“工程师正在变成魔法师” | OpenAI的Sherwin Wu谈软件开发的未来

对吧？

Right?

比如，你已经开始在AI创业领域看到这种趋势了，软件正变得越来越垂直化，像为某个垂直领域开发AI工具往往效果很好，因为你深入专研了这个特定领域。

Like one, you're kind of starting to see starting to see this play out in the AI startup scene where software's became a lot more vertical oriented, where like these verticals, like creating some AI tool for some vertical tends to work quite well because, you know, you really lean into that particular domain.

你真正理解了它的使用场景。

You like really understand the use case for it.

因此，如果继续发展AI，没有理由不能出现100倍数量的这类创业公司。

And so if you play out AI, there's no reason why you can't have like 100X more of these startups.

所以，我认为我们可能会看到一种局面：为了支持一个人打造十亿美元的初创企业，可能会有另外一百家小型创业公司，专门开发高度定制化的软件，来支持其他类型的个人十亿美元初创企业。

And so I think one world that we might end up seeing happen is in order to enable a one person billion dollar startup, there might be like a 100 other small startups building bespoke software that works extremely well to support other types of, you know, one person, you know, billion dollar startups.

因此，我认为我们可能会真正进入一个B2B SaaS以及软件和创业整体的黄金时代。

And so I think we might actually enter into a golden age of like B2B SaaS and just like software and startups in general.

所以，我认为这是一个非常有趣的趋势，因为随着软件开发越来越容易，运营公司也越来越简单，你可能会看到更多这样的初创企业涌现。

And so I think that's a really interesting trend to kind of see because as it gets easier and easier to build software, as it's easier and easier to run a company, you might actually just end up seeing way more of these startups.

因此，我一直在想，也许会出现一个人的十亿美元初创企业，也可能出现一亿美元级别的初创企业。

And so the way I've been thinking about it is like, yeah, there might be one person billion dollar startup, or there might be like a $100,000,000 startups.

可能会有数以万计的千万美元级别初创公司。

There might be tens of thousands of $10,000,000 startups.

作为个人，拥有一个一千万美元的业务其实非常好。

And as an individual, it's actually pretty great to have a $10,000,000 business.

到那个时候，这笔钱对你一生来说已经足够了。

Like that's like enough for yourself for life at that point.

因此，我们可能会看到这种方式的爆发式增长。

And so, we might really see an explosion in that way.

但感觉人们并没有真正推动这一点。

And feel like people aren't really pressing that in.

这还有另一个层面的连锁效应，你知道吗？

There's another kind of like third order effect to this, you know?

而且，随着预测越来越远，我认为其中充满了不确定性。

And again, all of these, like as you get to the further and further out predictions, I think there's a lot of uncertainty.

我认为，如果我们最终进入这样一个世界：出现许多微型公司，它们开发的软件仅供公司所有者——一到两个人——使用和运营。

I think if we end up moving to this world where you end up with these like kind of micro companies building software that works for one or two people who own the company and are working there.

我认为创业生态系统会发生变化。

I think the startup ecosystem will change.

我认为风险投资生态系统也会发生变化。

I think the VC ecosystem will change.

你知道吗，我们可能会进入一个世界，那里只有少数几个大玩家提供平台并支持所有这些初创公司。

You know, we might end up in a world where there's just like a handful of big players that are offering platforms and supporting all of these startups.

但你知道，那些能够带来百倍或千倍投资回报的风投规模型初创公司，可能会因为大量小型的、价值在1000万到5000万美元之间的公司出现而减少——这些公司虽然不适合风投回报模式，却非常适合那些高度自主、如今借助AI为自己打造事业的个人。

But, you know, the types of venture scale return startups that can really 100 or thousand x your your investment might actually end up shrinking if you end up having a bunch of these, you know, smaller 10 to $50,000,000 companies, which are not great for venture style returns but are great for the individuals, the high agency individuals who are now, you know, really leaning to AI to to to build these businesses for themselves.

我喜欢我们已经讨论了这么多层的连锁效应。

I love how many order, like, order effects we've been through.

每当你听到‘第四层效应’这个词时，谢尔文，我只是开个玩笑。

Whenever you hear the fourth order effect now, Sherwin, I'm just joking.

我实在没法理解，第四层效应太复杂了，超出了我的大脑容量。

I I can't I it's too fourth order is too too it's too gigabrain for me.

我真的没法想那么远。

I I can't I can't think that far ahead.

这就像《盗梦空间》，每深入一层，销售过程就变得越慢。

It's like inception where just everything gets slower every time you go deeper into selling.

是的。

Yeah.

对。

Yeah.

每一层都是如此。

Every layer.

好吧。

Okay.

至于十亿美元的初创公司，我经常思考这个问题，因为我做的东西根本不是风险投资规模，也没有很高的杠杆效应，但我确实经常收到一些极其荒谬的客户支持请求。

So the billion dollar startup, I've been I think about this a lot because I I'm not gonna be a billion dollar startup because what I'm doing is not venture scale in any way and not super high leverage, but just could see how many support tickets I get from just, like, the most ridiculous things.

我很难想象一个人能做到。

It's hard for me to imagine one person.

我对这种十亿美元的初创公司持悲观态度。

Like, I'm bearish on this billion dollar startup.

我只是想分享这个想法。

I just wanna share this thought.

仅仅因为支持成本，即使AI在帮你运营一家十亿美元的公司，除非你的客户平均消费额非常高，

Simply because of the support costs, even if AI is helping you at a billion dollars, just like unless your ACVs are, you know,

非常高

very high

而且客户数量很少，光是处理支持工作就够呛。

and you have very few customers, just dealing with support.

而且人们总是这样，你知道的，他们

And people are like, you know, like, they

能自己解决自己的问题，

can solve their own problems,

但他们还是会选择发邮件寻求支持。

but they're like, oh, email support.

问一些这样的问题。

Ask about this thing.

就我经验而言，单靠处理这些支持工作很难规模化。

Just dealing with that is hard to scale is in my experience.

所以，依我看来，除非你有一大群外包人员——我不知道，这算不算一个人的公司？

So unless you have in my opinion, unless you have a bunch of contractors, which I don't know, does that count as a single person company?

我觉得，要让一家十亿美元的初创公司规模化，却没有专人帮你至少处理支持工作，是极其困难的。

I feel like it's very difficult to scale a billion dollar startup and not have someone helping you with at least the support work.

而AI，我认为只能帮你走到一定程度。

And AI, I think, will only take you so far.

所以我觉得你说得对。

So I I I think that's true.

实际上，我对这个问题的看法略有不同：我觉得你的Lenny播客最终可能会成长为一家十亿美元的初创公司。

And actually, I think my view on it is is is slightly different, which is I think that your you know, Lenny's podcast might end up becoming a billion dollar startup.

但我认为可能发生的是，你不需要亲自去调度AI来处理和解决那些支持工单，反而可能会出现一大批专门为你的需求打造的初创公司。

But what I think might happen is instead of you kind of being the one person who has to dispatch an AI to solve and fix those support tickets, I think what might end up happening is there might be a whole smattering of other startups that are building software and super tailored towards what you might need.

因此，可能会有十到二十家初创公司专门开发面向播客和通讯简报的支持软件。

And so, there might be like 10 or 20 startups that build support software for podcasts and newsletters.

这可能是一个一个人的初创公司。

And that might be a one person startup.

它不需要很大。

Like it doesn't need to be a big one.

他们可以非常容易地编写出这个产品。

And they might be able to just code up this product very, very easily.

他们能够自己构建自己的东西。

They are able to kind of like build their own thing.

因为这个产品如此量身定制、独特，并且 hopefully 有用，它可能成为你作为一个人的十亿美元初创公司所购买的东西。

And because it's so tailored and unique and hopefully, you know, useful for you, it might be something that you purchase as the one person billion dollar startup.

我会。

I would

我会买这个。

buy that.

我会买这个。

I would buy that.

是的

Yeah.

这涉及到哪些事情应该由内部完成，哪些应该外包出去。

There's like a question of like what you in house and what you what you like kind of outsource.

我认为可能发生的情况是，由于编写软件和构建产品的成本急剧下降，你可能会将大量工作外包，从而缩小公司的规模。

And what I think might happen is because the cost of writing software and building products is collapsing so much, you might end up outsourcing a lot of this and in doing so, reducing the size of your company.

因此，这大概就是我认为可能会出现的世界。

And so that's kind of the world that I think might end up happening.

