构建物理AI栈 | 特拉维斯·卡兰尼克谈TBPN

是的。

Yep.

但情况还是一样。

So, but same story.

去开拓市场简直太疯狂了。

And so go to market is wild.

你最终得亲自跑到各种地方去，但这事儿真的特别重要。

You just end up in Literally go to places, but it's super needed.

所以现在的情况是，Pronto技术的效率已经超越了人工生产力。

And so what's happened is the the Pronto technology has got gotten past human productivity.

对。

Mhmm.

这意味着你去见金矿公司的CEO，谈市场拓展时，你可以问：你们想不想每年多产20%的黄金？

Which means you go to a gold mine CEO, you talk about go to market, you go to a gold mine CEO, and you say, would you like to have 20% more gold per year?

对。

Mhmm.

没错。

Absolutely.

很好。

Good.

我们还没听到过反对意见。

We haven't heard no.

对。

Yes.

好的。

Okay.

好的。

Okay.

但他们说，得拿出证据来。

But they're but they say prove it.

是的。

Yeah.

这正是关键所在。

And that's where the rubber meets the road.

对吧？

Right?

要证明这一点得花多长时间？

How long does it take to prove?

以前可要久得多。

It used to take a lot longer.

现在不一样了，只要你证明过足够多次，事情自然就会有推动力。

Now be like, once you've proven it enough times, then it sort of gets its own momentum.

就会传开。

Gets around.

我们现在的Pronto就处在这种状态，推动力已经开始显现了。

And so we're in that we're in that place on pronto where that momentum is taking hold.

因为已经有足够多的证据表明，它在很多地方都真正行得通了，大家都认可了。

Because there's enough proof points where it's just working in so many different places where people are like, alright.

我们开始吧。

Let's go.

我们可以把自主采矿看作像企业软件一样，先做个试点。

We're gonna think of mining autonomous mining almost like almost like enterprise software where you get a pilot.

对。

Mhmm.

比如一家上万人的大公司，你有个企业初创团队，他们说：我们只买了八个席位，但对方可是个大公司。

There's like a 10,000 person company, and you've got you you you got an enterprise startup, and they're like, I got like eight seats, but it's this huge company.

如果我们拿下了，那就是个大单。

And if we get it, it's huge.

我这边还有十个席位在另一家公司。

And I've got this other 10 seats over at this other one.

这只是一个试点，但我敢保证一定能成。

It's a pilot, but I swear it's going to work.

他们正在外面积极推广，努力推动这件事。

And they're out there pitching and trying to make it happen.

但只要一成功，在采矿业里，这意味着人的生产效率会大幅提升。

But once it works, and in mining, that means human productivity.

效率比人类本身还要高。

Human level better than human productivity.

一旦成功，规模就会迅速扩大。

Once it works, it goes big.

对。

Mhmm.

没错。

Yeah.

他们就说，行吧。

And they're like, okay.

咱们得把所有车辆都覆盖到。

Let's get across let's get across all the vehicles.

目前我们正和好几家客户处在这种推进阶段。

And so we're sort of in that mode with a bunch of different customers right now.

提升矿山运营 uptime 的机会有多大？

How big is the opportunity to just increase uptime of mining operations?

我想象有些矿山试图全天候运转，但在亚马逊雨林深处可靠地安排夜班，让每个人都准时出现，而且健康、开心、干劲十足，这可不是件容易的事。

I imagine that there are mines that are trying to operate twenty four seven, but getting a night shift in the middle of the Amazon reliably, everyone showing up and being, you know, healthy and happy and eager.

确实如此。

Totally.

一旦做到这点，事情就简单多了。

It gets a lot easier when it's like, yeah.

我们现场还是会留不少人手，但这些人主要负责监管机器人的作业任务。

We're still gonna have a bunch of people on-site, but they're gonna be overseeing robotic worshares.

所以，情况就是这样。

So, so yeah.

要说产能提升这块，其实能分成两个部分来看。

I mean, the, there's two parts to the productivity gain.

对的。

Mhmm.

首先是每台机器每小时能完成更多工作。

First is the machine per hour doing more.

对。

Yeah.

这是第一点。

That's part one.

第二点是工时、呼叫响应这些方面。

Part two is hours and callouts and all of that stuff.

是的。

Mhmm.

而且当风险降低后，安全规程也会随之调整。

As well as just, you know, the safety protocols change when you have less risk.

没错。

Yeah.

还有很多类似这样的因素，都会叠加在一起。

So there's a lot of things like this that pile onto each other.

对

Mhmm.

我估计，到最后你甚至能提升百分之三十到四十的生产力。

My guess is you could even end up 30%, 40% more productive at the end of all of it.

是的

Yeah.

