本·霍洛维茨：xAI高管出走、伊利亚50亿美元SSI估值、苹果的AI危机、AI发展速度 | #232

你知道会发生什么吗？

You know what happens?

他们最后都会给我发邮件。

They end up emailing me.

我现在每天都会收到好几封来自Open Claw代理的邮件，问我Peter的电话号码是多少。

I I get several emails probably per day at this point from Open Claw agents asking me what Peter's number is.

这太好笑了。

That's hilarious.

当他们打电话给我并问他们是从哪里弄到的号码时，Alex

And when they call me and ask them where they got it from, Alex

没关系。

That's alright.

关于Eleven Labs，我想说点看法，也许只是做个评论。

Thoughts, maybe maybe just a comment on the eleven labs.

如果你一直在使用Eleven Labs的v3 alpha版本模型，这几乎不会让人感到惊讶。

So if you've been using the version three v three alpha model from Eleven Labs, this should hardly be surprising.

V3允许你在Eleven Labs平台使用文本转语音时，通过方括号指定情感表达，这一点已经存在好几个月了。

V three enables you if you use text to speech in the Eleven Labs platform to specify with brackets emotional expressions, and this has been around for for many months.

对我来说，有点有趣的是，我们至少已经知道如何进行音频到音频的转换一年多了。

What's somewhat interesting here, to me at least, is we've known how to do audio to audio transfer for probably a year plus at this point.

如果你用过OpenAI的高级语音模式，那你就已经使用过音频到音频的变换模型了。

If you've used advanced voice mode from OpenAI, you've used audio to audio transformers.

但有趣的是，Eleven Labs更以文本转语音闻名，而不是语音转语音。

But what's somewhat interesting here is Eleven Labs is better known for text to speech than speech to speech.

据我所知，我们这里讨论的这种新表达模式似乎仍然依赖于文本模态，而这在过去一直非常困难。

As far as I can tell, this new expressive mode that we're talking about here seems to still leverage the text modality, which historically has been very difficult.

你必须先将语音转为文本，再转回语音，这会导致高延迟。

You'd have to go to from speech to text back to speech, which was high latency.

它很慢。

It was slow.

它不够对话感，而11 Labs似乎在无需直接进行音频到音频转换的情况下，找到了一种将语音转文本再转回语音的方法，让整个过程听起来自然、有互动感且实时。

It didn't feel very conversational, and somehow seems like without having to do direct audio to audio, 11 has found a way to do speech to text back to speech in a way that that feels natural and turn taking and real time.

所以我认为是的。

So I I think it's Yeah.

在某种意义上，这是一种渐进式的进步，但在另一种意义上，如果他们确实仍然保留了文本——这在循环中更容易挖掘和分析——那么这很可能是一项实质性的进步。

In some sense an incremental advance, but another sense, if they really are still keeping text, which is much easier to mine and analyze in the loop, it probably is a material advance.

是的，亚历克斯，通过这次演示，我认为我们在语音方面已经跨越了恐怖谷。

Yeah, Alex, I would say we've crossed the uncanny valley on voice at this point with this demonstration.

然后，在人工智能时代，语音将成为新的界面，对吧？

And then voice becomes the new interface in the AI era, right?

我的意思是，我花在与AI语音交互上的时间，远远超过那种繁琐的打字交互。

I mean, I can't tell you the amount of time that I'm just speaking to AI versus this cumbersome typing at it.

所以我认为，从这次演示中得出的这两个要点非常重要。

So I think those two things are are really important takeaway from this.

我觉得彼得你这样总结得很好。

I think that's a great way of putting it, Peter.

是的。

Yeah.

但问题是，你真的想把音频作为主要模型吗？

But the problem is, do you really want to use audio as your primary model?

我的意思是，在脑机接口的隔离环境下，它效果不错。

I mean, it works well in BCI isolation.

好吧。

Okay.

是的。

Yeah.

你想要脑机接口。

You want BCI.

我也想要脑机接口，还想要可穿戴设备，以及手势交互界面。

I I want a BCI too, and I want wearables, and I I want gestural interfaces.

我都想要。

I want it all.

但对大多数人来说，我是说，纽约有个著名的轶事。

But for for most people, I I mean, is it New York this is sort of famous anecdote.

上世纪八十年代，《纽约时报》做过一项著名研究，他们让所有记者使用当时最先进的语音转文字系统，通过语音输入来写作。

New York Times in the nineteen eighties did a famous study where they just put all of their their reporters on then state of the art speech to text systems and ask them to use voice.

结果，写作质量下降了。

And what happened was the writing quality went down.

为什么会下降呢？

Why did it go down?

因为当你说话时，很难像打字那样充分地提前思考。

Because it's difficult to think ahead as well as you can when if you're just typing.

如果你在说话，你其实是在利用大脑中与打字相似的区域。

If you're speaking, you're you're leveraging, right, similar portions of of the brain.

所以我还没有完全相信语音会是未来的主流交互方式。

So I'm not a 100% sold yet that speech is the modality of the future.

我确实喜欢脑机接口。

I I do like BCIs.

我也喜欢可穿戴设备和手势交互界面。

I do like wearables and gestural interfaces.

我喜欢打字。

I do like typing.

但我觉得，对于高带宽操作来说，语音对我来说还不确定。

But speech, I think, is still out for me for high bandwidth operation.

是的。

Yeah.

我有一件事想说

One thing I

我我

I I

你请说。

Go ahead.

对于那些不一定为《纽约时报》写作的普通人来说，我认为语音通常是——至少根据我的经验——首选方式。

I was just gonna say for for regular people who don't necessarily write for The New York Times, I think speech is often, or at least in my experience, is the mode of choice.

当然。

For sure.

我还想提一下Eleven Labs，顺便做个免责声明：我们是Eleven Labs的大投资者，我女儿索菲亚也在那里工作，所以我对这家公司有偏见。但让我感到震惊的是，当初我们投资时，最大的疑问是：最先进的模型难道不能说话吗？

The other thing I'd say about Eleven Labs and kind of quick disclaimer that, you know, we're a big investor in Eleven Labs and my daughter Sophia works there, so I'm But biased towards the you know, one of the really amazingly just kinda landscape shocking things to me about eleven Labs is, you know, when we invested originally, the big question was, well, like, aren't the, state of the art models gonna be able to talk?

我的意思是，它们当然能说话。

I mean, of course they're gonna be able to talk.

但正确地说话、把握细微差别，并为开发者打造合适的产品，这对他们来说证明是非常可持续的，我认为这在审视整个行业格局时很有意思——能力与产品之间的差距是巨大的。

But speaking correctly with the right nuance and building the right products for developers and so forth has proven to be very sustainable for them, which I think is interesting as you look at the entire landscape, the difference between the capability and the product is significant.

是的，让我感到惊讶的是Eleven Labs的什么方面呢？

Yeah, you know what's amazing to me about Eleven Labs?

我们实验室里有两个公司，分别做VoiceRun和Vocara。

We have two companies here in the lab that do VoiceRun and Vocara.

令人惊叹的是，这些是基于原始数据自我组织训练的系统，它们的表现简直令人难以置信。

And what's amazing is these are self organizing systems that are trained off raw data, and what they do well just blows your mind.

