本集简介
双语字幕
仅展示文本字幕，不包含中文音频；想边听边看，请使用 Bayt 播客 App。
我真心认为人工智能革命的热度被低估了。
I honestly believe that the AI revolution is underhyped.
那么，为什么这一切如此重要？
Now, why is this all important?
埃里克·施密特来了。
Eric Schmidt is here.
他是谷歌前执行董事长兼首席执行官。
He's the former Google executive chairman and CEO.
这些智能体将会非常强大，并且开始协同工作。
These agents are gonna be really powerful, and they'll start to work together.
很快我们将能让计算机自主运行并决定它们想做什么。
We're soon gonna be able to have computers running on their own deciding what they want to do.
现在我们迎来了新的非人类智能，其推理能力很可能超越人类。
Now we have the arrival of a new non human intelligence, which is likely to have better reasoning skills than humans can have.
如果你能当一小时世界皇帝
So if you were emperor of the world for one hour
我最重要的职责就是确保西方世界获胜。
The most important thing I do is make sure that the West wins.
女士们先生们，请欢迎埃里克·施密特。
Ladies and gentlemen, please welcome Eric Schmidt.
嗨。
Hi.
是的。
Yeah.
嗨。
Hi.
看起来不错。
Looking good.
致
To
很高兴见到你。
Good see to see you.
你看起来真精神，Eric。
You're looking schvelt, Eric.
你看起来真精神。
You're looking schvelt.
谢谢夸奖。
Thank you nice.
天啊，David。
Oh my god, David.
很高兴见到你。
Good to see you.
就像是
It's like
David Zach也在这里。
David Zach is here as well.
这就像是我们所有前公司的一次重聚，David。
It's like a it's like a reunion of all of our former companies, David.
你究竟为什么辞职？
Why did you quit after all?
是的。
Yes.
我以前的老板。
My old my old boss.
就是
It was
和年轻的弗里德伯格共事是怎样的体验？
What was it like working with young Friedberg?
带我们回到过去。
Take us back.
嗯，我能讲个故事吗？
Well, can I can I tell a story?
我们得去奥兰治县，他们说，嘿，我们要坐飞机去。
We have to come down to Orange County and they're like, hey, we're gonna take the plane.
那是埃里克的飞机。
And it was Eric's plane.
我们上了飞机后，他上去驾驶飞机。
We get on the plane and then he goes up and flies the plane.
我一个人坐在飞机后排。
I'm in the back of the plane by myself.
是比奇空中国王吗？
Was it King Air?
他做了什么
What did
我当时就像谷歌CEO一样，被专机送到橙县。
he I'm like the CEO of Google flying me down to Orange County.
那感觉太不可思议了。
It was incredible.
那是我第一次真正和埃里克一起出去玩。
That was my first time actually hanging out with Eric.
那是我的湾流飞机。
It it was my Gulfstream.
没错。
That's right.
他聪明过头了。
He was way too smart.
聪明过头了。
Way too smart.
聪明过头了。
Way too smart.
他专注吗？
Was he focused?
他有贡献吗？
Did he contribute?
他有推动进展吗？
Did he move the needle?
但他非常聪明。
But he was very smart.
好吧。
Well.
看。
Look.
看。
Look.
你们都很了解这家伙。
You guys know this guy well.
他确实那么聪明。
He's really that smart.
所以他教给我的东西比谷歌大多数员工都多。
So he taught me more stuff than most of any of the employees at Google.
然后你就离开了。
And then you left.
好吧，说说
Well, tell
你最近在忙什么。
us what you've been doing.
所以不。
So No.
不。
No.
等等。
Wait.
在那之前，我得先问你这个问题。
Before that, I gotta ask you this question.
斯坦福大学最近删除了一个视频。
There was a recently deleted video from Stanford.
哦，不。
Oh, no.
你曾有过一个清醒时刻，说过在谷歌，人们太注重工作与生活的平衡了。
You had a moment of clarity where you said, hey, you know, like, at Google, people are like too much work life balance.
他们需要更投入。
They need to commit.
他们需要更努力工作。
They need to work harder.
上次活动我们请了谢尔盖。
We had Sergei at the last event.
他要回去工作了。
He's going back to work.
所以谢尔盖收到了这个信息。
So Sergei got the message.
预测谢尔盖的行为是我可能会失败的事。
Predicting Sergei's behavior is something I can fail at.
