AI 吞噬世界？与本尼迪克特·埃文斯一起进行现实核查

他创办了一家公司，最后卖了一亿美元。

He built a company and sold it for like a $100,000,000.

他们确实有，但这就引出了第二个问题——这些问题已经超出了人工智能的范畴，变成了一半是AI、一半是别的领域的问题。

They've obviously does, but like, so that's kind of a second question, which is a question that's move outside of AI and they become sort of half AI questions, half something else kind of questions.

再往下是第三个层面，其实我早该提一下：这一切和以往的平台变革根本不同。当年3G、iPhone、互联网刚出现的时候，你不知道接下来会发生什么，但你至少清楚物理上的限制。

And then the third level, which is, think, and I probably should have said this earlier, the way that all of this is sort of fundamentally different from previous platform shifts, is that with three gs or the iPhone or the web or whatever it was, you didn't know what was going to happen next, but you knew the physical limits.

比如1995年，你知道电信公司不可能下周就给全世界人普及宽带，你也知道不可能人人都去买电脑，因为一台电脑要三千美元。所以你心里有数，哪些事能办到，哪些事根本不可能。

Like 1995, you knew that telcos weren't going to give everybody in the world broadband next week, and you knew that everyone in the world wasn't going to go out and buy a PC because a PC cost like $3,000 So you kind of knew the basic physical limits of what could and couldn't happen.

但生成式AI不一样，我们完全没底。

And with generative AI, obviously we don't.

我们可能刚结束这段对话，一打开手机，就收到推送：OpenAI发布了新模型，价格只有原来的2%，因为他们搞定了什么关键技术。

We might look at our phones when we get off this recording and there's a push notification that says OpenAI's new model is out and it's like 2% of the price because they worked something out.

我觉得现在这种情况可能性不大，但关键是我们根本不知道这类事情会不会发生。

I I don't think it's very likely at this, but we don't know those kinds of questions.

所以模型的规模到底会膨胀到什么程度？

So how much bigger will the models get?

能好多少？

How much better?

能快多少？

How much faster?

能便宜多少？

How much cheaper?

模型的表现会有什么变化？在哪些方面会不一样？

How much pick up, in what ways will the characters of models change?

我们不知道。

We don't know.

这跟以前的技术变革不一样，那时候我们至少知道基本的限制在哪里。

And that is different to previous platform shifts where you did know the sort of fundamental constraints.

所以这些问题也会随之变化。

And so that will kind of spin off your questions.

某种程度上，这其实我前面也提过。

In a sense, this is something I pointed to earlier.

我现在觉得，真正实现了产品市场契合的领域只有编程。

I said like the place that's got product market fit right now is coding.

目前没有任何其他领域能跟编程相提并论。

Nothing else has equivalent product market fit right now.

我觉得我说得挺有把握的，Anthropic去年年底的年营收才90亿美元，现在涨到470亿了，这全是软件带来的，对吧？

I think I'm pretty safe in saying that when Anthropic's gone from whatever it was, 9,000,000,000 run rate at the end of last year to $47,000,000,000 run rate now, that's all software, isn't it?

那要是别的领域也有人做出了类似的东西，会怎么样？

So what happens when someone else in some other field gets something working?

是的。

Yeah.

我的意思是，

I mean,

你指的是哪个领域？

Which you add field?

比如法律？

Like, law?

那法律、银行呢？

Well, I law, bank?

我不确定具体是哪个领域，但总得有个地方吧。

I don't know where, but something.

如果你要猜的话，除了编程之外，还有哪些应用场景可能带来日常的活跃使用？

If you had to guess, what are the use cases outside of coding that could potentially yield, you know, daily activity?

我应该提一下，几周前我发布过一个演讲，大致分为三个部分。

I should say the sort of presentation that I published a couple of weeks ago, there's sort of three sections.

其中一个部分讲的是资本、资本支出、基础设施、基础模型和差异化，就是我们之前讨论的内容。

And one of them is talking about capital and CapEx and infrastructure and foundation models and differentiation, which is the stuff we talked about.

第二个部分是，该怎么用这种技术来开发软件？它对软件行业意味着什么？你的软件会变成什么样？

And the second is, well, how would you build software with this and what does this do for the software industry and what would your software look like?

公司的利润率又会怎样变化？

What happens to the margins of the companies and everything else?

第三个部分我称之为变革，这正好说到点子上了。

And the third section I called change, which is kind of getting to this point.

我开头又提到了一个让某些人特别不爽的说法，就是瑜伽·贝拉那句老话：预测很难，尤其是关于未来的事。

And I opened it with, again, what appears to upset a certain category of person where I said, the Yogi Berra quote that predictions are hard, especially about the future.

我觉得这其实是个回溯点——想象一下，要是你在1997年问这些问题，关于互联网会怎么样。

And I think this is sort of a backtest point, which is imagine asking these kind of questions about the internet in 1997.

你能得到什么答案？

What would you have got?

你又会错过什么？

What would you have not got?

但我觉得，换个角度看，这其实是一种自动化——它让过去根本没法自动化的那些事，现在终于能被自动化了。

But I think one way you can look at this is to say, well, this is automation, that this makes a class of thing that people used to do that couldn't be automated, now you can automate that.

那这又意味着什么？

And so then, well, what does that mean?

我提出了三四种可能的应对方向。

And I propose like three or four way buttons to press.

第一个是：这不过是价格弹性的问题吗？

First one is, is this just price elasticity?

这其实就是杰文斯悖论的核心。

Which is really what the Jevons paradox is.

打个比方，如果做事的成本降低了，你是会用更少的钱做同样多的事，

Like if you make it cheaper to do stuff, do you do the same amount of stuff for less money?

还是用同样的钱做更多的事，

Or do you do more for the same money?

或者是花更多的钱做更多的事？

Or do you more do more for more money?

因为成本变得低多了。

Because it becomes so much cheaper.

有没有以前做不了、现在变得便宜了的事情？

Was there something that you couldn't do before that now becomes cheap?

有没有以前很贵、你把它当作行业门槛的事情，比如报纸拥有印刷机？

Was there something that was expensive and you were doing as a barrier to entry, like owning a printing press as a newspaper?

有没有以前是门槛、也是成本基础、现在却消失了的门槛？

Is there something that now was a barrier to entry and a cost base at a barrier to entry that now goes away?

有没有什么在你的商业模式或竞争格局中，因为这件事变便宜了而突然被打开的可能性？

Is there something that gets unlocked in your business model or in your competitive space because this thing became cheap?

最后一个问题是，有哪些事情以前因为计算成本太高，根本没人敢想？

And then the final question would be, what stuff was just completely impossible, computationally cost prohibitive, so that nobody even thought about it?

而现在这些事已经触手可及了。

And now that's within reach.

我以前常举的例子是，蒸汽机让火车成为可能，不管你买多少匹马，都不可能跑出火车或者特快列车。

And the example I used to give here was like, well, steam engines make trains possible, and it wouldn't matter how many horses you buy, you couldn't have a train or like an express train.

更贴近现在的例子，可以看看YouTube，或者直接说Spotify。

Much more contemporary example would be to point to something like YouTube or indeed point to Spotify.

Spotify自己说过，看看过去二十五年音乐产业的变化。

Spotify says, one of the Look at the last twenty five years of the music business.