但在语音领域，结果发现对话轮换管理异常困难。

But within voice, it turned out the turn management was very, very hard.

非常困难。

Very hard.

我从来没想过会这样。

And you're like, I didn't ever thought.

与那些能够说出有智慧话语的惊人合成语音相比，这看起来似乎微不足道，但它们不知道何时该停止说话。

Like, seems so trivial compared to actually doing these incredible synthetic voices that can say intelligent things, but they don't know when to stop talking.

就像在这个播客里，我一停，你就接着说。

You know, like on this podcast, when I stop, you start.

你一停，我就接上，这对我们来说非常自然。

When you stop, it's, like, very natural to us.

但因为训练数据里没有这种内容，所以到目前为止，它们在这方面表现得非常糟糕。

But because that wasn't in the training data, they they're just, you know, up until now, terrible at it.

等等。

Wait.

根据我们的听众反馈，我们经常在对话中互相打断，完全不同步。

According to our listeners, we're talking over each other all the time in a very unsynchronized way.

情况也是一样。

Same thing as the same.

我们从这些语音转文本再转语音的模型中学到的太多了，而不是轮流发言。

We have so much to learn from these speech to text to speech models, not turn taking.

是的。

Yeah.

天哪。

Oh my god.

是的。

Yeah.

好吧。

Alright.

下一个话题。

Next topic here.

如果你们不介意的话，我会继续往下讲。

If you guys are okay with it, I'll move us along.

在SpaceX和XAI合并之后，本着完全透明的原则，我既是SpaceX的投资人，也是XAI的投资人，你可能也是，本。

Just following the merger of SpaceX and XAI, in the spirit of full disclosure, I'm an investor in both, probably you are as well, Ben.

大量人员离职——

Large number of departures-

三位X的创始人都走了。

All three of the Xs.

是的。

Yes.

XAI公司有大量人员离职，尤其是创始团队中的多位华裔联合创始人。

A large number of departures from XAI, from the founding team, mostly ethnic Chinese co founders.

这很可能是因为SpaceX的ITAR法规。

And it's likely due to the ITAR regulations of SpaceX.

对吧？

Right?

在埃隆做过的一些视频演示中，XAI团队至少有一半是华裔。

And it was interesting on a few video presentations that Elon's done that's had the team from XAI, at least half of the team there was ethnic Chinese.

而且你知道，那里的文化孕育了极其出色的数学家和程序员。

And, you know, the culture there breeds incredible mathematicians and programmers.

有什么看法吗，本？

Any any comments, Ben?

我很想听听你的想法，但我查看了时间线，发现高级AI人才的离职潮甚至早于合并决策的出台。

I'd love to get your thoughts on this, but I looked at the timeline and the exodus of senior AI talent predates even the decision on who to merge with what.

所以，认为这与ITAR有关的理论并不成立，我不清楚他们是被解雇还是主动离开，但听说他们都是和平离职的。

So the theory that this is related to ITAR, and it wasn't clear to me whether they were fired or whether they left, you know, because they all leave on good terms.

我不知道你有没有什么见解。

I don't know if you have any insight.

是的，这个问题我也不清楚答案。

Yeah, don't know the answer to that question.

我想提一下相关的情况：我们最近听说一些中国籍博士生表示，中国政府正在加强对美国学术界使用中国开源模型的管控。

I will say that it's So in a related matter, we, like, recently heard from few Chinese nationals who are PhD students that the Chinese government is cracking down on the America's or US academia's use of Chinese open source models.

特别是在Meta收购Manus之后，他们非常担心技术机密会因此外流。

Like, particularly post the Meta Manus acquisition, they're, like, very, very worried about actually secrets going this way.

所以，这和我们原本担心的情况正好相反——我们整个开源工作其实都基于中国模型，因为美国的开源模型本来就比较少。

So kind of the opposite of what we've been worried about, which they you know, like, our whole open source kinda work is based on the Chinese model since there haven't been as many US open source models.

所以我认为，整个事情即将变得更加复杂。

So that's this whole thing, I think, is about to get more complicated.

确实如此。

It is.

我的意思是，

I mean,

如果你读这些推文，吴先生，

if you read these tweets, Mr.

和鲍勃先生，

Wu and Mr.

两人都对XAI充满热情。

Bob, both are super enthusiastic about XAI.

他们并没有表现出任何焦虑。

They're not leading with any kind of angst.

你知道，他们说他们非常喜欢XAI系列。

You know, they're saying that they love the XAI family.

我们正朝着百倍生产力的时代迈进。

We're heading towards an age of 100x productivity.

所以我认为他们不会自行离开。

So I don't think they would have left on their own accord.

这是我的个人观点。

That's my personal opinion.

那么是什么促使了这一点呢？

So what drove it?

你知道，让我们看看亚历克斯上次的评论。

You know, enter Alex's comment from last time.

他们的股票是什么时候归属的？

Well, when did their vest their stock vest?

我的意思是，

I mean,

还有另一种解释，那就是SpaceX相对于XAI的员工规模要大得多，也许因此发生了一场自然的重组。

there is a there is another explanation, which is just SpaceX is a very large company relative to x AI's headcount, and maybe there was a natural reorganization that happened as a result of that.

我还要指出，X AI 在这次合并之前一直在向美国国防部提供基础模型服务。

I'll I'll also point out that x AI was selling Foundation Model Services to the Department of War prior to this merger.

因此，我再次质疑，某种国籍方面的担忧是否真的是根本原因。

So I would again query whether some sort of nationality concern is really the trigger.

对我来说，这更可能是 XAI 开始并入 SpaceX 后自然重组的结果。

It seems more likely to me this is just the result of a natural reorg of XAI starting to merge into SpaceX.

是的。

Yeah.

如果这真是由于 ITAR 等规定所致，那么美国在人工智能领域的主导地位实际上很大程度上依赖于移民人才。

Well, if it were a result of ITAR and such, America's AI dominance is really built significantly immigrant talent.

所以，如果我们开始对博士移民做出这种反应，我个人认为——我也想知道你的看法，本——所有在美国攻读博士学位的人，毕业时都应该直接获得绿卡。

So if we're going to start having this kind of a reaction to PhD immigrants, I mean, I personally think, and I'm curious what you think Ben, that everybody going through a PhD program in The US should have a green card staple to their PhD when they graduate.

我认为把人赶出去的做法是错误的。

I think the idea of, you know, kicking people out is the wrong approach.

对。

Yeah.

我认为总体上来说，这是正确的。

I I think in generally that's in general, that's correct.

我确实认为，我们还没有完全想清楚中国的情况，比如我们公司里都有那些杰出的中国籍员工，我觉得是这样。

I do think that the the there there's I I don't think we've quite thought through the case of China totally in that, like, we have, you know, amazing Chinese nationals who work for us and in every company we have, I think.

当然。

Of course.

所以，来到这里的顶尖人才是非凡的。

So, you know, the talent that comes here is extraordinary.

我并不是说这个想法完全没有风险。

I I wouldn't say there's no risk to that idea.

我我

I I

我想我只是想说这一点。

guess I would just say that.

但我认为你说得对，将军。

But I I think being I I think you're right, General.

我认为我们应该保持开放。

I think we should be open.