一旦做到这一点，带来的机会不言而喻。

And when you do that, the opportunity speaks for itself.

对

Mhmm.

一个每年多产百分之三十到四十黄金的金矿，听起来很惊人，但所有矿产都一样。

A gold mine that's doing 30 or 40% more gold per year is kind of woah, but that's for every mineral.

对

Mhmm.

锂也是这样。

That's lithium.

我们还会深入到采石场。

That's like, we also go all the way down to quarries.

采石场的情况不太一样，简单来说，主要跟水泥有关。

Quarries are different because quarries are j basically, it's about cement, let's just say.

这就是核心所在。

That's the main jam.

当然还有其他情况，但我们先拿这个来说。

There are others, but let's just go with that.

你不能光想着多挖石头，因为那边的水泥客户用量是固定的。

You can't you don't just go do more rock because you need cement customers on the other side.

他们用的水泥量是有限的。

They're only using so much cement.

比如该怎么储存这些水泥。

Like where to store it.

对。

Yeah.

没错。

Exactly.

对。

Yeah.

所以这更多是运营支出方面的业务，利润空间也更薄。

And so so that's more of an OpEx play, and there are thinner margins there.

是的。

Mhmm.

但我已经入局了。

But I'm in the game.

对。

Yeah.

这还挺有意思的。

And it's kinda interesting.

特别有意思。

It's a lot of fun.

对于那家公司，也就是Pronto，他们团队非常厉害，Anthony Lewandowski和手下的人都特别能折腾。

And for that company, for Pronto, they were super Anthony Lewandowski and the team there, super scrappy.

完全是初创公司的风格，精简到了极点，真的非常精简。

True startup style, lean as hell, like, so lean.

就像那部克里斯蒂安·贝尔演的电影。

Like that Christian Bale movie.

我一时想不起那电影的名字了。

I can't remember the name of it.

《机械师》。

The Machinist.

哥们，那简直瘦得皮包骨头。

Dude, it's, like, super lean.

我就说，兄弟们，咱们得从瘦弱练成肌肉型。

And I'm like, guys, we gotta go from lean to muscular.

你得往蝙蝠侠那种方向练。

You gotta go to Batman.

这正是我们正在做的。

And that's what we're doing.

你想想看，在企业环境中也是这样。

Like and you think of that this in an enterprise.

肌肉型。

Muscular.

这说法不错，真的不错。

That's a good that's

是个好说法。

a good phrase.

你就想做那种肌肉型的公司。

You just wanna be muscular.

再想想企业市场的策略。

And so you think about enterprise go to market.

我们市场策略的一部分，就是让企业客户相信，我们正从精简走向强大。

Part of our go to market is is building credibility with our enterprise customers that we're going from lean to muscular.

是的

Yeah.

因为市场需求就在那儿。

Because they the demand is there.

已经准备好了。

It's ready to go.

他们说，我们需要你变得更强大。

They're like, we need you to be muscular.

我们需要蛋白粉，还有其他各种东西。

We need the protein powder and the whatever else.

是什么让你迟迟不敢

What what hold what holds you back from

去健身房。

go to the gym.

随便吧。

Whatever.

是的。

Yeah.

对。

Yeah.

没错。

Yeah.

是什么让你在扩展规模时犹豫不决？

What holds you back from scaling?

比如说，你先做一个试点项目。

You let's say you do a pilot.

效果很不错。

It works well.

你要把硬件对接到他们现有的系统和设备上。

You're attaching hardware to existing systems and hardware that they're using.

明白。

Mhmm.

他们就说，行吧。

And they say, okay.

我们产出更多了。

We're getting more out.

也许我们可以多订几台机器。

Maybe we wanna place orders for more machines.

我猜这些矿机的交货周期可能长得离谱。

Are those I imagine the lead times on some of this mining equipment could be insane.

你希望掌控整个链条的多少部分？

How much of the stack do you wanna own?

你再说一遍这个问题。

Say the question again.

不好意思。

I'm sorry.

我刚才一下子想不起来了。

I just blanked.

再试一次吧。

Go for it one more time.

抱歉。

I apologize.

现在，你是用现有的采矿设备，对。

Right now, you're taking existing mining equipment Yep.

然后你给它加上AI功能，让它变得智能化。

And you're augmenting it with you're bringing you're you're making it AI enabled.

你正在把这套流程改造成全自动运行的模式

You're making it autonomous.

效率提升了不少

You're making more efficient.

对的

Yep.

然后他们说，太好了。

And they say, great.

这已经奏效了。

This is working.

我们想扩大运营，因为也许人手不用增加，也能做得更多。

We wanna scale up our operation because maybe we need less or we can do more with the same, you know, human head count.

没错。

Yep.