我尝试了二十年。
I tried for twenty years.
我本质上并不赞成在家工作。
I am not in favor of essentially working at home.
我的意思是，你们很多人都在某种程度上在家工作，但你们的职业生涯已经稳定了。
I And the reason I mean, many of you guys all work at home at to some degree, but your careers are already established.
但想想一个20岁的年轻人，他们需要了解世界如何运转，比如从伯克利或达特茅斯毕业，受过良好教育。
But think about a 20 who has to learn how the world works and you know, they they come out of Berkeley or Dartmouth and they're very well educated.
当我回想在Sun公司时，光是听那些比我年长五到十岁的前辈们面对面争论，就学到了很多。
When I think about how much I learned when I was at Sun just listening to these elder people who were five or ten years older than I was argue with each other in person.
我不知道在这种新模式下要如何复现那种学习场景？
I don't how do you recreate that in this new thing?
而且我支持工作与生活的平衡——这也是人们选择为政府工作的原因。
And and I'm in favor of work life balance and that's why people work for the government.
抱歉。
Sorry.
跑题了。
Strays.
抱歉。
Sorry.
抱歉。
Sorry.
抱歉。
Sorry.
如果你要从事科技行业并想取得成功，就必须做出一些取舍。
If you're going to be in tech and you're going to win, you're gonna have to make some trade offs.
你们还记得，我们的对手是中国人。
And you remember, we're up against the Chinese.
中国人的工作生活平衡是996，即每周六天从早9点到晚9点。
The Chinese work life balance consists of nine nine six which is 9AM to 9PM six days a week.
顺便说一句，中国官方已明确表示这是违法的。
By the way, the Chinese have clarified that this is illegal.
然而大家都在这么做。
However, they all do it.
这就是你们要面对的竞争对手。
That's who you're competing against.
我把所有人都叫回办公室了。
I brought I brought everybody back to office.
这样好多了。
It's so much better.
那我们就接着这个话题继续。
So it Let's just pick up on that theme.
所以
So
你不需要为政府辩护。
You don't need to defend the government.
不。
No.
不。
No.
相信我，我觉得没必要这么做。
Believe me, I don't I don't see the need to.
我是政府无偿的兼职顾问。
I'm an unpaid part time advisor to the government.
但我们确实在与中国进行高科技竞争。
So But we are in this high-tech competition with China.
他们显然也很重视人工智能。
They obviously care about AI too.
他们正试图加速领先。
They're trying to race ahead.
我了解到你最近刚去过那里？
How do you I I understand that you recently made made a trip there.
你如何看待这场竞争？
How do you handicap this this this competition?
我们刚讨论过这个，是的。
You and I just talked about this Yeah.
作为你在白宫极其重要工作的一部分。
As part of your as your incredibly important work in the White House.
我曾以为中美在AI领域处于对等竞争，你们限制芯片的措施确实拖慢了他们的步伐。
I had thought that China and The United States were competing at the peer level in AI and that the good work that you have done and your predecessors did to restrict chips were slowing them down.
他们的做法和我预想的很不一样。
They're really doing something more different than I thought.
他们没追求疯狂的人工通用智能战略，部分是因为你们实施的硬件限制，部分因为他们的资本市场深度不足。
They're not pursuing crazy AGI strategies partly because the hardware limitations that you've put in place, but partly because the depth of their capital markets don't exist.
他们不能仅凭一厢情愿就筹集1亿美元或同等资源来建设数据中心。
They can't raise based on a wing and a prayer a 100,000,000 or a member equivalent to to build the data centers.
他们就是做不到。
They just can't do it.
因此结果是他们非常专注于将AI技术应用到各个领域。
And so the result is they're very focused on taking AI and applying it to everything.
所以我担心的是，在我们追求通用人工智能的同时——这虽然极其有趣且值得讨论，但我们所有人都将受其影响。
And so the concern I have is that while we're pursuing AGI, is incredibly interesting and we should talk about it, all of us will be affected by this.
我们最好也在日常科技领域与中国展开竞争。
We better also be competing with the Chinese in day to day stuff.
没错。
Right.
消费级应用。
Consumer apps.
这是你非常了解的领域，查马斯。
This is something you understand very well, Chamath.
消费应用、机器人技术等等。
Consumer apps, robots, so forth and so on.
我考察过上海所有的机器人公司。
I saw all the the Shanghai robotics companies.