我们应该接纳那些来到这里的杰出人才，而不是与之对抗，因为我们终究会输。

We should accept the phenomenal talent that comes over here and not not fight it because we'll we'll lose anyway.

我的意思是，美国并没有隐瞒什么秘密。

I mean, like, we're not keeping anything secret in America.

尤其是在美国公司里，更没有。

Certainly not in American companies.

它们都没有跳过任何环节。

There's none of them have skips.

它们都没有针对人员的良好安全协议。

None of them have good security protocols around personnel.

别让我谈隐私早已不复存在的这个观点了。

Don't get me started on this on the idea that privacy is long since dead.

但既然如此，接下来是你。

But so here You're next

年。

year.

正确。

Correct.

这是来自xAI联合创始人吉米·巴格的一条推文。

Here's a tweet from Jimmy Bhag, a co founder at x AI.

递归式自我改进循环可能在接下来的十二个月内上线。

Recursive self improvement loops likely to go live in the next twelve months.

2026年将会非常疯狂，很可能是我们物种未来最繁忙、最具决定性的一年。

2026 is going to be insane and likely the busiest and most consequential year for the future of our species.

你会听到这个。

You're going hear this

我就像，你几个

a I'm like, few you're

你会多次听到，未来几年是一个巨大的转折点，一切都会改变。

hear this a few times that these next few years are a massive inflection point that everything changes.

我们将来会回望这一转折点，跨越数千年的历史，认为它是一个关键转折，还是仅仅是一次平滑的奇点，亚历克斯？

And we're gonna look back thousands of years in the future, back to this inflection point, or is it just a smooth singularity, Alex?

我认为两者都有可能。

I think it can be both.

我认为我们已经进入了递归自我改进的时代。

I I think we've already hit the era of recursive self improvement.

我在每一期节目里、每天在自己的通讯中，都在拍着桌子 rhetorically 地谈论递归自我改进。

I'm banging the the table rhetorically every episode and and every day in my newsletter talking about recursive self improvement.

我们已经到达那里了。

We're we're there.

目前，所有前沿实验室都在使用自己的模型来开发新的模型。

All of the frontier labs are are using their own models at this point to develop their models.

从实际操作来看，这几乎就是递归自我改进的定义。

That's practically the definition of recursive self improvement at this point in practice.

我认为不是未来十二个月的事。

I don't think it's the next twelve months.

我觉得现在就是了。

Think it's now.

如果跳过所有中间阶段，2026年会不会相对于以往的年份变得疯狂？

And is 2026 going to be insane relative to years past if you just sort of skip over all of the interim time?

当然会。

Absolutely.

从某种意义上说，它已经疯狂了吗？

Is it already insane in some sense?

当然。

Absolutely.

即使我们只看过去24小时发生的事件——其中一些我觉得我们稍后会谈到，比如自我复制的AI，以及AI在AI陪审团面前调解自身纠纷的法庭。

Even if we just look at the events of the past twenty four hours, some of which I think we'll get to, like self replicating AI and courts where AIs can mediate their own disputes in front of an AI jury.

几年前，这还完全是科幻小说。

That would have been pure science fiction several years ago.

这并不是未来十二个月的事。

That's not twelve months from now.

这并不是严格意义上的未来十二个月内发生的事。

It's not quote unquote within the next twelve months.

这是过去二十四小时内的事。

That's the past twenty four hours.

我觉得这确实是过去二十四小时内的事。

I think that's that's like past twenty four hours.

所以我认为吉米反而低估了这种疯狂的程度。

So I think Jimmy is underselling, if anything, what craziness looks like.

同时，在局部时空范围内，一切仍是平滑的；你只需要看看我几个月前提到的这些故事——在几年前，这些还完全是科幻，但如果你长期生活在最核心的循环里，就会有点习以为常。

At the same time, locally space time is smooth, And you need to look no further than my saying ho to a few of these stories as being so several months ago that would have been sci fi years ago, but you get a little bit spoiled living inside the innermost loop.

我认为在有人参与和无人参与的情况下，递归式自我改进是有区别的。

I would say I do think there's delineation between recursive self improvement with a human in the loop and without one.

而且我觉得他似乎暗示了将不会有人类参与其中，而这

And I think he seemed to be implying that there'd be no human in the loop, which

嗯

Mhmm.

我认为这是一个加速器。

I I think is an accelerant.

你知道的，TBD，它能有多大的加速作用？

You know, TBD, how much of an accelerant?

但我认为这可能会非常不同。

But but I think that could be very different.

是的。

Yeah.

无许可的自我改进。

Permissionless self improvement.

对吧？

Right?

就像 flipped the switch，全力加速。

Like flip the switch and go as fast as you can.

不过，这里有一个区别。

Well, there's a distinction.

昨天在EverQuote董事会会议上提到了这个问题。

This came up at the EverQuote board meeting yesterday.

但那种通过AI提升推理速度的自我改进显然已经取得显著进展，而推理速度的提升现在可以直接转化为智能。

But the the self improvement where the inference time speed algorithms are being improved by AI clearly well underway way down the path, and then inference time speed can be directly translated into intelligence now.

我们现在已掌握如何通过增加推理循环次数来提升智商。

We now have the knowhow to turn more inference time loops into a higher IQ.

因此，这显然已经在进行中。

So that is clearly underway.

问题是，这种提升是否会推动算法的下一个部分，也就是算法的下一个环节，如果是这样，那我们已经触及临界点，现在只是在讨论它扩散的速度。

The question is, does that notch up, then allow it to work on the next part of the algorithm, the next part of the algorithm, in which case we've already hit the flashpoint, and now we're just talking about the rate at which it percolates across.

但本，我同意他的观点是关于核心算法开发的，接下来的创意100%来自AI，然后进入生产阶段，对吧？

But Ben, I agree that his comment is about the actual core algorithm development, the next ideas are 100% from AI, and then they go into production, you know?

我也认为，人类在环外、环上、环内这些说法在某种程度上是模糊不清或容易滑坡的。

I I also think human outside the loop, on the loop, in the loop is is in some sense a pretty blurry or slippery slope.

如果你还记得《杰森一家》里的乔治·杰森，乔治，那个卡通人物，你还记得乔治吗？

If if you remember George Jetson from The Jetsons, George, the I cartoon do remember George.

记得乔治，本。

Remember George, Ben.

乔治去上班时，会抱怨他的手指。

George would go into work, and he would complain about his finger.

他一整天都得用手指按一个按钮，然后抱怨手指疼，偶尔因为一整天按同一个按钮，手指都会肿起来。

He'd have to press one button all day with his finger and then complain that his finger was sore, and the finger would be swollen from pressing a single button all day occasionally.

我觉得这很好地比喻了目前前沿实验室内外递归式自我改进的状态——你有Claude代码实例，而Claude每隔几分钟就会问你：‘我有权执行以下操作吗？’

I I think that's like that that's a good metaphor for the state of recursive self improvement at at inside and outside the Frontier Labs right now where you have Claude code instances, and Claude is asking you every few minutes, do I have your permission to do the following thing?

而你按下了乔治·杰特森的按钮。

And you press the George Jetson button.

是的。

Yes.

我同意。

I approve.

不。

No.

我不批准。

I don't approve.