同样，这里存在很大的不确定性，但最终结果可能仍然是由一个人主导一个高度杠杆化的庞大公司，甚至可能达到十亿美元的估值。

Again, there's like high uncertainty in what might play out here, but the end result still might be a one like, one person driving this, like, high high massive leveraged company that might actually reach a billion dollars.

我能理解这一点。

I could see that.

我还会想到Clawbot/Moldbot/OpenClaw的Peter，他现在被各种请求、邮件、通知、私信和公关信息淹没，而他甚至还没从这个项目中赚到一分钱。

I also think about Peter at Clawbot slash Moldbot slash OpenClaw of just, like, how he barraged he is right now by all these asks and emails and pings and DMs and PRs just like, oh, I'm curious to and he's not even making any money off this thing.

是的

Yeah.

我无法想象他现在是什么感觉。

I I can't imagine what it's like to be him right now.

这一定简直疯狂至极。

It's it must be, like, absolutely insane.

这大概就像我们发布ChatGPT之后的那几个月，那种疯狂劲儿，是的。

It it it's probably, like, you know, like, the the the months after we launched ChatGBT, the craziness that was Yeah.

就一个人。

As one as one man.

是的。

Yeah.

顺便说一下，他一周后就要出舱了。

He's coming out of the pod, by the way, in in a week.

哦，这真令人兴奋。

Oh, that's exciting.

是的。

Yeah.

也许第四阶效应是分发变得越来越重要，因为有太多东西在争夺你的注意力。

Maybe the fourth order effect is distribution becomes increasingly important because there are so many freaking things trying to get your attention.

所以，拥有受众和平台的人会变得越来越有价值，这很好，是的。

So people with an audience and platform, I think, become more and more valuable, which is good Yep.

很棒的内容。

Good stuff.

好的。

Okay.

我其实想回到你之前谈的管理话题。

I wanted to come back actually to your management stuff.

我非常欣赏你关于花更多时间与顶尖员工相处非常有效的见解。

So I I really loved your insight about spending more time with top performers has been really successful to you.

想想你作为一支团队的管理者，这支团队正在构建一个平台，几乎整个AI经济都基于你的API运行。

Just thinking about you as a manager of a team that is building the platform that powers basically the entire AI economy, like every AI startup is building on your API.

显然，你做得非常出色。

Clearly, you're doing a great job.

你还学到了哪些其他重要的管理经验？

What other kind of core management lessons have you learned?

作为工程师和团队的管理者，你认为什么是成功最关键的因素？

What do you find is really important and key to your success as a manager of engineers and just people?

是的。

Yeah.

我认为我在这里学到的很多经验，不一定只适用于OpenAI API或我们的一些企业产品。

I think a lot of the lessons that I've learned here, I don't know how specific it is to the OpenAI API or some of our enterprise products in particular.

我的管理理念当然随着时间有所变化，但我觉得它保持不变的部分可能比变化的还要多。

I think my management philosophy has obviously changed over time, but I think it's probably stayed the same more than it's changed over time.

其中一个原则就是我之前跟你提到过的，那就是花大量时间与顶尖员工在一起，说得具体一点，就是超过50%的时间要放在你的顶尖员工身上，可能是你前10%的优秀员工，并尽全力支持他们。

One of these principles is is kind of what I talked to you about before, which is, you know, spending a lot of time with with top performers, like actually spending and like to be very concrete, like, it's like more than 50% of your time with your top performers, with maybe your top, like, 10% performers and really, really trying your best to empower them.

我看待这个问题的方式，是回到《人月神话》这本书里提到的一个类比，就是把软件工程师比作外科医生。

The way that I think about it is is is is kind of come back to this analogy of software engineer as as as a surgeon, which comes from the the mythical man month book.

这其实挺有意思的。

So it is actually it is funny.

所以我从这本书中引用了这个观点，但书中实际上描述了一个世界，我认为他们当时是在预测未来，因为这本书大概是七十年代写的。

So I I pull it from the book, but in the book, they actually describe this world where I think they were like predicting the future because because I think the book was written like in the seventies or something.

他们说，软件工程可能会发展成一个软件工程师如同外科医生的世界，在手术室里，只有一个人在做实际工作，比如一个人负责切开或进行所有手术操作。

They said that software engineering might end up moving into a world where that software engineers are like surgeons, where in a surgery room, there's one person doing the work and there's the one person cutting or whatever and doing all the surgery.

而房间里的其他所有人都只是在支持他。

And everyone else in the room is there to just support them.

对吧？

Right?

就像护士、助手、住院医师和实习医生一样。

It's like the nurse and like the assistant, the resident and the fellow.

然后外科医生说：我需要一把手术刀。

And then the surgeon's like, I need a scalpel.

他们就把手术刀递给他。

They give them scalpel.

然后他说：我需要这个工具和这台机器，他们就会把东西拿过来。

And then they're like, I need this tool and this machine and they'll bring it over.

所有人都在支持那位主刀医生。

Everyone's there to just like support the one surgeon.

因此，《神话般的猛犸象》实际上预测了软件工程将朝这个方向发展。

And so the Mythical Mammoth actually predicted that that is kind of the direction that software engineering is going to go.

我认为这并没有完全实现，因为现在的工作方式更加协作，并非只有一个人在做事，但我一直非常喜欢这个比喻。

I don't think that's exactly played out where like, you know, it's much more collaborative and like, it's not only one person doing the work, but I've always really liked that analogy.

这个比喻实际上也是我管理哲学中努力效仿的：软件工程并不是像外科手术那样只有一个人在工作，但我作为管理者，希望以这种方式对待团队成员——赋予他们权力，让他们感觉自己就是主刀医生。

And that analogy is actually what I strive to kind of emulate in my own management philosophy, which is software engineering isn't really like surgery where it's not just one person doing work, but the way in which I like treating the people on my team and the way that I act as a manager is I want to empower them, make them feel like they're a surgeon.

我会尽力支持他们，确保他们拥有完成工作所需的一切。

And in so far as like making sure that I'm supporting them and making sure they have everything that they need to do their work.

这让他们感觉背后有一支庞大的团队在支持他们，提前预见问题，并在他们需要时提供一切，而实际上，这一切都只是我这个管理者在运作。

And it feels like they have an army of people kind of supporting them and looking around corners and giving them everything that they need when it's really just me as the manager.

我举的例子是，提前预见障碍并消除阻碍，尤其是在组织层面，这极其有用。

And so like the example that I give is looking around corners and unblocking people, especially from an organizational perspective, is extremely, extremely useful.

再说回到人工智能的话题，如今这显得更加重要了，对吧？

And again, going back to the AI conversation, it's even more important nowadays, right?

比如，当人们不断提交拉取请求时，真正阻碍进展和交付成果的往往是组织或流程层面的问题。

Like if people are just like cranking PR after PR, the main thing, bottlenecking progress and shipping something tends to be organizational or process oriented.

如果你作为管理者能提前预见并为团队扫清障碍，比如外科医生需要手术刀时，管理者已经提前准备好了一把，那就是

And if you as a manager can kind of look around corners and kind of unblock the team, if you can, like if the surgeon needs scalpel, but the manager kind of already has a scalpel ready for them, that's

最理想的情况。

the best case scenario.

这正是我对待管理、尤其是工程管理的方式。

That's kind of the way that I approach management and especially engineering management.

因此，这种理念多年来一直深深印在我的脑海中。

And so that's something that's really, really stuck with me over time.

尽管软件工程师并不是真正的外科医生，但这个比喻在我的整个职业生涯中始终萦绕在我心头。

And even though, you know, software engineers aren't exactly surgeons, that metaphor has always kind of stayed in my mind as of the rest of my career.

我非常喜欢这个说法。

I love that.

我觉得我在想，AI是否能帮助我们提前预见，比如某个工程师会被某个决策卡住。

And I I feel like I wonder if that's something AI can help with is look around corners and predict here, this engineer is gonna be blocked by this decision.

我们需要弄清楚这个问题。

We need to figure this out.

我们需要搞定。

We need to get Yeah.

这其实是个非常好的观点。

That's actually a really good point.

我还没试过，但我很好奇，如果我让 ChadGPT 连接公司知识库，去查询一下当前的阻碍因素，会怎么样？

I haven't tried this yet, but I wonder what would happen if I ask ChadGPT hooked up to company knowledge, you know, like, are the active blockers?

查看所有 Notion 文档。

Look through all the Notion docs.