这些企业正试图在机器人领域复制他们在电动车领域的成功。
And these guys are attempting to do in robots what they've successfully done with electric vehicles.
对吧？
Right?
他们的职业道德令人难以置信。
And they're they're work ethics incredible.
他们资金充足。
They're well funded.
不像我们在美国看到的那些疯狂估值。
It's not the crazy valuations that we have in America.
没错。
Right.
他们虽然无法筹集资金，但能在这方面取胜。
They can't raise the capital, but they can win across that.
中国人正在做的另一件事——我想强调这是个重大地缘政治问题——我的背景观众都知道，开源意味着开放代码。
The other thing the Chinese are doing, and I wanna emphasize this as a major geopolitical issue, is that my own background audience, you all know open source means open code.
开放权重意味着开放训练数据。
Open weight open weights means open training data.
中国正通过开放权重和训练数据竞争，而美国主要聚焦于封闭权重和封闭数据。
China is competing with open weights and open training data and The US is largely and majority focused on closed weights, closed data.
这意味着世界上大多数国家——可以类比一带一路倡议——将会使用中国模型而非美国模型。
That means that the majority of the world, think of it as the Belt and Road Initiative, are gonna use Chinese models and not American models.
我个人认为西方民主国家是正确的，我更希望大型语言模型的扩散和学习基于西方价值观。
Now I happen to think the West and democracies are correct, and I'd much rather have the proliferation of large language models and learning be done based on Western values.
埃里克，我们曾与Meta有过重大开源合作，他们资产负债表惊人、技术实力雄厚，但似乎执行失误，现在正退后一步，按照你提到的思路重新制定更偏向闭源的方案。
Eric, we had a major open source initiative with Meta, you know, incredible balance sheet, tremendous technical firepower, but they seem to have mis executed and now are taking a step back and reformulating something to your point that looks a little bit more closed source.
目前还不明确。
It's not clear.
你知道，Alex Wang是个好朋友。
You know, Alex Wang is a good friend.
他进来了。
He's come in.
他接手了。
He's taking over.
他显然能力超群，非常出色。
He's obviously incredibly incredibly capable.
我不会说他们完全封闭了。
I would not hold keep I I would not say that they're going fully closed.
而且我认为他们也搞砸了，因为DeepSeek团队做得太出色了。
And I think also they got screwed up because the DeepSeek people, r one, did such a good job.
对吧？
Right?
如果你看看DeepSeek的推理模型，尤其是他们能反复进行强化学习的能力，这真是个重大成就。
If you look at the reasoning model in DeepSeek, and in particular, their ability to do reinforcement learning forward and back, forward and back, and forward and back, this is a major achievement.
而且看起来他们在数值精度上比美国模型用的更低。
And it appears that they're doing it with less precision than numeric precision than the American models.
从技术角度讲，有种东西叫fp64、fp32、fp16。
As a bit of technical things, there's something called f p 64, f p 32, f p 16.
美国模型通常使用16位精度进行训练。
The American models are typically using 16 bit precision for their training.
中国团队正在推进8位甚至4位精度。
The Chinese are pushing eight and now even four.
有
Is there
美国那些大公司是否需要在开源领域采取更多行动，这样我们才能真正应对这个问题？
is there something that the American, you know, bigger companies need to be doing in open source so that we can actually combat this?
嗯，许多大公司都表示他们也想成为开源领域的领导者。
Well, a number of the large companies have said that they wanna be leaders in open source as well.
山姆·奥特曼曾表示o3模型的最小版本将会发布，我相信是开放权重的，他们也确实这么做了。
Sam Altman indicated that the smallest version of the o three model would be released, I believe, open weights and they have done so.
总之他告诉我这个模型规模远小于10的26次方。
And he told me anyway that this model is much smaller than 10 to the 26.
它训练起来容易得多，而且能适配甚至可以在手机上运行。
It's much easier to train and it will fit or or can fit on your phone.
所以一种路径是让超级计算机来负责AGI，但这永远会极其昂贵等等。
So one path is to say that we'll have these supercomputers doing AGI which will always be incredibly expensive and so forth.
但我们还必须确保这些手机端模型的扩散处于美国控制之下，无论是OpenAI、Meta、Gemini还是其他公司。
But we also have to watch to make sure that the proliferation of these models for handheld devices is under American control, whether it's OpenAI or Meta or Gemini or what have you.
最近你接手了相对论太空公司。
Recently, you took over Relativity Space.