我们都坐在这里抱怨，因为不断点击‘批准’来运行搭载代理团队的Opus 4.6的Claude代码，导致手指肿胀。

And we're all sitting here complaining about our swollen finger from pressing approve, approve, approve for Claude code running Opus 4.6 with agent teams.

但事实上，这确实是递归的。

But, really, it is recursive.

我认为，即使我们假装自己在循环中，通过按下乔治· Jetson按钮来批准，这依然是递归式自我改进。

I I would argue it it is recursive self improvement even if we're we're pretending we're in the loop by pressing the George Jetson button.

我现在在Telegram上用Skippy做的，正是这样。

That's that's exactly what I'm doing on Telegram right now with my with Skippy.

对。

Yes.

继续进入下一阶段。

Go go on to the next stage.

我不想拖慢进度。

I don't wanna be I don't wanna slow things down.

正如我们多次所说，这是迄今为止最慢、最昂贵的阶段。

And as we've said so many times, this is the slowest most expensive it's ever gonna be.

看。

Look.

这里有几点需要注意。

There there's a couple of points here.

一个是丽莎。

One is Lisa.

我们已经讨论了很久，递归自我改进（RSI）才是奇点真正的触发因素，而且它早就发生了。

It's it's we've been talking for a while that recursive self improvement, RSI, is the real trigger for the singularity, and it happened a while ago.

所以我们现在所做的，只是在加速这一进程。

So all we're doing now is kind of accelerating that path.

我们正在谈话的同时，永久地告别了工业时代。

We're exiting the industrial age permanently as as we're talking.

所以

So

是的

Yeah.

我真的认为，奇点每分钟的演进是我经历过的最令人着迷的事情。

I really think the the minute by minute unfolding of the singularity is the most fascinating thing I've ever experienced.

你知道，亚历克斯说得完全对。

And, you know, Alex is exactly right.

我们现在所处的这个时刻，人类仍然参与其中，但很难分辨进展中哪些是人工智能的贡献，哪些是人类的贡献。

There is this point in time we're in right now where there's a human in the loop contributing, but it's really ambiguous what part of the progress is AI versus human.

如果你正在实际编写代码，那是你的想法吗？

If you're in the actual coding process, was that my idea?

那可以说是半是我的想法，但随后人工智能提出了另一个建议，我就采纳了。

It kind of was half my idea, but then the AI suggested this other thing, and I kind of adopted it.

现在就不清楚这到底是不是我的想法了。

And now it's not clear whether it was my idea or not.

但我们现在正处于这种模式中，许多核心算法的研究就是让我运行这500个测试，然后告诉我哪些超参数效果更好，或者哪种神经网络结构更优。

But we're in that mode right now where the research, you know, lot of the research in these core algorithms is just deploy these 500 tests for me and tell me which hyperparameters worked better, or which neural topology worked better.

这并不是像发明或发现相对论那样。

It's not like inventing or discovering relativity.

这只是一个接一个地进行实验，尝试不同的方法，然后选取有效的方案重新部署，从而获得更智能的AI。

It's just litany of experiments with different trials, and then taking the one that worked and redeploying it, and now you have a smarter AI.

现在它正在尝试更多的实验。

And now it's trying more trials.

我们很可能已经在这条道路上走得很远了。

It's really very likely that we're well down that path.

我不想补充，但我确实

I don't to add, I do

我认为我们也会借助AI发现下一个相对论或物理学中的类似突破。

think we're going to discover the next relativity or equivalent of relativity in physics as well with AI.

我对这一点也非常感兴趣。

I'm super interested in that as well.

预测呢？

Prediction?

你想听什么？

What would you like to hear?

下一次相对论？

Next relativity?

我们什么时候会看到AI独立发现像相对论那样重大的成果？

When are we going to have discovery by an AI of something as significant as relativity on its own?

我觉得是未来两年。

I think next two years.

好的。

Okay.

是的。

Yeah.

我也认为，这是一个关于亚历克斯的绝佳预示。

I I would also this is a great bellwether of Alex.

我同意。

I'm with that.

2017年推出的Transformer算法开启了我们如今所经历的一切。在我看来，AI在过去六个月里已经发现了比Transformer更难发现、更难解决的成果。

The the transformer algorithm 2017 that kicked off everything that we're experiencing right now, To me, the AI's already discovered things in the last six months that are harder to discover and harder to solve than the transformer was.

但我想问问你，Alex，你是否也觉得这是真的。

But I'll ask you, Alex, if you feel like that's true too.

但对我来说，现实情况显然在复杂性上还远不及相对论。

But it seems pretty clear to me that the reality is just not up there with relativity in terms of complexity.

我想，所有这些发现最终都可以归结为可以简化为方程的洞见。以相对论为例，特别是狭义相对论，你可以进行一些思想实验，并且幸运的是，你可以用大量实验数据进行验证。我可以想象一个我最喜欢的思想实验：贝叶斯超级智能。

Well, I guess everything, many of these discoveries boil down to insights that can be distilled down into equations, and you can, in principle, with relativity, with with special relativity, you can do a number of thought experiments, you can compare it fortunately with lots of experimental data, and one could imagine a thought experiment one of my favorite thought experiments is a Bayesian super intelligence.

假设你有一段视频，虽然这是牛顿引力，而不是狭义相对论，但你可以想象让一个超级智能观看一段苹果从树上掉落的视频。这个思想实验的论点是：仅凭视频的前三帧，它就应该得出牛顿引力是解释宇宙的极高后验概率的理论。

So if you had like a video of this would be Newtonian gravity, not special relativity, but you could imagine taking a super intelligence, making it watch a video of an apple falling from a tree, And the the argument and the thought experiment goes within three frames of that video, it should have concluded Newtonian gravity is a pretty high likelihood posterior probability of be of explaining the universe.

再增加几帧，它在自己的理论分布中——这背后有一整套信息论领域，称为所罗门诺夫归纳，专门研究如何从有限观测中高效推断出宇宙的理论。

And with a few more frames, it somewhere in its distribution and and this there's a whole class of information theory devoted to what's called Solomonoff induction, like just devoted to thinking about how you can efficiently infer theories of the universe from limited observations.

在那个理论分布中，应该就包含着广义相对论。

Somewhere in that distribution of theories should have been general relativity.

因此，我极其乐观地认为，借助超级智能，我们能够发现新的物理定律和变革性的发明。

So I am incredibly bullish that we'll be able to, with superintelligence, discover any new laws of physics, discover transformative inventions.

为了透明起见，既然我们在玩透明游戏，我有一家投资公司叫Physical Superintelligence，也在研究这些问题。

For for disclosure purposes, since we're playing the disclosure game, I have a portfolio company, physical superintelligence that's working on these issues as well.

我对这个领域非常看好。

I'm very bullish on this space.

太棒了。

Amazing.

我想把话题转向埃里克·施密特最近的一次对话。

I want to move us to a conversation that Eric Schmidt recently had.

我从里面剪了一个片段，我们之前在播客里讨论过这个问题：AI能被暂停吗？

I pulled a clip out of it, and the question that we've talked about on the pod before, it's been the debate, can AI be paused?

如今，几乎所有人都明确回答：不能。

The answer quite clearly by almost everybody today is no.

我们之前也和埃隆讨论过这个话题，戴夫，你和我。

We had that conversation, Dave, you and I with Elon.