前半段是，当你不用花十五美元买一张CD就能听到那首歌的时候，会发生什么？

The first half is what happens if you don't have to buy a $15 CD to get that track.

后半段是，如果每月十五美元就能拥有所有音乐，这在过去是完全不敢想的事。

But then the second half is what if $15 a month gets you all the music that there is, which is something that was just completely impossible.

这类预测的问题在于，一方面你说的可能听起来挺聪明、挺显而易见的，但你其实并不清楚它对每个行业具体意味着什么。

The problem with these kind of making predictions like this is that on the one hand, you're going to say stuff that's kind of clever and obvious, but you don't actually know what it's going mean industry by industry.

如果我们回到上世纪九十年代末，说互联网会摧毁实体分销的价值，结果发现这对报纸和电影公司的影响完全不一样。

So if we'd been back in the late '90s and we'd said, internet destroy the value of physical distribution, it turned out that meant completely different things for newspapers and movie studios.

报纸因此彻底垮了，而电影公司却没怎么改变。

Newspapers got completely screwed by this and movie studios have of not really changed very much.

所以，这还得看具体情况。

So again, it depends.

但另一方面，有些领域我觉得还是可以问出更有价值的问题。

But the other kind of part of this is, there are some places where I think you kind of can ask more useful questions.

让我特别感兴趣的是：这会如何改变广告、电商、品牌和营销，以及我们买的所有东西？

And the one that sort of intrigues me is to say, well, how does this change advertising and e commerce and brands and marketing and everything that we buy?

广告市场价值一万亿，零售市场高达两万五千亿美元，这可是个相当大的市场空间。

Advertising is a trillion dollars and retail is $25,000,000,000,000 So, this is a reasonable sized TAM.

我一直觉得，谷歌、Meta 和亚马逊其实并不真正清楚这个产品到底是什么。

So the thing that I've always used to think about was that Google and Meta and Amazon don't really know what that product is.

他们只知道这是个商品编码。

They know it's a SKU.

他们知道出版商在元数据里填了什么内容。

They know what the publisher typed in the metadata field.

他们也知道买了这个的人通常也买了那个，但不明白背后的原因。

And they know that people who bought this also bought that, but they don't know why.

他们根本搞不清这些商品到底是什么，所以才会出现那种笑话——‘亚马逊，我买了个马桶座圈。’

And they don't really know what those things are, which is why you get these jokes about, Hey, Amazon, I bought a toilet seat cover.

我又不是在收集马桶座圈。

I'm not collecting toilet seats.

因为亚马逊根本不懂马桶座圈是啥，也不知道没人会买两个。

Because Amazon doesn't really know what a toilet seat is and doesn't know that people don't buy two.

其实他们早就该知道这一点。

Actually, they should know that.

这明明可以通过购买频率分析出来，但他们就是没做。

That should be frequency analysis, but they don't.

而有了大语言模型，原则上你就能明白这些东西到底是什么，人们为什么买，还会买些什么。

And with an LLM, in principle, you would know what those things are and why people buy them and what other things people buy.

当然，'不'这个词用起来挺棘手的。

And obviously no is like a difficult, tricky term to use.

你说的'不'，到底指的是什么？

What do you mean when you say no?

但至少，AI能实现的统计关联水平完全不同了，这正是谷歌和脸书每个季度广告数据和转化率飙升的原因——他们把这些全都整合进了广告系统、推荐引擎和预测算法里。

But at a minimum, a very different level of statistical correlation of what an AI system would be able to do, which is of course why you see the ad numbers and the conversion rates shooting up in every quarter from Google and Facebook, because they're rolling all of this into their ad systems and their recommendation engines and their prediction algorithms.

你会看到更多你感兴趣的东西，你卖的广告也更可能符合你的购买偏好。

And like you get shown more stuff that you would like, and the ads that you're selling are more likely to be things you'd like to buy.

于是他们的广告收入突然迎来了爆发式增长。

And so they have these enormous, this sudden acceleration in their ad revenue.

这一切都说明，现在的系统通常只会说：'买过这个的人，也可能买这个。'

And all of which is to say, you kind of look at how these systems work, and right now they say, well, people who bought that could buy this.

而你现在应该能说，就像我演示文稿里那张图一样：'这是件大衣。'

And you should now be able to say, like the slide I had in the presentation was like, here's a picture of a coat.

这怎么运作的？

What does it work?

我能买这个吗？

Can I buy that?

五年前，甚至十年前，这根本行不通。

And like five years ago, that really wouldn't ten years ago, that certainly wouldn't work.

五年前，大概也行不通。

Five years ago, probably wouldn't work.

现在应该可以了。

Now that should work.

然后你可以说：推荐十件类似的外套，价格不同，告诉我哪里能买，再分析每件的优缺点。

And then you can say, okay, suggest 10 other coats like that with different prices and tell me where I can buy them and suggest the pros and cons of each one.

这种功能现在也都能实现了。

And you'll kind of get that too.

再进一步，你可以说：看看我的Instagram，给我推荐一件冬季外套，能改变我的风格，但别改太多。

And then you can push one step further and say, look at my Instagram and suggest a winter coat I should buy that will change my look, but not too much.

再往前推三年，这简直就像科幻小说里的情节。

And again, like three years ago, that would have been total science fiction.

但现在你就会想，嗯，这种东西应该能做出来。

And now you think, yeah, you could probably build something like that.

这种功能大概能行得通。

That would kind of work.

还有就是，计算机能知道什么、能自动化什么、能给出什么建议，这些能力都在发生转变。

And those kinds of shifts in like what the computer knows, what it can automate, what suggestions it can make.

回到最开始，每当出现一项新技术，人们第一反应总是把原来的事做得更多。

Going back right to the beginning, whenever you get a new technology, you start by doing the old thing, but more.

需要更多的表格、更多的幻灯片、更多的邮件，更好的邮件系统。

You need more spreadsheets, more PowerPoints, more email, better email.

但真正重要的不是把原来的事做得更多，而是做些用旧技术根本做不到的新事。

But the important stuff is not doing the old thing, but more it's doing something new that you couldn't have done with the old thing.

这话说起来挺普通的，但人很容易就忘了这一点。

I mean, it's a pretty banal observation, but you kind of lose sight of it.

那么，相比只是自动化旧有的工作，我们现在能做哪些以前根本做不到的新事情呢？

And so what are the new things that you can only do with this as opposed to automating the old stuff?

我的意思是，企业版的终极形态可能是：所有与客户的Zoom通话都被记录下来，所有进出Salesforce的邮件流都被追踪，你能看到用户使用我们产品的全部数据、指标和分析结果。

I mean, I think at the end of the enterprise version of this would be, you've got all Zoom calls with clients recorded, and you've got all the flows of emails in and out of Salesforce, and you can see all of the telemetry and the metrics and the analytics of how people use our product.

那我们该怎么调整定价，才能降低客户流失率呢？

So how should we change our prices to improve our churn?

而且，这类事情是大语言模型可能办到的，跟简单地对客服电话做情绪分析、判断哪些客户生气了，完全是两码事。

And again, that's something that an LLM might be able to do, which is very different to saying, do sentiment analysis on calls into the call center and tell me which customers are angry.