我想对于不了解的人来说，这家企业的雄心实际上是要与SpaceX竞争。
And I think for for the people that don't know, this is a business that effectively whose ambition is to compete with SpaceX.
我想你是第一个投资者或最早期的投资人
I think you were the first investor or the earliest investor in
是的
I was.
对不起。
I'm sorry.
没关系。
It's okay.
我损失了一些钱在
I lost some money in the
第一次发生？
first happened?
他被压榨了吗？
Did he get crammed down?
没有。
No.
没有。
No.
没有。
No.
我是说，
I mean,
我们所有人都损失了。
the All of us did.
我们所有人。
All of us.
我是说，看。
Mean, look.
是的。
Yeah.
我是说，我一直非常高兴能成为SpaceX、Swarm和Starlink的投资者。
I mean, I've been very happily, you know, an investor in SpaceX and Swarm and Starlink.
相对论
Relativity
顺便说一句，Swarm可是个大项目。
was By the way, Swarm is a big deal.
创建
Created
是一个
was a
所以谢谢你。
So thank you.
对。
Yeah.
Swarm对他们来说是个巨大成功，我认为对世界也是。
Swarm has been a really great success for for them and I think for the world.
但我想问的是，带我们回顾下航天市场的演变过程。
But what I was gonna ask you is, walk us through the evolution of the space market.
在众多公司中，你拥有可以任意投资的资本基础，为什么选择这个领域？
Why you decided of all the companies, you have the capital base to kind of put your money anywhere.
为什么选这个？
Why did you pick that?
你为什么选择那个行业？
Why did you pick that business?
为什么是现在？
Why now?
火箭非常酷也非常难造。
Rockets are really cool and they're really hard.
正如你所知，我是一名飞行员，对喷气机很了解，我曾以为火箭技术已经像喷气发动机一样成熟。
I had, I'm a as you know, I'm a pilot and I know lots about jets and I had assumed that rockets were as mature as jet engines.
但事实并非如此。
They're not.
这既是艺术也是科学。
It is an art and a science.
这些事情做起来非常困难。
These things are very hard to do.
所需的动力极其巨大——以我们的火箭为例，推力达到400万磅。
The amounts of power, mean, our case, the rocket is 4,000,000 pounds of thrust.
测试时必须固定住它，甚至金属装置都难以承受。
You have to hold the thing down to test it and you can't even hold it with metal things.
还需要其他辅助装置来固定。
You have to have other things to hold it down as well.
产生的力量太强大了。
There's so much force.
否则它就会直接起飞。
Otherwise, it will take off.
关于火箭的另一个有趣事实是，粗略来说，火箭重量的2%是有效载荷，18%是火箭本身，80%是推进剂。
Another interesting thing about rockets is that a rough number is that 2% of the weight of the rocket is the payload, 18% is roughly the rocket, and 80% is the propellant.
作为一个新人，我的反应是：你是在告诉我你们无法做得更好吗？
And my reaction as a new person is, you're telling me you can't do any better?
物理学家们表示，经过六十年的物理学发展，这就是我们摆脱地球引力的最佳方案了。
And the physicists say, after sixty years of physics, that's the best we can do to get out of the gravitation of of the earth.
因此我认为火箭既有趣又充满挑战。
And so I think rockets are interesting and they're challenging.
这里永远存在竞争的机会。
There's always an opportunity for competition.
在相对论空间领域，本质上就是近地轨道的竞争者。
In relativity spaces area, it's essentially a LEO a LEO competitor.
比如近地轨道卫星这类事物。
So low Earth orbit satellites, that sort of thing.
订单已经排满了。
The order book is full.
我们只需要发射火箭就行。
We just have to launch the rocket.
这次太空探索的契机出现了——我不确定这件事的知名度，所以你可以尽情深入探讨。
And and this entry into space happened, and I'm not sure how well known this is, so you can go as far as you wanna go into this.
但你在下一代战争领域也做了很多工作。
But you've done a lot as well in next generation warfare as well.
您想让我们聊聊这个话题吗？包括您如何进入该领域以及它扮演的角色？
Do you want us to just to talk about that and how you ended up there and what role that plays?
给我们展示一幅风景吧。
And just give us a landscape.
也许大卫问过中国的问题，但它们之间多少都有些关联。
Maybe David asked about the China question, but it's they're all kind of almost interrelated.