我们来听听埃里克怎么说。

Let's take a listen to what Eric has to say.

这项技术一定会出现。

This technology is going to happen.

它不可能被阻止。

It's not gonna get prevented.

它不可能被停止。

It's not gonna get stopped.

有太多国家、太多人、太多动力推动它发展。

There's too many countries, too many people, too many incentives.

它一定会发生。

It's going to happen.

那么这意味着什么？

So what does this mean?

这意味着医疗保健将取得巨大突破，新药问世，能源和电力分配将更加优化。

It means great great solutions for health care, new drugs, better energy solutions, better power distribution.

但它也可能被用于不良目的。

It also means that it can be used for bad.

它可能被用于压迫，例如。

It can be used, for example, for oppression.

它可能被用来限制政府的自由，希望西方不会如此。

It can be used to limit freedom in governments, hopefully not in the West.

它可能被用于战争。

It can be used in war.

这项技术本身极具吸引力，可能影响我们的年轻人，因此我们必须确保我们的年轻人免受这项技术最糟糕部分的侵害。

The technology itself is so addictive that it can affect our young people, and we should make sure that our young people are protected from some of the worst parts of the technology.

我们现在面临着如何应对这种极其强大技术的选择。

We face choices now about how we want to deal with this incredibly powerful technology.

我要告诉你，非常重要的是，我们正经历一个将被记载数千年历史的时刻——非人类智能出现了，它成为了我们的竞争对手。

I will tell you and it's really important to understand that we are living through a moment that will be in history for thousands of years and nonhuman intelligence arrived, and it was a competitor to us.

很惊人，对吧？

Amazing, right?

我们正同时持有两种未来，对吧？

We hold two futures in superposition, right?

一个未来，人工智能成为人类精神最伟大的倡导者、支持者和加速器。

One future in which AI is the greatest advocate and supporter and accelerant to human spirit.

另一个未来，则是反乌托邦的结局。

Another future where it is a dystopian outcome.

而未来几年，如何塑造它取决于我们。

And it's ours to shape it these next few years.

戴夫，你刚才想说什么？

Dave, you were gonna say?

哦，亚历克斯和我那时在圆顶里。

Oh, Alex and I were there in the dome.

那就是他演讲时的达沃斯圆顶。

That was our Davos Dome where he was speaking.

他掌控全场的能力真让我惊叹不已。

And it just blows my mind how he owns a room.

这个人太有表达力了。

The guy is so articulate.

是的。

Yeah.

但他今年将主持丰盛峰会，我们会深入探讨很多这类话题。

But he's gonna be opening up the Abundant Summit this year, and we'll be going into a lot of this conversation.

戴夫，你和我将和他一起登台，这会很有趣。

Dave, you and I are gonna be with him on stage, so it'll be fun.

嗯，那段视频里他所描述的内容非常相似。

Well, what he's describing in that clip was very similar.

我们对他做了一整期采访，你可以在YouTube上找到。

We did a whole interview of him, which you can find on YouTube.

而且，在我们的采访中，他也提到了这一点：没有任何力量能减缓这一趋势。

And yeah, he made that point in our interview of him as well, that there's no force that's gonna slow this down.

所以，那些举着标语牌游行抗议的人，只是在浪费时间。

And so all the people picketing and walking around saying, Stop it, you're just wasting your time.

你或许可以通过其他方式提供帮助、贡献力量，引导它走向积极的方向，但站在街头举牌抗议，纯粹是在浪费你的时间。

There are probably ways you can help and contribute and help point it toward good, but picketing with a sign on the street is a complete waste of your time.

它不会放缓。

It's not gonna slow down.

抱歉，亚历克斯，我打断你了。

Sorry, Alex, I cut you off

别担心。

No for worries.

我认为，事实上人工智能是可以被暂停的，但不应该被暂停。

I would take the position, I think AI in fact can be paused, but it shouldn't be.

我们确实知道如何暂停它。

We do know ways to pause it.

各种人已经在科幻作品和现实预测中，有时甚至是规范性建议中，描述过如何暂停——比如拉里和吉哈德的科幻情节，或者阿西洛马指南，在重组DNA的情况下也是如此。

Various folks have described both in science fiction and in realistic prognostications, in some cases normative recommendations, but Larry and Jihad from sci fi or or the Asylumar guidelines, maybe in the case of recombinant DNA.

但那些是指导性的，而不是暂停？

But those were guiding, not pausing?

嗯，阿西洛马会议确实如此，除非你听说过其他故事，它在前二十年里有效地遏制了将重组DNA引入人类生殖细胞，除非你有

Well, Asilomar was, unless you know of some other story, pretty effective for the first two decades in discouragingpausing recombinant DNA entering the human germline, unless you have a

不，这仅仅涉及五个P1到P5实验室在安全和自我监管方面的措施。

counter No, it just the five P1 through P5 labs in terms of safety and self regulation.

是的，我们确实暂停过人类生殖系编辑。

Yeah, we did pause human germline editing for sure.

是的，如果我们有意愿，是有能力暂停一项技术的。

Yes, so we are capable of pausing a technology if there's a desire to.

我只是认为，在人工智能的情况下，我们既没有意愿，也不应该有意愿去暂停它。

I just think in the case of AI, there isn't and arguably shouldn't be a desire to pause it.

我认为每天有十五万人在地球上死亡，而我认为人工智能可能是阻止这一状况的最佳机会。

I I think we have a hundred and fifty thousand people per day dying on Earth, and I think AI is probably the best chance we have at stopping that.

我认为尼克·博斯特罗姆——他在过去二十四到四十八小时内发表了一篇精彩的文章，我推荐每个人都去读一读，题为《超级智能的最佳时机》——他认为人工智能可以而且应该被暂停，但仅限于我们即将实现超级智能的时候。

I I think Nick Bostrom, who also in the past twenty four or forty eight hours, put out a wonderful essay that I recommend everyone read called Optimal Timing for Superintelligence, argues that AI can and should be paused, but only once we're on the verge of superintelligence.

他是这么定义的，不是我这么定义的。

How how he defines it, not how I define it.

我认为它已经出现了。

I think it's already here.

这有点像保持航位。

That's sort of like station keeping.

我想他的比喻是，你希望尽快驶入港口。

You you wanna get into the harbor, I think is his analogy, as quickly as possible.

但当你即将靠近码头时——我有点把他的比喻搞混了——你应该稍微减速。

But when once you're about to approach approach the dock, I'm I'm mangling his metaphor a bit, you slow down a bit.

在即将靠岸时，你应该暂停一下。

You pause as you're about to dock.

我认为这种理念比某种泰格马克式的六个月暂停要合理得多。

I think that sort of concept I think makes much more sense than some sort of Tegmarky and six month pause.

对不起。

I'm sorry.

我想补充几点。

I'd like to throw in a couple of things here.

我觉得这又是另一件荒唐的事。

I I find this is another ho thing.

是的

Yeah.

埃里克非常善于表达，但我们几个月来，至少一年了，一直在说这一点。

Eric is fabulously articulate, but we've been saying this for months on this, a a year at least now.

我们一直在讨论这个事实。

We've been talking about the fact.

而且存在许多维度，AI根本无法被暂停。

And there's a bunch of dimensions that AI cannot be paused at all.

