lot of iOS users. Tell us a little bit how you think about partnerships for ChatGPT to meet the user base, and maybe specifically on those two.

我认为合作是将两个产品结合在一起的好方式，也能让可能从未接触过 ChatGPT 的人了解它。在考虑合作时，我最关心的是用户体验，我们能否让它变得出色？因为归根结底，看看市场上的情况，你可以让用户点击任何东西，让他们点击任何产品，尤其是当它看起来像他们认识的产品时。但如果体验不够好，用户就会流失，至少不会像我们在 ChatGPT 上幸运地留住他们那样留存。因此，我对这类路径非常感兴趣，但它必须很棒。它必须对用户有益。我们很幸运拥有一个强大的品牌和许多人都认识的产品，我希望确保我们所做的任何事情都能为这一切增值。

I think partnerships are a great way to bring two products together and to expose something like ChatGPT to people who might not otherwise have encountered it. The thing I care about most when considering a partnership is what is the user experience and can we make it amazing? Because at the end of the day, when you look at what's going on in the market, you can get users to click on things, you can get them to tap any sort of product, especially if it looks like a product they recognize. But if the experience isn't truly awesome, people will churn or at least not retain in the way we've been lucky to retain them on ChatGPT. So for that reason, I'm super interested in paths like that, but it needs to be great. It needs to accrue to the user. We are very lucky to have a great brand and a recognizable product for many folks, and I want to make sure that anything we do is accretive to all that.

Nick，你是权衡取舍的大师。你现在肯定在做很多权衡。跟我们说说你正在做的一些权衡，也许是一个外界不太理解的选择。

Nick, you are a master of trade-offs. You must be making a lot of trade-offs right now. Tell us about some of the trade-offs you're making, maybe one that people don't appreciate from the outside.

确实有很多权衡，原因各不相同。我经常遇到的是，在改进现有产品用例与将带来全新用例的颠覆性技术产品化之间做权衡。因为回想 ChatGPT 的诞生，它完全是一个开放式的产品。它基本上就是围绕一个技术突破打造的用户体验。我们当时无法预知人们会发现它的所有价值，但把它推出去非常重要，因为它让我们和世界一起发现我们能做什么。然后发布之后，我们当然可以系统地改进人们实际想用它做的事情。当你身处一家公司，既拥有现有产品的惊人吸引力，又在研究方面有最令人震撼的突破时，你必须找到平衡：一方面让现有核心产品在延迟、可靠性等所有重要方面变得更好，让用户来的用例变得非常出色；另一方面还要提供对颠覆性技术的访问。我们努力找到正确的平衡，但团队很小，并不总能做对。因此，这是我必须处理的最困难的权衡之一。

There are a lot of trade-offs indeed, for different reasons. What I encounter a lot is trading off delivering on the use cases that exist in the product today and making them better versus productizing step-change technology that's going to generate a whole new set of use cases. Because when you think about how ChatGPT came to be, it was a totally open-ended product. It was basically a user experience around a technical breakthrough. And we couldn't have told you all the ways that people find it valuable, but putting it out there was really important because it allowed us and the world to discover what we can do. And then post-launch, we can obviously very systematically go and improve on the things that people actually want to use it for. When you're at a company in this moment where you both have such amazing traction with what exists today and the most mind-bending breakthroughs on the research side, the balance you have to strike is making the core product you have better today with all the things that matter — latency, reliability, making the use cases really great that people come to — while providing access to the step-change technology. We try to get the balance right, but we're a small team and we don't always get it right. For that reason, it's one of the most difficult trade-offs I have to deal with.

Nick，我想你们在这里做的最艰难的权衡之一就是那些在 ChatGPT、编码器研究之间分配的 GPU。你们是如何分配 GPU 的？

Nick, I imagine one of the hardest trade-offs you guys make here is those GPUs that are melting between ChatGPT, between Codex research. How do you allocate the GPUs?

这是个非常好的问题。等我想明白了再告诉你。开个玩笑。我们在这方面已经进步了很多。

That is a very good question. I'll let you know when I figure it out. Just kidding. We've gotten a lot better at this.

顺便说一句，我真的希望有一天能到达一个点——我还没到——我们不必面对这种权衡，因为真实用户对你无法服务的产品的需求真的很痛苦。就像，如果你只做过软件，那完全是种不寻常的动态，对吧？你只是，或者你被这种零和资源所限制。市场有这种动态。但我认为纯软件没有这种动态。

I really hope, by the way, to be at a point one day, and I've yet to reach that point, where we don't have to face this trade-off because it's really painful to have real user demand for products that you can't serve. Like, if you only ever worked in software, that's an entirely unusual dynamic, right? Where you just, you know, or you were limited by this zero-sum resource out there. The marketplaces have it. But I think pure software doesn't really have that dynamic.

