OpenAI 联合创始人格雷格·布罗克曼：在准备好之前就开始用 AI 构建

我们将看到创业的巨大复兴。我觉得它正在到来。我认为它会在未来一到两年内真正开始。

We're going to see this massive renaissance of entrepreneurship. Like I think it is coming. I think it's really going to kick in over the next year to two years.

这是 Greg，OpenAI 的联合创始人，ChatGPT 背后的公司，每周有超过十亿用户使用。

This is Greg, co-founder of OpenAI, the company behind ChatGPT, which more than a billion users use every week.

我很擅长写软件。我现在不写软件了。我指挥 AI 来写软件。我提供大量的反馈和指导。

I'm pretty good at writing software. I don't write software anymore. I direct an AI for it to write software. I provide lots of feedback and guidance.

当人们构建软件时，你如何制定策略来构建不会被新模型淘汰的东西？我正要给你展示我最近构建的东西。嗯，我对 AI 的野心够大吗？

When people are building software, how do you strategize around building something that's not going to be made obsolete with the new models? I'm just about to show you something that I've built recently. Um, am I ambitious enough with AI?

嗯，我认为对于“野心够大”这个问题的答案永远是不够，因为我们总能做更多。

Well, I think the answer to being ambitious enough will always be no because there's always more we can do.

对于愿意提升智能体水平的人，他们有一个小时。这个周末他们从哪里开始？

For someone who is willing to step up their agent game and they have 1 hour. Where do they start this weekend?

我想说首先，

I'd say first of all,

本视频由 HubSpot 赞助。

This video is sponsored by HubSpot.

我想谈谈所有实际的事情。我最近和诺贝尔经济学奖得主交谈，一个主题不断出现：我们有如此强大的 AI 模型，但我们缺乏应用。你认为你今年推出的东西将如何帮助改变这一点，并帮助人们意识到 AI 的力量？

I want to talk all things practical. I was recently talking to Nobel Prize laureate in economics and the one theme kept coming up that we have such capable AI models but we lack applications. What do you think that you're launching this year will help change that and help people realize what's the power of AI?

嗯，首先看到有多少人使用 ChatGPT 非常有趣。你知道，我们有 12 亿周活跃用户。每周有大约 3 亿人使用聊天进行健康查询，对吧？来帮助他们个人生活。同时，这些用例中有多少只是触及表面？我们在指标中看到，如果人们将聊天用于三个不同的用例，即你意识到聊天可以为你做的三件不同的事情，你就会成为高级用户。你会留存，你会继续使用聊天，并且你想更多地使用它。但达到这三个用例，是一个如此艰难的挑战。

Well, it's been very interesting to see first of all how many people use ChatGPT. You know, we have 1.2 billion weekly active users. We have like 300 million people who use chat every single week for health queries, right? to help them in their personal lives. At the same time, how many of those use cases just scratch the surface? And we see this in the metrics that if people use chat for three different use cases, so three different things that you realize chat could do for you, you become a power user. You become you retain you you keep using chat and you want to use it more. But getting to those three use cases, it's such a hard challenge.

在某些方面，这有点道理，因为好吧，这个 AI 可以做任何事情，但它能做的一件事是什么，对吧？就像有时候

And in some ways, it kind of makes sense because okay, so here's this AI that can do anything, but what's one thing it can do, right? It's like sometimes

神奇的事情是什么？

what's the magic thing?

正是如此。对吧。所以真正找到那种对你说话的神奇之处，但同时，我认为它展示了这个巨大的机会，这个巨大的空间，几乎是一个缺口，你正在与 AI 交谈，AI 知道它能做什么，它可以向你建议，嘿，你让我为你做这件事。实际上，我可以做更多，对吧？这里有一个想法。我可以做一个完整的演示文稿或电子表格。所以这实际上是我们一直在探索的事情。但我觉得这里有一些东西，关于你希望从 AI 中得到的是简单性，对吧？你不想要一个有按钮、滑块和模型选择器的界面。就像没有这些。那不是我们被承诺的 AI。

Exactly. Right. And so really finding that magic that speaks to you, but also at the same time, I think it shows this massive opportunity, this massive space that's almost a gap where you have you're talking to an AI and the AI knows what it's capable of and it can suggest to you, hey, you asked me to do this thing for you. Actually, I could do even more, right? Here's an idea. I can make a whole presentation a spreadsheet. So that's actually been something we've been exploring. But I feel like there's something here about the fact that what you want out of AI is simplicity, right? You don't want an interface that has buttons and sliders and model pickers. Like none of that. That's not the AI we were promised.

你想要一个可以交谈的东西，可以解决问题，可以帮助你，可以帮助解决目标，但实际上甚至可以主动，可以对你说：“嘿，你收件箱里有这封邮件，你已经忽略这个人好几天了，实际上我做了一些研究，发现这是问题的正确答案，我已经起草了一个答案。你批准吗？”

You want something you can talk to that can solve problems, that can help you, can help solve goals, but that can actually even be proactive, that can say to you, "Hey, you got this email that's in your inbox that you've been ignoring this person for the past couple days and that actually I did some research and it turns out that here's like the right answer to the question and I've drafted an answer. Do you approve?"

顺便说一下，我刚刚提到的用例，不是理论上的。实际上，就在我们开始这个播客之前，我正好遇到了这个用例。所以我认为我们现在处于一个阶段，AI 实际上能够更加主动，并消除很多负担，比如弄清楚这个系统能做什么，真正帮助你，然后帮助你实现你想做的任何事情。

And by the way, the use case I just mentioned, it's not theoretical with dots. I actually got that exactly uh this this this use case happening to me right before we we started this podcast. And so I think we're at a point now where AI is actually able to be much more proactive and to remove so much of the burden of figuring out like what is this system capable of and to really help you and then help you be able to achieve whatever it is that you want to do.

有没有其他一年前无法想象、现在可能、人们应该尝试的用例？嗯，我认为有很多，对吧？有一整个谱系，从……我认为我一直推荐去发现的方式是你必须玩这个系统，对吧？你必须看到它能做什么。所以，当涉及到编码时，像这些模型，它们现在写软件比基本上我认识的任何人都好，

Are there any other use cases that were unimaginable a year ago that are possible now and that people should be trying out? Well, I think so many, right? There's a whole spectrum of them from and I think the way that I always recommend going to to discovery is you got to play with the system, right? You got to see what it's capable of. And so, when it comes to coding, like these models, they're now better at writing software than basically anyone I know,

对吧？我很擅长写软件。就像，我不再写软件了。我指挥 AI 来写软件。我提供大量的反馈和指导。所以，总是很有趣，比如做一个小游戏，然后快速看到，好吧，交付那个花了多长时间？然后要求更复杂的东西。我认为我们现在处于一个阶段，Astra 在 Blender 中如此出色，在 3D 建模中如此出色。所以，要求它生成照片的 3D 模型之类的事情很有趣。你可以看到它使用像 Microsoft Paint 这样的软件来勾勒照片。但你也可以要求更真实的东西。比如我知道一个朋友，他正在设计或想建一个客房，使用 Astra 实际生成 CAD 图纸，然后把这些带给承包商，现在正在建造。

right? I'm pretty good at writing software. Like, I don't write software anymore. I direct an AI for it to write software. I provide lots of feedback and guidance. So, it's always fun to like make a little game and just kind of get this this quick hit on just seeing like, okay, how long did it take to deliver on that? and then ask for something more complex. I I think that we're at a point now where Astra is so good at Blender, so good at 3D modeling. And so it's kind of fun to ask it to produce 3D models of a photo and those kinds of things. You can see it using software like using Microsoft Paint to to sketch out a photo. But you can ask for more real things, too. Like I know for example a friend who was designing or wanted to have a like guest house built and used Astra to actually produce CAD plans and then brought those to a contractor and now it's being built.