首先，我是个搞软件的人，不是搞硬件的。
Well, first place, I'm a software person, not a hardware person.
我向人们解释过，硬件人员和软件人员接受的是不同体系的教育。
I I explain to people that hardware people go to different schools than software people.
他们的思维方式略有不同，所以我在这些新兴行业里总是存在局限。
And they think slightly differently, so I'm always at a limitation in these new industries.
我曾为国防部长工作，拥有最高机密权限之类的资格。
I had worked for the Secretary of Defense and have a top secret clearance in all that.
我还因为试图帮助五角大楼重组而获得过勋章等等。
I was given a medal, etcetera, for trying to help the Pentagon reorganize itself.
当乌克兰战争爆发时，我观察着局势并想到，这是个观察一个没有海军、也没有空军的国家如何运用自动化的机会。
And when the Ukraine war started, I was watching and I thought, well, here's an opportunity to see a country that has no navy, and no, and no air force, how they do this with automation.
事实上，这堪称创新领域的惊人成功——尽管在动能武器、军备和动员力等方面以三倍劣势对抗，乌克兰表现得相当顽强。
And indeed, it has been a spectacular success as a matter of innovation and outnumbered three to one with huge differences in kinetic strength, weapons, mobilization, and so forth and so on, Ukraine has held on really quite well.
现在正在发生的，本质上是一个全新军事国家安全体系的诞生。
And what's happening now is you're seeing essentially the birth of a completely new military national security structure.
可以这样理解：我们都...首先，我亲眼见证过现场，实话告诉你，真实战争比你见过最残酷的战争电影还要惨烈得多。
One way to think about it is that we all so first place, and I've I've seen it live, and I will tell you that real war is much worse than the worst movies you have ever seen about war.
我就说这么多。
And that's all I'll say.
这确实令人毛骨悚然。
It's really horrific.
要不惜一切代价避免这种情况。
It is to be avoided at all cost.
然后就是正确的。
And then right.
出于显而易见的原因。
For obvious reasons.
我最喜欢那些说'要小心好战言论，因为对方也有发言权'的人。
And the the and I love all these people who say, well, well, you know, the warmongering talk, be careful what you wish for because the other side gets a vote.
当我开始研究并试图理解乌克兰的行动时，俄罗斯已被击退。
When I started working and and trying to understand what Ukraine was doing, you Russia was pushed back.
但他们以非常非常强势的第二和第三轮攻势卷土重来。
And they've come back with a very, very strong second and third round.
所以在这种局势下敌人也有发言权。
So the enemy gets a vote in this situation.
但战争的大致演变方向是：首先作战方式必须高度机动化，绝不能固守阵地。
But to but to go on, the rough way in which war will evolve is first things will have to be very very mobile and very much not in fixed places.
这将淘汰世界上现存的大部分军事基础设施。
This takes out most of the military infrastructure that exists in the world.
像坦克这样的装备——尽管美国现在正在建造更多更强的坦克——在武装无人机只需两公斤载荷就能摧毁坦克的世界里毫无意义。
Things like tanks, of which we're now building a whole bunch more, even stronger tanks here in America, don't make any sense in a world where a two kilogram payload from a well armed drone can destroy the tank.
这就是所谓的杀伤比率。
It's called the kill ratio.
那款无人机的零售价是5000美元，4000美元。
And that drone costs retail $5,000, $4,000.
坦克，美国坦克的造价是3000万美元。
The tank, the American tank costs $30,000,000.
你可以看到，你可以派出大量无人机去摧毁那些坦克。
You can see the the you can send an awful lot of those drones to destroy those tanks.
可能的演变过程大致是这样的。
The likely evolution goes something like this.
首先，人们认识到无人机就像步枪和大炮一样。
So first, people learn that drones are like rifles and like artillery.
所以现在使用无人机比使用迫击炮、手榴弹、大炮更高效。
So it's more efficient to use drones now than to use mortars, grenades, artillery.
这很明显。
That's clear.
如果你只看经济学，从成本或所谓效能的角度来看。
If you just look at the economics, economics in terms of cost or effectiveness as it's called.
接下来发生的是双方都发展无人机能力，这正是你现在看到的，然后演变成无人机对抗无人机的战争。
The next thing that happens is that both sides develop drone capabilities which is what you're seeing now and each then becomes a war of drone against drone.
所以你会看到无人机对抗反无人机。
So you have drone against anti drone.