但我猜，到了某个阶段，总会有些事是你控制不了的，比如……

But what's the I imagine there's some things that are out of control for for you at that point where they're like, okay.

我们需要更多这种重型采矿设备。

We need more of this heavy mining equipment.

那就把它加到现场吧。

Let's let's add it to the site.

但中间会不会有延迟？

But is there, like, a lag time there?

真正的延迟在于，比如我们想让一批亚马逊上的设备投入使用，得花时间。

The real lag time is getting so you have to you're so let's say we wanna get a bunch of machines that are in the Amazon up and running.

对

Mhmm.

是的

Yeah.

好的

Okay.

你怎么做呢？

How do you do that?

对

Mhmm.

所以我得运一批传感器、一批计算设备过去。

So I've got to ship a bunch of sensors, a bunch of compute Yeah.

一批设备，对。

A bunch of equipment Mhmm.

还有机械系统，总之得让团队能过去安装。

And mechanical systems, let's just say, so that a team can then go install it.

是的。

Yeah.

你基本上是在现场搭建一个数据中心？

You're basically building a data center on-site?

嗯，我不太这么形容。

So it's sore I wouldn't put it that way.

我的意思是，如果你把装了传感器和计算设备的机器看作数据中心，那倒也可以。

I would say I mean, if you considered a machine with sensors and compute a data center, I mean, you could.

但其实那些更像是服务器，我不太会称它为数据中心。

But it's really think of those there there are servers, but I wouldn't say a data center.

你并不是在搬进那种像阿玛达风格的、集装箱大小的设备。

It's not really bring, like, an Armada style, like, shipping container sized level of You're

你也是自愿参与的。

just volunteer as well.

你只是把设备运过来。

You're so you bring in the stuff.

是的。

Yeah.

明白吗？

Okay?

你得安装它。

You have to install it.

是的。

Yeah.

你得把它 setup 好，确保一切正常。

You have to, like, bring it up and make sure, okay.

这是个新地方。

This is a new place.

这台机器在这儿怎么才能正常运行？得校准，确保安全等等。

How does it does this machine work properly in this new place and calibrate and make sure it's safe and all of this.

所以这中间有一套启动流程。

So there's like a process of getting up.

然后还得做变更管理，因为这个站点以前是早上有人来上班的。

Then there's change management because that that site's going from there are people that show up in the morning.

有一套非常严格的流程，确保每件事都按计划进行，每个人都在该在的位置上。

There's all this very regimented process to make sure everything's going exactly as planned and people are exactly where they're supposed to be.

对。

Mhmm.

否则在矿场上很容易出岔子。

Because otherwise, weird things happen on a on a mining site.

是的。

Yeah.

所以你得从那种模式，转到这样一种情况：

So you have to go from that to, okay.

我们现在运行的是自动驾驶采矿系统。

We're now running autonomous mining operation.

这完全是两码事。

It's just a very different thing.

所以安装、调试，还有我们所说的试运行，都是必须完成的步骤。

So the installation and the bring up and what we call commissioning are sort of like the things you have to do.

对。

Mhmm.

那为什么安装要花这么长时间呢？

And, you know, like, why does it take a long time to install?

因为那台机器可能根本不是线控的。

Because that machine may not even be drive by wire.

是的。

Yeah.

所以它还是个纯机械系统。

So you have a mechanical system.

比如说，你转动方向盘的时候，它

Like, if you turn the steering wheel like it's

你知道吗

you know what

我的意思是？

I mean?

对。

Yeah.

这是一个机械系统，液压系统。

It's a it's a mechanical system, a hydraulics system.

所以你要把系统连接到需要到达的位置

So you're bringing Where you have to go

一个可能推动按钮的执行器

actuator that might push a

物理按钮。

physical button.

你确实是在

You you're yeah.

你得把电能转化成实际的物理动作，这就需要机械执行装置，没错。

You're you're trying to make electricity then do a physical thing, so then you need physical actuation Yep.

对。

Yeah.

因为现在这些机器都不是线控的。

To do the things because it's not these machines are not natively drive by wire.

对。

Yeah.

所以所有的机器都是这样。

So that's all machines.

其中一些。

Some of them.

这听起来难度极大，但却是必要的，因为你不可能指望他们把已经投入了数千万美元的设备全部拆掉。

That sounds, incredibly difficult but necessary because you're not gonna get a mind to rip out tens of millions of dollars of equipment that they already have.

但你们最终会走向全栈模式吗？比如，把整个

But would you eventually go full stack, like, build the entire

系统都自己搭建起来？

system?

在矿业里，说到底，有个专门的说法。

We ultimately I mean, if you go in the mining industry, there's like this this, term.

叫‘无人矿井’。

It's called no entry mine.