所以，这是一件非常赋予能力的事情。通常你必须雇别人，比如我不知道如何制作 CAD 图纸，但现在我们口袋里有这种智能，可以为我们做这些事情。对我来说，我真正喜欢做的事情之一是尝试在我生活的不同维度推动 AI。所以我认为语音模式使 AI 如此易于访问，对吧？你只需进入语音模式，然后开始与它交谈。语音模式现在连接到工具。所以你可以要求它做任何你的聊天可以做的事情。如果你连接到你的电子邮件、日历，现在你可以问它，比如我今天的日程是什么，或者我应该……你知道，我想腾出一些时间进行深度思考。我该怎么做？它实际上可以获取你日历中每个部分正在发生的事情的上下文。所以，我认为在我脑海中，有这种活动，就是退后一步，思考我可以要求我的 AI 做的比它迄今为止为我做的稍微更有野心的事情。让我们看看它是否有效。

And so, it's a very empowering thing. Normally you'd have to hire someone else like I don't know how to make CAD plans, but now we have this intelligence in our pocket that can can actually do these kinds of things for us. To me, one of the things I really like to do is to try to push the AI in different dimensions of of my life. And so I think that voice mode is something that makes AI so accessible, right? You just go into voice mode and you start talking to it. Voice mode now is hooked up to tools. So you can actually ask it to do anything that your chat can do. And if you're hooked up to your email, to your calendar, now you can ask it about like what's my schedule for today or like what should I, you know, I want to free up some time for some deep thought. How can I do that? And it can actually go and have context on what's going on in each part of your calendar. So, I think that in my mind that there's this activity of just stepping back and thinking about what's a slightly more ambitious thing I can ask my AI to do than it's done for me to date. Let's see if it works.

是的。我正要给你展示我最近构建的东西。我认为我们处于一个时代，每个人都可以构建个人生产力应用。多个。呃，我构建的这个，所以我基本上要求它创建一个应用，为我准备播客。我告诉它我在采访谁，多长时间，我的成功指标是什么，基本上是 YouTube 上的观看次数，然后它继续创建一个播客。

Yeah. And I'm just about to show you something that I've built recently. I think we're in this era where everyone can build a personal productivity app. Multiple. And uh this one that I've built, so I basically asked it to create an app that's going to prepare me for the podcast. I tell it who I'm interviewing for how long and what's my success metric and it's basically the views on YouTube and then it goes ahead and creates a podcast.

它会做研究，建议讨论话题。这很棒。但我怎么让它从我的经验中学习？还有哪里我可以进一步推进？因为我要录制这段对话，它会到这里来检查——现在它在检查我之前的对话。但你觉得这里缺了什么？我对 AI 的野心够大吗？

It does the research, suggests topics to discuss. This is great. But how do I make it learn from my experience? And is there anywhere I can push this further? Because I'm going to record this conversation, it's going to come here, and it's going to check—now it's checking my previous conversations. But what do you think is missing here? And am I ambitious enough with AI?

嗯，我认为“野心够大”的答案永远是不够，因为总有更多我们能做的，对吧？这就是 AI 的美妙之处——它真正关乎解决问题，帮助你实现你想实现的目标。你知道，你能想象的东西没有上限，对吧？所以，这真的只是关于调整——你能实现一件事，意味着也许你可以梦想得更大一点。所以，在我看来，是的，我觉得这是一个超级、超级酷的应用。

Well, I think the answer to being ambitious enough will always be no, because there's always more we can do, right? And that's the beauty of AI—it's really about solving problems, helping you achieve what you want to achieve. You know, there's no ceiling to what you might imagine, right? And so, it's really just about sort of tuning that you were able to achieve a thing means maybe you can dream a little bit bigger. And so, in my mind, yeah, I think this is super, super cool app.

所以，首先，就做背景研究而言——也许在构建这个的时候，他们已经去看了我所有的采访，做了——

So, first of all, in terms of just doing the background research—and maybe when this was being built, they already went and looked at all my interviews that I've ever done and did—

啊，这就对了。做了所有背景研究。

Ah, there we go. Did all the background research.

是的。它做了背景研究。唯一的问题是它还不能连接社交媒体，但我认为这是一个插件的事情。是的。

It did. It did the background. The only thing is it still can't connect to social media, but I think it's a plug-in thing. Yep.

但然后是的，它浏览了你的对话。

But then yes, it went through your conversations.

这就对了。然后所有 OpenAI 过去的新闻稿。哦，我们有一个问题标签页。我喜欢这个。

There we go. And then all of OpenAI's past press releases. Oh, there we got a questions tab. I like that.

是的，我们有一个问题标签页。没错。标题和缩略图已经在这里了。这些是我可以复制的问题。

Yeah, we got a questions tab. That's right. And that's the title and thumbnail already here. And these are the questions that I can copy.

是的。我很好奇它在后期处理和最终视频的实际制作中能帮上多少忙。对吧。因为这东西有关于我、关于 OpenAI、关于你在过去剧集中涵盖的各种话题的所有这些上下文，然后你想怎么剪辑？就像，我听说有很多很好的用例，Astra 帮助视频后期制作，并能主动做到这一点。所以，实际上我很好奇，如果你能有一个应用，就像你唯一要做的就是基本上查看问题，确保你喜欢它们，如果不喜欢，提供一些反馈，它就能做得更好。你出现，实际进行采访，然后其他一切都被处理好了。你现在从采访部分到实际在 YouTube 上发布，要投入多少工作？

Yes. I'd be curious how much it helps with the post-processing and the actual production of the final video once it's done. Right. Because this thing has all this context on me, on OpenAI, on the kinds of things that you've covered in your past episodes, and then how do you want to cut it? Like, I've heard there's so many good use cases of Astra helping with post-production for videos and proactively being able to do that. And so, I think actually I'd be very curious if you could have an app where it's just like the only thing you have to do is basically look over the questions, make sure you like them, and if you don't, provide some feedback, and it can do better. You show up, you actually do the interview, and then everything else is taken care of. How much work do you put in right now from the interview part to something actually being live on YouTube?