你实际上是在多个层面上提升了分析的抽象程度。

You get kind of multiple shifts in the layer of abstraction around what analysis you can do.

当然，这也会催生新公司，淘汰旧公司，创造出全新的商业模式等等。

And of course that then creates new companies and destroys old companies and creates new businesses and everything else.

但我们现在就像在1997年，试图预测Uber和Airbnb会怎么出现。

But again, we're in 1997 and I'm trying to predict Uber and Airbnb.

如果我真的能准确预测这些，那说明我们活在一个平行宇宙里。

If I could actually do that, there's a sort of general point here, which is if we could actually predict what was gonna happen, we live in a parallel universe.

风投的命中率就不会是十分之一，而是百分之百了。

VCs would have, it wouldn't be a one in 10 hit rate, would be a 10 out of 10

命中率。

hit rate.

看起来我们现在问的问题是，以前贵得离谱、根本做不了的事，现在哪些变得可能了？

It seems like, yeah, one of the questions we're now asking is what or or or your sort of follows is sort of what was unreasonably expensive to do before that now is is possible?

比如说，像从零开始重做YouTube，或者从头重写Linux这样的事。

And may you know, is it so I don't know, something crazy like rebuilding YouTube from scratch or rewriting Linux from scratch or

是啊，这挺有意思的。

Yeah, it's funny.

当然，一个常见的误区是，新东西一出来，就想着用新东西去重构旧东西。

Mean, the paired fallacy, of course, is the new thing comes along and says, well, we're gonna build the new thing with the old thing with the new thing.

我们当然要基于开源技术来打造Office。

Of course, we're gonna we're gonna build Office with open source.

我们要把它完全重建在网页上。

We're gonna rebuild it on the web.

你知道吗，结果发现，猜怎么着？

And, you know, it turns out, like, you know, guess what?

看看谷歌文档。

Look at Google Docs.

它占了大约百分之二十的市场，但那根本不是重点。

It's got, like, 20% of the market, because that's actually not the point.

真正有意思的是做点别的，做点全新的东西。

What's interesting is to do something else, is to do something new.

要改变吸引力的层面，去发现那些以前根本不存在的问题。

And it's to shift that level of attraction and to spot problems that have never existed.

你在风投公司整天听项目演示时会发现，有些东西听起来挺有用。

It's the experience you get sitting in pitches all day at a venture firm is there's some stuff where you think, well, sounds kind of useful.

还有一些东西，你根本想不明白它凭什么能成。

And the stuff where you think, well, I'm not quite sure why that would work.

但总有一些东西，是填补了宇宙中的一个空缺。

But there are some things that fill a hole in the universe.

一旦有人给你解释清楚了，你就会想：天啊，怎么以前就没人想到这么做呢？

As soon as somebody explains it to you, you think, wow, why did nobody do that before?

怎么就没人注意到这件事其实早就存在呢？

Why did no one see that that thing existed?

这正是看初创公司有趣的地方，人们也会用这种方式去操作。

And that's where part of the fun part of looking at startups is, and that's what people will do with this.

人们会突然发现一种方法，能解决那个一直存在的问题——连那些深受其苦的人都没意识到这竟然是个问题。

People will suddenly work out a way that you could turn, that that will work out that that problem existed, and no one, including the people who have that problem, no one realized that problem existed.

然后他们就会去开发一个产品来解决它。

And then they'll go out and make a thing to solve it.

顺便说一句，回到之前的话题，这正是我不看好这个模型的原因。

Is also incidentally, going back to an earlier point, this is why I don't see the model.

我觉得问题就在于，以为模型能包办一切的想法根本站不住脚。

I think that this is the problem with the idea the model will do the whole thing.

如果你回头想想自2016年C版以来见过的那些案例，有多少是行业内的人早就清楚存在的问题？

And if you kind go back and think about all the pictures you've seen since the draw date of '16 C, how many of them were things where people in the industry knew that was a problem?

很多时候，答案其实是否定的。

Quite often the answer is actually no.

事实上，行业内根本没人觉得这是个问题，花了整整两年时间才让他们明白，这个问题确实存在，而这个新东西能帮他们解决它。

Actually, no one in the industry thought that was a problem, and it actually took like two years to explain to them and persuade them that that problem actually existed at all, and that this new thing would fix that for them.

这就是为什么说让金融部门的中层管理者用这个工具来解决整个行业的重大问题，想法本身就有问题。

And that's kind of the problem with the idea that middle manager in finance is going to use this tool to solve this big global industry problem.

根本不是这么回事，因为连这个行业问题是否存在都没人知道，更别提知道该怎么设计工具来解决了。

Like, no, because no one knew that industry problem was there, let alone could work out the right way to build a tool to solve it.

这是否意味着，AI出现后，SaaS环境会比以前更分散？

Does this imply a less consolidated SaaS environment than than before AI?

也许会出现更少的整合，也不会再有像微软企业版这样的庞然大物了。

Maybe less bundling or single behemoths like the Microsoft Enterprise?

天哪。

Gosh.

你这话可真把我拉回现实了。

Way to bring me back down to Earth.

这个SaaS行业未来会变得更分散，本尼迪克特。

Like, this is a SaaS industry going to be less consolidated, Benedict.

这当然很好。

That's all great.

但你跟我们说说股票吧。

But, like, tell us about the stocks.

我们到底能打下哪些基础？

What are the kind of building blocks that we can put down here?

很明显，开发软件会便宜得多，也快得多。

So obviously, it's going to be way cheaper and quicker to build software.

肯定会出现很多以前根本不可能做到的软件应用。

Obviously, there's going be a bunch of stuff you could do with software that you just couldn't do before at all.

所以竞争会更加激烈。

And so there will be more competition.

当然，这也会带来新的利润结构，但就像我们之前聊的，我们还不清楚这个利润结构到底会是什么样子。

And of course, this comes with a new margin structure, but as per our conversation earlier, we don't really know what that margin structure is gonna look like.

你打算改用按成果收费的模式吗？

Are you going to go to outcome based pricing?

要把企业软件里每个按钮的操作都跟利润表挂钩，真的很难。

I think it's really hard to tie each button press in a piece of enterprise software to the P and L.

有时候在Salesforce这类系统里倒是能行。

Sometimes you can, in Salesforce or something.

毕竟用的软件太多了。

It was an awful lot of software.

要说是今天我做的工作提升了每股收益，所以该付我多少钱，这根本说不通。

Would be really hard to say, well, work I did today did this to the EPS, therefore this is what we should pay for it.

要为这一块软件定个价，也不现实。

This is what we should pay for that piece of software.

我觉得这种做法行不通。

I don't think that makes sense.

不过话说回来，未来的定价结构跟现在相比会是什么样？

Anyway, what does the pricing structure look like over time versus now?

竞争会越来越激烈。

There will be more competition.

开发软件会变得更简单、更快捷。

It will be easier to build stuff and quicker to build stuff.

我之前想过，或许可以从两个角度来理解这个问题，这样比较有帮助。

The way that I sort of thought about, suppose there's maybe kind of two framing to think about this that are kind of useful.

一个角度是，看看现在企业软件的整体格局，大致可以分成三类。

One of them is to say that if you think about the enterprise software fleet today, you've got three buckets.