于是重点转向如何探测敌方无人机，并在它摧毁你之前摧毁它。
And so then the shift moves to how do you detect the enemy drone and how do you destroy it before it destroys you.
因此最终的战术原则是无人机在前，人员在后。
So the doctrine ultimately is the drones are forward and the people are behind.
例如在基辅，我亲眼目睹乌克兰人通过星链指挥作战，可以说在远距离战争中他们表现得非常非常高效。
And I've seen operations in, for example, sitting in Kyiv where the Ukrainians are commanding things over Starlink, might add, in the distance in the distant war and they're very very effective.
因此我们已解决了该区域的延迟问题、时序问题等等。
So we've solved the latency problems, we've solved the timing problems and so forth in that area.
最终状态非常有趣，我认为没有人预见到这一点。
The ultimate state is very interesting and I don't think anyone has foreseen this.
如果你回顾我们关于强化学习与规划的讨论，这正是你现在在AI领域看到的。
If you go back to our conversation about RL and planning which what you're seeing with AI.
假设我们这边有百万架无人机，对面也有百万架无人机。
Let's say that that we're on one side and we have a million drones and there's another side over here that has another million drones.
双方都将运用强化学习AI策略来制定作战计划。
Each side will use reinforcement learning AI strategies to do battle plans.
但双方都无法洞悉对方的作战计划。
But neither side can figure out what the other side's battle plan is.
因此，相互攻击的威慑力会非常高。
And therefore, the deterrence against attacking each other will be very high.
当今军事策划者的运作方式是清点武器数量。
Today, the way military planners operate is that they count weapons.
他们会说：你有这么多，我有这么多，你可以进行某种战术机动等等。
They say, well, you have this many and I have this many and you can do this kind of a maneuver and so forth.
但在运用强化学习的AI世界里，你无法计算对方的计划。
But in an AI world where you're doing reinforcement learning, you can't count what the other side is planning.
你看不见、不知道对方的计划，我相信这将遏制我认为人类有史以来最可怕的暴行——战争。
You can't see it, you don't know it and I believe that that will deter what I view as one of the most horrendous things ever done by humans which is war.
因为除非双方达到完美平衡，否则无人机的供应将相互摧毁，就像传统战争中任何炮兵库存一样，最终幸存的一方将获胜。
Because unless there's a perfect balance between either side, there will be some mutual destruction of the drone supply like there would be with any artillery stock in traditional warfare and whoever's left ends up winning.
就像，它们只是
Like, they're just
重要的是要明白，战争中没有赢家。
Well, it's very important to understand that there's no winners in war.
当你进行我所描述的那种规模的无人机战争时，你方的整个基础设施将被摧毁，对方的整个基础设施也将被摧毁。
By the time you have a drone battle of the scale I'm describing, the entire infrastructure of your side will be destroyed, the entire infrastructure of the other side will be destroyed.
这些都是双输的局面。
These are lose lose scenarios.
难道不存在某种平衡吗？由于这种相互确保毁灭，可能会形成一种缓和状态，还是说...
Isn't there like an an equilibrium though that that can also create where there because of that mutually assured destruction, it there's a detente or is that
我认为这不是缓和状态。
Well, I'm arguing that it's it's not a detente.
这是一种威慑。
It's a deterrence.
对吧。
Right.
威慑可以理解为：我想打击你（虽然我并不想）。
That as deterrence can be understood as I want to hit you, which I don't.
是的。
Yeah.
但我非常想打击你，可如果我这么做，惩罚将超过打击你带来的价值。
But I wanna hit you so much, but that if I do that, the is the penalty is greater than the value of me hitting you.
对。
Right.
这就是威慑的方式，这就是威慑的方式
And that's how deter that's how deter
这看起来像是向无人机和自动化转型的巨大优势和好处，这是我们目前所不具备的。
that seems like an that that seems like a great advantage and upside of this move to sort of drones and automation that we don't have today.
嗯，我认为转向无人机和自动化有许多优势。
Well, I there are many advantages to moving to drones and automation.
第一，它们成本要低得多。
One, they're much much cheaper.
对吧？
Right?
它们成本要低得多。
They're much much cheaper.
是的。
Yeah.
第二，第二，你可以储备算法。
And two and two, you can stockpile algorithms.
你基本上可以不断学习再学习。
You can essentially learn and learn and learn.
记住，你还可以构建训练数据，对吧，是合成的，所以你可能比其他人做得更好。
And remember, you can also build training data, right, that's synthetic so you could be even better than the others.