是的。所以我们尝试做的一件事显然是优先考虑现有用户。我们希望提供快速可靠的产品，这是关键和基本要求。然后当你看到新能力时，那种天真的商学院做法可能是看每 GPU 的增量收入之类的。但这更像是一门艺术而非科学，因为我们经常有全新的突破性能力，完全是从零到一。深度研究就是其中之一。我们无法事先告诉你是否会有消费者对研究产品的需求，但如果你不把它产品化去发现，你永远不会知道。所以在这里我们需要深思熟虑，如何平衡那些人们会喜欢的不言而喻的事情和全新的想法。然后在研究方面，马克有他的工作是有原因的，因为他工作的一大部分是决定资助哪些研究，显然 GPU 是其中的重要部分。所以这是一个非常微妙的话题，我们一直在不断改进，但对我来说，用户永远是第一位的。

Yeah. So one thing we try to do obviously is prioritize our existing users first. We want to provide a fast, reliable product, and that is critical and table stakes. Then when you look at new capabilities, the sort of naive business school thing to do would be to probably look at revenue incremental revenue per GPU or something like that. But this is where it's more an art than a science because we often have new breakthrough capabilities that are entirely zero to one. Deep research was one of those. We couldn't have told you if there is going to be consumer demand for a research product, but if you don't productize it to find out, you will never know. So this is where we have to be a little bit thoughtful on how we balance things that are no-brainers that people are really going to love with things that are brand new ideas. Then obviously on the research side, there's a reason that Mark has the job he has because a big part of his job is figuring out what research to fund, and obviously GPU is a big part of that. So it's a very nuanced topic that we're continuously getting better at, but for me the priority is always on our users.

是的。我的另一个收获是，你看不到什么时候不会再有这个问题。

Yeah. The other takeaway that I had is you don't have line of sight to a time when you won't have that problem.

这太迷人了。因为我们显然非常幸运地遇到了越来越多的用户想要使用我们的技术，而且我们能为每个用户提供的价值也在上升，GPU 消耗与这个价值相当相关。当你只看每个用户的 token 消耗时，尤其是在企业领域——这是一个巨大的机会——你会看到很多非常消耗 GPU 的工作流，需求甚至在价格下降时还在上升。

It's been so fascinating. Because we obviously have been incredibly lucky to encounter more and more users who want to use our technology, but then the value that we're able to provide for each user is going up as well, and GPU consumption correlates pretty well with that value. And when you just look at token consumption per user, especially in the enterprise too, which is a massive opportunity, you see a lot of very GPU-hungry workflows, and demand keeps going up even as prices go down.

这是一个迷人的见解。人们过去认为人类，你知道，你不能很好地制造更多人类，需要九个月然后 19 年，但你说这实际上比 GPU 更有限。

This is a fascinating insight. People used to think that humans were, you know, you can't make more humans well, it takes nine months and then 19 years, but you're saying that's actually a more finite resource than GPUs.

是的，我的意思是，在人力方面，你可以雇佣更多的人，显然我们一直在忙于这样做，并吸引跨职能的最佳人才加入 OpenAI。在智能体的世界里，你也可以从每个人身上获得更多杠杆。你可以让你的员工在工作中非常高效，做更多事情，但 GPU 是零和的，如果你没有更多的 GPU，你真的必须想办法做出非常艰难的取舍，我讨厌为用户做出艰难的取舍。

Yeah, I mean on the human side you can hire more humans, and obviously we've been busy doing that and bringing the best talent across functions to OpenAI. In the world with agents, you can also get more leverage per human. You can make your humans very effective at their job to do more, but GPUs are zero sum, and if you don't have more GPUs, you really have to figure out how to make very hard trades, and I hate making hard trades for users.

是的。因此渴望拥有更多 GPU。但在做规划时，从最零和的权衡开始是有用的。所以我认为从 GPU 开始倒推通常是个好主意。

Yeah. Hence the desire to have more GPUs. But it's useful to start with the most zero-sum trade-off when you do your planning. So I think starting working backwards from GPUs is usually a pretty good idea.

是的。你知道，我们有所有这些外部数据源，用于用户、使用量、活动和留存率的图表，所有这些。我们没有的是每个用户随时间变化的 token 数。我敢打赌那个图表是一条漂亮的上升曲线。我认为内部数据相当不错。我们的内部员工是即将发生的事情的一个很好的指标。是的，这些图表令人难以置信。

Yeah. You know, one of the things we have all these external data sources for charts of users and usage and activity and retention, all those things. What we don't have is tokens per user over time. And I bet that chart is like a sweet line going this way. I think internal is pretty good. Our internal employees is a pretty good indicator for what's about to happen. And yes, the charts are mindboggling.

是的。迷人。好的，在进入格局之前，快速问几个关于当下的话题，那就是购物。我们刚搬进新房子。我们拍了一些照片，希望所有家具都能神奇地出现，ChatGPT 帮我们粉刷。但最近有很多关于 ChatGPT 购物的更新。跟我们说说吧。你在想什么？

Yeah. Fascinating. Okay, a couple of quick ones on the present before we go into the landscape, which is shopping. We just moved into a new house. We took some photos and we were hoping that all our furniture would magically appear that ChatGPT helped us paint. But a lot of recent updates on ChatGPT shopping. Tell us about it. What are you thinking?