第一类是大型通用系统，比如SAP、Workday、客户关系管理、资本管理、薪酬管理这些软件。

You've got your Big Iron horizontal systems, so SAP and Workday and your CRM and your capital management software and your payroll management software and so on.

第二类是行业专用软件，一家典型的美国大公司通常会用三百到四百个SaaS应用，另外还有一千多个自己开发或采购的内部应用，部署在本地服务器上。

And then you've got vertical software, and typical big US company has like 300 to 400 SaaS apps, and then like another thousand apps that they've built, bought or built themselves internally, running on prem.

第三类是介于两者之间的模糊地带，比如Excel、邮件和共享文件系统这些临时拼凑的工具。

And then in the middle, you've got this kind of fuzzy, improvised space of Excel and email and a shared file system.

这些工具之间经常互相流转、替代。

And stuff moves back and forth between those.

原则上，每个SaaS应用做的事儿，你本来都可以用SAP或者Excel来完成。

And in principle, every SaaS app is doing something that you could have done in SAP or you could have done in Excel.

你本来可以用Workday管研究生招聘，但如果你是普华永道，每年要招几千名毕业生来培训成会计师，你大概率会自己开发一套专门的软件，或者请埃森哲帮你做——而且你肯定对它恨得牙痒痒。

You could have managed your graduate recruiting in Workday, but at a certain point, if you're PwC and you hire however many thousand graduates every year to train to be accountants, you probably got a piece of dedicated software that you built for yourself, or maybe you hired Accenture to build, you probably hate it.

不过不管怎样，你总归有一套专门用来招聘的软件，要么是买的，要么是自己弄的。

But anyway, you've got this piece of dedicated hiring software, or you bought something.

如果你公司每年只招五个毕业生，那你肯定就是用邮件加一个共享的Google表格搞定，谁会为这点事儿去买软件呢？

If you are a company that hires five graduates a year, you're doing that in email and a shared Google Sheet, because why would you buy software for that?

而在中间，还有一片模糊地带。

And then there's a space in the middle.

你是用Workday来处理呢？

Do you do it in Workday?

还是用Excel？

Do you do it in Excel?

或者用一个专门的应用？

Do you do it in a dedicated app?

现在你再把ChatGPT加进来。

And now you add ChatGPC to that.

你会不会用大语言模型来做这件事？

Do you do that in an LLM?

有没有什么大语言模型工具，让你能在Salesforce里做以前做不到的事，或者在你那个垂直应用里实现以前无法完成的功能？

Is there an LLM tool that means you can do that in Salesforce where you couldn't do it before, or you can do it in your vertical app that you couldn't do before?

你会不会用大语言模型自己搭个工具？就像有些公司部门还在用十五年前有人做的一个10兆的Excel表格，没人懂它，但大家照样在用。

Do you use the LM to build yourself a tool for that, just as you might have a company department that runs on a 10 meg Excel spreadsheet that someone built fifteen years ago, and no one knows it but they're still using that.

所以它就这样出现在这个庞大、分散又复杂的环境里，又多了一种完成这项任务的选择。

So it arrives within this broad, fragmented, complicated landscape, and it's another set of options for how you would do that task.

这是一种看待问题的方式。

So this is one framing to think about it.

我觉得另一种思路是，大语言模型该放在技术栈的顶层还是底层？

I think the other framing to think about is does the LLM go at the top of the stack or the bottom of the stack?

一方面，底层就是Salesforce里的一个功能。

So on the one hand, bottom of the stack is a feature inside Salesforce.

你在Salesforce里，看看这个客户的沟通记录，回顾我们之前所有的销售通话，结合公司的业务目标，帮我想一封邮件，或者告诉我现在该怎么做，该跟客户发什么内容。

So you're in Salesforce, look at the history with this customer, look at the context of every other sales call we've done, look at our business objectives and suggest an email, or suggest me what I should do here and what I should send the call to the customer.

这就像一个功能按钮，有专门的工具和安全机制，完全围绕这个具体场景设计。

So it's a feature, it's a button that's controlled and has tooling and guardrails and everything else that are driven by that particular use case.

但另一种思路，就像我之前说的，去整合Salesforce、Workday、我们的所有邮件、Google Analytics等等数据，然后生成一些以前根本做不到的洞察。

But the other way to look at it is the example I gave earlier, which is go look at Salesforce and Work Day and all of our email and Google Analytics and, and, and, and then synthesize something that you couldn't have done before.

这两种方式的核心矛盾在于：把那些可能出错的概率型软件放在哪儿，又把那些无法回答这类问题的确定性软件放在哪儿？

So the tension in both cases is where do you put the probabilistic software that can make mistakes and where do you put the deterministic software that can't answer these kind of questions?

数据库和大语言模型，到底谁该在上层，谁该在底层？

So where do you put the database and where do you put the LLM?

到底是哪个在上，哪个在下？

And is it which is at the top and which is the bottom?

答案很可能是两者都有，具体要看你做什么、用在什么地方。

The answer is probably both, depending on what you're doing and where it goes.

说了这么多，归根结底是问：这对软件意味着什么？

All of which is a long way of saying, what does this do to software?

答案是需要更多的软件，多得多的软件。

And the answer is more software, like way more software.

说真的，所有软件公司存在的目的，都是为了解决其他软件公司制造的问题。

Mean, all software companies exist to solve problems created by other software companies.

这在安全领域是个笑话。

That was a joke in security.

所有安全软件都是为了应对其他安全软件引发的问题而存在的。

All security software exists to solve problems created by other security software.

我们之前在SaaS领域就经历过这一套。

And clearly that's what we went through with SAS.

SaaS带来了十倍甚至百倍的软件增长，我们大概率也会在这一波看到同样的情况。

SAS gave us an order of magnitude, two orders of magnitude more software, and we should probably expect that with this.

所谓的SaaS末日就是，所有投资人都在盯着这些公司，琢磨着到底哪家会被这场变革拖垮。

What that gets to the SAS apocalypse is all the investors are kind of looking at all these companies and saying, well, we don't really know which of these companies are gonna get screwed by all of this.

肯定有一部分公司逃不掉。

Some of them must be, obviously.

肯定有一部分，等这场风波过去，会有百分之几十的SaaS公司被彻底淘汰，但没人知道到底是哪些。

There must be some, go through the end of this and X percent of all the SaaS companies that are out there are gonna get wiped out by this, but you don't know which ones.

所以你大概不该直接把整个软件行业打五折，但很明显，你现在会想：在搞清楚到底发生了什么之前，我暂时不打算押注软件了。

So you probably shouldn't de rate the whole thing by 50%, but clearly you're gonna go, I'm not sure I'm gonna be long software at the moment until I have some idea of what the hell is going on.

是的。

Yeah.

你跟本·汤普森聊的时候说过，软件就是有人坐下来设计了一个工作流程，然后说：从今往后，就这么干。

You said in your talk with Ben Thompson that software is someone sat down and designed a workflow and said, this is the right way of doing this from now on.

但你也提到，实际的流程往往是企业自己运行中慢慢长出来的。

But you also said that a process grows out of the way just a business runs.

这只是需要时间吗？

Does that just take time?

还是说，我们需要这些垂直领域的AI初创公司多做些尝试和迭代，才能打造出真正适合未来的软件？

Or do you think we need more experimentation, iteration from these vertical AI startups to get this the right shave of software for the future?