看看可能的 Slack 消息。

What are maybe Slack messages.

你知道，这些信息 probably 就藏在 Slack 的某个地方。

You know, it's probably in Slack somewhere.

我的团队目前有哪些主要阻碍？

What are the active blockers on my team?

有什么我可以帮忙的吗？

And is there something I can do to to help?

这真的很有趣。

Now that's very interesting.

我之前没想过这一点，但你说得对。

I have not thought about that, but you're right.

我刚刚有了一个想法。

Just had an insight right here.

是的。

Yeah.

是的。

Yeah.

是的。

Yeah.

更有趣的是，你预计这位工程师或这个团队在未来几个月会遇到什么障碍？

And it's I think even more interestingly, what do you anticipate will be a blocker for this engineer or this team in the in the coming months?

是的。

Yeah.

你让模型、你让AI去处理第二和第三层次的事情了。

You asked the you asked the model you asked the AI to do the second and third order things.

提前想想吧，老兄。

Anticipate that, man.

也要预测一下下个月博主们会做什么。

Anticipate what the vloggers will be next month too.

我们很棒。

We're great.

我觉得我们这里已经有了一个好想法。

I think we've got a we've got a good idea right here.

是的。

Yeah.

是的。

Yeah.

本集由Datadog赞助播出，Datadog现已整合Epo，这是领先的实验与功能开关平台。

This episode is brought to you by Datadog, now home to Epo, the leading experimentation and feature flagging platform.

全球顶尖公司的产品经理使用Datadog——与工程师每天依赖的同一平台——将产品洞察与产品问题（如漏洞、用户体验摩擦和业务影响）联系起来。

Product managers at the world's best companies use Datadog, the same platform their engineers rely on every day, to connect product insights to product issues like bugs, UX friction, and business impact.

它从产品分析开始，产品经理可以观看回放、分析漏斗、深入研究留存率，并探索增长指标。

It starts with product analytics, where PMs can watch replays, review funnels, dive into retention, and explore their growth metrics.

当其他工具止步不前时，Datadog却更进一步。

Where other tools stop, Datadog goes even further.

它能帮助你真正诊断漏斗流失、漏洞和用户体验摩擦所带来的影响。

It helps you actually diagnose the impact of funnel drop offs and bugs and UX friction.

一旦你知道该聚焦哪里，实验就能验证哪些方法有效。

Once you know where to focus, experiments prove what works.

当我还在Airbnb时，我亲身体验过这一点，我们的实验平台对于分析哪些方法有效、哪些环节出错至关重要。

I saw this firsthand when I was at Airbnb, where our experimentation platform was critical for analyzing what worked and where things went wrong.

而打造Airbnb实验平台的同一团队，也创建了Epo。

And the same team that built the experimentation at Airbnb built Epo.

Datadog 还通过会话回放功能，让你超越数字本身。

Datadog then lets you go beyond the numbers with session replay.

通过热力图和滚动图，精确观察用户如何与产品互动，真正理解他们的行为。

Watch exactly how users interact with heat maps and scroll maps to truly understand their behavior.

所有这些功能都由与实时数据关联的功能标志驱动，使你能够安全发布、精准定位并持续学习。

And all of this is powered by feature flags that are tied to real time data so that you can roll out safely, target precisely, and learn continuously.

Datadog 不仅仅是工程指标。

Datadog is more than engineering metrics.

它是优秀产品团队快速学习、智能修复并自信发布的地方。

It's where great product teams learn faster, fix smarter, and ship with confidence.

前往 datadoghq.com/lenny 申请演示。

Request a demo at datadoghq.com/lenny.

就是 datadoghq.com/lenny。

That's datadoghq.com/lenny.

好的。

Okay.

我现在要转向谈谈你们所有人所构建的API和平台。

I'm gonna shift to talking about the API and the platform that you all build.

所以你与许多公司合作，帮助它们基于你的工具实施你的API和平台。

Some so you work with a lot of companies implementing your API, your platform building on on your on your tools.

你告诉我，许多公司实际上在AI部署上的投资回报率为负，这我觉得正是很多人阅读、感受和认为的情况，而你亲眼看到这一点，这很有趣。

You told me that you find that a lot of companies actually have negative ROI on their AI deployments, which, I think is what a lot of people read about and feel and think, and it's interesting you're actually seeing that.

那里到底发生了什么？

What what's going on there?

他们做错了什么？

What are they doing wrong?

在AI和我们的部署方式中，世界上正在发生什么？

What do you what's happening in the world of AI and deployments in our way?

是的。

Yeah.

为了明确一点，我并没有明确看到这方面的量化数据。

So to be clear, I don't explicitly see quantitative numbers around this.

实际上，衡量这些事情非常困难。

It's actually really hard to measure these things.

但通过观察一些公司尝试使用AI，我对许多AI部署实际上具有负投资回报率这一点并不感到惊讶。

But especially from observing some companies kind of trying to do AI, I would not be surprised if a lot of AI deployments are actually, you know, negative ROI.

我的意思是，这部分原因也在于，我认为全国范围内，尤其是科技圈之外的人，普遍觉得AI被强加给了他们。

I mean, part of this too is I think there's also general sentiment from folks around the country, like, basically outside of tech that AI is being forced onto them.

我认为这在一定程度上是某些AI部署投资回报率为负的体现。

And I think part of this is is is probably a symptom of some negative ROI AI deployments.

关于这一点，我观察到一些情况。

A couple of things I've observed around this.

其中一个方面是，我想我一再回到这个问题上。

So one one thing is and I think I I come back to this again and again.

我觉得我们在硅谷的人常常忘记我们生活在一个泡沫里。

Like, I think we in Silicon Valley just forget that we live in a bubble.

我们太沉浸于推特了，抱歉，是X这个泡沫了。

Like, we are so like, Twitter is a bubble sorry, X is a bubble.

硅谷是一个泡沫。

Silicon Valley is a bubble.

软件工程是一个泡沫。

Software engineering is a bubble.

世界上大多数人，美国的大多数人并不是软件工程师，也没有被AI深深影响，也不会关注每一个模型的发布。

Most people in the world, most people in The US are not software engineers, are not very AI pilled, are not following every single model release.

因此，我们完全不了解如何使用这项技术。

And so we're just highly out of the loop on how to use this technology.

所以我们总是谈论关于Codex的所有最佳实践，以及OpenAI内部那些沉迷于Codex的人。

And so, we always talk about all these best practices for Codex, all these Codex Pill people within OpenAI.

我敢肯定，X上发帖的每个人都是这些AI工具的超级用户。

I'm sure everyone on X who posts are like crazy power users of these AI tools.

他们热衷于提升技能，热衷于使用智能代理。

You know, they lean into skills, they lean into agents.

MD。

MD.

MCPs。

MCPs.

是的。

Yes.

对。

Yeah.

所有这些。

All of that.

当我与一些公司交谈，并与实际使用这些工具的员工交流时，发现他们试图做的都是最基础的事情。

And when I talk to some of these companies and I talk to the actual employees using these, it's like the most basic thing that they're trying to do.

他们对这项技术的工作原理几乎一无所知。

And they like have very little understanding of exactly how this technology works.

因此，这对我来说是一个重要的观察：他们只是在向这些工具提出非常简单的问题。

And so that's kind of like one big observation for me, which is like, they're asking very simple questions of these things.

他们目前还没有真正去深入挖掘它。

They're really not pushing it just yet.

因此，这又回到了我认为更多公司应该做或实际该做的事情，以及更理想的AI部署模式是什么样子。

And so that kind of goes back to, that's kind of ties into what I think more companies do or like what should do, or what a more ideal AI deployment setup looks like.

这也是我们在OpenAI内部运作的方式。

And this is kind of how we've run things within OpenAI too.

我认为真正取得良好成效的公司，都具备自上而下的支持。

The companies where I think it started to work really well, have a combination of both top down buy in.

也就是说，高管层明确表示：我们要成为一家以AI为先的公司。

So it's like the C suite, like, we want to become an AI first company.

因此，他们不仅认同并采购了这些工具，还获得了管理层的支持，同时也实现了自下而上的采纳与认同。

So there's buy in, they buy the tools, they have exec support, but it also has bottoms up adoption and buy in.

我的意思是，有真正的员工在积极使用这项技术，他们对它充满热情，愿意学习、推广、建立最佳实践，并在组织内部分享知识。

And so what I mean by that is it has like actual employees doing the work who are really excited about this technology and are willing to learn, evangelize, build best practices and kind of like knowledge share within the organization.