是的。关于 ChatGPT 作为购物助手。购物是今天 ChatGPT 中自然存在的用例之一，而且效果不错。你可以向 ChatGPT 询问任何你计划购买的东西，并获得相当出色的建议。但这也是一个例子，说明今天聊天中的体验并不是你想要的完美体验，因为购物非常视觉化。所以你实际上想看到产品和图片，能够比较和对比，而不仅仅是阅读大段文字。人们关心来源，比如，我在哪里可以了解更多关于某个产品的信息等等。所以有很多工作要做，让这种发现变得非常好，让人们能够使用聊天作为助手找到合适的产品购买。这就是我们的重点：让它变得非常好，并且以一种对我们的零售合作伙伴也有效的方式做到这一点。因为正如我之前提到的，生态系统中有巨大的兴趣成为 ChatGPT 旅程的一部分。而做好发现部分，是迄今为止这里最有前景的重点。

Yeah. On shopping as ChatGPT as a shopping assistant. Shopping is one of those use cases that exists organically in ChatGPT today, and they work. You can ask ChatGPT about any purchase you might be planning and get pretty excellent advice. But it's also one of those cases where the experience that exists in chat today is not the perfect experience that you would want because shopping is very visual, for example. So you're going to want to actually see products and images and be able to compare and contrast, not just read walls of text. People care about the sources of, you know, where can I learn more about a given product, etc. So there's a lot of work to do to make this discovery really really good and allowing people to use chat as an assistant to find the right product to buy. And that's where our focus lies: making that really great and making that really great in a way that works for our retail partners as well. Because as I mentioned earlier, there's huge appetite from the ecosystem to be part of the ChatGPT journey. And nailing the discovery piece has been the most promising focus here to date.

Nick，在 ChatGPT 上你一定看到了广泛的信息。你一定看到了人们使用 ChatGPT 的各种用例。告诉我们一些关于世界低估了 ChatGPT 什么，你可能感到惊讶或听众可能感到惊讶的事情。

Nick, on ChatGPT you must see a breadth of information. You must see a breadth of use cases that people are doing with ChatGPT. Tell us something about what the world underestimates about ChatGPT that you have maybe been surprised by or a listener might be surprised by.

在过去一年左右的时间里，人们对 ChatGPT 的看法发生了真正的变化，它越来越像人们真正的思想伙伴。它不仅仅是一个回答你问题的东西，而是一个你可以用来作为陪练伙伴，真正与之一起思考的东西。这出现在各种领域，从生活建议开始，如果你有感情问题，你实际上可以从 ChatGPT 那里获得很多价值，它帮助你思考如何处理以及如何与你的伴侣谈论它。

There's been a real change in the way that people think of ChatGPT over the last year or so, where it's increasingly like a true thought partner to people. It's not just a thing that answers your question, but it's a thing that you can use as a sparring partner that you can actually think things through with. And that shows up in all kinds of domains, ranging from life advice where, if you have a relationship problem, you can actually get a lot of value from ChatGPT helping you think through how to handle it and how to talk to your partner about it.

一直到工作场景，你在做分析、琢磨如何构建框架或尝试构建某样东西时，ChatGPT 真的就像一个第二大脑一样出现。

All the way to a work setting where you're working on an analysis or trying to figure out how to frame something or trying to build something, and ChatGPT really shows up as a second brain of sorts.

我认为这在人们心中占据的心智模型上是质的区别。从使用模式和使用案例中就能看出来。我觉得我们越是在主动性这类事情上做得好——我们之前聊任务时提到过——它就越会像工作场所的队友和家里的超级助手。我认为这将深刻改变人们的使用场景。

I think that's qualitatively different in terms of the mental model it occupies with people. And you see it in the usage patterns and the use cases that exist. I think the more we nail things like proactivity, which we talked about earlier in task, the more it's going to feel like a teammate in the workplace and like a super assistant at home. And I think that's going to meaningfully change the use cases that people come for.

是的。你知道吗，我用 ChatGPT 做过的最高风险的事是——我们刚有了宝宝。凌晨 3 点宝宝哭的时候，我就问：“ChatGPT，怎么回事？”

Yeah. You know, the most high-stakes thing I do with ChatGPT is we have a new baby. And when the baby's crying at 3 in the morning, 'ChatGPT, what's going on?'

首先，恭喜。其次，我生活中所有父母都跟我提过这一点。ChatGPT 作为思考伙伴已经变得不可或缺。这很合理，对吧？如果你有一个非常具体的场景，或者你觉得这个场景很个人化，ChatGPT 真的能帮上忙，帮你建立信心。我觉得这非常赋能。新手父母并不总是对正确做法充满信心，如果 ChatGPT 能让你感到自己有掌控力和主动权，我认为这非常有价值。

First of all, congrats. Second of all, I've heard this from all parents in my life. ChatGPT has become indispensable as a thought partner. And it makes sense, right? If you have a really specific scenario or you think it's a specific scenario to you, ChatGPT really comes through and can help you build confidence. I think that's such an empowering thing. New parents aren't always the most confident about what is the right thing to do, and if ChatGPT can make you feel like you have agency and control, I think it's really valuable.

是的，太重要了。感谢你创造了 ChatGPT。它真的让我每天多睡一个小时。

Yeah, it's huge. Well, thank you for making ChatGPT. It's literally getting me an extra hour of sleep every day.