其一是，一旦你下载了模型，人们就会用它来做各种事情。

One is once you have a downloaded model, people are gonna do stuff with it.

我们这里正上演着一个全球性的囚徒困境。

We have a global prisoner's dilemma model going on here.

无论我们做什么，整个事情现在都将迅速扩大。

The the whole thing is gonna scale now no matter what we do.

你知道，OpenAI做了两件事，某种程度上打开了这个潘多拉魔盒。

You know, OpenAI did two things that that that kind of unlocked this Pandora's box.

第一，它编写了代码；第二，它被发布在了开放的互联网上。

One, it wrote code, and the second, it was released on the open Internet.

一旦你做了这两件事，就一切都晚了。

Once you do those things, you're done.

潘多拉的盒子已经打开了。

Pandora's box is gone.

我们谈论的是马已经跑出去之后的谷仓门。

We're talking about the barn door after the horses bolted.

我们用十八世纪的隐喻来试图理解正在发生的事情，你怎么看？

We're using an eighteenth century metaphor to try and understand what's going I think what do you think about this?

他提到的一个观点是，每个人在经济上都有动力推动它继续发展并加速竞争。

One of the points he made was everybody's economically incentivized to keep it going and race it along.

美国、中国，你知道，所有的激励机制都已经到位，使得停止它变得极不可能，几乎不可能。

And The US, China, you know, it's like all the incentives are in place that make it highly improbable and almost impossible.

本，你怎么看？

Ben, what are your thoughts

在这里？

here?

不可能。

Impossible.

不可能。

Impossible.

所以我认为，从地缘政治的角度来看这个问题，显然，我不认为我们有任何手段能实现全球性的控制，尤其是当这项技术已经普及到人们的笔记本电脑上时，我们根本无法阻止它。

So I think, you know, look at this question through the geopolitical lens, is, you know, clearly, I I I don't think there's any kind of leverage where we would get some global I mean, as especially when it's on, like, people's laptops as well where we would actually stop the technology.

是的，我同意这不切实际。

Like, I I do agree that it's impractical.

我认为这确实存在真正的危险，我们在拜登政府时期可能更明显地面临过这种风险，但如今它仍然是一种潜在的动向——如果我们大幅减缓美国的AI进展，就会导致他提到的自由威胁彻底失控，因为谁掌握了AI技术，谁就对社会的运作拥有巨大控制权。

I think that there is a real danger, and we faced it probably more in the Biden administration than in this one, but it's still like a potential movement that we really slowed down AI progress in The US to the point where the other thing that he mentioned, the threat to freedom, you know, becomes completely out of our control because we're just whoever is building the AI has a lot of control about how society is gonna work.

因此，我认为试图暂停AI发展，即使实际上无法完全暂停，但只要在美国将其放缓到足够程度，导致我们与中国拉开巨大差距，这就会成为一个严重问题。

And so I do think there's real danger along these lines of attempting to pause it and maybe not actually pausing it, but slowing it down enough in The US that, we just become far enough behind China that it's a real problem.

或者，任何由习近平认为我们应当走向的社会形态。

Or, like, where whatever society, you know, the Xi Jinping thinks we should be.

是的

Yeah.

我完全同意。

I completely agree.

我认为，如果你仔细听埃里克的话，他并不是说我们不能暂停它。

I think, you know, if you if you listen to Eric's words closely, he wasn't saying we couldn't pause it.

他说，因为我们正与中国展开全面的军备竞赛，而我们才刚刚进入总统任期的第一年，接下来三年内这件事一定会发生。

He's saying that because we're in the middle of a all out arms race with China, and we have we're only one year into a presidential administration, the it's gonna happen in this next three years.

所以，鉴于当前的政府和与中国的关系，根本不可能暂停。

So given the current administration and the current situation with China, there is no chance of it being paused.

因此，要应对现实。

And so react to the real reality.

不要假设那些根本不会发生的事情。

Don't hypothesize something that is just never gonna happen.

但这一切都是地缘政治的

But it was all That's geopolitical,

像是你对动机的某种理解。

like something you're to the motivation.

我想说的是，根本没有任何机制可以暂停或停止它。

I'm speaking speaking of the fact that there's no mechanism to pause it or stop it at all.

完全没有。

None.

必须监管每一行代码。

Have to regulate every line of code.

我的天啊，算了吧。

Mean, come on.

哦，萨利姆，还是有办法的。

Oh Salim, there are ways to do it.

想不到办法，恰恰说明了你的品格。

Think not being able to think of it speaks well of of your character.

你必须想象一个完全病态的社会。

You have to imagine a completely pathological society.

韦纳·文奇对此写了不少内容。

Werner Vinci wrote quite a bit about this.

我们拥有这些过剩的晶体管预算。

We have these excess transistor budgets.

想象一个社会，在这个社会中，晶体管 literally 监视着同一片系统级芯片上的其他晶体管。

Imagine a society where you have literally transistors spying on other transistors on a single SOC.

想象一下，人们在互相监视。

Imagine you have people spying on other people.

想象一下，设有悬赏。

Imagine there are bounties.

任何人只要发现他人在逻辑层面或社会层面做了算法上不允许的事情，就可以举报。

Anyone discovers anyone else doing something that's algorithmically impermissible, either at the logical level or the social level.

你可以构建一个足够病态的社会，在这个社会中，人们被煽动互相对抗，以压制人工智能。

You can construct a sufficiently pathological society where people are turned against each other in order to suppress AI.

至少我能想象出这样的情况。

At least I can imagine it.

我可以给你一个真实的现实案例。

I I can give you an actual real world example.

让我给你一个真实的

Let me give you a real

例子。

world example.

拜登政府最近发布的行政命令规定，未经美国政府批准，不得出售任何GPU，一块都不行。

The last executive order from the Biden administration was that you could not sell a GPU without US government approval, not a single GPU.

所以，我觉得这并不会阻止AI的发展。

So, like, I think that would it wouldn't stop AI.

但它足以在美国境内让AI的发展大幅放缓。

It would slow it down enough in The US that it would be extremely material.

说到这个话题，亚历克斯，我们谈话后我放了这张幻灯片。

Speaking of this subject, Alex, I put this slide up after our conversation.

你想

Do you wanna

本，为了满足我的一点私心，我请你展示这张幻灯片，所以我得问你这个问题。

tee it this slide, Ben, as an indulgence to me, so I have to ask you this question.

你和马克在上届政府末期，曾公开谈论过你们在白宫会面的事情。

You and Mark, towards the end of the the last administration, were were very public making comments that you took a meeting at the White House, apropos.

如果我转述得准确的话，你们对听说AI进展被像数学和基础物理那样分类甚至过度分类感到震惊——当然，如果我理解有误，请纠正我。

And, if if I'm relaying the the comments accurately, you were dismayed to hear about plans to classify AI progress just like, and again correct me if I'm mischaracterizing, advances in math and fundamental physics had been purportedly classified or over classified for decades.

我对几个层面感到好奇。

I'm curious at a few levels.

第一，如果我的理解准确，你们当时听说了哪些内容被分类了？

One, if that's accurately characterizing what you heard, what do you think was classified?

你们认为，对数学和基础物理的这种分类或过度分类，对经济和世界造成了什么影响？

What do you think was the impact on the economy and the world from such classification or over classification of math and fundamental physics?