哦，很多工作。我有一个大团队，但也是——我们记录了很多，我做这个已经 12 年了，所以我们有所有我们参考的文件，所以现在用 AI 构建更容易，但每集仍然至少 40 小时。

Oh, a lot of work. I have a big team, but it's also—we've documented a lot and I've been doing this for 12 years, so we have all the files that we're referencing, so it's easier to build with AI now, but it's still every episode is at least 40 hours.

哇。如果你能拿回那段时间，对吧，那样所有机械性的工作都被处理好了。这样你可以花很多时间真正思考什么是正确的采访，以及所有其他投入其中的事情，或者真正思考如果你能扩展，你可以做更多的采访，就像，是的，你会用那额外的时间做什么？

Wow. And if you could get back that time, right, so that kind of all of the mechanics are taken care of. So you can spend so much time really thinking about what the right interview is and all of the other things that go into this or really thinking about if you could scale you could do way more interviews like yeah what would you do with that additional time?

这是给你的问题，对于那些将 AI 应用于日常工作的小企业，企业主的角色是什么，他们应该更多地思考什么，现在企业的瓶颈实际上在哪里？

That's the question to you with small businesses who that are applying AI for their day-to-day jobs what is the role of the business owner what should they be thinking more of and where's actually the bottleneck for the business now

我认为有一些根本性的东西。所以首先，我们看到很多小企业正在被创立，因为人们因为拥有 ChatGPT 而感到安心，对吧？有很多专业知识，如果你想启动一个企业，这很难，

I think that there's something fundamental. So first of all, we see so many small businesses that are being started because people feel secure with the fact that they have ChatGPT, right? That there's so many pieces of expertise that if you want to get a business off the ground, it's hard,

对吧？

right?

但你实际上拥有一个博士级别的智能，一个世界专家，商业顾问。你现在口袋里免费拥有所有这些。这相当了不起，对吧？如果你愿意，如果你有能力支付，你可以获得更多的深度和更多的算力来解决任何你想解决的问题。所以小企业比以往任何时候都更容易启动，我们非常具体地看到这一点。然后人们能够——当你与许多这些小企业主交谈时，有太多任务他们外包给 AI，你知道，当然有他们知道必须做的，但有太多任务以前没有人会做，所以他们商业策略中的小漏洞，思考你实际上如何制作这份文件。另一件事是，小企业主甚至没有意识到他们可以做什么。比如我们的一个员工告诉我，他在一家古董手表店买手表，他开始和店主谈论 AI，店主说：“哦，AI，我不能用它做任何事。”就像，你知道，我是一家手表店。他说：“不，不，AI 可以帮助每个人。让我展示给你看。”所以他谈到了他的一些问题，他说：“嗯，是的，他实际上没有网站。”所以你就——

But you actually have a PhD level intelligence, a world expert, business consultant. You have all these things in your pocket now for free. It's like pretty remarkable, right? And if you're willing, if you're able to pay, you can get even so much more depth and so much more compute towards any problem you want. So small businesses easier to start than ever and we're seeing it very concretely. And then people are able to—when you talk to many of these small business owners, there's just so many tasks that they outsource to an AI that you know certainly there's the ones that they knew that they had to do but there's so many tasks that they would just no one would have done it before and so little holes in their business strategy are thinking about how do you actually produce this document. The other thing too is small business owners who don't even realize what they could be doing. Like one of our employees was telling me that he was shopping for watches in this antique watch store and that he started talking to the owner about AI and that the owner was like, "Oh, AI, I can't use that for anything." Like, you know, I'm a watch store. And he was like, "No, no, AI can help everyone. Let me show you." And so he talked about some of the problems he has and he's like, "Well, yeah, he doesn't really have a website." And so you just—

没错。对。

Exactly. Right.

就像 ChatGPT 语音说：“给我建一个网站。建一个很棒的网站。”然后店主说：“嘿，我也有每块手表都有自己的故事，对吧？”所以这就变得很容易，就像：“好的，和 AI 谈谈每块手表背后的故事。”你最终得到所有这些美丽的可视化。你只是意识到，技术以这种方式变得鲜活，而通常这对许多人来说几乎是无法触及的。所以我认为这真的在推动——如果你有世界上最好的程序员在你的工资单上，你会让他们做什么？现在你可以做——我想你在某个地方提到过，实际上小企业必须开始努力的是野心，对吧？

And just like ChatGPT voice being like, "Build me a website. Build an awesome website." And then the owner said, "Hey, I also have each watch has its own story, right?" And so it's just like then becomes very easy like, "Okay, talk to the AI about here's the story behind each watch." And you end up with this beautiful visualization of everything. And you just realize that there's this way that technology becomes alive that normally it's almost inaccessible to so many people. And so I think that this really pushing on if you had like the world's best programmer on your payroll, what would you put them to work on? And now you can do—I think you mentioned somewhere that it's actually ambition that small businesses have to start working on, right?

是的。是的。我认为——

Yes. Yes. And I think that—

我觉得这很迷人。

I think it's fascinating.

我同意，我认为这真的是关于我们都能提高野心的上限，对吧？我们都能完成更多，真正投入其中的人。我认为我们已经开始看到非常了不起的事情。顺便说另一个例子，我们有——有两个盲人兄弟使用我们的技术，他们在日常生活和工作中使用它，甚至用于像将电缆插入电脑正确端口这样的任务，对吧？你可以使用 ChatGPT 来基本上提供这些信息，而通常他们必须向别人求助。所以我认为这种赋能，你知道，能够帮助人们变得更独立，能够以他们想要的方式生活，所有这些，正在发生。它现在正在大规模发生。

I agree and I think it is really about we can all raise a ceiling of ambition, right? We can all accomplish more people who really lean into that. I think we're already starting to see really remarkable things. I'll tell you another example by the way is that we have—there are these two brothers who are blind who use our technology and they use it to both in their daily life and for their work and even for tasks like being able to plug the cable into the right port on a computer, right? You can use ChatGPT to actually basically provide that information when normally they would have to ask someone for help. And so I think this kind of empowerment, you know, being able to help people be able to be more independent, be able to live their life the way that they want, all of that, it's happening. It's happening right now at massive scale.

当有人想在软件领域做点什么，而 OpenAI 又在围绕工作和生产力做这么多东西时，你觉得一个应用凭什么值得付费？

When someone is thinking of building something in software, when OpenAI is building so much around work and productivity, what do you think makes an app worth paying for?