这是集体的功劳。但这是个很棒的指标。它应该成为北极星指标：增加的睡眠小时数。

It took a village. But that is a great metric. That should be the north star metric: incremental hours of sleep.

这个指标真好。

That's a great one.

增加的睡眠小时数，增加的快乐小时数。

Incremental hours of sleep, incremental hours of joy.

没错。

There you go.

我是说，你在开玩笑，但我们经常讨论这个，因为从精神层面来说，这非常接近我们希望达成的目标：帮助你实现你认为的自我实现，无论是睡眠、快乐还是任何其他目标。

I mean, you joke, but we talk about this a lot because spiritually that is pretty close to what we hope we can do: help you reach whatever you consider self-actualization, whether that's sleep or joy or any other goal you might have.

是的。感谢这个集体。

Yeah. Well, thank you to the village.

我们换个话题，聊聊行业格局。这个领域有很多事情在发生。你会如何向外界描述 ChatGPT 的差异化？市面上有很多不同的产品。

We're going to switch gears and talk about the landscape. There's a lot going on in the field. How would you frame ChatGPT's differentiation to people out there? There's a lot of different products out there.

你看，现在是历史上消费者享受科技最好的时代。确实如此。因为你有选择，竞争也很激烈。我认为这很美好，实际上对我们也有好处，因为如果你要预演为什么像 OpenAI 这样的公司无法实现其使命，那很可能是因为专注问题——当你接近 AGI 时，涌现的机会实在太多。竞争和选择的存在也迫使我们专注于客户，专注于真正重要的事情，而这些不总是最炫酷的东西。有时是延迟、可靠性、用户体验的质量。所以我认为这是件好事。

Look, it's the best time in history to be a consumer of technology. It is indeed. Because you have options and the competition is intense. I think that's beautiful and it's actually good for us too because if you were to premortem why a company like OpenAI does not achieve its mission, it's probably focus because of the sheer number of opportunities that become possible when you approach AGI. Having competition and options out there forces us to focus on our customers too, and on the things that really matter, which aren't always the most flashy things. Sometimes it's latency, reliability, the quality of the user experience. So I think it's a really good thing.

是的。

Yeah.

我认为 ChatGPT 最大的差异化在于背后的团队。因为我们不是静止的。我们构建的任何东西都会被复制，有时是精工细作的方式，有时是打勾清单式的方式。对我们来说，引领品类进化、构建我们一直想象的超级助手非常重要。我有信心能比被复制的速度更快地实现这一点，是因为我们拥有一支出色的团队，涵盖研究、工程、设计以及打造卓越产品所需的所有职能。我们独特的能力在于将这些职能整合在一起，在那一刻构建出处于有用与可能交汇点的东西。所以我的最佳答案是：我们不断向前推进，并希望拓展人们对这款产品的认知。

I think the biggest differentiation of ChatGPT is the team behind it. Because we're not static. Anything we build will get copied, sometimes in ways that are high craft, sometimes in ways that are sort of checkboxes. It's really important to us that we evolve the category and build the super assistant we've always imagined. The reason I have confidence that that's possible at a speed that outpaces being copied is that we have an amazing team across research, engineering, design, and all the different functions it takes to make something amazing. Our unique ability has been to bring those functions together to build something that is at the intersection of useful and possible at that moment. So my best answer is we keep pushing forward and we hope to expand what people think of this product as.

去年冬天我们经历了所谓的“红色代码”。谷歌推出了一个很棒的模型。有很多讨论。马克·贝尼奥夫非常高调地转向 Gemini，而我们推迟了广告、健康助手和购物等功能。基本上暂停了一切，专注于让 ChatGPT 变得更好。跟我们聊聊那个时刻吧，包括是什么导致了它，以及当时发生了什么。

Last winter we had what was called Code Red. Google had a great model. There was a lot of talk about it. Mark Benioff switching very vocally to Gemini, and us delaying ads, health agents, and shopping. Basically hit pause on everything, making ChatGPT better. Talk to us about that moment, both about what led to that and what was happening in that moment.

是的。首先，红色代码是我们用来创造专注力的工具。可以想象，在 OpenAI 这样的地方工作，有太多不同的事情在同时进行。这是一个研究实验室；我们在追求许多不同的想法。有些时刻我们希望整个公司团结起来，跨越边界解决一个问题，无论你的项目是什么。去年年底，我们就遇到了这样一个时刻，觉得需要为用户挺身而出。我们需要专注于基础：可靠性、性能、与模型对话的体验、让个性化变得非常出色。所有这些用户关心的要素。我很喜欢它，因为这是一个机会，让我和平时不常合作的同事一起把产品做得更好。我们刚刚结束了红色代码——我们知道会结束——推出了 GPT-4o，这对日常用户来说是一个很棒的模型，对话体验很好；还有 GPT-4，如果你要做真正的知识工作，它是一个得力工具。毫无疑问，只要我们想创造专注力，就会继续使用红色代码这个工具。但我很兴奋，因为我认为 ChatGPT 处于一个很好的位置。