军方问我的最后一个问题是：传统陆军的作用是什么？
The final question I've been asked by our military is what's the role of the of a traditional land army?
我多么希望能说所有这些人类行为都能在人类不面临风险的情况下发生。
And I wish I could say that all of these human behaviors can occur without humans being at at risk.
我不这么认为。
I don't think so.
我认为机器人战争，本质上是无人机战争，将会以毁灭性浪潮的形式出现。
I think that the way robot war, essentially drone war will occur is there will be destructive waves.
但最终，人类将不得不跨越一条界限。
But eventually, humans are going to have to cross a line.
他们将会
They're going to
耗尽它们。
have depleted them.
所以你在投资这项无人机技术，那你认为乐观主义者和人形机器人会是下一轮，你知道的，这场
So you're investing in this drone technology and then do you think optimists and humanoid robots are the next, you know, volley in this
新型战争中的主力吗？
new warfare?
距离我们看到电影里整天出现的那种人形机器人还有很长一段时间。
It's gonna be a long time before humanoid robots, which what we see in the movies all day.
对吧？
Right?
我们距离看到那种场景还有很久很久。
It'd be a very long time before we see that.
你将看到的是非常非常快速的机动解决方案。
What you're gonna see is very very fast mobility solutions.
对吧？
Right?
本质上就是空中领域，基于空中的解决方案以及高超音速技术。
Air essentially, air based solutions and also Hypersonics.
高超音速。
Hypersonics.
还有水下领域。
Also, things underwater.
这方面有很多进展。
There's a lot of that going on.
这是个不同的领域。
It's a different domain.
如果你看看乌克兰使用的马古拉号和其他船只，他们基本上是用无人艇摧毁了黑海的俄罗斯舰队。
If you look at the Muguro and some other boats that the Ukrainians used, they have essentially used USVs to destroy the Russian fleet in the Black Sea.
这对他们至关重要，因为他们需要通过敖德萨出口粮食，这约占他们经济的6%到10%。
This was crucial for them because they needed to be able to export the grain from Odessa around and it's like 6% or 10% of their economy.
所以影响非常大。
So very big deal.
他们是用无人机做到的。
And they did that with drones.
埃里克，你似乎有一个贯穿始终的世界观。
Eric, it seems like there's this overarching world view that you have.
也就是说，你对人工智能有独到的见解。
Meaning, you have this view on AI.
你们现在在无人机、战争和火箭技术方面做了这么多事情。
There's all this stuff you're doing now in drones and warfare, in rocketry.
坦率地说，这一切都会汇聚在一起，因为在未来五到十年内，这些事情都将成为现实。
It all converges quite honestly because in the in the next five or ten years, these things will all come to pass.
比如说，你如何看待这个世界？
What is the like, how do you view the world?
比如说，美国的角色是什么？
Like, what is the role of America?
作为资本家、技术专家，甚至是政治家，你的角色是什么？
What is your role as a as a capitalist, as a technologist, as like a statesman?
我希望美国能赢。
I want America to win.
对吧？
Right?
我站在这里是因为美国梦，那些投资我的人，比如伯克利等等，他们给了我机会。
I am here because the American dream, the people who invested in me, in my case Berkeley and so forth, people took a chance on me.
我希望下一代也能拥有这样的机会。
I want the next generation to have that.
我还想提醒大家，就在檀香山举行的二战投降仪式上，他们谈到了与暴政抗争。
I also want you all to remember, was just in in as part of the World War two surrender ceremony in in Honolulu and they talked about fighting tyranny.
对吧？
Right?
我们忘记了我们的祖先或曾祖父母们曾为捍卫自由主义和民主而奋战。
We forget that our ancestors or great grandparents or whatever fought the great war to keep liberalism and democracy alive.
我希望我们能实现这个目标。
I want us to do that.
作为美国人，我们该如何做到这一点？
How do we do that as Americans?
我们要发挥自身优势。
We use our strengths.
我们的优势是什么？
What are our strengths?
我们虽然混乱、令人困惑、吵闹，但你要知道，我们很聪明。
We're chaotic, confusing, loud, you know, but we're clever.
我们善于明智地配置资本。
We allocate capital smartly.
我们拥有非常深厚的金融市场。
We have very deep financial markets.
我们拥有由大学和企业家构成的庞大产业基础，这些都体现在这里。
We have this enormous industrial base of universities and entrepreneurs which represented here.