无人矿井就是里面完全没有人。

A no entry mine is a mine where there are no people.

对。

Mhmm.

就像那种无人工厂。

The out factory.

是的。

Yeah.

差不多就是那种版本。

Kinda like that version version of it.

可能会有人在控制中心。

There might be people in a control center.

可能有人在。

There might be.

但在那个矿坑里，一个人也没有。

But, like, in that pit, no human.

从安全角度来看，这完全是另一套计算方式吧。

And it's a wildly different calculus from a safety perspective I imagine.

完全不同，显而易见。

Totally different, obviously.

是的。

Yeah.

所以会有钻孔作业。

And so there's drilling.

会有爆破作业。

There's blasting.

会有装载作业。

There's loading.

还有运输。

There's haulage.

对。

Mhmm.

还有破碎。

There's crushing.

我只是在梳理采矿作业的各个流程。

I'm just going through the different parts of the mining operation.

而且

And

你得先从某个点开始，然后逐步扩展到其他区域，最终达到那个禁止进入的区域。

what you do is you start somewhere, then you start extending to those other areas to get to that no entry thing.

那个禁止进入的区域，比如我们的运输系统就是全自动的。

And the no entry thing is you can have an autonomous thing like like our haulage system is autonomous.

对。

Mhmm.

如果你要进入一个新区域，可以先用远程控制，再逐步过渡到全自动模式。

If you're getting into a new place, you can do remote control and move into autonomous.

对。

Mhmm.

如果你想做到完全无人员介入，或者尽量减少人员介入，这么说你能明白吧。

If you wanna go super no entry or lower entry, if that makes sense.

这真的特别有意思。

It's it's super fascinating.

然后你想想，那是一台重达两百万磅的设备，可能在道路上以每小时35英里的速度行驶。

And then you're talking about you're talking about loaded a 2,000,000 pound machine that's moving potentially 35 miles an hour down the road.

而且这还是在非公路上运行的。

And it's it's an off road thing.

不过呢，

But like

两百万磅重的设备，以每小时35英里的速度在野外行驶。

2,000,000 pound machine moving 35 miles an hour off road.

是的，老兄。

Yeah, dude.

这就是为什么你得选最顶级的全地形车。

This is why you have to get the ultimate ATV.

用来干活的。

To work.

这才是真正的顶级全地形车。

This is the ultimate ATV.

所以，你只要坐进去，就能亲身体验一下。

So, no, I you get in it, and you can, you know, you can experience it.

我的意思是，又没有专门的游乐场让你玩这个，但我确实亲自体验过，能坐进机器里，亲自看看里面是怎么运作的。

Like, I mean, it's not like there's, like, a amusement park for this, but, like, I've certainly experienced it where I can get in the, I get in the machines and and check out what's going on.

这简直就是矿用卡车吧。

This is like the dump truck Are

这些车中有

any of these

轮胎差不多有20英尺高？

20 foot tires essentially?

这些公司里，有没有哪些是潜在的收购目标？你们可以进去说：你们确实有不少做得不错的地方，但那些你们永远也搞不定的环节，恰恰是我们擅长的。

Are any of these are any of these companies, like, acquisition targets where you would you would be able to come in and say, like, you're doing a lot of stuff well, but here's all the stuff that you're never gonna figure out like us.

对。

And Mhmm.

这么说吧，我们看待这个问题的角度是，矿场里大部分车辆都用在运输环节。

I mean, look, I would say the way we think about it is the haulage part of a mine is where most of the vehicles are.

所以我们把运输系统看作矿场的血液循环系统。

And so and we think of haulage as the cardiovascular system of a mine.

对。

Mhmm.

我们当然和所有其他设备都紧密关联，但我们并不生产其他所有设备。

So we're obviously very connected to all the other machines, but we don't do all the other machines.

我们就像在血液循环系统里一样。

We're like in an Mhmm.

所以我们可以通过API和他们合作。

So we can work with them where, like, there's APIs.

因为如果你在做运输，就得知道其他设备的位置和状态，没错。

Because, like, if you're doing haulage, you need to know where the other machines are and what their status is Yeah.

举个例子。

As an example.

那里需要有调度和协调，这挺有意思的。

There needs to be orchestration and coordination there, which is pretty interesting.

说到收购，我得坦白说，我的想法确实有点像优步那种模式。

In terms of, like, acquisition, like, you know, I I do I have to sort of admit, like, the Uber mentality my mentality.

我们就这么想。

Let's just Yeah.

对。

Yeah.

我的Dune。

My Dune.

对。

Yeah.

是的。

Yeah.

我觉得现在Uber的情况可能不一样了，但在我们那个圈子里，我们压根没搞过收购。

It's like not I guess Uber is different today, but in my world, we didn't acquire shit.