Yeah. First off, Code Reds are a tool we use to create focus. As you can imagine, when you're in a place like OpenAI, and this is what makes it special to work here, there are so many different things going on. It's a research lab; we are pursuing many different ideas. There have been moments where we wanted the company to come together to solve a problem across boundaries, no matter what your project might be. At the end of last year, we had one of those moments where we felt we need to show up for our users. We need to focus on the basics: reliability, performance, the way talking to the model feels, making personalization really great. All these elements that our users care about. I loved it because it was an opportunity to work with a bunch of folks I don't normally get to work with on making the product great. We just exited the Code Red, which we knew we would, with the launch of GPT-4o, which is a great model for the everyday user, great to talk to, and GPT-4, which is a workhorse if you're trying to do real knowledge work. Undoubtedly we're going to continue to use the tool of a Code Red whenever we want to create focus. But I'm excited because I think ChatGPT is in a great spot.

所以红色代码现在结束了？

So Code Red is over now?

是的。

That's correct.

它不是新常态？

It's not the new normal?

它不是新常态。我们希望它是一件特别的事情，但它是一个工具，我猜我们会继续使用。

It's not the new normal. We want it to be a special thing, but it is a tool I suspect we will continue to use.

太好了。也许具体来说，红色代码如何改变了 ChatGPT，或者运营方式，或者团队的运作方式？

That's great. Maybe tangibly, how did Code Red change ChatGPT or the ops or how the team operates?

我试图在团队中培养的就是专注力。

The thing I try to foster with the team is focus.

嗯，我们确实比六个月前更专注了，专注于我们真正想做好的事情，其中一些是非常后台的，比如延迟、可靠性这类事情。

Um so we certainly more focused than we were six months ago on the things we really want to nail and some of those things are very behind the scenes like latency reliability those kind of things.

好的。其中一些是经过深思熟虑的努力，比如将 ChatGPT 整合到超级助手中。所以专注是主要的持久成果。嗯，正如你想象的那样，当这个领域有这么多事情发生时，有时很难保持专注，但这就是艰巨的工作，你之前问过我关于权衡的问题。让团队专注于对用户真正重要的事情肯定是其中之一。这总是值得的。

Okay. And some of those things are like very considered efforts like you involving ChatGPT into the super assistant and so focus is the main lasting artifact. Um and as you imagine it's hard to stay focused sometimes when there's so much going on in the space but that's the hard job and you asked me about trade-offs earlier. Getting the team to focus on the things that really matter to users is certainly one of them. That's always worth it.

是的。你知道吗，我问你这个问题时，心里想的是所有其他正在竞技场上的创始人。只是想提醒一下，嘿，Code Red 是你的工具。我们过去称之为“战时平衡”也是一种工具。

Yeah. You know, in the back of my mind as I ask you that question is all the other founders that are in the arena right now. And just a reminder that hey, Code Red is a tool for you. Wartime at balance as we used to call it is a tool.

是的，我认为每家公司做事的方式都不同。但我认为拥有有意义的术语非常有价值，它能向人们发出信号：放下其他事情是可以的，一起专注于这件事也是可以的，即使那不是你原来的工作。

Yeah, I think you know every company does it differently in terms of how you get stuff done. But I think it's really valuable to have terminology that means something, that signals to people it's okay to drop your other stuff and it's okay to focus on this thing together even if that wasn't your original job.

是的。

Yeah.

所以我认为这在 OpenAI 这样的地方效果很好。但我想初创公司也会有类似的做法。

So I think it worked really well at a place like OpenAI. But I imagine startups would have an equivalent.

是的。你知道，我们团队中让所有人充满想象的一件事是 Peter 在 OpenClaw 所做的工作。将所有工具整合在一起非常强大。显然，Peter 是一位出色的构建者。恭喜你让 Peter 加入团队。跟我们说说 Peter 在做什么，以及 ChatGPT 上的数十亿用户什么时候能看到一些成果？

Yeah. You know, one of the things that caught everybody's imagination on our team was what Peter was doing at OpenClaw. Incredibly potent to put all the tools together. Obviously, Peter's a great builder. Congrats on bringing on Peter to the team. Tell us a little bit about what Peter's working on and when might the billions on ChatGPT have something to see there.

嗯，首先，我非常高兴 Peter 能来这里。我很高兴团队里又多了一位德语使用者。他是奥地利人，我是德国人。所以我们互相说“Guten Morgen”。但 OpenClaw 非常鼓舞人心，因为它在很多方面实现了一个我们曾以不同形式拥有的愿景：一种完全具身的 AI，存在于不同的用户界面中，可以为你做事，拥有状态，交互模式更像与人交谈。因为你可以用非常自然的方式互动，来回发送很多文本，而且非常简洁。所以 OpenClaw 有很多元素，我认为对整个行业的人来说都很有启发性。但我非常兴奋能向 Peter 学习，把他带入公司，看看我们能一起做什么。所以未来还有很多值得期待。

Well, first of all, I'm very excited for Peter to be here. I was excited to have another German speaker in the house. He's Austrian. I'm German. So we were exchanging Guten Morgans. But the OpenClaw is so inspiring because it brought to life in many ways a vision that we had in different forms around this kind of AI that is fully embodied, that exists across different UIs, that can do stuff for you, that has state, that has an interaction pattern that feels a little bit more like talking to a human. Because you can interact in a very natural way where you can send many texts back and forth and it's very curt. So there's a lot of elements of OpenClaw that I think were very clarifying to folks across the industry. But I'm super excited to just learn from Peter and bring him into the company and figure out what we can do together. So there's a lot more to come.