从某种角度说，换个角度看，这其实既是战略顾问做的事，也是软件公司做的事——他们观察企业内部的运作，然后说：这做法太烂了。

Well, in a sense, I mean, maybe kind of an interesting turn on this is this is both what strategy consultants do and software companies do, is they kind of look at what's going on inside a company and say, well, this is a crap way of doing it.

这样做的效果会更好，更能实现你的目标。

This would be a better way of doing it that would achieve your objectives better.

软件公司会把这套方法固化成系统，而咨询公司则把它变成流程、组织架构、培训方案和绩效目标，有时还会建议客户开发相应的软件来落地。

And a software company kind of encodes that in software and a strategy consultancy kind of encodes that in workflows and org charts and processes and training and objectives and maybe tells them to provide some software to do that thing.

或者现在越来越多的情况是，他们自己直接把软件也给做了。

Or maybe now increasingly maybe builds in that software as well.

我觉得这里还有一点值得讨论：企业内部其实有很多事情都是心照不宣的，没写下来，也没被纳入训练数据，连公司里的人自己都很难坐下来画个清晰的流程图，跟你讲清楚到底是怎么运作的。

I think another thing to talk about here is how much of what's done inside an organization is implicit and not documented and not in the training data and not something that anybody in that company could actually kind of sit down and draw you a neat flowchart of and explain to you.

贝恩、波士顿咨询、麦肯锡这些公司真正有价值的地方，就在于他们有资格走进企业，跟所有人谈话——包括那些平时根本接触不到的跨部门员工，而且还不用担心被开除。

That's a big chunk of the value of Bain, BCG, McKinsey, is that they have license to come into a company and talk to everyone and talk to the people they're not allowed to talk to that are in a different org and not get fired.

他们能搞清楚事情实际是怎么运行的，而不是纸上谈兵的官方说法，也能发现为什么员工明明知道该怎么做，却偏偏不执行——原来他们的奖金目标根本就鼓励他们别按策略来。

And to go and work out how this actually works as opposed to how it's supposed to work and why it is that people aren't doing the strategy because they actually guess what their bonus targets depend on them not doing the strategy.

他们把这些都理清楚，然后带着一套现成的解决方案进来，最后你只需要把责任推给他们，或者直接用这个现成的答案就行。

And work all of that out and be a team that's ready to come in from the outside and give you the answer, and then you can blame them or have that kind of, that pre baked solution.

这些都属于组织管理中的深层问题，涉及人的行为、沟通和表达方式，非常难被写成文档，也很难直接塞进像Claude这样的AI技能里，然后说：‘来，给我做个PPT。’

These are sort of problems in organizational management and how people function and how people can explain what they do that are very hard to write down and very hard to kind of bake into a Claude skill and say, there you are, make me a PowerPoint.

所以这里还有一个更广泛的问题：怎么才能让人真正使用这些技术？

And so there's a sort of broader how does this always work challenge here of how do you get people to use these technologies?

人们是怎么接受新工具的？

How do people adopt new tools?

你怎么帮人适应新工具，又怎么想清楚能用它们做些什么新事？这其实和当年云、网页、手机、互联网、个人电脑、电子表格出现时的情况一模一样。

How do you work out how to help people adopt new tools and work out what new things you would do with them, which is also what happened with cloud and the web and mobile and the internet and PCs and spreadsheets and so on.

说到网络，你觉得AI原生软件和新型界面之间会不会相互影响、共同演进？

To the net, do do you think there's some kind of coevolution between AI native software and new types of interfaces?

比如，一些新的客户服务AI平台可能根本不需要面向人类的界面，或者某些系统级软件连前端都没有，因为它的主要用户其实是AI代理，直接去查询它。

For example, new customer service AI platforms that might not have had as much human facing UI or system of record software being built without a front end at all because its primary user will be AI agents querying it directly?

我觉得这些想法挺有意思的。

So I think these are kind of interesting ideas.

但我很难对它们有明确的看法，毕竟这些事离企业采购基础设施的细节还太远了。

They're things I struggle to have a strong opinion on because, you know, they're not kind of not not deep deep into the weeds of of how enterprise infrastructure gets bought.

我倒是在想，这些问题到底有多新？

I wonder how new some of these questions are.

你还记得克里斯·迪克森十年前、十五年前说过的话吗？他说API就是新的商务拓展，软件公司只要把API开放出来就行——结果呢，老东西又成了新潮流。

I mean, remember Chris Dixon saying like ten, fifteen years ago that APIs is a new BD and software companies could just open up your APIs and like, well, what's old is new.

现在你根本不需要什么API了。

You don't need an API anymore.

你只需要搭一个MCP服务器，别人直接连上去，智能代理自动接入就行。

You just have an MCP server and people will just plug in, the agent will just plug into that.

说实话，我觉得这些问题的难点在于，几乎所有决策都得靠处理异常情况。

I don't know, I think the challenge with a lot of this stuff is that all the decisions are really exception handling.

关键问题始终是：哪些事情是没法自动化的？

The question is always what can you not automate?

哪些事必须有人来做判断、拿主意、表达观点？因为这些事可能从来没被写下来过，以前也没发生过，或者跟过去的情况不太一样。

What requires someone to make a decision and some judgment and have an opinion about it because maybe that hasn't been written down or that didn't happen before, or it doesn't look quite the way it happened before.

我觉得，对于区分哪些该自动化、哪些不该自动化，其实有好几种思路。

I think there's a sort of, there are various ways of kind of think about separating out what gets automated and what doesn't.

我之前在演示文稿里用的方法，是把‘任务’和‘工作’区分开来看。

The way that I used in the deck was to talk about what's a task versus what's a job.

完成某个工作所用的具体任务，可能会变，但工作本身，或者客户所购买的核心价值，并没有太大的变化。

And we often the tasks that are used to accomplish the job might change without the job itself changing very much or without the thing that the job is selling to the client changing very much.

想想五十年前的会计和现在的会计，他们几乎不再做同样的事情了。

If you think about what accountants did fifty years ago and what accountants do today, they do spend almost none of their time doing the same things.

但对客户来说，感觉还是那回事。

But to the client, it's kind of the same thing.

只是完成的方式完全不同了，用了一大堆全新的任务。

It just gets done in a completely different way with a whole bunch of different tasks.

关于这个问题，有一种更深刻或更抽象的思考方式：你希望把平均化放在哪里？

And one of the more sort of either profound or kind of abstract ways to think about this is where is it that you want the average?

你希望在哪一个环节，大家的做法都趋于一致？

Where is it that what you want is the way that everybody will do this?

那就是所有人都会采用的方式。

That's the way everyone would do it.

这才是每个人都会认同的做法。

That's what anyone would say.

每个人都会这么想。

That's what anyone would make.

每个助理都会这么拿。

That's what any associate would make.

任何人给我都会是这个答案。

That's what anybody would give me.

任何人回答都会是这样。

That's the answer anyone would give.

那什么时候你并不想要这样？

Versus where is that not what you want?

什么时候你想要的是一个新问题的答案，或者不同的观点？

Where is it that you want the answer to a new question or a different answer or a different idea?

因为大语言模型在那些你能清楚描述人们如何操作、而你想要的是普遍做法的场景下，会非常擅长。

Because LLMs are going to be very good at anything where you can describe how people do it and where what you want is the way anybody would do that.