我们在内部已经多次看到这种情况。

We've seen this a lot internally.

显然，OpenAI一直希望成为一家以AI为核心的公司，但真正开始蓬勃发展，是在Codex和这些工具推出之后，员工们能够亲自将它们应用到自己的工作中。

So like obviously OpenAI has always wanted to be a very AI centric company, but when it really started taking off was with the introduction of Codex and these tools where people then like actual employees themselves could start applying it to their work.

我认为你真的需要这样，因为归根结底，每个人的工作都截然不同。

And I think you really need this because at the end of the day, everyone's work is very different.

非常独特。

It's very unique.

软件工程不同于金融，也不同于运营、市场推广和销售。

Software engineering is different than finance, is different than operations, different than go to market and sales.

因此，工作中有很多最后一步的细节，必须通过自下而上的方式来完成。

And so there's like a lot of these last mile intricacies of work that needs to really be done in a bottoms up fashion.

所以我的感觉是，许多AI部署都没有实现自下而上的采纳。

And so my sense is a lot of these AI deployments don't have bottoms up adoption.

这就像一项强制指令，完全是自上而下的，与实际工作内容完全脱节。

It was like an exact mandate and it's extremely top down and is very divorced from what the actual work looks like.

结果就是，你面对的是一个根本不懂这项技术的庞大员工群体。

And as an end result, you end up with a giant workforce that doesn't really understand the technology.

就像，我知道我应该用这个，也许它还和我的绩效考核有关，但我并不确定该怎么做。

It's like, I know I'm supposed to use this and maybe it's like on my performance review too, but I'm not sure what to do.

他们环顾四周，发现没人这么做。

And they look around, no one else is doing it.

没有人可以学习。

There's no one else to learn from.

因此，我建议那些推动这一举措的公司，找一个或甚至组建一个全职的内部团队，作为这样的内部特遣小组，去全面探索这项技术的能力，将其应用到具体的工作流程中，进行知识分享，并激发那些可能想使用这项技术的人的热情。

And so my recommendation for companies kind of pushing this is find, or maybe even staff a full time team internally that is this kind of tiger team internally that can explore the full extent of the capabilities, apply to specific workflows, do the knowledge sharing, create excitement within folks who might wanna use this technology.

因为如果没有这样的团队，这项技术实际上很难上手。

Because in the absence of that, it's very difficult to it's actually very difficult to pick up.

你会把谁放在这个特遣团队里？

And who would you put on this TIGR team?

是工程师主导的吗？

Is it like engineer led?

根据你的经验，你发现是这样吗？

Do you find in your experience?

这是一个跨职能的团队吗？

Is it a cross functional sort of team?

是的，这很有趣。

Yeah, it's interesting.

因为很多公司根本没有软件工程师。

Because so also a lot of companies don't have software engineers.

所以我看到的模式是，通常是那些与软件工程相关、本质上是技术人员但不是软件工程师的人。

And so the pattern I've seen is it tends to be these like software engineering adjacent, like basically technical people, but are not software engineers.

我认为这些人最有可能对这项技术感到兴奋。

I think those are the ones who get tend to get most excited around this.

就像是，你知道的，也许是那个不写代码但非常喜欢使用这些工具的支持团队运营主管，或者是个Excel高手之类的人。

It's like, you know, maybe the it's like maybe the like, you know, support team operations lead who doesn't code, but loves using these tools and is like an Excel wizard or something.

所以他们是技术相关或编程相关，而且相当懂技术。

And so it's like technical adjacent or coding adjacent and pretty technical.

我看到的这些公司里，正是这类人真正被点燃并对此感到兴奋。

Those are the kinds of people I've seen in these companies who just really light up and get excited around this.

你通常可以围绕这样的人组建一个团队。

And you can usually build a team around that.

但确实，通常并不是软件工程师。

But yeah, it's like oftentimes not software engineers.

软件工程师可能会理解这一点，但并不是每家公司都有软件工程师。

Software engineers, I think will understand this, but not every company has software engineers.

实际上，这种情况还挺罕见的。

It's actually kind of a rarity.

很难找到他们。

They're hard to find.

他们很贵。

They're expensive.

所以是这些其他类型的人。

And so it's these other types of folks.

我听到的是，反模式是自上而下的。

What I'm hearing is the anti pattern is top down.

这是CEO直接找高管团队说：我们要全面拥抱AI。

This is very the CEO found an exec team just like, we are gonna go AI first.

我们要以AI为引领。

We're gonna lead into AI.

每个人的表现都将通过AI工具来评估，看你因AI而提升了多少生产力。

Everyone's gonna be judged on their performance using AI tools, how much your productivity is increasing thanks to AI.

但如果只是自上而下推行，而不建立一个自下而上、传播理念的团队，你会发现这种方式行不通。

And without, with that being just top down and not creating a team that is bottom up, spreading the gospel, you find that doesn't work.

是的，没错。

Yeah, exactly.

没错。

Exactly.

建议是：找到那些最热情的人，不要让他们各自分散在组织中，真正有效的方式是组建一个小型的AI传道者团队，他们。

And the advice is, find the people that are most excited and instead of kind of having them spread out through the organization, you're what you find works is create a little t AI kind of evangelist team that Yeah.

会探索如何使用AI，并将其推广到整个工作中。

Finds ways to use it and kind of spreads it across the work.

是的。

Yeah.

我的意思是，这就像你把之前的话重新说给我听一样。

I mean, another it's kinda like hearing you you play back to me.

另一种思考方式，跟我自己的管理理念相关，就是找到那些在AI应用上表现突出的人，并给予他们支持。

Another way to think about it, kinda tying back to my own management philosophy is find the high performers in AI adoption and empower them.

你知道的。

You know?

让他们组织黑客马拉松。

Let them build hackathons.

让他们举办研讨会，进行知识分享，从内部培育起兴奋的种子。

Let them, you know, hold seminars, do knowledge sharing, kind of create the seeds of excitement internally.

好的，太棒了。

Okay, amazing.

我想听听你的一些大胆观点，一些我见过你分享过的看法。

There's a couple of hot takes I wanna hear from you, something that I've seen you talk about and share.

其中一个观点是，你曾提到，与客户交谈并倾听客户的意见，并不总是AI领域的正确策略，有时反而会误导你。

One is, you've shared that talking to customers and listening to customers is not always the right strategy in AI and it might often lead you astray.

我不确定这算不算一个很颠覆的观点。

I don't know if it's that hot of a take.

我认为这里最关键的是，你当然应该和客户交流。

I think the main thing here is so obviously you should talk to your customers.

和客户交流确实很有用。

Like, it's it's like useful to talk to customers.

我只是觉得，AI领域，尤其是我过去三年在开发API并见证其不断演变的过程中，这个领域和模型本身变化得太快了。

I just think the AI field, especially what I've seen over the last kind of like three years working on the API and seeing kind of all that evolve, is the field and the models themselves are just changing so, so quickly.

它们往往会自我颠覆，特别是在工具和框架层面。

They tend to like disrupt themselves, especially around the like tooling and the scaffolding space.

所以这周早些时候，我读到一句话，来自一个叫尼古拉斯的人在X平台上的文章，他是初创公司Fintool的创始人，他分享了许多在为金融服务构建AI代理过程中积累的最佳实践，我想他最初就是在Fintool做的。

So there there's this quote that I read actually earlier this week from a it's from an X article by this guy named Nicholas, who's who's the founder of a a startup called Fintool, where I think he was sharing a lot of the best practices that he has learned through building AI agents for financial services, I think at a start at Fintool.

他有一句话我觉得特别好，就是‘模型会把你搭建的框架当早餐吃掉’。

And he had this phrase that I thought was really good, which is the models will eat your scaffolding for breakfast.

如果你回溯到2022年，也就是ChatGPT刚发布的时候，这些模型还很原始。

Like, if you look if you rewind back to 2022, right when ChatGPT launched, these models are pretty raw.

当时在开发者领域，有大量产品架构和工具，主要是为了引导模型、围绕它构建架构，以让它完成你想要的任务。

And there was like all this product scaffolding and and things, especially in the developer space to basically try and steer the model and build a scaffolding around it to get it to do what you want.

比如智能体框架、向量存储，那时候这些都非常流行。

Like agent frameworks, there's vector stores, think was like really popular back then.