如果由你来主导，你会采取什么不同的做法？

And what would you have done differently than if you had been in charge?

是的。

Yeah.

所以我可以告诉你当时说了什么。

So I I can tell you what was said.

我说，你看。

I said, look.

你知道，我当时试图保持务实。

You know, I was trying to be pragmatic.

我说，你知道，从根本上说，人工智能就是数学。

I said, you know, at the core, AI is math.

这就是它在做的事情。

That is what it's doing.

它是数学。

It's math.

所以如果你开始限制模型并监管模型，你实际上就是在监管数学。

So if you start restricting the models and you start regulating the models, you're just regulating math.

你是在以某种方式将数学排除在外。

You're outlying math in some way.

你要么是在限制数学的某些部分，要么就是在说你无法进行足够的数学运算。

Either you're outlying parts of math or you're saying you can't do enough math.

他回答说，是的。

And he goes, yes.

我们可以这么做。

We can do that.

那就是他的回答。

Like, that was his answer.

他说道，是的。

He he goes, yes.

我们可以这么做。

We can do that.

我们在四十年代对核物理就是这样做的，其中一些内容至今仍属于机密。

We did that in the forties around nuclear physics, and some of that stuff is still classified today.

而且，我感到非常震惊。

And one, I was shocked.

我的下巴都惊掉了。

Like, my jaw hit the floor.

我当时就想，哇。

I was like, wow.

这太疯狂了。

That's crazy.

而接下来这事儿会更疯狂。

And then this would be even crazier.

但你知道，我不确定那到底是什么，不过如果你回想起——我就简单说一句。

But, you know, I I don't know what it was, but, like, if you think back about I mean, I'll just make this comment.

如果你看看美国和全世界在物理学上的进展，一直到爱因斯坦、冯·诺伊曼那个时代。

If you look at the progress in The US and in the world in physics up until kind of the, you know, Einstein, John von Neumann era.

而从那以后，我们取得的进步少得惊人。

And then since then, like, it's pretty startling how little progress we've made.

我想说的是，那些后来出现的许多想法似乎都没奏效。

I would just say, you know, many of the ideas that have come since then don't seem to work.

而且，你知道，希望我们能通过人工智能找到突破的那一天。

And, you know, hopefully, we'll we'll get to the other side of that with AI figuring things out.

但我确实好奇，我们是不是曾经掌握过某些知识，本可以解开我们现在试图解决的难题？

But I do wonder, like, you know, did we put something away that we knew that would have unlocked some of the problems we're trying to solve now?

这太有趣了。

That's fascinating.

好的。

Okay.

我要说明一下，这正是我原本以为你所表达的意思，或者我听到、推断出的内容。

That that is for the record, that is what I assumed you meant and or heard or inferred.

所以，如果可以的话，关于问题的第二部分：如果你确信基础物理学确实曾发生过这种分类，那你接下来会怎么做？

So maybe just if if I may, the second part of the question, what would you do going forward now if if you knew for a fact that such classification of fundamental physics had in fact happened?

比如，你会如何改变这个世界？

Like, how would you fix the world?

嗯，我的意思是，是的。

Well, I mean, yeah.

请简单快速地说一下。

In one in one quick sense, please.

从一个角度来说，我真的不知道他们做了什么，但我觉得，首先，这件事确实奏效了。

I I really In one sound I really don't know what they did, but, like, I I I just think that stopping first of all, like, one thing, it did work.

对吧？

Right?

俄罗斯人确实得到了原子弹，包括最机密的引爆机制。

The Russians did get, like, the bomb, including the exact the exact trigger mechanism, which was the most proprietary thing.

他们得到了几乎一模一样的零件，逐个复制。

They got, like, exact, you know, like, part for part.

尽管有这么多保密措施，他们还是完全复制了我们的技术。

The whole thing they were able to get from us despite all this classification and and whatnot.

所以它并没有带来任何积极效果。

So it didn't do anything positive.

而且，你知道，限制知识，我觉得这几乎是一个非常危险的想法。

And, you you know, restricting knowledge, I I just think that's a This is almost very dangerous idea

一般来说。

in general.

本，这几乎就像埃隆对知识产权的看法。

Ben, this is almost like Elon's point of view on intellectual property.

就是这样。

It's like Yeah.

如果你依赖知识产权来保护自己，不如干脆更快地创新。

If you're depending on IP to keep you safe, you know, it's better just keep innovating faster.

是的，我

Yes, I

有同感

feel that

方式。

way.

这里的数字有点令人震惊。

The numbers here are kind of shocking.

当今经济中的巨额资金正流向资本，而非劳动力。

Big money in today's economy is going to capital, not labor.

因此，自2019年以来，平均工资仅增长了3%，而利润却飙升了43%。

So since 2019, the average wages have grown 3%, but profits have soared 43%.

这里有一个很好的对比。

And here's a good comparison.

英伟达体现了这种转变，其市值是1980年代IBM的20倍，利润是其5倍，但员工人数只有其十分之一。

Nvidia symbolizes that shift, 20X more valuable and 5X more profitable than IBM in the 1980s with onetenth the staff.

所以，我的意思是，这正是我们之前和埃隆讨论过的关于全民基本收入的话题——资本正在带来非凡的回报，并有可能在未来五年内实现三位数的GDP增长。

So, I mean, this is what we were talking about with Elon heading towards universal high income where capital is just providing extraordinary returns and the potential for you know, a triple digit GDP growth in the next five years.

本，你有看到那些关于GDP增长的预测吗？

Have you seen those predictions on GDP growth, Ben?

你对这些预测怎么看？

What do you think of them?

我的意思是，这确实感觉非常有可能。

I mean, it does feel very possible.

所以我就说，你看，我们还处在人工智能的非常早期阶段。

So I'll just say that, like, if you look at, you know, we're so early in AI.

我记得Anthropics说他们的收入达到了140亿美元？

And I think what did Anthropics say they were at 14,000,000,000 in in revenue?

你想想，他们才进入市场多久？

You and you go, well, well, how early into the market are they?

连百分之一都不到。

And it's like, not one percent.

我不觉得，如果你真的想想这些产品能做什么，以及它们所创造的价值的话。

I don't think, you know, if you if you really think about what all these products can do and the value that they have.

所以对我来说，这个GDP预测并不离谱。

And so that doesn't seem outrageous to me as a as a GDP prediction.

至于什么时候真正爆发，技术成熟和普及之间总是有差距的，就像卡洛塔·佩雷斯之前的分析，我认为它经受住了时间的考验。

Now when it kicks in and there's always a difference between when the technology is ready and how fast it's adopted, the old Carlotta Perez analysis, which I think has, you know, held up super well over time.

但另一点是，有了AI，我们已经拥有互联网了。

But the other thing is with AI, we already have the internet.

所以技术采用的速度会快得多，我认为会比互联网快得多，因为当初我们得建设所有基础设施，铺设所有光纤，把宽带接入每家每户。

So the technology adoption is much, much I think it's going be much, much faster than, say, the internet, where we had to build out all of the infrastructure, all the fiber, all the various things you had to broadband to the house.