过去做应用的一部分工作确实正在被商品化。它变得人人都能上手，正在被大规模民主化。我觉得这是件非常重要、非常美好的事，对吧？就是可以有更多软件。于是“差异化是什么”这个问题就变得更加重要了。我认为，能以独特方式创造价值的软件，空间会比以往任何时候都大。而这其中一部分，是要真正深入地理解问题；一部分是要真正深入地理解你的用户，对吧？比如你想为教育做点东西，有老师、有家长、有学生，对吧？有太多利益相关方需要你去认真考虑，还有行政人员。你怎么做出一个真正提供正确信息、有正确护栏、以正确方式有用的产品，以及你要卖给谁？所有这些问题。所以我认为这些人的因素——业务的核心是什么？你发现了什么？你的洞察是什么？你看世界的独特方式是什么？而至于你用什么机制去实现它——你必须用 C++ 还是 Swift 还是别的什么——谁在乎呢？那从来就不该是门槛。

There's definitely parts of what app building used to be that's becoming commoditized. It's becoming accessible to everyone. It's becoming massively democratized. And I think that's a really important thing, a beautiful thing, right? That there can just be more software. And so then the question of what is differentiation becomes even more important. I think that there is going to be more room than ever for software that adds value in unique ways. And some of this is really about deeply understanding the problem. Some of this is deeply understanding your users, right? You think about something like if you want to build something for education, there's teachers, there's parents, there's students, right? There's so many stakeholders that you have to really think about. There's administrators. How do you build a product that actually gives the right information, has the right guardrails, that is useful in the right ways, and who are you selling to? All these questions. So I think that these human factors of what is the core of the business? What is it that you have discovered? What is it that's your insight? What is your unique way of looking at the world? And I think that the mechanics of how you instantiate that — do you have to write in C++ or Swift or something — who cares about that? That was never what the barrier should have been.

对，我看到 YC 最近几批创业公司有 60% 是 B2B，而我的直觉告诉我，有了这些可用的工具，如果我想成为创业者，我在消费领域看到了太多问题。你觉得我们会不会进入一个越来越多人做消费业务的时代？

Yeah, I've seen that YC startups' recent batches are 60% B2B, and my intuition is telling me with these tools available to us, if I want to become an entrepreneur, I've seen so many problems in consumer. Do you think we're going to enter this era when more and more people are building consumer businesses?

嗯，我认为企业和消费者之间的界线会变得模糊。我认为“公司”是什么，在某些方面会变得不那么明确。或者说，它不再那么像——好吧，公司就是一大群人，以某种方式组织起来。我认为人们会被极大地赋能，对吧？我们会看到一场大规模的创业复兴。我觉得它正在到来。我们正在看到它的前沿，但我认为它真正全面爆发会是在接下来一到两年。至于你是卖给消费者还是卖给企业，有些东西在区分上仍然成立。比如我想到企业销售时，它本质上关乎关系。但如果你把它看作关系，那好，你之所以没法真正对消费者做企业销售，是因为消费者实在太多了，对吧？规模太高了，成本太贵了。但如果你有 AI，能真正放大你建立关系的能力——建立真正有意义的关系，并在大规模上真正照顾好人们——那么实际上，也许企业销售这套打法，对更多消费业务来说也变得合理了。但一家企业的运作机制，我认为会演变。你和用户建立关系的方式，我认为会演变。但根本的东西——你想要交付价值，你想和接受这份价值的人建立关系，并让它成为双向的交换——所有这些，无论哪个领域，都会变得更加核心、更加突出。

Well, I think that the line between what enterprise and consumer are is going to blur. I think that what a company is is going to become just less well-defined in some ways. Or it's going to be less like, okay, a company is like these big masses of humans that are organized in certain ways. I think people are going to be so empowered, right? We're going to see this massive renaissance of entrepreneurship. Like, I think it is coming. I think we're seeing the leading edges of it, but it's really, I think, really going to kick in over the next year to two years. And I think that the question of are you selling to consumers or enterprises, there's some things that will remain true in terms of the split. Like I think that when I think about what enterprise sales is, it's fundamentally about relationships. But if you think about it as relationships, it's like, well, the reason that you can't really do enterprise sales to consumers is because there's just too many consumers, right? It's just too high of a scale. It's too expensive. But if you have AI that really amplifies your ability to have relationships, really meaningful ones, and really take care of people at massive scale, then actually maybe the enterprise sales motion becomes something that makes sense even for more consumer businesses. But the mechanics of what a business is, I think, will evolve. I think that how you relate to your users will evolve. But the fundamentals — that you want to be delivering value and you want to build a relationship with the people who are receiving that value and have it be a bidirectional exchange — all of that is going to become much more front and center regardless of which domain.

Greg 在讲，如今即便是一个很小的团队，你也能做多少事。如果这让你想“也许我可以开始做那个生意”，我希望你想想你的第一个客户。你要倾听、理解他们需要什么，并把事情做好。同时你还有笔记、跟进事项和一张要更新的表格。我们的赞助商 HubSpot 能帮上忙。它全新的智能 CRM 是一个能自我更新的 CRM，把你的客户信息和对话都放在一起。它的 AI 笔记助手会记录客户通话、起草跟进内容，并建议下一步行动和客户记录的更新，供你批准，这样你就能把更多注意力放在你为之打造这门生意的人身上。通过描述里的链接了解 HubSpot Smart CRM。感谢 HubSpot 赞助本期节目。现在，回到 Greg。

Greg is talking about how much more you can do now even with a tiny team. If that makes you think maybe I can start that business, I want you to think about your first customer. You want to listen, understand what they need, and do a great job. And you also have notes, follow-ups, and a spreadsheet to update. Our sponsor, HubSpot, helps with that. Its new smart CRM is a CRM that updates itself, keeping your customer information and conversations together. Its AI notetaker captures customer calls, drafts follow-ups, and suggests next steps and updates to customer records for you to approve, so more of your attention can go to the people you're building this business for. Check out HubSpot Smart CRM through the link in the description. And thanks HubSpot for sponsoring this episode. Now, back to Greg.

当人们在开发软件时，你如何从战略上考虑，做出不会被新模型淘汰的东西？

When people are building software, how do you strategize around building something that's not going to be made obsolete with the new models?