还有一大堆各种各样的工具。

And just like a whole smattering of tools here.

随着你观察到这个领域的发展，模型已经发生了巨大变化，变得好得多，以至于它们确实会直接‘吃掉’一部分架构。

And as you've kind of seen the field play out that the models have just changed so much that and gotten so much better that they end up, yeah, literally eating some of the scaffolding.

我认为这一点在今天也同样成立。

And I think this is even true today.

所以我认为，尼古拉斯的这篇文章提到的当前流行的架构，是基于技能文件的上下文管理。

So I think the article from Nicholas actually is, you know, the current scaffolding, which is fashionable is skills files based context management.

我能想象，未来某一天，这种方式可能不再有用，模型自己就能管理所有这些内容，或者可能会转向某种新范式，不再需要这种基于文件的技能系统。

I could see a world where at some point, you know, that's no longer useful, where the model can actually, you know, manage all that themselves or like, you know, or there might be, know, it's hard to predict, but like might move on to some new paradigm where you no longer need this file based skills type thing.

你确实亲眼见证了这种演变，对吧？

You have literally seen this play out, right?

像代理框架，我觉得现在用处小多了。

Like the agent frameworks, I think are a little less useful now.

在2023年左右，我们曾认为向量存储会成为向模型注入组织性上下文的主要方式。

There was a period of time like 2023 where we thought vector stores is gonna be like the main way for you to bring organizational context into the models.

你需要将整个语料库的每一部分都进行向量化和嵌入。

And you need to vectorize and then embed every bit of your corpuses.

然后你还要花大量精力去研究向量搜索，优化它，以便在正确的时间提取出正确信息。

And then you do all this work to figure out the vector search, to optimize that, to pull out the right information at right time.

所有这些都属于临时搭建的结构，因为当时的模型还不够强大。

All of that is scaffolding because the model was not good enough.

结果发现，随着模型变得更好，更优的做法其实是去掉大量这类逻辑，信任模型本身，并给它提供一组搜索工具。

And turns out, in this case, it turns out as the models get better, a better approach is actually to take out a lot of that logic and trust the model and give it a set of tools for search.

它并不需要一个向量存储。

It doesn't need to be a vector store.

你完全可以把它连接到任何类型的搜索系统上。

You could actually just hook it up to any type of search.

它完全可以是文件系统上的文件，比如 skills 和 agents MD，用来引导它。

It could literally be files on a file system like skills and agents MD to kind of steer it as well.

当然，向量存储仍然有其用武之地。

Obviously, there's still a place for vector stores.

我知道很多公司仍在使用它，但围绕它构建的整个基础设施、围绕它建立的整个生态系统，以及认为这就是唯一需要的基础设施的观念，已经发生了巨大变化。

I know a lot of companies are still using it, but the entire scaffolding around that and building an entire ecosystem around that and assuming that's the only scaffolding that you need has really changed.

因此，回到这一点：你并不总是必须听从客户的意见，因为这个领域变化太快，很多人现在都处于局部最优状态。

And so tying this back to the like, you don't always have to listen to your customers Because the field is changing so much at any point in time, a lot of people are kind of in this local maximum.

如果你盲目听从客户，他们会说：是的，我想要一个更好的向量存储。

And if you just blindly listen to your customers, they'll be like, yeah, I want a better vector store.

比如，我想要一个更好的代理框架来解决这个问题。

Like, I want a better agent framework for this.

如果你只一味追随这条路径，实际上会让你去构建一个再次陷入局部最优的东西。

And if you had just kind of only chased down that path, it actually would have led you to build something that again is the local maxima.

而随着模型变得越来越好，我们必须重新发明并重新思考围绕这些模型构建的正确抽象、工具和框架。

Whereas as the models get better, we've had to reinvent and kind of rethink the right abstractions and the right tools and frameworks to build around these models.

最酷、最令人兴奋却又有点烦人的是，这是一个移动的目标。

And the coolexcitingkind of crazy annoying part is it's a moving target.

因此，目前这些零散的工具和框架很可能需要随着模型变得越来越智能而发生显著的演变和改变。

And so, like the current smattering of tools and frameworks right now will likely need to evolve and change pretty significantly over time as the models get smarter and better.

但这就是构建这个领域的真实本质。

But that is just the nature of building this space.

我认为这正是它令人兴奋的原因。

Think that's what makes it exciting.

但这也意味着，当你与客户交流时，你需要平衡他们提出的具体反馈与你对模型未来走向以及未来一两年内趋势的判断。

But it also means when you talk to customers, you kind of need to balance the exact feedback that they want with where you think the models are going and where you think things will trend over the next one or two years.

有趣的是，这正体现了AI和机器学习领域学到的苦涩教训：你越少过度复杂化，越少在机器学习和AI中添加人为逻辑，它就越能扩展和成长，索性全部剥离，只让系统纯粹地计算，只需赋予它更多算力即可。

It's interesting how this is, the bitter lesson is, you know, this big lesson that AI and ML folks learned, which is just like, don't the less you overcomplicate, the less logic you add to to machine learning to AI, the more it'll be able to scale and grow and just like take it all away and let it just compute basically, just give it more power to get

是的，这实际上是苦涩教训在AI构建中的一个版本——我们曾经试图围绕它设计所有这些架构，结果发现模型自己就把这些都消化掉了。

Yeah, there's smarter on its literally a version of the bitter lesson applied to like building with AI where, you know, we were trying to architect all this stuff around and turns out the models have just kind of, you know, eat it all away.

老实说，OpenAI API团队也犯过类似的错误，我们曾经在不该转弯的时候走了弯路。

And and and and honestly, like OpenAI API team has like been guilty of this where we kind of like took some, you know, left and right turns when we shouldn't have.

但模型仍在不断进步，我们每天都在反复学习这个苦涩的教训。

But, the models still end up models get better and we're all learning the bitter lesson day in and day out.

那么，对于那些正在使用API或构建智能体、目前不得不围绕这些技术做一些开发的人来说，关键启示是什么？

So what would be the the key takeaway for folks building on, say, the API or just building agents and, you know, having to build a little bit of this around for now?

就是是的。

Is it just yeah.

有什么建议吗？

What would be the advice?

我的总体建议是，我长期以来一直这样告诉别人，而且我认为今天依然适用：确保你为模型未来的发展方向而构建，而不是为它们今天的现状而构建。

My general advice, and I've been giving this to people for a while and I think it's still true today, is make sure you're building for where the models are going and not where they are today.

你知道，这明显是一个移动的目标。

You know, the the it's it's clearly a moving target.

我认为我见过的许多公司，尤其是初创企业，之所以能做得特别好，是因为它们为一种理想能力构建了产品，而这种能力今天大约已经实现了80%。

And I think a lot of the companies that I've seen, startups that I've seen really, really do well, is they build a product for an ideal type of capability that is like maybe 80% of the way there today.

于是，它们最终推出的产品看起来差不多能用，但就是还差那么一点点。

And it like, they end up having a product that like kind of works, but it's like just almost there.

但随着模型变得越来越好，突然间一切就贯通了，他们的产品也因此变得极其出色，因为现在它真的能用了——比如在某个时刻，它突然就运行得起来了。

But then as the models get better, suddenly it might click and then their product now is incredible because it works, like like maybe with like three at some point, it suddenly works.

但到了5.1、5.2版本，它突然就解锁了。

But 5.1, 5.2 suddenly it unlocks it.

但他们构建这些产品时，始终考虑着模型能力的持续提升。

But they're building these products with the model capability improvements in mind.

这样一来，你最终创造出的体验，远比一开始就假设模型能力固定不变要好得多。

And with that, you end up creating an experience that's way better than if you had assumed that it's it's static in the first place.

因此，这就是我的总体建议：要为模型未来的发展方向而构建，而不是基于它们今天的状态。

And so that would be my general advice, which is, you know, build for where the models are going and not where they are today.

你会打造出更好的产品。

You end up building a better product.

你可能需要稍微等一等，但要知道，模型的进步速度实在太快了，你通常并不需要等太久。

You may need to, you know, like wait a little bit, but like, know, the models are getting so much better so quickly, you often don't need to wait that long.

那么顺着这个思路，未来六到十二个月，API的发展方向会是什么？

So to follow that thread, where are like in the next six to twelve months, where is the API heading?

平台的发展方向是什么？

Where's the platform heading?

模型的发展方向是什么？

Where are the models heading?