好了，现在进入最有趣的环节：快速问答。准备好了吗？

All right. So now on to the most fun section. Rapid fire. You ready?

当然。

Sure.

我们先从我最喜欢的游戏开始，叫“Long Short”。选一个你喜欢的、非常看好的想法、初创公司、业务或产品。

We'll start with my favorite game, which is Long Short. Pick an idea, a startup, a business, a product that you love, you think you're very bullish on.

是的，如果今天我要创办一家公司，我对那些深入企业、亲力亲为、用 AI 有效提供专业服务的公司非常兴奋，因为我们已经饱和了所有电子邮件，你需要贴近问题。所以我在关注的就是这类公司。

Yeah, if I were starting a company today, I'm really excited about these companies that are going into companies and getting extremely hands-on and doing effectively professional services with AI, because we've saturated all the emails and you need to get proximate to the problems. So it's those companies that I'm paying attention to.

有意思。所以这是一个例子，就像，嘿，你要么收购一家有规模且运转良好的运营公司，要么进入其中。

Fascinating. So this is an example. This would be like, hey, you're going and either acquiring or going inside an operating firm that has scale and a humming engine.

正是。

Exactly.

然后让它变得更高效。

And making that a more efficient engine.

是的。或者就像，你为有真正难题的客户做合同，然后实际介入并承诺解决问题。结果导向。

Yeah. Or just like, you know, you're doing contracts for customers that have really hard problems and you're actually going in and committing to solving the problem. Outcomes.

是的。

Yeah.

因为我认为我们在数学和编码上取得了很大进展，但在许多其他领域却没有，原因在于这些领域是我们这些在实验室工作的人所贴近的。而还有其他各种领域我们并不那么贴近。如果你贴近了，我认为你可以构建出变革性的东西。而且我认为现在这更重要，正是因为简单的问题已经被解决了，显而易见的问题已经被模型解决了。

Because there's a reason I think that we made so much progress on math and coding but not on many other domains, because those are domains we are proximate to, as people who work in labs. And there's all kinds of other domains that we are not as proximate to. And if you get proximate, I think you can build something transformative. And I think this is more important now precisely because the easy problems have been solved, the obvious problems have been solved by the models.

该表扬就表扬，我认为 NotebookLM 很棒，有差异化，能帮我学习新东西。我觉得它很棒。

Credit where credit is due, I think NotebookLM is awesome and differentiated and helps me learn new stuff. I think it's great.

它太好了。

It's so good.

是的。

Yeah.

它太好了。

It's so good.

我认为这是一个例子，说明你可以创新，可以构建完全不同的东西。太棒了。

I think this is the example of you can innovate and you can build something totally different. It's awesome.

是的，是的，是的。它太好了。尤其对于一些更技术性的学习，我发现它是一种非常容易上手的方式，可以彻底学习。而且真的很酷。我觉得 AI 一个被低估的能力就是能把事物转换成不同的媒介。

Yeah. Yeah. Yeah. It's so good. Particularly for some more technical learning, I found it to be a very approachable way to totally learn. And it's really cool. I feel like an underrated capability of AI is to just transform things into a different medium.

嗯。

Mhm.

而且我认为这对学习非常重要。我们刚刚推出了这些动态数学模块，让你能在 ChatGPT 内直观地理解数学。学习显然也是我们的一个重要用例。我认为能够将事物从文本转换为视觉，很快从视觉转换为视频，以及所有这些不同的媒介，这太神奇了，因为人们处理信息的方式各不相同。有些人是听觉学习者，有些人是视觉学习者，有些人喜欢阅读。所以我认为这真的很神奇，也是一个很好的角度。

And I think that's so important for learning. We just launched these dynamic math blocks which allow you to visually understand math inside ChatGPT. Learning is obviously a big use case for us too. And I think just being able to transform things from text to visual, soon from visual to video, and all these different media is amazing because people have such different ways of processing information. Some people are auditory learners, some people are visual, some people like reading. So I think that's really magical and a great angle to take.

是的。太棒了，太棒了，太棒了。你知道，我经常思考的一件事是教育，以及现在在校孩子的教育。世界变化如此之快。我不确定我们的教育系统也在同样快速地变化。

Yeah. Amazing. Amazing. Amazing. You know, one of the things I think about a lot is education and education for kids now in school. The world's changing so fast. I'm not sure our education system is changing that fast.

是的。

Yeah.

你对现在在校的学生有什么建议？他们可能必须比周围系统适应得更快。

What advice would you have for students who are in school now? Who might have to adapt faster than the system around them might adapt?