但在你无法解释为什么这么做、或者你刻意用不同于常人的方式处理时，它们就不太行了。

And not so good at where you can't really explain why you did it like that and where you're doing it differently to the way people would normally do it.

包括谷歌CEO在内的不少人表示，投资不足的风险比投资过度还要大。

Various people, including CEO Google, said that the risk of under investing is riskier than over investing.

有没有一个资本支出的临界点，让这句话不再成立？

Is there any level of CapEx where that stops being true?

我们现在是不是已经接近那个点了？

And are we getting there now?

问题是，微软、Meta和谷歌今年的资本支出都打算占到营收的50%以上，这在财务上是个大问题。

Well, there's a financial gravity problem in that Microsoft, Meta, and Google are all in line to spend over 50% of revenue on CapEx this year.

我们通常觉得电信行业是资本密集型的。

And we think of telecoms as being capital intensive.

电信行业一般只把营收的15%到20%花在资本支出上。

Telecom spends sort of 15 to 20% of revenue on CapEx.

今年，这四家巨头公司的资本支出总额预计将达到7000亿美元。

So, 700,000,000,000 is the guidance from the big four companies this year.

电信行业总共3000亿，移动业务2000亿，加起来电信部分就是3000亿。

Well, telecoms is 300, mobile is 200, total telecoms is 300.

石油和天然气行业，具体看怎么算，年支出大概在七千亿到一万亿美元之间，我印象中是这样。

Oil and gas, depending on which definition or which bits of it you're counting, is anything from 700,000,000,000 to a trillion dollars, I think, from memory, think.

这得看你怎么问、问谁。

Depends on which exactly who you ask.

每年七千亿美金，这数字其实并不算离谱。

So $700,000,000,000 a year is not an impossibly large amount of money.

这正是全球大型基础设施项目的正常开销。

It's what big global infrastructure costs.

就是钱花得太多了。

It's just a lot of money.

显然，这些公司明年不可能花上一万五千亿美元，就算真花了，也得靠借钱，更不可能长期维持这种支出水平。

Clearly, those companies could not spend 1 and a half trillion next year, Or if they did, they'd have to borrow it and they certainly couldn't sustain that level of spending for any length of time.

所以，当资金跟不上时，增长就不得不放缓了。

And so there's a certain point at which that growth has to slow down because there isn't any more money.

当然，你也可以谈投资回报率，谈这些投入能带来多少收益。

Now, clearly you can talk about ROI and your ability to produce returns from that investment.

显然，资本市场愿意为这个投入资金，但到一定程度就到头了，随便说个数吧。

Clearly the capital markets are willing to fund that up to a point, but pick a number at random.

我们不可能每年花一万亿美金在人工智能基础设施上，因为根本没那么多钱可花。

We can't spend $10,000,000,000,000 a year on AI infrastructure because there isn't $10,000,000,000,000 a year there to spend on it.

所以，可用资金的上限就像物理定律一样，是客观存在的。

So there are kind of laws of physics caps on the amount of money that's available.

目前我还不太敢说比这更具体的东西。

I'd hesitate to say something more tangible than that at the moment.

说白了，我差不多又回到开头那句话了：我们现在面对的是一堆倍数关系。

I mean, I kind of almost go back to what I said at the beginning, that we've got a bunch of multiples.

需求远远超过供给。

So there's far more demand of supply.

但另一方面，效率正在大幅提升。

On the other hand, the efficiency is increasing massively.

我们还不知道下一个模型会是什么样子。

We don't know what the next model will be.

目前还不清楚边缘计算和开源技术会在什么时候介入。

We don't know where edge or open source come in yet, when edge and open source come in yet.

与此同时，你总是在追赶下一个模型。

And meanwhile, you're always chasing the next model.

所以贯穿这一切的一个关键点是，模型的生命周期只有三到六个月，或者六到九个月，具体多久看你怎么说。

And so this is kind of the line that runs across all of it is the model is only relevant for three to six months, six to nine months, whatever you want to say.

开发一个模型要花多少钱？

And the model costs how many billion dollars?

要支撑这个模型，需要多少基础设施？

And how much infrastructure do you need to do that?

我觉得这个规模还没真正明朗起来。

I don't think that mass is really shaken out yet.

当然，有很多非常聪明的半导体分析师，花大量时间试图给这些数字做估算。

I mean, obviously there's a bunch of very clever semiconductor analysts who spend lots of time trying to put numbers on this.

这有点像上世纪九十年代末，试图估算互联网带宽的规模一样。

It is kind of like trying to put numbers on bandwidth, internet bandwidth in the late '90s.

你知道表格的行大致是什么，但真说不清那些数据到底在哪。

You kind of know what the rows of the spreadsheet are, but you don't really know where the values are.

你只能说，这根本不可能。

All you could really say is, well, look, it can't be.

这中间肯定有物理上的极限。

There's clearly physical limits on this.

换个角度想，如果你是谷歌、Meta、微软，或者某种程度上亚马逊、苹果，这简直就是生死存亡的问题。

Think another way to answer the question is like, if you're Google or Meta or Microsoft, some extent Amazon, some extent Apple, this is sort of an existential problem.

你同时面临错失恐惧症：一方面，目前的投资回报率非常高。

And you have a sort of a FOMO problem in that, So on the one hand, your returns on the investment at the moment are hugely positive.

但另一方面，你又不能眼睁睁看着别人抢跑而自己不参与，否则公司就完了——你不想重蹈微软2000年代、IBM90年代，或者Meta2010年代的覆辙，那时候他们总被苹果压着打。

On the other, you can't let other people get away with this without you participating because then your company's gone and you don't want to end up like Microsoft in the 2000s or IBM in the 90s, or indeed Meta in the 2010s where they are kind of continually getting shafted by Apple.

所以，如果这就是计算的未来，那你必须参与进去。

So if this is the future of compute, then you need to be participating in it.

但与此同时，CFO肯定在那儿问：没错，这听起来不错，但具体你打算投入多少？

But obviously at the same time, the CFO is sitting there saying, well, yeah, that's great, but how much participation are we talking about here?

我觉得，到了某个阶段，这条曲线肯定得平缓下来，毕竟它已经没多少上升空间了。

And I don't think we've, you know, it's clearly at a certain point that curve is going to have to taper off because like, there's nowhere else it can go.

你觉得会在令牌使用上限上出问题吗？

You think there's going to be a reckoning around token maxing?

会不会有公司过度使用AI了？等他们做一回真正的投资回报分析，就会开始收缩了？

Is it possible that companies have been overshooting AI usage and when they do proper ROI studies, they'll pull back?

很明显，有些人拿最贵的模型在上网瞎玩，就跟2010年智能手机刚出来那会儿一样。

Well, obviously, you know, you've had people, like, using the most expensive model to dick around on the internet, which is kind of what happened with mobile in 2010.

结果账单来了上万块，你才反应过来：等等，我不是买了无限套餐吗？

Like, you know, you got a $10,000 bill and you wish said, wait, wait, I thought this was a flat rate bundle.

这到底是怎么回事？

Like what happened?

所以，确实有不少让人哭笑不得的糟心故事。

So there's like, obviously you've got a bunch of like sillypainful stories.