你能分享多少就多少吧，我知道这里有很多机密，也许你更兴奋的是那些，或者你觉得人们应该开始为哪些方面做准备？无论你能分享多少都行。

As much as you can share, I know there's a lot of secrets here that maybe you're more excited about or do you think that people should start to prepare for and however much you can share?

我的意思是，很明显的一个问题是，这些模型能连贯地完成多长的任务。

I mean, so the obvious one is, how long of a task, these models can do coherently.

比如有一个叫Meter的基准测试，它追踪软件工程任务，看看这些模型在50%的时间和80%的时间内能完成多长的任务。

So there's like the the meter benchmark that that I think tracks software engineering tasks and how long, you know, like, how long of a task can these models do 50% of the time, 80% of the time.

我认为，目前前沿模型已经有50%的概率能够完成数小时的软件工程任务。

I think we're at something like multi hour tasks being able to be done by software engineering tasks being able to be done by these frontier models 50% of the time.

而80%的概率对应的大概是不到一小时。

And then I think 80% is something like just under an hour.

但关于这张图表令人清醒的一点是，它还把所有之前的模型都画在了上面。

But the the the sobering thing about that that chart is they plot all the previous models on this chart as well.

所以你能清楚地看到这个趋势。

So you can really see the trend of this.

这正是我非常兴奋的一点，我觉得如今的产品其实主要优化的是模型能连续执行几分钟的任务。

That's something that I'm really excited about, which is, you know, I actually think products today really optimize for tasks that the model can do for like minutes at a time.

即使是代码工具和类似的东西，我觉得也主要是在命令行界面中。

Like even codecs and like the coding tools, I'd say like, you know, it's in the CLI.

你正在看到它变得更具交互性。

You're kind of like seeing it be interactive.

它确实非常优化，适合最多十分钟左右的任务。

It's really, you know, quite optimized well for like maybe at most ten minute type tasks.

我见过有人把代码工具推到极限，执行长达数小时的任务。

I have seen people push Codecs to the limit and do like multi hour long tasks.

但同样，我认为这更多是例外情况。

But again, I I I think that that's more of the exception.

但如果你关注这个趋势，我觉得在未来十二到十八个月内，我们可能会看到模型能够非常连贯地执行数小时长的任务。

But I if you follow this trend, like, I think, like in the next twelve to eighteen months, we could see models that could do multi hour long tasks very, very coherently.

在某个时候，它可能会达到每天六小时的任务长度，你可以将任务分配出去，让它自己独立运行一段时间。

At some point it might reach like, you know, six hours a day long task where you kind of like dispatch it and have it do things on its own for a while.

围绕这一点构建的产品将会非常不同。

The types of products you build around that will look very different.

你想给模型提供反馈。

You wanna give the model feedback.

你显然不希望它一整天完全失控，也许你希望，但大概率你并不希望。

You obviously don't want it to completely run wild for a day, maybe you do, but you probably don't.

然后，你能让模型完成的事情范围将大大扩展。

And then the universe of things you can have the model do really expand.

所以，这是我非常期待看到的进展。

So that's something that I'm really excited about seeing.

在未来十二到十八个月里，另一个让我非常期待的方面是多模态模型的改进。

Another thing over the next twelve to eighteen months where I think would be really cool is improvements in the multimodal models.

说到多模态，我主要指的是音频方面。

So and and actually by by multimodality, I'm mostly thinking about audio.

在这里，模型在音频方面已经相当不错了，我认为在未来六到十二个月内，它们在音频上的表现会大幅提升，尤其是原生的多模态模型和语音到语音的模型。

Here where the models are pretty good at audio, I think they're gonna get a lot better at audio over the next six to twelve months, especially the likes, native multimodal model, the speech to speech ones.

我认为在多模态音频方面，围绕新型模型和架构的研究也在进行中。

I think there's also interesting work being done around new types of models and architectures on the multimodal audio side as well.

但音频，尤其是在企业与商业环境中，仍然是一个被严重低估的领域。

But audio, especially in the enterprise and in a business setting, think is a hugely underrated domain still.

每个人都谈论编程。

Like everyone talks about coding.

一切都在文本中。

It's all text.

但我们实际上是在用音频交流。

But we're talking in audio.

世界上大量的商业活动都是通过音频完成的。

A lot of the world's business is done via audio.

许多服务和运营都是通过交谈和音频进行的。

A lot of services and operations are done via talking and audio.

所以我认为在未来十二到十八个月内，这个领域将会变得非常令人兴奋。

And so I think that that area is gonna look very exciting in the next twelve to eighteen months.

我认为在音频模型方面，我们还能解锁更多可能性。

And I think there will be even more unlock for what we can do with with audio models there as well.

太棒了。

Amazing.

简单总结一下。

So quick summary.

预计智能体和AI工具将沿着这一趋势持续发展，音频和语音将变得越来越重要，成为更主流、更原生、更出色且核心的体验。

Expect agents and AI tools to run longer to that that trajectory to continue to increase, and then audio and speech becoming a bigger deal, more first party and and native and better and core to the experience.

是的。

Yeah.

非常酷。

Extremely cool.

好的。

Okay.

我想回到你之前的一个大胆观点，另一个我见过你讨论过的观点。

I wanna go back to one of your hot takes, another hot take that I've seen you discuss.

你对商业流程自动化在人工智能时代的机会非常乐观。

You're big you're very bullish on business process automation as an opportunity in the world of AI.

谈谈这个。

Talk about that.

是的。

Yeah.

这其实回到我之前说过的一点：我们生活在硅谷的泡沫里。

This go this goes back to the thing that I said previously, which is we we live in a bubble in Silicon Valley.

我们所从事的很多工作——比如软件工程、产品管理、开发产品——其形态与支撑整个经济运转的工作截然不同。

And a lot of the work that we do, that we're used to, software engineering, product management, building products, is very differently shaped than the work that goes on that runs our entire economy.

我在与客户交流时经常看到这一点。

And I see this in and out when I talk to customers.

如果你跟任何一家非科技公司交谈，就会发现他们有大量的业务流程。

If you talk to any company that's not based in, it's not a tech company, there's a lot of business processes.

我所说的这意味着，一般来说，我会将其区分为：软件工程本质上是一种开放式的知识工作，对吧？

And so what I mean by this is, you know, I generally delineate it as, you know, there's like like software engineering is kind of like open ended knowledge work, right?

这就是为什么我认为像Codex这样的工具表现得相当出色，因为它是在探索，而你给它的是这些开放性的问题。

It's like and this is why I think tools like Codex tend to be quite quite good because it's exploring and you're giving it these like open ended things.

但软件工程本质上是非常开放的，并且不太可重复，对吧？

But software engineering is fundamentally like pretty open ended and it's not very repeatable, right?

所以你开发一个功能时，并不是试图一遍又一遍地构建完全相同的功能。

So like you build a feature, you're not trying to build the exact same feature over and over again.

很多科技类工作都属于这一类。

And a lot of like tech jobs are in the space.

我认为数据科学也属于这一范畴。

I think like data science is kind of in this space as well.

甚至一些战略财务工作也是如此。

Even some of the like strategic finance stuff.

但当你越来越远离软件工程以及科技的核心领域时，许多工作其实就是业务流程。

But as you move further and further away from software engineering and like what is core and tech, a lot of jobs are just business processes.

这些是可重复的事情、可重复的操作，由公司里的某个经理不断优化过。

They're like repeatable things, repeatable operations that some manager at a company has kind of like iterated on.

通常都有标准操作流程，人们希望遵循它，不太愿意偏离。

There's usually a standard operating procedure that people wanna do and you don't wanna deviate from it that much.

在软件工程中，创造力体现在不偏离，但世界上很多工作实际上只是在执行这些流程和操作。

There's like in software engineering, the ingenuity isn't deviating, but a lot of the work being done in the world is actually just running through these procedures and operations.

比如，如果我拨打客服热线，他们就是在执行其中某一个流程。

Like if I call a support line, they're running through one of these.

如果我联系公用事业公司，他们对我能做和不能做的事都有明确的流程。

If I call my utility company, there's a bunch of processes and things that they can and cannot do for me.

因此，我对这一大类工作极度看好，我认为它被低估了，因为它和硅谷的思维方式截然不同。

And so I'm just extremely bullish on this general category of like, and I think it's underrated because it's so different from what we think about in Silicon Valley.