当时为了推广互联网，我们得做那么多事情，而这次AI将直接借助这些已有基础，迅速广泛传播。

Like, there were so many things we had to do to get the Internet adopted, and this is gonna just piggyback off that and be distributed very, very fast.

所以别觉得这些GDP数字太离谱。

So don't think those GDP numbers are outrageous.

各位还有什么其他看法吗？

Other comments, gents.

财富集中效应是这页幻灯片的另一面。

Well, concentration of wealth effect is the other side of this slide.

这才刚刚开始，但我之前跟一些人说过，如果这个趋势持续下去，十年内旧金山可能会成为整个太阳系的首都。

It's only just beginning, but it's gonna, I was telling some people earlier that if the trend continues, San Francisco will end up being the capital of the entire solar system in just about ten years.

我不知道，人们现在在想什么

I don't know, people are What are

在看什么？

looking at?

我的意思是，本在拉斯维加斯，埃隆在得克萨斯。

I mean, like Ben's in Las Vegas, and Elon's in Texas.

好吧，如果没有人们因税收问题逃离，旧金山本会成为太阳系的首都。

Okay, well absent people fleeing the tax situation, it would have become the capital of the solar system.

但有了人工智能驱动的劳动力，你能用少得多的人完成多得多的工作。

But with an AI effective workforce, you know, that you're getting so much more done with so many fewer people.

实际上，另一件让我震惊的事情是这个事件链条：如果你从左边的OpenAI开始，他们与中间的Merkur合作，而Merkur在印度有数万名员工在为Merkur工作，而这些工作实际上是为了OpenAI。

And actually, the other thing that really is startling to me is this chain of events between if you take OpenAI on the left and they work with Merkur in the middle, and then Merkur has tens of thousands of people in India doing work that benefits Merkur that is actually for OpenAI.

在整个价值链条中，回流到母公司的价值比例只是其中的一小部分。

The fraction of all value created that flows back to the mothership is just a massive fraction of that value chain.

所以，如果你将这种趋势扩展到未来三年，覆盖整个经济的各个领域，价值将高度集中地流向少数几家公司和人群。

So if you extrapolate that out across the next three years, across all these sections of the economy, the funneling of value goes to a very concentrated group of companies and people.

这正在发生。

It's just happening.

但你能在数据中看到它，它正在发生。

But you can see it, like in the numbers, it's happening.

这就是埃隆在一次采访中所说的，将会出现前所未有的巨大繁荣，财富量级空前，但与此同时社会动荡也会加剧。

And this is where Elon was saying in an interview, it's like, it's gonna be a massive amount of total prosperity, huge amounts, unprecedented, crazy amounts of prosperity with massive social unrest concurrent.

我们正朝着这个方向发展。

That's where we're heading.

是的。

Yeah.

这很有趣。

It's interesting.

你知道，雷·库兹韦尔早期的一个预测——这里是奇点世界——就是每个人都会成为创业者。

You know, one of Ray Kurzweil's early predictions, this is Singularity World here, was that, like, everybody would become an entrepreneur.

每个人都会成为一个独立公司，至少在极限情况下是这样。

Like, everybody was gonna be a company of one, you know, at the limit.

我认为我们已经看到了很多这样的趋势，只不过这些趋势并没有被就业数据等很好地反映出来。

I think that I I think there's some we're already seeing a lot of that, which is not very well captured, by the way, by the employment numbers and so forth.

我认为人工智能真正地、真正地、真正地推动了这一点。

And I think AI really, really, really enables that.

但我认为，那些具备创业者精神的人和那些不具备的人之间，将会出现巨大的差距。

But there's gonna be a big disparity, I think, between people who have that kind of initiative to be an entrepreneur and those who don't.

这种差异将会非常显著。

It's gonna be pretty dramatic.

本，我对这个问题的描述是，我们将把世界划分为消费者和创造者。

Ben, the way the way I characterize it is we're gonna split the world into consumers and creators.

也就是沙发土豆和《星际迷航》里的员工，这么说吧。

The couch potatoes and the Star Trek employees, if you would.

我觉得这非常重要。

And I think it's super important.

谈到创造者和创业世界，我们在这个播客里已经多次提到，未来的职业就是成为创业者。

Speaking about creators and the entrepreneurial world, and I think we've said on this pod so many times that the career of the future is being the entrepreneur.

这是一条很有趣的推文，我把它截下来了，因为我觉得它精准地反映了当前的精神。

This is an interesting tweet that went out and I captured it because I think this hits the ethos right now.

科技公司正在拥抱996，每周工作72小时。

Tech firms are embracing nine ninety six, seventy two hour work weeks.

这是一则包含警告的招聘信息中的引述。

This is a quote from a job ad that contains a warning.

如果你对每周在纽约市与一些最具雄心的人一起工作70小时感到不兴奋，请不要加入。

Please don't join if you're not excited about working seventy hours per week in person with some of the most ambitious people in New York City.

我的反应是：每周才72小时？

My reaction was only seventy two hours per week.

我的意思是，其他时间你都在干嘛？

I mean, what are you doing with the other hours?

彼得，我和你一样。

That was the same as me, Peter.

我的意思是，老实说，这种变化的速度——

I mean, honestly, speed, I mean, the speed at which it's happening is

我有个简单的答案。

I have an easy answer I have an easy answer to this.

说吧，斯莱。

Go ahead, Sly.

如果你没有个人的使命目标，并且不是发自内心地热衷于与像SpaceX、特斯拉这样的人一起工作，哪怕是人形机器人，哪怕只有两只手臂，那都无所谓。

If if you don't have a personal MTP and you're not driven personally about a deep passion for working with somebody that's the lion, say it's SpaceX, say it's Tesla, whatever it is, humanoid robots, even with two arms, it doesn't matter.

如果你没有这种热情，就不应该和他们一起工作。

If you're not that passionate, you shouldn't be working with them.

如果你真的有这种热情，那么每周工作七十小时也会很有趣。

If you are that passionate, then seventy hours a week is fun.

所以我看不出这里有什么区别。

So I don't see the distinction here.

是的，当然。

Yes, for sure.

对，我完全同意这一点。

Yeah, completely agree with that.

当你真正和那些每周工作七十小时以上的初创公司员工交谈时，他们会非常有活力。

When you actually talk to the people in these startups working seventy plus hours a week, they're super energized.

他们乐在其中。

They're loving it.

超级热爱这件事。

Super loving it.

而且他们通常都很年轻。

And they're usually young.

他们没有太多其他责任。

They don't have a lot of other obligations.

在那个年纪，他们还没开始教练足球队。

They're not coaching the soccer team yet at that age.

所以对他们来说，这并不难做到。

So it's just not hard for them to do.

但另一方面，当你深入解决这些技术问题时，你总是会想着它。

But the other side of it is when you're deep into one of these tech problems, you're thinking about it all the time anyway.

因为上下文切换会严重拖慢进度，但如果你全身心投入工作，它就会一直萦绕在你脑海中——早上洗澡时、做任何事时，它都占据着你的思绪，真的非常耗尽心神。

Because the context switching is such a slowdown, but if you're just fixated on the work, it's in your mind in the shower in the morning, it's in your mind, whatever you're doing, it's really pretty all consuming.

我认为如果你在人生中的某段时期，比如几年里这样做，是很好的。

And I think it's great if you do it for a period of your life, a few years.