但我觉得更值得探讨的是，就像我之前说过的，我们现在正处在巨大的失衡状态，价格必须跟成本对齐，使用量也得跟价格和投资回报率匹配。

Think there's also a point of like, I think maybe what's slightly more interesting as the question is, and clearly there's going to be a point in which, as I've said several times, we're at a moment of massive disequilibrium, and the pricing has got to get back into alignment with the cost, and the usage has got to get it into alignment with the pricing and the ROI.

问题是，这有点难说清楚。

The challenge is it's a bit tricky.

在这么早期的阶段，很难判断投资回报率到底在哪。

This early stage is quite hard to know what the ROI is.

这就像是在90年代末把互联网发给所有人，然后说：去吧，提高效率。

It's rather like giving everybody the internet in the late 90s and saying, okay, go off, be more productive.

如果你看看德勤的调查，还有我演示文稿里提到的美联储的报告，当你去问CFO们，你们在哪儿看到了收益？

And if you look at, there's a survey from Deloitte, there's also a survey from the Fed that's in my presentation, if where you go and ask CFOs, where have you seen the benefits?

到目前为止，大部分收益都是一些很难量化的方面。

And most of the benefits so far have been stuff that's pretty hard to measure.

比如分析能力提升了，客户支持变好了，工作效率提高了，能更快做出更多幻灯片，更快完成分析。

So like better analytics, better customer support, more productivity, you could make more slides more quickly, you could do the analysis more quickly.

这些很难用金钱来衡量。

It's kind of tough to put a financial value on that.

它们当然有经济价值，但跟‘我们用AI做出了新产品，带来了多少收入’或者‘帮我们省了多少钱’这种直接数字不一样。

It has a financial value, but it's not the same as saying, well, we made this new thing with AI and it had this revenue or it saved us this much money.

很明显，事情花的时间更长。

Obviously, things take longer.

要开辟一条新的收入渠道，比把这工具发给大家让他们更快做表格要难多了。

It's harder to build a new revenue line than to give this to everybody and have them use it to make spreadsheets more quickly.

所以大家难免会问：这到底得花多久？

So there's a little bit of like, well, how long does this take?

我觉得这个问题的另一个答案是消费者剩余，就像当年Excel出现时那样——你猜怎么着？

I think the other answer to a problem here, of course, is consumer surplus, which is to say that this is kind of what happened with Excel, in that, guess what?

如果做一个DCF模型要花一周，那你可能一年就做一两个。

If a DCF takes you a week, then you probably only do one or two DCFs.

但如果一个DCF模型只要十秒，那你就能做五十个。

And if a DCF takes you ten seconds, then you do 50 DCFs.

你根本没法为此多收一分钱。

You really can't charge any more money for that.

所以有些时候，这些工具会变成必备品，人人都得买、都得用。

So some of what happens is that these things become competitive necessities, and everybody has to buy it and use it.

但你从这里面节省的成本或提升的效率，很快就会被市场竞争掉，所以你没法因此涨价。

But the cost saving or the productivity gain that you get from it just gets competed away, so you don't get to charge more for it.

如果你在麦肯锡、贝恩或者BCG，以前做这份分析要一周，现在只要一天，你很可能会多做五倍的分析，但还是收客户同样的钱。

If you're at McKinsey or Bain or BCG and doing that piece of analysis used to take a week and now it takes a day, you'd probably do five times more analysis and charge your customer the same.

你的成本结构也没变，这恰恰说明了投资银行和金融分析领域发生过什么。

And your cost base hasn't changed either, which is exactly the way to think about what happened with investment banks and financial analysis.

你用可能更少的人做了多得多的分析，却还是收客户同样的费用。

You just did way more analysis with probably fewer people and charged customers the same amount of money.

你整个观点的一部分是，模型最终会变成大宗商品，但眼下，融资最快、金额最大的，恰恰是这些基础模型公司。

Part of your your a big part of your thesis is this idea that models are gonna end up as commodities, and yet the, you know, the layer that's raising the most money, you know, in the in the fastest time in history is these foundation model companies.

既然这样，你对他们有什么建议？是整体上该怎么做，还是我们可以挑一家具体公司来说说，他们该怎么调整？

So given that, what advice might you have for them, either either collectively or we can pick on someone individually in order to in order to adapt.

我不是说它们一定会变成大宗商品。

It's not that I know that they're gonna become commodities.

我的观点更偏向于现在这个阶段。

My position is more one more now.

嗯，听我说。

Well, like, hey.

我来给你梳理一个逻辑：从发展趋势看，这些技术注定会变成大宗商品，你倒是说说，为什么它们不会？

Here's a here's a chain of argument that says that deterministically, it looks like these things will be commodities, and explain to me why they won't be.

我就说到这儿，不再进一步断言了。

And that's as far as I would commit to that.

一想到他们融了这么多钱，我就想起移动通信行业的事了——虽然这事儿没法预测，但确实值得提一提：移动行业规模庞大，砸钱建基础设施，利润却很低，真正赚钱的都是别人做的那些应用。

Think raising all this money, I kind of get back to my point about mobile, which again has no predictive value, it's a worthwhile observation, is that the mobile industry is very big and spends a lot of money on infrastructure and isn't very profitable and all the cool stuff is done by somebody else.

那你说，资本回报率怎么样？

And then you can do, well, what's the return on capital?

答案是：得看你在哪儿，美国、欧洲、印度还是中国，情况都不一样。

And the answer is, well, it depends whether you're in America or Europe or India or China.

不过话说回来，这事儿确实干得值，有人赚了钱，但最后控制权没落在他们手里，反而是别人从里面拿走了更多价值。

But meanwhile, that was a worthwhile thing to do and it produced a return for somebody, but then it ended up not controlling the whole thing and other people ended up getting more value from that than they did.

具体数字我记不清了。

I don't have the number in my head.

谷歌去年的净利润是多少？五千亿美金左右？

Was Google's net income last year was what, dollars 50,000,000,000 or something?

整个电信行业的净利润是多少？

What's net income for the total telecoms industry?

我真该订个彭博社的订阅服务。

I should really subscribe to Bloomberg.

那样的话，这些问题我都能马上答上来。

Then I could just answer these questions instantly.

可以肯定的是，谷歌、Meta、亚马逊、微软和苹果的利润加起来，比整个电信行业还要高。

But like a pretty safe bet that Google, Meta, Amazon, Microsoft, Apple produced more profits than the entire telecoms industry.

这就成了一个难题：你正推动前沿发展，

So this is a puzzle is you're driving the frontier forward.

却陷入了一个陷阱——你必须不断竞争，否则别人抢先一步，你就被甩在后面了。

You're kind of caught in this trap that you have to keep competing because otherwise they'll do it and you'll fall behind.

还有一件事我们完全没提到：我们是不是其实就在建造通用人工智能？

You've also got this thing that we haven't talked about at all, which is, hey, aren't we just building AGI?

我们正在造一个装神明的盒子，有些人真这么信，虽然很难分析，但也许吧。

We're going to build God in a box, which some people they do believe, although it's hard to analyze, but maybe.

你还是会继续搞这些东西。

You're going to carry on building this stuff.

但实际的问题是，你怎么做出那些不是给程序员用、但普通人真想用的产品？

But the practical question is, well, how do you get things that people want to use that aren't software development?