我认为有些活动会随着模型变聪明而很好地扩展，而有些活动则是在绕开模型的局限。第二种很难持久，所以我认为在不同时期都有创业公司专注于它。拿模型来，加点提示词，做一些非常手工定制的东西。这并不是说在某个时间点它没有价值。有时甚至关乎界面，对吧？在每一个模型能力水平上，几乎都会自然浮现出一种新界面。2022 年，我们有了一个足够个性化的 AI，值得去读它的回答——GPT-4。所以当然，你会想做一个 ChatGPT 单元，在 3.5 和 GPT-4 上。而 2025 年底，我会说，是编程智能体时代真正开始的时候。那时 AI 从占你软件产出的 20% 变成了 80%。于是你就想要一个非常不同的产品来抓住这件事。这是所有人都会用的东西。现在，我认为我们开始进入个性化智能体的时代。所以你会有一个基于云的 AI，而它正以多种不同的形态被确立下来。在 OpenAI，我们刚发布了 Dots。我超级兴奋。我认为它背后有最聪明的模型支撑这类技术，而且它太有用了。它对我有用。我认为它对很多人都会有用。所以我认为核心是思考：随着模型变聪明，你会在哪里扩展。再说一次，这其中一部分是理解行业，一部分是建立关系。回到医院，对吧？那里有太多不同的利益相关方，对吧？有患者、医生、医院管理者、保险公司。有太多不同的参与方。所以你怎么在每一方之间建立信任，做出对他们有用的产品，然后真正能够——即便你有一个更好的模型，如果你已经身处那个行业，你拥有这些客户，你的产品实际上会因此变得更好、更有防御性。

There's some activities I think scale well as the models get smarter, and some activities that are kind of working around limitations in the model. The second one is pretty hard to be durable, and so I think that there have at various points been startups that really focus on it. Take the model, we add some prompts, we have some very handcrafted whatever. And it's not to say that there's a point in time where that is valuable. And sometimes it's even about the interface, right? That at each level of model capability, there's almost a new interface that suggests itself. In 2022, we had an AI that was personalized enough that it was worthwhile reading the responses — GPT-4. So it's like, of course, you want to make a ChatGPT unit at 3.5 and GPT-4. And end of 2025, I'd say, is when the coding agent era really started. And that was like you had AI that kind of shifted from being 20% of your software production to 80% of your software production. And so you just want a very different product that captures that thing. This is the thing everyone's going to utilize. Now, I think we're starting to enter this era of the personalized agent. So you're going to have a cloud-based AI, and this is being ratified in a number of different form factors. At OpenAI, we just released Dots. I'm super excited about it. I think it's got the smartest model backing up this kind of technology, and it's so useful. It's useful to me. I think it'll be useful for lots of people. And so I think that the core thing is thinking about where will you scale as the models get smarter. And again, some of this is about understanding the industry. Some of this is about building relationships. Back to if you think about hospitals, right? There's so many different stakeholders there, right? There's the patient, the doctor, the hospital admin, there's the insurance provider. There's so many different parties. And so how do you build trust across each of those and have a product that's useful to them, and then actually be able to — even if you have a model that's just even better, if you are already in that industry, you have these customers, that actually your product gets better and more defensible as a result.

有没有什么你特别期待看到初创公司用模型来构建的东西？

Is there anything you're particularly looking forward to seeing startups use models to build something?

我认为领域专业知识也非常关键，因为容易忽略的一点是，像 OpenAI 这样的公司，我们拥有的这项技术非常广泛，对吧？它几乎触及经济增长和发展的每一个方面，人们的个人生活、工作生活。每个人都能找到适合自己的东西，能真正帮助他们实现目标的东西。但通常在任何特定垂直领域，我们都远未达到最优，对吧？如果你把经济想象成这种分形的东西，对吧？你放大一看，哦，这是医疗保健领域，然后你再放大，真正看看不同公司的数量和不同的做事方式，无论你选择哪个领域，都有太多唾手可得的成果。实际上，在构建 OpenAI 的过程中，对我来说最有趣和惊讶的事情之一就是意识到经济中存在如此多的次优之处，你可以拥有更高效的流程，让人们节省时间，真正交付价值，而能够跨不同领域连接并不是我们经常做的事情，对吧？就像人们专业化一样。

I think the domain expertise is also really key because one thing that's easy to miss is that as something like OpenAI we have this technology that's so broad, right? It kind of touches every single aspect of economic growth and development, people in their personal lives, work lives. There's something for everyone, something that can really help them, something to help them achieve their goals. But often in any specific vertical we're so far from optimal, right? That if you almost think of the economy as this fractal thing, right? Where it's like you zoom in, you're like okay here's a sector of like healthcare and then you zoom and you really look at the number of different companies and the number of different ways of doing things and no matter what sector you pick, there's just so much low hanging fruit. That's actually been one of the most interesting and surprising things for me in building OpenAI is to realize that there's just so much suboptimality in the economy in terms of how you can actually have much more efficient process that gives people their time back, that actually delivers the value, that being able to connect across different fields is not something we do very much, right? It's like people specialize.

现在把它们构建出来在经济上是合理的，因为 5 年前它们可能只是大公司的一个功能，现在你可以围绕这个功能建立一个小公司。

And now it makes economic sense to build them out because 5 years ago they would be like a feature of a big company, now you can build a small company around that feature.

完全正确。我认为即使是这些经典行业和领域，也真的是因为我们工具的限制。你再想想，如何成为任何领域的专家？就像你去上学，你专业化，随着人类学到更多东西，专业化的含义变得越来越狭窄。比如你现在是生物学家，你读了博士，你的博士论文可能只研究一个特定的反应，对吧？它与整个系统没有联系。我认为在创业时，问题在于你是否能拥有这些更加混合的东西。比如教育科技就是一个例子，你把懂教育的人和懂技术的人结合起来，他们就能创新，做新事情。我想知道我们是否会找到其他我们从未想过的领域，它们实际上能很好地融合在一起。

Exactly. And I think that even these like classic sectors and domains are really because of the limitation of our tools. Again you think about how do you become good, like an expert in any field is like you go to school, you specialize and as humanity has learned more things, what it means to specialize becomes narrower and narrower. Like if you're a biologist now, you get a PhD, there's like one specific reaction that you got your PhD on, right? And it's not connected to the whole system. And I think that when it comes to building a business, a question of can you have these much more hybrid things. So you think about things like edtech as an example of like you take people who understand about education, they understand technology, they mash them up and then they're able to innovate and do new things. I wonder if we're going to find that there's going to be other domains that we've never thought about that actually mesh together super well.

完全正确。这就是创业的未来，匹配这些领域并找到正确的解决方案。你能跟我谈谈你有一个足够复杂的设置，让我可以比较一下，因为有时人们在评论中抨击我问像你这样的人关于晨间简报智能体的问题。所以，我们不要做晨间简报智能体。我们做一些帮助你运营公司并提高效率的事情。这是人们看到后会认为这是他们生产力的顿悟时刻的东西。

Totally. That's the future of entrepreneurship, matching those domains and finding the right solution. Can you talk to me about a setup that you have that is sophisticated enough for me to compare this because sometimes people bash me in the comments for asking people like you about morning briefing agents. So, let's not do morning briefing agent. Let's do something that helps you run a company and be efficient. It's something that people see, they will think it's an aha moment for their productivity.