人们往往忽视这一点，但如何将AI以及我们现有的工具和框架应用于这种业务流程自动化，自动化并简化那些具有高度确定性、并与企业数据、业务决策和不同系统深度集成的可重复业务流程呢？

People tend to not think about it, but how can we apply AI and some of the tools and frameworks that we have towards this business process automation, towards automating and making easier repeatable business processes with high determinism that is fully integrated with business data and business decisions and different systems within an enterprise.

我们该如何真正改善这一过程？

And how can we actually make that that process better?

因为我真的认为这个领域有很多机会和大量工作要做。

Because I actually think there's a lot of opportunity and a lot of work to be done in that area.

我们只是不谈论它，因为它稍微超出了我们的专长范围。

And we just we just don't talk about it because it's it's a little bit less in our wheelhouse.

为了确保我完全理解，你的观点是：你认为在工程之外，AI 对企业生产力以及从事这些重复性、易自动化任务的员工的工作影响会更大吗？

Your take here, just to make sure I fully understand it, is you think there's a much bigger opportunity outside of engineering for AI to impact productivity of companies and also jobs of these folks that are doing these kind of repetitive, easily automated tasks?

影响工作，同时也影响工作的方式。

Impact jobs and also just impact how work is done.

很多工作都是以这种方式完成的。

So much of work is done in this way.

你想想，经常与客户交流，大型企业，AI 将如何改变我的公司？

You think about what a Basically, talk to customers all the time, big enterprises, how will AI transfer my company?

在拥有 AI 的世界里，二十年后它会如何运行？

How will it run-in a world with AI in like twenty years?

软件工程只是故事的一部分，但在业务流程方面还有更多内容。

Software engineering is part of the story, but there's so much more on the business process side.

而且我认为，在业务流程方面，情况可能会显得更加不同，那里的工作量也相当大。

And I actually think it might look even more different on the business process side and the work there is pretty substantial.

这其实挺有意思的。

Actually interesting.

我不知道，从绝对比例或绝对基数来看，我不确定它是否比软件工程更大或更小。

I don't know, like, from an absolute percentage or absolute base, I don't know if it's bigger or smaller than software engineering.

软件工程本身也非常庞大和广泛，但业务流程的规模确实非常巨大。

Like software engineering is pretty huge and pretty extensive as well, but it is pretty massive.

而且它肯定比人们在X平台或推特上谈论或不谈论时所认为的要大得多。

And it's definitely bigger than, you know it's it's bigger than you would think it is based off of how how people talk about it or don't talk about it on X or Twitter.

好的。

Okay.

换个稍微不同的方向，既然你构建了一个平台、开发了API，让别人基于API进行开发，人们最关心的问题总是：我该如何避免被OpenAI碾压，让他们复制我的创意，然后摧毁我创造的市场？

Going in a slightly different direction, having built a platform, building the API, people building on API, the biggest question on people's minds is always just, how do I not have OpenAI squash my idea and build their own thing and then, you know, destroy this market I created?

总体的政策是什么？

What's the general policy?

初创公司应该如何思考OpenAI不太可能涉足的领域，其总体理念是什么？

What's the general philosophy of how startups should think about where OpenAI is unlikely to go?

我的总体回答是，这个市场太大了，太庞大了。

My general answer here is is the market is so big and so massive.

实际上，我认为初创公司根本不需要过度担心OpenAI或这些实验室会往哪个方向走。

Like, I actually think you know, startups should just not overly think about where OpenAI or these labs are going.

我接触过很多初创公司，有的失败了，有的做得非常好。

I've talked to a lot of startups, you know, that have, you know, not worked out, startups that are doing really well.

我见过的每一个最终失败的初创公司，都不是因为OpenAI、BigLab、谷歌之类的大公司来抢走它们的市场。

Every startup that I've seen that is kind of fizzled out is not because OpenAI or, you know, BigLab or Google or something has has come to swatch them.

而是因为它们打造的产品根本没能引起客户的共鸣。

It's because they built something and it like really didn't resonate with with the customers.

而那些成功的公司，即使在像编程这样竞争激烈的领域，比如Cursor，现在已经非常庞大了。

Whereas the ones that take off, like even in very competitive spaces like coding, like Cursor is huge at this point.

这是因为它们打造了人们真正喜爱的产品。

And it's because they build something that people really love.

所以我的总体建议是，别为此太过焦虑。

And so my general advice is like, don't overly stress about this.

只要打造人们喜欢的东西，你就能在这个领域占有一席之地。

Just build something that people like and you will have a space in this.

我再怎么强调现在的机会有多大都不为过。

I can't overstate how big of an opportunity there is right now.

就像利用AI进行创业的机会空间实在太大了。

Like the opportunity space and building with AI is so big.

一个很好的例子是，这个领域如此广阔，以至于风投认为什么是可接受、什么是不可接受的舆论边界已经彻底改变了。

A good example of this is like the space is so big that the Overton window of what is acceptable and not acceptable for VCs to do has completely changed here.

风投们正在不断投资各种竞争性公司。

VCs are like investing in like competitive companies left and right.

这整个领域之所以如此广阔，是因为机遇前所未有。

It's just like the space is so big because the opportunity is unlike anything that we've seen before.

尽管这影响了风投的运作方式，但从创业者的角度来看，这是世界上最令人振奋的事，因为即使你只是打造了某些人真正、真正喜爱的产品，你也终将拥有一个价值巨大的企业。

And while that affects how VCs operate, from a startup perspective, it's the most empowering thing in the world because even if you just build something that some people really, really love, you will end up with a massively valuable business.

所以这就是为什么我会告诉人们，别太在意你怎么想。

And so that's why I tell people, don't know what you think about it.

另一件我认为很重要、至少从OpenAI的角度来看的是，我们始终非常重视的一点，即萨姆和格雷格也从高层不断强调：我们本质上把自己视为一个生态系统平台公司。

The other thing I also think is important to remember, at least from an OpenAI perspective, one thing that we've always held very near and dear, which both Sam and Greg help, you know, reinforce from the top as well, is we actually view ourselves fundamentally as a like ecosystem platform company.

API 是我们的第一个产品。

The API was our first product.

我们觉得，培育这个生态系统并持续支持它、而不是压制它，对我们来说至关重要。

We think it's really important for us to foster this ecosystem and continue to, you know, support it and and not squash it.

所以如果你仔细看看我们所做的决策，这一切都是我们一贯坚持的。

And so if you kind of look at the decisions we make, it this is all we've we've through it.

我们发布的每一个模型，都会通过API对外提供。

Every single model we've released in one of our products gets released in the API.

比如，我们现在发布的这些Codex模型虽然更针对Codex框架做了优化，但它们最终都会进入API。

Like even, you know, we release these codex models now that are a little bit more optimized for the codex harness, but they always find their way into the API.

而且，我们所有的客户最终都会把这些功能用到极致。

And like all of our, you know, customers end up abusing those.

我们对此从不保留。

We don't hold back on any of that.

我们认为保持平台中立非常重要。

We think it's really important to keep our platform neutral.

所以，不要封锁竞争对手。

And so, don't block competitors.

我们允许人们使用我们的模型。

We allow people to have access to our models.

我们也希望像最近一样，更多地测试登录ChatGPT产品。

We also want, like we've recently been testing more of like the sign in with ChatGPT product as well.

因此，我们希望培育这个生态系统。

And so we want to foster this ecosystem.

我认为这样做非常重要。

I think it's really important that we do so.

关于这一点的普遍看法是，潮水涨起，所有船只都会随之上升，而我们可能是一艘航空母舰。

The general thinking about this is a rising tide lifts all boats and we might be an aircraft carrier.

我们目前规模已经相当大了，但我们认为提升整体水位很重要，因为每个人都会受益，我相信我们自己也会从中受益。

We're pretty big at this point, but we think it's important to raise the tide because everyone kind of benefits and I think we'll benefit as well.

我们的API本身已经显著增长，这正是因为我们的这种做法。

Like our API itself has grown pretty significantly because we act in this way.

因此，我真心鼓励大家不要把OpenAI看作一个会把别人挤出市场的存在，而应该专注于创造有价值的东西。

And so I'd really encourage people not to view OpenAI as this kind of you know, thing that'll just shove people out of the way, but instead focus on building something valuable.

我们始终致力于打造一个开放的生态系统。

And we, you know, remain committed to providing an open ecosystem.

为什么这对OpenAI如此重要？为什么你们如此专注于构建平台，为人们创造创业机会？

Why is that important to OpenAI, just this focus on building a platform, creating a way for people to build businesses?