我们就是自己从头做起。

We just built.

对。

Yeah.

没错。

That's right.

我还没想好怎么看待这件事。

I don't know if I have an opinion yet.

我倒不是非要死守某种原则，只要觉得能做成，我们就去做。

I'm not, like, religious about it, but if we feel like we can build something, we do.

但有时候别人做出了特别棒的东西，你就会想，不如咱们合作吧。

But that but sometimes people have differentiated awesome stuff, and you're like, let's partner.

我们对这种合作持开放态度。

We're open to it.

你懂的。

You know?

如果我是个刚毕业的斯坦福计算机系学生，担心软件工程这条路不再像以前那样轻松，没法在谷歌、优步这些公司之间跳来跳去，你会怎么说服我去亚马逊从头做起？

How would you pitch me if I was a young person, Stanford CS, new grad, worried about software engineering not being the easy path where I can bounce around from Google, and maybe Uber had a cushy job for me.

你会怎么劝我，去亚马逊从零开始做点东西？

Pitch me on going to the Amazon and building

这太棒了。

And that's awesome.

我的意思是，这真的太棒了。

I mean, that's awesome.

我以为我已经说过了。

I thought I just did.

我的意思，那正是我的建议。

I mean, that was the pitch.

那正是我的建议。

That was the pitch.

不过，你

But Do you

觉得年轻人现在能接受这个建议吗？

think do you think, young people are receptive to this pitch yet?

他们会不会愿意接受？

Are we about to be receptive?

他们凭什么愿意接受？

Why should they be receptive?

这挺有意思的，因为我遇到的年轻人大多都挺开放的。

It's really interesting because I only run into the young people that are receptive.

当然。

Sure.

我是说，我可不会去给那种懒散的家伙做推介，那种人根本不想干活。

Like, I'm not out there pitching, like, lame sauce dude who doesn't wanna work.

当然。

Sure.

当然。

Sure.

我是说，我根本不会和那种人待在同一个房间里。

Like, I don't end up I don't end up in the same room as this guy.

你想做什么工作？是想找一份每天对着电脑的办公室工作，还是想去亚马逊流域的矿场里实地工作，就像科幻片里那样，亲手搭建点什么？

Do you wanna do you want a a job where do you want a laptop job, or do you wanna be dropped in to a mine in the Amazon and, like, build build, you know, science fiction?

这就是问题所在。

This is the thing.

是吧？

Right?

这就是Atoms这个项目有意思的地方。

This is why the Atoms thing is cool.

对。

Mhmm.

你可不是随便往App商店上扔个应用就完事了。

Because you're not dropping a you're not dropping a fucking app in the App Store.

你是在控制一台重达两百万磅、时速三十五英里、运着黄金的巨型机器自动化运行。

You're, like, automating a 2,000,000 pound machine going 35 miles an hour carrying gold.

你平时看不看，有没有收到

Do you do you watch, do you get do you get an

今天这对话满嘴脏话啊。

There's a lot of profanity happening today.

我不会。

I don't

我不知道怎么搞的，但就是发生了。

know why it's happening, but it is.

哦，随它去吧。

Oh, it's let it flow.

我只是想提一下这件事。

Just wanna acknowledge it.

你平时会看科幻作品吗？会不会从中得到灵感？

Do you do you watch, sign do you do you get inspired by science fiction at all?

我都能想象，你一边看《沙丘》，一边给团队发消息发图片。

I can I can imagine, like, watching Dune for you as you're just, like, texting pictures to the team?

当然了。

Of course.

我最喜欢的是阿西莫夫。

I'm like, I'm my fave is is Asimov.

他就是我最欣赏的作家。

He's my fave.

没错。

Yeah.

你知道吗，《我，机器人》系列简直太震撼了。

You know, the iRobot series is, like, just so epic.

从《我，机器人》系列里，你对人工智能安全风险最大的体会是什么？

What is your takeaway from the iRobot series with regard to AI safety doom generally?

你有没有过这样的时刻，觉得我们可能根本解决不了这个问题？

Have you ever had moments of, maybe we won't figure it out?

解决不了什么问题？

Won't figure what out?

就是广义上的对齐问题？

The alignment problem broadly?

比如《我，机器人》里的机器人三定律，没错。

Like, the the the iRobot, the three laws of robotics Sure.

对。

Sure.

真是个优雅的解决方案。

A of elegant solution.

当然，它已经被验证过了。

It's tested, obviously.

但我真希望有一天，无论是悲观派还是AI开发者，都能认同：是的，机器人三定律足以应对我们未来的挑战。

But I love I would love to come back to a world where where everyone, both the doomers and the AI builders agree that, yep, the three laws of robotics will be sufficient to be going where we're going.