我不会太贬低晨间简报智能体，因为我认为晨间简报智能体有很深的深度，对吧？有一种是，好吧，是的，它给了我一些信息，但说实话，我自己滚动通知也能看到。还有一种版本是，它说，这是待处理的事情，对吧？比如这些你还没回复的 Slack 消息。嗯，你真的应该按这个优先级顺序回复这些人。再下一个层次是，我已经做了研究。你知道，比如关于我们在某个特定市场的策略应该是什么，我实际上已经去看了所有的演示文稿，这是你需要关注的事情，以便做出这个决定。第三件事就是我认为我们现在所处的阶段，对吧？就是 AI 真的能够跨一堆不同的事情做非常复杂的研究，并提供我应该关注的内容的非常压缩的版本。实际上，我非常兴奋的下一步，我们还没有完全达到，就是所谓的晨间简报智能体，它就像，嘿，我实际上打了电话，你知道，我打电话给这些人，我和你的这些同事谈了谈，我和这个外部合作伙伴谈了谈，我把一切都安排得井井有条，我只需要你批准这个采购订单，我们就可以开始新的什么了。真正主动并能代表你做事的 AI，那即将到来，我认为有了 DOTS，我们实际上有技术来实现它，所以我认为我们会开始看到人们真正开始利用它。我认为我们还有改进要做等等。但我想说的是，在我看来，任何这些用例，都有很深的深度。对我来说，这就是 AI 的关键。如果你只看一个写软件的 AI 的标签，好吧，当然，我们已经有了写软件的 AI 五年了，但今天写软件的 AI，与五年前写软件的 AI 完全不同。所以我对可能的应用深度非常欣赏，即使罐子上的标签是一样的，每个人都听过 100 遍了。

I would not poo poo the morning briefing agent too much because I think the morning briefing agent has so much depth to it, right? There's the like, okay, yeah, it's like it kind of gave me some info where it's like, honestly, I could have just scrolled the notification myself. There's the version of it where it says like, here are the things that are pending, right? Like here are these slacks that you have not answered. Um, and you really should get back to these people with this prioritization order. There's the next level of I have like done the research. You know, there's this question about like what our strategy should be in some particular market and like I've actually gone and I even like, you know, sort of looked at all the presentations and like here's like the things you need to like be paying attention to to make this decision. That third thing is like where I think we're at, right, is like the AI that really can do very sophisticated research across a bunch of different things and provide a very compressed version of what I should be paying attention to. Thing I'm actually very excited about as a next step that we're not quite there yet is the morning briefing agent so to speak that is like hey I actually called like you know I called up these people like I talked to you know these these co-workers of yours I talked to this external partner and like I got everything into perfect order like I just need you to approve this purchase order and like we're ready to go on the new whatever AI that really is proactive and able to do things on your behalf like that's coming and I think with DOTS we actually have the technology for it so I think we're going to start seeing people really start to utilize that. I think we have improvements to make and things like that. But I'd say in my mind, any of these use cases, there's so much depth to them. And and that is to me the key thing about AI. If you just look at the label of an AI that writes software, okay, sure, we've had an AI that writes software for 5 years now, but AI that writes software today, totally different from the AI that wrote software 5 years ago. So I have a lot of appreciation for the depth of application that's possible even if the like label on the tin is like the same thing and kind of everyone's heard it 100 times.

当你在 OpenAI 内部看这些用例时，你有没有一个内部指标来衡量模型的进步实际上如何转化为人们的生产力？

When you look at these use cases inside OpenAI, do you have like an internal metric how advances in models are actually translating into people's productivity?

我们也发表了关于这个的博客文章。所以我们做了相当多的工作，研究 token 使用量，以及它如何与代码行数或提交和 PR 等相关。再说一次，所有这些都是代理指标。它们并不完美，但我认为另一种判断 AI 是否真的在改变生产力以及改变了多少的方法是，如果你遇到内部故障，比如内部人员无法访问模型，嗯，他们会有多沮丧。每个人都说：“我没法工作了。”你会说：“好吧，6 个月前你还没有这种东西，现在你说没有它两个小时就没法工作。”这怎么可能？但我们一直看到这种情况。太酷了。我自己感觉，在晨间简报智能体访问方面，我经常做的事情是，每当我对 OpenAI 有疑问，我就直接问我的智能体，我不知道为什么我还要做其他事情来开始，因为它太高效了。就像经常，有时我甚至觉得自己有点懒，比如有人给我发了一条消息，里面有一个文档。我只是试图模糊地记得它在哪里，然后

We published blog posts on this as well. So we've done a decent amount of work looking at token usage and looking how it correlates with like lines of code or commits and PRs and things like that. And again, all these are proxy metrics. They're not perfect, but I think that another way that you can tell is AI really changing productivity and how much is if you ever have an internal outage, like if internally people can't get access to a model, um how much they feel upset. Everyone's like, "I can't do my work." And you're like, "Okay, 6 months ago you had nothing like this and now you're saying you can't do your work when you don't have it for two hours." Like, how is that possible? But we see this all the time. It's so so cool. And I feel for myself like the kinds of things on the morning briefing agent access that I do a lot of is I like anytime I have a question about OpenAI I just ask my agent like I don't know why I would do anything else to start because it's just so efficient. It's like often and sometimes I even feel like I'm like being a little bit lazy of like there's like some message that someone has sent me with a doc in it. I'm just like trying to vaguely remember where was it and

所以只要把它连接到一切。它可以处理邮件、Slack，无论你用什么，它都有所有信息。所以每当你需要什么，你只要问它，它就有所有这些。

So just connect it to everything. It can it's going to do a mail slack whatever you're using and it has all the information. So whenever you need something, you just ask it and it has all this.

是的。

Yes.

有了 computer use，它真正打通了连接我所有企业上下文的最后一公里。这让它变得非常有用，因为我真的有了一个幕僚长，而且我实际上还有一位人类幕僚长，她也很出色，对吧？我能够给她更大的杠杆，因为她不必再回答那些基础问题，现在我的 AI 就能回答。而其中一些所谓的基础问题其实也相当复杂，比如询问不同的业务指标，对吧？所以它实际上必须去查询我们的数据仓库，然后得出结果，比如你在看某个市场的增长，或者留存曲线之类的。我记得在我上一份工作，也就是在 Stripe 的时候，我们有一个数据团队，你给他们发邮件，一周后他们才会带着那个查询的答案回来。而现在我直接问 AI，它就直接做了。这太不可思议了。

And with computer use, it really closes the last mile of being able to connect to all my enterprise context. And so then that makes it so useful because I really have this chief of staff, and I actually have a human chief of staff who is also incredible, right? And I'm able to give her so much more leverage by not having to ask these basic questions that now my AI can answer. And some of these so-called basic questions are also quite sophisticated, like for example asking questions about different business metrics, right? So it actually has to then go and issue queries into our data warehouse and come up with, you know, you're looking at the growth in some market or the retention curves or something like that. And I remember in my last job when I was at Stripe, we had a data team that you would send an email to, and then a week later they would come back with an answer to that query. And now I just ask the AI and it just does it. It's incredible.