这是一个非常好的问题，也是我自己思考了很多的事情。我认为这个时代最重要的永久技能是好奇心。因为如果机器能回答你所有的问题，你最好有好的问题。

It's a really good question and something that I've thought a lot about myself. And I think the most important perma skill in this era is curiosity, I think. Because if the machine can answer all your questions, you better have good questions.

我认为，提出好问题的唯一方法，就是从很小的时候开始，并贯穿一生，去追求那些你真正热爱的事情。

And the only way to have good questions, I think, is to pursue the things you were actually excited about from an early age and throughout your entire life.

是的。

Yeah.

我回想这些，是因为我之所以能在这里做这些事情，唯一的原因就是在面试过程中被“技术狙击”了，觉得这太酷了。所以无论你在做什么，保持好奇并学会持续好奇，都是一项重要的技能。我相信，如果你培养这种技能，你就会知道如何适应不断变化的工具、AI 和工作环境。这就是我的建议。

And I reflect on this because the only reason I'm here and working on this stuff is because I thought it was neat when I got nerd sniped in the interview process, right? And it was like, that's right. This is so cool. And so no matter what you're doing, I think that's an important skill is to be curious and learn to stay curious. And I think I'm confident that if you foster that skill, you will know how to adapt to an evolving landscape of tools and AIs and jobs. So that would be my advice.

是的，好奇心一直是最稀缺的尺度。我们的朋友 Bill Gurley 在他的书《Running Down a Dream》中写过这一点。但好奇心，我得去看看。

Yeah, curiosity has always been the poorest scale. Our friend Bill Gurley wrote about it in his book Running Down a Dream. But curiosity, have to check that out.

嗯。

Yeah.

随着 AI 越来越好，AGI 到来，什么工作会变得更有价值，而不是更少？

What is a job that gets more valuable, not less, as AI gets better, as AGI arrives?

嗯，也许简单的答案是成为创业者。因为就实现自我想法而言，这是有史以来最好的创业时机。

Well, I think maybe the easy answer is being an entrepreneur. Because it's the best time to build ever in terms of being able to self-actualize your idea.

是的。

Yeah.

但也许一个不那么明显的答案是，我认为写作实际上非常重要。不是因为 AI 不会写——AI 会像在其他领域一样变得非常擅长写作——而是因为写作技能迫使你非常清晰地表达自己的想法。尽管提示工程显然会消失，而且在很大程度上已经消失了，但向机器表达你想要什么，需要你成为一个相当好的、非常精确的写作者。所以我认为，任何涉及清晰写作（因此也是清晰思考）的职业都前景良好。

But maybe one that is non-obvious is I think writing actually is very important. And it's not because the AI can't write — AI will become amazing at writing just like any other domain — but because I think the skill of writing forces you to be very clear of what you have to say. And even though prompt engineering is obviously going to go away and has gone away to much extent, the idea of expressing what you want to a machine requires you to be a pretty good writer and a very precise writer. So I would say that any profession that involves very clear writing and therefore thinking is well set up.

是的，100%。老实说，这就是关于 slot 的全部，对吧？还有很多——这是另一回事。我认为对高质量、可信、权威的内容将会有永久的需求，像 CHAP 这样的工具可以帮助你发现这些内容。

Yeah, 100%. Honestly, I mean this is the whole thing about slot, right? There's so much — that's the other thing. I think there's going to be a permanent need for high quality, trusted, authoritative content and tools like CHAP can help you discover that content.

是的。

Yeah.

但我认为对精彩内容的需求也将持续存在。

But I think the need for amazing content is also here to stay.

最后一个问题，你的 AGI 时刻是什么？你什么时候感受到了它？

Final question, what has been your AGI feel the AGI moment? When did you feel it?

老实说，我有很多这样的时刻，而且肯定没有停止。我加入 OpenAI 几周后，GPT-4 完成了训练。我记得试用它时，它完全没有给我留下印象，那周其他人也没有，因为它基本上不工作。这是因为我们还没弄清楚如何进行后训练。看到它从“等等，这真的行吗？还是 GPT-3 就这样了？”变成“哇，这完全是阶跃变化”，而当时对 AI 几乎一无所知的我看来，这只是一些调整或最后的收尾工作，这让我深感谦卑。因为你可能会意识到，我们看起来可能并不接近真正强大有用的 AI，但我们很可能已经接近了。

I've had so many honestly. And it's definitely not stopped. A few weeks or so after I joined OpenAI, GPT-4 had finished training. And I remember trying it out and it actually didn't impress me at all, nor anyone else that week, because it kind of didn't work. And it's because we hadn't figured out how to post-train it. And I think seeing it go from kind of 'wait, is this really a thing or was GPT-3 kind of it' to 'wow, actually this is an entire step change' with what felt to me at the time — who didn't understand much about AI at all — as just some tweaks or a little bit of final stretch work was profoundly humbling. Because you can realize that it might not look like we are close to really powerful useful AI, but we probably are.

嗯。

Um.