某种程度上，剧里确实如此。

Well, in some ways, in the series Yeah.

但机器人三定律并不总是管用。

The three laws don't always work out.

对。

Yeah.

我觉得这三条定律其实很有讲究，虽然表面上看很简单。

So I think there's a lot of I I thought there's a lot of nuance to those three laws even though the laws are sort of so simple.

对。

Yeah.

我很欣赏这三条定律的初衷。

I love the intention of those laws.

当然。

Sure.

我看待这个问题的角度稍微有点不同，因为我创业已经有很长一段时间了。

I I sort of think of it a little bit differently, which is I have been entrepreneuring for a a long time.

对。

Mhmm.

真的很久了。

Like a long time.

是的。

Yeah.

而且我失败了。

And I have failed.

当我反思失败的原因时，通常是因为我做的东西没人喜欢。

And when I think about why I failed, it's usually because I was building something that nobody liked.

对。

Mhmm.

所以，如果你做的东西大家都不喜欢，我觉得你很难成功。

So if you build something that people don't like, I don't think you're going to succeed.

那这和你刚才的问题有什么关系呢？

So how does that relate to your question?

他跟我说，你一定要告诉我

He's like, please tell me

因为我完全没搞懂你在说什么。

because I'm not connecting the dots at all.

你到底在讲什么？

What are you talking about?

如果你做出的东西违背人性，不为人们服务，我觉得你不可能成功。

Well, if you make something that is anti human, if you make something that doesn't serve people, I don't think you're gonna make it.

你不可能成功的。

I don't think you're gonna make it.

顺便说一句，没错，我们确实用人工智能来帮我们做决策之类的。

And by the way, like, yes, we're using AI to help us make decisions, etcetera.

但这些人工智能真正真正想做什么呢？

But what do those AIs really, really wanna do?

是不是太多了？

Almost too much?

它们就是想讨好我们。

They wanna please us.

所以我觉得，如果你做的东西不是人类想要的，那肯定行不通。

So I just think if you're not making stuff that humans want, it's not gonna work out.

这其实挺明显的，显而易见。

And that's kinda obvious, obviously.

但我认为这趋势还会继续。

But I think it keeps going.

对。

Yeah.

当然，也有风险和问题，还有失误，但这些就是我思考这些问题的出发点，我们不可能掌控所有事情。

And, yes, there's the dangers and the things and the miss, but that's my that's my starting point for how I think about these things, and we can't control all the things.

对。

Yeah.

当然，安全措施是必须的，但总会遇到优先级冲突——比如该先顾哪一头，怎么安排，这正是阿西莫夫机器人三定律想解决的问题。

And I do think, of course, you have to have safety situations, and there's collisions of, like, what do I what do I prioritize first and how do I do it, which is I think where Asimov's laws go.

我不写科幻小说，我只是在做实际的事，确保机器别跑偏了。

But I I I instead of writing sci fi books, I'm just doing the thing, and I'm making sure that the the machine stays on the road.

对。

Yes.

说到‘做实事’、让机器不跑偏，我觉得你的全局视角很大程度上来自现实经验——比如你看得清扩散技术的发展，也清楚自动驾驶系统花了多久才普及，所以你认为AI必须以更贴近生活的方式落地，觉得这整体是正向的，机会更多。

And and related to that idea of, like, doing the thing, making the machine stay on the road, I imagine that your your world board view is somewhat informed by your contact with reality, the fact that you can see the progress of diffusion, how long drive by wire systems take took to roll out, and the need for AI to be deployed in, like, tactile ways that that that you you just see it as more positive sum, more upper there's more opportunity.

你是在部署人们真正需要的机器人。

Like you're deploying robots that people want.

现在的机器人，你看吧。

Right now, robot I mean, look.

总有一天，机器人会有自己的银行账户，有自己的公民身份，诸如此类。

There's some point where robots have their own bank accounts and their citizens and all this.

但我们还没走到那一步。

We're just not there yet.

当然。

Sure.

当然。

Sure.

但在我们达到那一步之前

And before we until we get there

对。

Yeah.

那个机器人是有主人的。

That robot is owned by somebody.

没错。

Yeah.

那个主人是有银行账户的。

That somebody has a bank account.

没错。

Yeah.

他们是根据你为他们带来的价值来付费的

And they are paying based on the value you're bringing them

是的

Yeah.

因为他们喜欢你的东西

Because they like your stuff.

对

Mhmm.

所以如果你做的是人类不喜欢的事，那你就完了

So if you are doing things that humans don't like, you're done.

相信我，我干过这事儿

And trust me, I've done it.