你有没有看到过人们在哪些领域没有充分利用 token？你想告诉大家，比如在这方面多用一些 token 吗？

Do you ever see areas where people are underutilizing tokens? Do you want to tell people like use more tokens for this?

哦，绝对有。你看，我觉得这有两个维度。第一，曾经有一个“token 拉满”的时代，对吧？就是人们盲目地说往里扔更多 token，这显然不是首要目标，但它是一个很好的代理指标。如果你不去刻意优化 token，你可能会看到生产力提升与 token 数量增加相关。我们确实看到了这一点。我们有排行榜，我们会看这些。我觉得那些使用方式最成熟的人，就是能做出真正了不起的事情。当然，这都取决于领域等等，但我很喜欢看到，比如我们的财务团队，他们真的完全 AI 优先了。我觉得有太多事情，比如过去结账要花很长时间的工作流，现在快得多，对吧？而且这些都不是……每个人都能感受到生产力的提升，以及它有多大帮助。我们有一些来自传播团队的早期采用者。我觉得他们能够做到，比如你有一个活动，一群记者要来，你在排座位表，试图搞清楚每个人的饮食偏好，而你一个人就能管理所有这些，因为你有 Codex 实际上帮你处理了大量后端工作。这些从今年年初就一直在发生。所以我们在各个地方都能看到这类知识工作生产力的用例。当然，还有一件事我觉得也很容易被忽视，而且我认为很快会惠及所有人，那就是科学发现这一侧，对吧？我们投入了大量算力去解决 Navier-Stokes，你知道，这个千禧年大奖难题，那大概是 10,000 个智能体跑了若干天，算力很大，我们能够为人类创造新知识。你想想，实际上在更小的规模上，我认为我们仍然看到很多突破在发生。所以我有一种感觉，那些真正把算力指向正确问题的人，对吧，有一些判断力，真正思考过什么才重要、算力该用在哪里，然后愿意把 token 预算推高的人，我们真的看到了这种超乎寻常的影响。

Oh, absolutely. And look, I think there's two dimensions to it. So, one is that there was this era of token maxing, right? Where it's just like people were just blindly saying throw more tokens at it, which is clearly not a primary objective, but it is a good proxy metric. If you're not trying to optimize tokens, you're probably going to then see improved productivity correlated with increased number of tokens. And we really do see it. We have leaderboards. We kind of look at that. And I think that we see like the people who tend to be the most sophisticated in their usage are just able to do really incredible things. Again, it all depends on the domain and things like that, but I love seeing, for example, in our finance team like that they've really gone full AI first. And I think that there's just like so many things where like workflows that used to take a long time for closing the books that are now just so much faster, right? And none of those are like everyone just feels that increased productivity and how helpful it is. We have some early adopters from our comms team. I think that they're just like being able to manage like you have an event with a bunch of journalists coming and you're putting together a seating chart and trying to figure out everyone's dietary preferences and being able to manage all of that as one person because you have Codex that's actually taking care of so much of the backend for you. Like that's all been happening since early this year. So we see these kinds of use cases for knowledge work productivity everywhere. And then of course the thing that I think is also very easy to miss and I think is actually coming soon for everyone is the scientific discovery side of things, right? That we put a ton of compute into solving Navier-Stokes, you know, this big millennium problem, and that was like 10,000 agents for some number of days, like it's a lot of compute, and we're able to create new knowledge for humanity. And you think about actually at much smaller scale, I think it still is the case that we're seeing lots of breakthroughs happening. And so I just have this feeling of like, I think that people who really point compute at the right problem, right, have some judgment and some real thought on like what really matters, where is the compute going to be utilized, and then are willing to push that token budget high. We're really seeing this sort of outsized impact.

我觉得这是很多第一次开始使用智能体和自动化的人都会有的一个大问题。你从什么开始？你怎么想这件事？

I think it's a big question that a lot of people have who are starting using agents and automating for the first time. What do you start with? How do you think about that?

这个白纸问题在某种程度上是 AI 里最难的问题，而利用 AI 就是这个问题：你从哪里开始？你用什么？我觉得我的答案永远是：从小处开始。就试试看，随便打点什么，按回车，看看会发生什么，对吧？我觉得一旦你有了那个反馈回路，你就可以迭代，对吧？你可以改进。所以我觉得，找到你生活里的一个小痛点，对吧？也许是你收件箱里一封你一直害怕回复的邮件，或者你知道要花很多功夫才能处理好的事情。然后第一个问题就是：我怎么把这封邮件给 AI？对吧？一个答案是你复制粘贴。另一个答案是你连接了 Gmail 连接器。所以我觉得你可以非常机械地开始，自己走一遍流程，然后意识到，好，我已经做了三次，我已经做了五次，现在我明白这个过程是怎么运作的了。我们来把它自动化吧。顺便说一句，这就像一条通用的软件工程格言。我一直都是这么处理的：什么时候该构建软件，什么时候只是机械地做，就是当你已经做了五次之后，大概就是开始自动化的好时机。所以我觉得

This white page problem is like the hardest problem in AI in some ways, and utilizing AI is this question of where do you get started? What do you use? I think my answer is always start small. Like just try something, type something random, push enter, see what happens, right? And like I think that then as soon as you get that feedback loop then you can iterate, right? You can improve. And so I think that find like a little pain point in your life, right? Whether it's maybe it's an email in your inbox that you know that you have been dreading replying to, or you know that it's going to take a lot of work to do to get there. Um and so then the first question is well how do I get the email to the AI? Right? And so there's one answer which you copy paste it. There's another answer where you have your Gmail connector connected. And so I think that you can start very mechanically, very much going through the effort yourself and then realize, okay, I've done this three times, I've done this five times, and now I understand how this process works. Let's automate it. And that, by the way, is like a general software engineering maxim. Like that's very much how I've always approached when you build software versus when you just kind of do it mechanically is after you've done it five times, it's probably a good time to start automating. And so I think

我喜欢这个“五次”的指标。

I like that the five times metric.

这是个相当不错的指标。

It's a pretty good one.

没错。是的。我觉得对 AI 也这样做，因为这样你能一直把握问题的脉搏，我觉得这非常重要。归根结底，回到小企业主用 AI 做什么这个问题，责任仍然在他们身上，对吧？企业仍然以某种方式运营，客户要么注册要么不注册，要么付费要么不付费，所以他们真的应该关心那个结果，并且深入细节。

Exactly. Yeah. And I think I think that doing that for AI because that way you kind of keep your your finger on the pulse of the problem uh which I think is very important like at the end of the day you know back to the question of what are small business owners doing with AI accountability is something that remains with them right it's still the business is operating a certain way and the customer either signs up or doesn't and you know either pays or doesn't and so they should really care about that outcome and be deep on the details

这个问题的另一面是，你如何决定什么不该自动化？我想结合你们的使命来问你。在 OpenAI，你们如何决定下一步不构建什么？

Another side of this problem is how do you decide what not to automate and I wanted to ask you about with you know your mission in mind. How do you at OpenAI decide what not to build next?