然后真正让我感觉像 AGI 的时刻——GPT-4 做了两件事。一是它能写诗，我原本认为 AI 模型不可能写诗，这在根本上、哲学上都不在范围内。二是它能生成实际可运行、可编译的代码。接下来我盯着天花板惊叹的时刻，是当我意识到 GPT-4 可以模拟整个计算机终端——一个完整的计算机，带命令等等。我想，“等等，这怎么可能被注入到一个语言模型中？”从那以后，这样的时刻还有很多。比如推理就是一个时刻。有一次，我和 Mark 在全公司面前演示推理。那时我们还在努力寻找足够难的用例，让推理能发挥作用。现在我们早已过了那个阶段。但当时，我们得在大家面前做一个谜题。其中一个让我完全感受到 AGI 的时刻是，演示进行到一半时，大家都笑了，我心想，“等等，有什么好笑的？”然后我盯着屏幕，因为我们正在展示模型流出的思维链。模型骂了一句，说：“哦，该死。我必须调整，因为我意识到我在谜题中犯了一个错误。”它这样做的事实，尤其是它完全是从强化学习过程中涌现出来的，彻底震撼了我，让我对这些模型还能做什么感到非常谦卑。所以那是其中一个时刻。

And then the moment that really felt like AGI to me — there were two things that GPT-4 did. One is it could do poetry, and I didn't think it was possible for an AI model to do poetry. It just kind of fundamentally philosophically didn't feel like in scope. And then the other one was it could produce code that actually worked and compiled. And then my next moment where I stared at the ceiling just in awe was when I realized GPT-4 could just simulate an entire computer terminal — a full computer with commands etc. And I'm like, 'Wait, how would this be imbued in a language model?' And there've been so many moments since then, honestly. Like reasoning was a moment. One of the moments was when I think Mark and I were giving a demo of reasoning in front of the whole company. And this was a moment where we were still trying to find use cases that were hard enough for the AI for the reasoning to make a difference. We're way past that point now. But at the time, I think we were having to do a puzzle in front of everyone. And one of the moments that made me totally feel the AGI is like we were in the middle of the demo and everyone started laughing, and I was like, 'Wait, what is funny?' And then I stared at the screen because we were showing this chain of thought as it was streaming out of the model. And the model swore and said like, 'Oh, damn it. I have to adjust because I realized I had made a mistake in the puzzle.' And the fact that it did that, but in particular the fact that it did that in a way that was entirely emergent from the RL process, completely blew my mind and made me feel quite humble about what else these models might be able to do. So that was one of those moments.

是的。

Yeah.

然后最近，看着人们使用 Codex——比如看着人们带着打开的电脑走来走去，因为他们不想让任务结束。看着从未写过代码的人创造东西，把想法变成现实——那感觉就像 AGI 时刻。所以老实说，对我来说，这种感觉在加速，而且完全没有消退。显然每个人有不同的时刻，但这些是我的一些时刻。

And then most recently, watching people use Codex — like watching people walk around with their computer open because they don't want the task to end. Watching people who have never coded in their life make stuff and bring ideas to life — that feels like an AGI moment. So honestly, it's just accelerating for me and it doesn't wear off at all. And everyone has a different thing obviously, but those were some of mine.

是的。你知道，10 年前有一个叫 Kite 的产品，我不知道你是否记得，它是给软件工程师用的，像一个 AI 编程产品。那是我感受到对个人 AI 的渴望的时候。然后你知道，10 年什么都没发生，然后在过去 10 个月里，一切都发生了。

Yeah. You know, it's 10 years ago there was a product called Kite. I don't know if you remember it, it was for software engineers. It was like an AI coding product. That's when I felt the hunger for personal AI. And you know, nothing happened for 10 years, and then everything happened in the last 10 months.

时机问题真的很难，因为我认为，就你将拥有的产品形态而言，预测最终结果其实相当可能。但要确切知道什么时候发生，对我来说，很难对“最终”和“三个月内”之间的任何事情做出断言。

The timing thing is really hard because it's actually quite possible to predict where things will end up, I think, in terms of the kind of product form factors you're going to have. But to know when it happens, it's really hard for me to make statements on anything between eventually and in three months.

是的，因为所有的不确定性——嗯，那是一个足够紧的时间窗口，你知道，现在和三个月之间是一个足够紧的时间窗口。

Yeah, because of all the ambiguity around — well, that's a tight enough window, you know, now and three months is a tight enough window.

三个月还算可以。我尽量大致坚持三个月计划。不过，我的团队可能会告诉我我们没有，但我尽力了。但是的，三个月和最终之间的任何时间点都很难。

Three months is pretty okay. Try to stick to the three-month plan more or less. Though, my team would probably tell me we don't, but I try. But yeah, anything in between three months and eventually is difficult.

是的。是的。好的，谢谢你来做这个。你有很多事情要做。这真是一次享受。

Yeah. Yeah. Yeah. Well, thanks for doing it. You've got a lot going on. This was a total treat.

我们非常期待看到你为我们发布的所有优秀产品。我们愿意提供任何帮助，请告诉我们。

We're so excited to see all the great products you release for us. We can do anything to be of help, let us know.

太棒了。非常感谢。谢谢邀请我。

Awesome. Thanks very much. Thanks for having me.

当然，伙计。这很有趣。提醒大家，这只是我们的观点，不是投资建议。

Of course, man. This was fun. As a reminder to everybody, just our opinions, not investment advice.