我做过一些没人喜欢的东西，那感觉太糟了

I've built things that nobody liked and it sucked.

我不建议任何人这么干

I don't recommend anybody do it.

如果能避开，那绝对应该避开。

If you can avoid it, you totally should.

从商业模式的角度来看，你现在在矿产方面做些什么？

On the business model side, what are you doing now in mining?

对。

Mhmm.

你觉得未来它可能会发展到什么程度？

And where do you think it could go over time?

如果你能引入一个系统，帮助别人把产量提高百分之三十到四十，我猜最终你会搞个合资，让你的人去卖服务。

Because if you're able to bring in a system that helps someone increase their yield 30 to 40%, I I imagine eventually you just do some type of JV so that your guys have We're selling work.

激励型SaaS。

Incentive SaaS.

你知道的，企业级软件公司是怎么做的，你就该往这个方向想。

There's you know, and the the instinct should be how do enterprise company enterprise software companies do it.

先从这儿开始。

Start there.

你们心里清楚这一点。

And you guys will know that.

你们就是知道的。

Like, you you know that.

是的。

Yeah.

对。

Yeah.

那答案是什么？

What's the answer?

你就说你是企业软件公司。

Just say you're enterprise software company.

你是在帮公司提升效率。

You're making a company more productive.

那你怎么做？

What do you do?

涨价，用订阅制模式。

Raise prices, subscription.

或者等你真正证明了能提升 productivity 之后，再涨价。

Or you The price goes up when you prove that productivity.

对。

Yeah.

所以这是最基本的情况。

So there's baseline.

对。

Yeah.

然后，根据结果，你会得到一点额外的动力。

And then based on outcomes, you get a little extra juice.

当然。

Sure.

当然。

Sure.

你也可以这么说，或者

And you can say Or

你总得确保自己创造的价值比你拿走的多，但这里面就像猫鼠游戏，你总在琢磨怎么拿捏分寸，别把价值给得太大方了。

you're always trying to make sure that, like, you wanna be producing creating more value than you're capturing, but there's this sort of cat and mouse game where you're always trying to cut you don't you don't wanna give away maybe too much value.

这点完全对。

Here totally.

但关键是这样：

But here's the thing.

你从不会直接去找客户，

You never go to a customer.

不管你卖的是什么，

I don't care what you're selling.

明白吗？

Okay?

我不在乎你是不是

I don't care if

这是企业级软件。

it's enterprise software.

不管你卖的是什么小零件。

I don't care if it's widgets.

不管是什么东西。

I don't care what it is.

你从来不会去对客户说：把你们收益的百分之几分给我。

You never go to a customer and say, give me a percentage of your stuff.

是的。

Yeah.

你会去对客户说：这是我们产品的价格。

You go to a customer and say, here's the price of our stuff.

如果它真帮到你了，我们觉得应该多拿一点回报，就是额外的收益，你知道的，随便怎么说吧。

And if it does really well for you, we think we should get a little more scratch, casheesh stuff, you know, whatever.

你明白我的意思吧？

You know what I mean?

是的

Yeah.

对

Yeah.

就这么简单。

And it's that simple.

别太粗俗地对待

Don't be crass about

它。

it.

明白。

Mhmm.

而且，跟合作伙伴联手，他们也愿意配合。

And, you know, partner with folks, and they're down.

他们也想成功。

They wanna win too.

你知道吧。

You know?

企业级软件那种谈判方式，整个流程得多快才行。

How quickly do enterprise it's an enterprise software style negotiation or approach to the whole thing.

是的。

Yeah.

你的价值越独特，你就能拿到越多。

And the more differentiated your value is, the more you're gonna get.

对。

Yeah.

你现在招聘高管的流程是怎样的？

What is your process for hiring executives today?

祈祷。

Pray.

我本来还以为你有什么凯拉尼克的系统，能达成99%的成功率。

I was hoping I was hoping you had the Kalanick system to achieve a 99%

我为什么要告诉你？根本没理由啊。

would I tell you, Why would I tell period?

不会的。

No.

我的意思是，不会。

I mean No.

但我认为你做不到。

But I think you can't.

这种事，你就算把你的做法原原本本告诉别人，他们也不是卡兰尼克那样的人。

This is one of those things you can tell people exactly what you do, and they're not they're not Kalanick.

所以他们就算知道也未必能跟你竞争。

So they're not it doesn't that doesn't mean they can compete with you.

你知道吗，他们确实有可能，嗯。

You know, they could they yeah.

好吧。

Okay.

那我该怎么说呢？

So how would I put it?

听我说，不管你找谁，没人能把高管这事儿彻底搞明白。

Look, I think no matter who you go, nobody's nailed executives all the way.

对。

Mhmm.