那可是个不错的生意。

Mean, that's a good business.

就只有这一种生意吗？

Is that the only business?

有个领域，把软件行业的生产率提升上去，就能撬动几千亿美元，这没问题。

There's an area, take a number of hundreds of billions of dollars it is to make the software industry more productive, great.

然后呢？

Then what?

这可能值一万亿美元，但之后呢？

And that's worth a trillion dollars maybe, but then what?

你该怎么把这种模式推广到整个经济，推广到其他人身上？

How do you expand this into the rest of the economy, into everybody else?

所以才会出现这些关于与私募股权、咨询公司合作的讨论，就像我们刚才说的，你猜怎么着？

Which is why you get these conversations about partnering private equity, partnering with consultancies where exactly as we've been discussing, guess what?

如果你真在经营一家公司，其实很难搞清楚该怎么用这些东西。

It's actually quite hard to work out what to do with this stuff if you're actually running a real company.

于是你就去找贝恩、波士顿咨询、麦肯锡，或者 Infosys、认知服务、IBM、埃森哲这些公司，或者找私募股东。

So you go to Bain, BCG, McKinsey or Infosys and Cognizant and IBM and Accenture or private equity shareholders.

一方面，我边说边在想这个问题，一方面你确实在不断打造越来越大的模型，总觉得必须继续做下去。

So there is this sort of sense of like on the one hand, sorry, I'm sort of trying to work out the answer as I speak, but on the one hand, you're building these bigger and bigger models and you kinda feel like you've got gotta keep doing it.

但另一方面，没错，可大家到底拿它来干什么呢？

But on the other hand, yes, but what are people doing with it?

是啊。

Yeah.

为什么大多数人看到 ChatGPT，却想不出现在能拿它做什么？

Why do most people look at chat GPT and not really think of anything to do with it today?

对。

Yeah.

最后一个问题是。

Last question.

从这次演讲中，还有什么是你希望听众一定要记住的吗？

Is there anything else from the presentation that you wanna make sure listeners leave with?

我去年用过，今年又用了一次，是IBM在五十年代初的一则广告，画面里是一群工程师举着计算尺。

The thing that I used last year and I I used again is an IBM ad I found from the early fifties, which has got a picture of a sea of engineers all holding up slide rules.

广告上写着：一台IBM电子计算器，相当于为你增加了150名工程师。

And there's an IBM ad and it says, an IBM electronic calculator gives you 150 extra engineers.

你见过多少次a16z的路演，用的就是这种说法？

And that's like how many pictures have you seen at a16z where that was the pitch?

我们其实都记得，每十年、十五年或者二十年，就会迎来一波根本性的技术变革。

And we kind of remember, we go through these waves of these fundamental technology changes every ten or fifteen or twenty years.

每一次都令人惊叹，彻底改变一切，而且跟以往任何事都截然不同。

And they're all amazing and change everything and are completely unlike anything that's happened before.

所以人工智能真的很了不起，带来了翻天覆地的变化，跟以往任何技术都完全不同。

And so AI is amazing and transformative and completely unlike anything that's happened before.

移动互联网当年也是一场大变革。

Mobile was quite a big deal too.

互联网是，个人电脑是，计算机技术本身也是，每一个都影响深远。

And so was the internet, and so were PCs, and so was computing.

这些技术刚出现的时候，谁都说不准未来会变成什么样。

Those were all also very big deals where it was hard to tell what was going to happen.

所以我们不妨默认一个基本前提：这种浪潮还会再来一次。

And so we should sort of presume as a base case, okay, well, we're going to go through that again.

而它会带来不少让人生活受挫的事，让很多人失业，也会出现一些我们并不喜欢的后果。

And that will produce a bunch of things that ruin people's lives, and it will put a bunch of people out of work, and there'll be a bunch of stuff that we're not very happy about.

但同时，也会有很多让我们都觉得特别棒的变化。

And there'll be a bunch of stuff that we all think is great.

二十年后，我们可能就完全想不起来，曾经还有个世界是电脑做不到这些的。

And then in twenty years' time, we'll forget that there was a world when computers couldn't do that.

我们开了一个小时的会，电脑居然没崩溃。

Here we are, we've been on this call for an hour and our computers didn't crash.

而且我们还在互相传输高清视频。

And we're streaming HD videos to each other.

这事儿好像本来就应该这么顺利。

And it's like, well, of course that worked.

事实上，我用iPhone也在做同样的事。

In fact, I'm also doing it with my iPhone.

我的iPhone正通过Wi-Fi把视频流传输到我的Mac上。

So my iPhone is streaming to my Mac over WiFi streaming video here.

一切都很顺畅，根本不用操心。

And it just works.

这简直像魔法一样，可我们早就习以为常了。

It's magic and we don't notice it anymore.

我觉得，这恰恰最能概括这一切最终会变成什么样。

And I think that's really my kind of one line description of how all of this is going to end up.

这将会变得神奇。

It's going to be magic.

二十年后，人们只会说，当然就是这样啊。

And in twenty years' time, we'll just say, well, of course, that's how it is.

电脑本来就是这么做的。

Computers have always done that.

是的。

Yeah.

这真是个完美的收尾。

That's a great place to great place to wrap.

这个演讲叫《AI吞噬世界》。

The presentation is called AI Eats the World.

在本尼迪克特·埃文斯的网站上。

It's on Benedict Evans' website.

非常精彩。

It is excellent.

还有很多内容我们没来得及聊，一定要去看看。

There's a lot more that we didn't get to, so definitely go check it out.

本尼迪克特，这次对话真的很棒。

Benedict, this has been a great conversation.

非常感谢你来参加这个播客。

Thanks so much for coming to the podcast.

嘿。

Hey.

谢谢。

Thanks.

聊得很开心。

Great to chat.

感谢收听这期的a16z播客。

Thanks for listening to this episode of the a 16 z podcast.

如果你喜欢这期节目，别忘了点赞、评论、订阅，给我们的节目打分或写评价，也推荐给你的朋友和家人。

If you like this episode, be sure to like, comment, subscribe, leave us a rating or a review, and share it with your friends and family.

想看更多节目，请前往YouTube、Apple Podcasts和Spotify。

For more episodes, go to YouTube, Apple Podcasts, and Spotify.

在X上关注我们，账号是a sixteen z；也可以在a16z.substack.com订阅我们的Substack。

Follow us on X at a sixteen z, and subscribe to our Substack at a16z.substack.com.

再次感谢收听，我们下一期再见。

Thanks again for listening, and I'll see you in the next episode.

提醒一下，本节目内容仅供信息参考，不构成法律、商业、税务或投资建议，也不应用于评估任何投资或证券，且并非面向任何a16z基金的投资者或潜在投资者。

As a reminder, the content here is for informational purposes only, should not be taken as legal business tax or investment advice, or be used to evaluate any investment or security, and is not directed at any investors or potential investors in any a sixteen z fund.

请注意，a16z及其关联方可能仍持有本节目所讨论公司的投资。

Please note that a sixteen z and its affiliates may also maintain investments in the companies discussed in this podcast.

如需更多详情，包括我们的投资列表，请访问a16z.com/disclosures。

For more details, including a link to our investments, please see a 16z.com forward slash disclosures.