这从一开始就是你们的愿景吗？

Just like is that just that's been the vision from the beginning?

我们希望这能成为一个平台。

We want this to be a platform.

这从一开始就是我们的愿景。

It's been the vision from the beginning.

这其实可以追溯到我们的章程，也就是我们的使命。

It comes goes back to our charter actually, like our mission.

所以，OpenAI的使命一直以来都是第一点：开发通用人工智能。

So the the OpenAI's mission has always been to one, to build AGI.

所以，我们显然正在做这件事。

So, you know, we're obviously doing that.

但第二点是，要让全人类都能受益于它。

But then the second thing is to like spread the benefits of it to all of humanity.

这其中的关键部分，就是‘全人类’这一点。

And there's kind of like a lot of, you know the main part there is all of humanity.

比如，ChatGPT 就在努力实现这一点。

Like and obviously, ChadGPT is trying to do this.

我们正试图触达尽可能多的人，整个世界。

We're trying to reach however many, the whole world.

但早在早期，这就是我们为什么在2020年左右就推出API的原因。

But very early on, and this is why we launched the API back in, I think it was like 2020 or something, like really early.

我们不认为作为一家公司，我们能够触及全人类，对吧？

We don't think we as a company will be able to reach all of humanity, right?

世界上每一个角落都相当深远。

There's, I don't know, every corner of the world is pretty deep.

因此，我们实际上认为，为了实现我们的使命，我们需要一种平台式的思维，赋能其他人去为播客主和通讯录作者构建客服机器人，因为我们自己无法做到这一点。

And so we actually feel like in order for us to fulfill our mission, we need to have some platform style think here where we can empower other people to build the customer support bot for podcasters and newsletter hosts because we're not gonna be able to do it ourselves.

因此，我们已经很大程度上看到API在这方面发挥了作用。

And so we've largely seen this play out with the API.

这就是为什么我们与这么多客户交流，并且非常乐于看到基于我们平台构建的多样化应用。

This is why we, know, we talk to so many of our customers and and and really, you know, love seeing the diversity of of things built on.

但事实上，从第一天起我们就一直如此，因为我们将其视为一种授权的体现。

But, yeah, it it's been there since day one because it's it's it's kind of we view it as an expression of permission.

你甚至还没提到你们正在推出的App Store，即ChatGPT App Store。

And you haven't even mentioned the the App Store that you guys are launching, ChatGipitsi App Store.

是的。

Yeah.

顺便问一下，这是在你们的管辖范围内，还是另一个俄勒冈团队？

Is is that under your umbrella, by the way, or is that a different Oregon team?

这是一个不同的团队。

It's a it's a different team.

所以它属于ChatGPT。

So it's under ChatGPT.

我们显然与他们紧密合作，而且他们开发了一个应用SDK，是与我们的团队密切协作完成的。

We obviously collaborate very closely with them and, you know, they built like an apps SDK, which is built in close collaboration with our team.

但那更多是在ChatGPT的范畴内。

But that is more within the ChatGPT umbrella.

但这也是另一个类似的例子。

But that is also another like, that's another example of this.

对吧？

Right?

就像ChatGPT，我们有大约8亿周活跃用户，他们不断回来使用。

It's like ChatGeeBT is like we we we we we kind of like have these 800,000,000 weekly active users who are just coming over and over again.

这作为一个企业来说是个巨大的优势，但话说回来，如果我们能允许其他公司也进来，利用这一点，为这个受众群体开发产品，岂不是更好？

Like, it's a great asset to have as a business, but like, man, would it be better if we could somehow allow other companies to come in and take advantage of this as well and build for this audience as well.

然后，最终我们认为这也有助于我们扩大这个群体。

And and then ultimately, we think it'll help us expand that that that group as well.

对吧？

Right?

所以这一切其实都回归到我们的使命，我们发现作为开放平台会在这里起到积极作用。

And so it's all it all kinda comes back to the mission, and we find that being a platform being open tends to help here.

就这个数字，8亿，我认为是月活跃用户吗？

Just that number, 800,000,000, I think it's m MAs?

不是的。

Just like No.

每周。

Weekly.

每周。

Weekly

每周活跃，是的。

act Weekly act yeah.

这太疯狂了。

It's crazy.

十亿人每周使用。

Billion people using weekly.

简直难以置信，我们现在对这些数字都习以为常了，但这真的太惊人了。

Just like, it's absurd how many how these numbers we're just used to now, but that's in insane.

史无前例。

Unprecedented.

是的。

Yeah.

从规模的角度来看，说实话，这让我觉得难以想象。

It's it's mind boggling for me to think about from a scale perspective, honestly.

我是这么想的：占全球人口的10%，而且还在快速增长，这数字简直在飙升。

And the way I think about it is like 10% of the world and growing by the way, like it's just, it's it's shooting up.

来使用ChatGPT，每周都用一次。

Come to chat GPT and and use it every day, or sorry, every week.

在这一点上，我想再强调一下你提到的观点。

And this point, I just wanna double down on this point you're making.

OpenAI的使命是让人工智能惠及全人类。

OpenAI's mission was to make AI available to all of humanity.

有些人只是觉得，哦，这要花钱。

And I think some people just that, they're like, oh, know, costs money.

但事实上，ChatGPT有一个免费版本，任何人都可以使用，这个版本与世界上最强的AI模型相差无几，而且免费、开放、人人可用。

And it's like, like the fact that it it's there's a free version of ChatGPT that anybody can use that is not so different from the most powerful AI model that exists in the world for free, that's not gated, that anyone can use.

比如，即使你是亿万富翁，你能从AI中获得的收益，也未必比非洲一个村庄里的人多多少。

Like, if you have if you're a billionaire, there's only so much more you can get out of AI than what someone, you know, in a village in Africa can get.

我知道这一直对OpenAI至关重要。

And I know that's always been really important to OpenAI.

是的。

Yeah.

是的。

Yeah.

我的意思是，你看，这就是为什么我们认为我们专注于健康领域是正确的。

I mean, look, that's why I think we've leaned into the health work.

我们还关注教育，这将是一个非常有趣的领域。

We've leaned into like education is gonna be a very interesting here.

这里另一个疯狂的趋势是，免费模型随着时间的推移变得越来越智能。

The other insane kind of trend here is is the free model has gotten so smart over time.

比如，2022年的免费版本虽然当时还不错，但跟今天的东西完全没法比，因为现在你能用到的是GPT-5。

Like the free model back in 2022 was, you know, like, well, it was good at the time, but it's like nothing compared to what you get today because you get GPT-five today.

所以，提升全球整体水平，这正是我们真正努力在做的事情。

And so the like, you know, raising the floor across the world is kind of, you know, something that we're really trying to do.

我们认为这是我们的使命的一部分。

We view it as part of our mission.

顺便说一下，另一方面，就是关于那些亿万富翁之类的话题。

The other flip side of this, by the way, is like, you know, kind of talking about like the billionaires or whatever.

我知道有人会说，你用的iPhone和史蒂夫——抱歉，是马克·扎克伯格那样的亿万富翁用的一样。

I know people are saying like, you're using the same iPhone that like, you know, Steve or sorry, like Mark Zuckerberg's probably using or like the billionaires are using.

但只要你每月花20美元，你实际上就在使用和亿万富翁一样的AI。

But for like $20 a month, you're basically using, you know, like using the same AI that, you know, the billionaires are using.

每月花200美元，你就能获得和所有亿万富翁一样的专业版，但他们可能并不会全程都用专业版。

For like $200 a month, you get the same pro model that, you know, all the billionaires are using, but they're probably not using pro for everything.

他们大概只是用高级版来应付日常需求。

They're probably just using the plus tier ones for their day in and day out.

因此，这种技术的民主化，以及让这种福利惠及全球，对我们来说意义重大，也是推动我们很多工作的动力。

And so, this kind of like democratization and just like spreading of this benefit, like across all of the world has seen us really meaningful to us and something that drives a lot of what we do.

最后一个问题是，给那些正在考虑基于API开发，或者想着‘等等，我可以用OpenAI的模型和API做点酷炫东西’的人。

One last question, just for folks that are thinking about building on the API or just like, oh, wait, I could do cool stuff with OpenAI's models and APIs.

你的API和平台能让人做些什么呢？

What what does your API and and platform allow people to do?

我知道你可以在平台上构建智能代理。

Like, I know you can build agents on top of the platform.