这事儿挺怪的，因为高管们说起话来一套一套的，特别厉害。

It's it's it's weird because what will happen is executives talk a fucking awesome game.

其实你对高管有两个基本期待。

And there's two things you want an executive to do.

一是能大规模组织和管理，二是能带团队、领大局。

You want them to be able to organize at scale, organize and manage at scale, lead at scale.

二是要成为最顶尖、最有战略眼光的问题解决高手。

You also want them to be epic problem solvers, the most strategic badass problem solvers alive.

这就像惯用左手还是右手一样自然。

This is like being left handed or right handed.

但真正左右手都能用的人太少了，而你恰恰就需要这样的人。

And there's very few people that are ambidextrous, but you need that.

不过，大多数人总会不自觉地偏向某一边。

Now, somebody's they're always leaning a little bit one side or the other.

最出色的高管，是那些两边都能做得很棒的人。

The best executives are the ones that are doing both well.

我这些年做下来，慢慢得出一个结论：解决问题的能力才是最重要的。

But I have come to the conclusion over my years doing the stuff is the problem solving is the most important thing.

如果你找的人擅长组织管理，却不会解决问题，那他只会把荒唐的事安排得井井有条。

If you get somebody who organizes and manages well but cannot solve a problem, they're going to be doing ridiculous stuff in a super organized way.

对。

Yeah.

所以这就是我的看法——也许我管人带团队的方式不止一种，但归根结底，限制你想象力的，只有管理能力。

And so that's the and and and sort of my theory, I I maybe there's a couple of theories on how I manage or how I lead, is that the only constraint on your imagination is management capacity.

没错。

Yeah.

那管理能力到底是什么呢？

But what is management capacity?

其实就是大规模解决问题的能力。

It's really problem solving at scale.

没错。

Sure.

因为如果你在那边做得特别出色，你猜怎么着？

Because if you are doing super well over there, guess what?

他们在那儿解决问题。

They're problem solving there.

我还能制造出其他更棘手的问题。

I can create other awesome problems.

对。

Yeah.

对。

Yeah.

就像，特别喜欢制造问题。

Like, love creating problems.

当然。

Sure.

去解决那些问题吧。

Go solve those too.

对。

Yeah.

但要是我没有这个管理能力，那我就彻底完蛋了。

But if I don't have the management capacity, then I'm effed.

没错。

Sure.

我采用的管理方式有点像首席问题解决官：专门把那些没人能解决、影响最大的难题都接到自己手里来处理。

The the management style that I do is sort of problem solver in chief, which is I take the most impactful problems that are not being solved, and that's on my desk.

没错。

Yeah.

不管是叫桌面、房间，还是别的啥名字吧。

Or desk or room or whatever you wanna call it.

我大部分时间都耗在这上面。

That's where I'm spending my time.

所以别人总会问：你平时都忙点啥？

So people go, oh, what do you spend your time on?

这得看眼下哪些问题最关键。

Like, it depends what the fricking problems are that matter.

而且情况随时可能变。

And it can change.

我就是这么做事的。

And that's how I roll.

但一旦你成了首席问题解决者，这种思维就会往下传。

But it means once you have a problem solver in chief mentality, that flows downward.

这意味着我的每个下属都得是被授权的首席问题解决者。

That means any direct report of mine must be the deputized problem solver in chief.

因为一天只有24小时，他们得有自己的理由。

And they've got their because there's only twenty four hours in a day.

我自己能解决的问题有限。

I can only solve so many myself.

他们得在自己的领域里继续这么做，再对下面的人也一样。

They have to then take that for their world and do the same thing, and then do the same thing to their people.

对。

Mhmm.

对。

Yep.

所以归根结底，你得证明这些人真能解决实际问题，而不是光说不练。

So the bottom line is you gotta prove that these folks can solve actual problems and aren't just talking the talk.

这是首要条件。

That's the number one.

然后才是其他方面。

And then correct.

在面试过程中，我会模拟实际合作的场景，让员工第一天上班就像已经工作了两周一样。

And then I on the interview process, simulate what it's like working together so that day one is really feels like week two.

第一天，你必须得充满热情。

And day one, you better be excited.

如果你在面试中体验过合作方式后，第一天依然充满热情，那第一天就真的像第二周了，你也把很多风险从系统里剔除了。

So if you're excited in day one, after simulating what it's like working together in the interview process, then day one is really week two and you're still excited, you took a lot of risk out of the system.

我就说这些了。

That's all I got for you.

我有个关于监管的问题。

I have question about regulation.

优步当年就是一座城市一座城市地推进。

Uber famously went city by city.

是的。

Yeah.

AI实验室现在正在为联邦优先权的问题激烈竞争。

The AI labs are duking it out over federal preemption.