我认为非常重要的一点是，人类要保持控制和主导，对吧？目标来自人。这项技术的意义是帮助人，对吧？帮助赋能每个人，帮助人们实现更多他们想做的事。所以我觉得那种“人类优先、为人类服务”的气质非常重要，是我们使命的核心。这真的渗透到我们思考事情的方式里。所以当我们思考要构建什么时，我们总是想：如何设置正确的护栏？如何建立信任？如何确保有良好的监督、可监控性？这从一开始就一直是我们在构建所有这些技术时的核心关注点。

I think it's very important that humans remain in control and in charge, right? That the goals come from people. Point of this technology is to help people, right? To help empower everyone to help people achieve more of what they want to do. And so I think that that human first team humanity kind of vibe is very important, very core to our mission. And that's something that really bleeds through to how we think about things. And so when we think about what to build, we always think about how do you put in the right guard rails? How do you build trust? How do you ensure that there's good oversight, monitorability, and that that is something that has been a very core focus of building all that technology for us for really since inception?

对于愿意提升自己智能体玩法、并且有 1 小时的人，这个周末他们该从哪里开始？

For someone who is willing to step up their agent game and they have 1 hour, where do they start this weekend?

我会说，首先，注册 ChatGPT。

I'd say first of all, sign up for ChatGPT.

我觉得这就是一个非常酷的新形态。你可以说话，聊几句，感受一下它能做什么。但把它连接到合适的上下文，它就能真正干活，对吧？它有自己的云计算机，也可以选择连接到你的本地电脑。我们过去对编程智能体的印象就是这种基于任务的系统，你得像微管理者一样盯着它，它跑在你的笔记本电脑上。你合上笔记本，它就停止工作。显然，那从来就不是被承诺的 AI，对吧？那从来不是我们 5 年前想象的 AI。所以我觉得现在我们有了一个真正像我们一直想象的那种 AI。所以，从一个小任务开始，比如让它帮你买东西，或者试试——我很喜欢给我的那个发短信，对吧？就去设置一些连接器，让它感觉像是，好吧，我现在有了一个和它沟通的渠道，然后让它做一些重复性任务，或者让它做一些研究。非常有趣的是，DOTS 确实可以——它相当复杂，对吧？它由 GP6 Astra 驱动。所以它实际上拥有一个不可思议的大脑，一个聪明、不可思议、最智能的模型来驱动这类助手。这意味着它能在事物之间建立联系。比如你问它，嘿，这个特定地区发生了什么？它会意识到这是你关心的事情。然后它可能会主动浮现，如果有关于那个地区的新新闻报道，也许它会说，嘿，我在为你做一些额外的思考，我真的很感激。

I think it's just a really cool new form factor. Again, you can talk, chat a little bit, get some sense of what it does. But connecting it to the appropriate context can really do work, right? It has its own cloud computer and it can optionally hook to your local computer. And I think that this picture that we've all had of the way coding agents work is this task-based system that you kind of micromanage that lives on your laptop. You close your laptop, it stops working. Clearly that was never the AI that was promised, right? That was never the AI we pictured 5 years ago. And so I think we now have something that really looks like the AI that we were always picturing. So again, start with a little task like ask it to go buy something for you, or try to—I really like texting mine, right? Just go and really set up some connector so that it feels like, okay, I've got a communication channel to this now, and ask it to do some recurring task, or ask it to do some research. And the thing that's very interesting is that DOTS can really—it's quite sophisticated, right? It's powered by GP6 Astra. So it's actually got basically this incredible brain, this smart, incredible, most intelligent model for any of these kinds of assistants powering it. And that means that it makes some connections between things. Like you ask it a question about, hey, what's going on in this particular geography or something? And it will realize that that's something you care about. And then it might proactively surface if there's a new news report about that region, then maybe it will say, hey, you're doing some extra thinking for you, which I really appreciate.

你觉得这就是工作的未来吗？还是说 5 年后它会变成我们现在根本无法想象的东西？

Do you think that's the future of work or do you think in 5 years it's going to be something we can't even imagine now?

嗯，我认为 AI 总是令人惊讶。所以，我认为无论如何，基本面会是：我们会有更智能的 AI，更易获取、更赋能，诸如此类。但你利用它的方式会相对于我们今天所预期的令人惊讶。所以，我确实认为形态会继续变化。而且我认为——如果你想想，我们真的会拥有这种能够解决未解数学问题的 AI，在每个人手中。那么人们会用它做什么？会有什么样的突破成为可能？我想到医学，对吧？药物发现将会不可思议。材料科学，我们将生活在一个人们能够解决以前需要一小群专业专家才能解决的问题的世界。就像你能够做到。你能够解决那些别人从未想过的挑战，而且会有一些没人面对过的挑战，你将能够成为第一个解决它们的人。我认为我们将作为一个广泛的社区来做这件事，所以工具方面，我认为会有更多的协作，这些 AI 将成为代表你的东西，你将拥有一个为你利益工作的 AI，公司会有自己的 AI。所以我认为需要构建很多基础设施，并以可信、可观察、真正提升人类的方式构建它，我认为这是我们应共同做的核心。

Well, I think AI is always surprising. So, I think that there's going to be somehow that the fundamentals will be we'll have smarter AI that's like more accessible and more empowering and all those things. But that the way in which you'll utilize it will be surprising relative to what we expect today. So, I do think that the form factors will continue to change. And I think that the—if you think about really we're gonna have this AI that's able to solve these unsolved math problems in everyone's hands. And so what are people going to do with that? Like what are the kinds of breakthroughs that are going to become possible? And I think about that for medicine, right? For drug discovery is going to be incredible. Material science, we're just going to live in a world where people are going to be able to solve problems that before required a very small set of specialized experts to be mobilized behind them. Like you will be able to do that. You will be able to solve these challenges that no one else has ever thought about and there going to be some challenges that no one's ever faced before that you're going to be able to just like be the first person to solve and I think that we're going to do this as this broad community and so the tools for that I think there's going to be so much more collaboration and these AIs are going to be something that will represent you that you are going to have an AI that is kind of there working for your interest companies will have their own AI so it will be I think there's like a lot of infrastructure to be built and building this in a way that's trustworthy and that is observable and that really uplifts humanity like that is I think very core to what we should collectively be doing.

我觉得这是一个很棒的答案，我想这个周末我们会有功课要努力提升我们的雄心，而不仅仅是智能体。

I think that's an amazing answer and I think we'll have homework to work on our ambition this weekend not just agents.

好的。

All right.

非常感谢你，Greg。太精彩了。

Thank you so much Greg. That was amazing.

谢谢。谢谢。

Thank you. Thank you.