展开剩余字幕（还有 150 条）
我们应该为此感到自豪。
We should celebrate this.
我们应该继续添柴加火。
We should stoke it.
我们应该让它发展得越来越快。
We should make it go faster and faster.
因为乌克兰的事情，我在欧洲待了很长时间。
I spent lots of time in Europe because of the Ukraine stuff.
他们非常嫉妒我们。
They are so envious of us.
当你在亚洲时，他们会嫉妒我们。
When you're in Asia, they are envious of us.
伙计们，别搞砸了。
Don't screw it up, guys.
这正是我想要的
That's what I want to
工作 我能问一下吗，除了这个外部冲突，我们今天和Alex Karp聊过，昨天Tucker Carlson也在这里？
work Can I can I just ask you, outside of this external conflict, we had a conversation with Alex Karp today and we actually had Tucker Carlson here yesterday?
部分对话围绕西方世界的——我不确定用词是否准确——‘衰落’展开，认为西方可能存在一些正在酝酿的社会问题，可能从内部伤害我们。
And some of the dialogue was around the I I don't know if the right term is the erosion of the West, that there may be social issues that are brewing in the West that may be hurting us from the inside.
你花多少时间关注这些问题？
How much do you observe or spend time on these issues?
现在常被引用的指标是西方出生率下降，我们的人口...稍后我们会和埃隆讨论这个话题。
And the metric that often is cited now is declining birth rates in the West and that our population and we're gonna talk with Elon in a few minutes about this.
哦，抱歉。
Oh, sorry.
噢，我们刚写错了。我的错。
Oh, we just wrote My bad.
我的错。
My bad.
惊喜。
Surprise.
哎呀。
Oops.
抱歉。
Sorry.
抱歉。
Sorry.
你的惊喜客人来了。
There's your surprise guest.
抱歉。
Sorry.
抱歉。
Sorry.
抱歉。
Sorry.
抱歉。
Sorry.
滑倒。
Slip.
但是
But
他是一位好朋友，正在亲自直接解决人口问题。
is a good friend and he's addressing this issue of population directly himself.
为此。
For it.
为你高兴，亲爱的。
Good for you, love.
是吗
Is it
这是否反映了某些正在发生的事情？
is it a reflection of something going on?
纽约出现了马姆达尼当选人数上升的情况。
There's a rise of Mamdani getting elected in New York.
西方的一些历史价值观似乎正处于某种转型状态。
Some of the historic values of the West seem to be on, you know, kind of under a state of transformation right now.
衡量一个社会成功与否的标准之一是其繁衍能力。
One metric of the success of a society is its ability to reproduce.
因此我认为这是西方社会合理的担忧。
And so I think this is a legitimate concern of the West.
亚洲的情况要糟糕得多。
It's much worse in Asia.
中国每对父母的生育率约为1.02。
The the Chinese number is about one point o for two parents.
韩国的数字现在已降至每对父母0.78。
In in Korea, it's now down to point seven eight for two.
所以我们必须认识到，作为人类，我们正在集体选择减少人口，这些数字令人震惊。
So it's really important to recognize that we as humans are collectively choosing to depopulate And the numbers are staggering.
对吧？
Right?
想象一种情况，不是增长而是萎缩。
And imagine a situation where instead of having growth, you have shrinkage.
而且，他们正在老龄化。
And furthermore, they're getting older.
因此作为企业，突然间你的收入在下降却无能为力，因为客户越来越少就无法创新。
And so as a business, all of a sudden, your revenue is declining and there's nothing you can do because you can't innovate with fewer and fewer customers.
所以如果仅从商业角度考虑，忽略那些非常现实的道德问题，这本身就是糟糕的。
So if you just put it in a business context, ignoring the moral issues which are all very real, it's just bad.
对吧？
Right?
所以我们必须解决这个问题。
So we have to solve that problem.
我个人总体上支持移民政策，因为我认为移民有助于解决这个问题，但作为全球机制，我们必须正视它。
I happen to be in favor broadly of immigration because I think immigration helps us solve that problem, but as a global mechanism, we have to address that.
无论如何在我看来，这些问题必然存在，但美国是围绕'美国例外主义'理念构建的。
In any case, from my perspective, you're going to have these issues, but America is organized around the concept of American exceptionalism.
只要我们明白进步之道在于投资正确的人、正确的企业，拥有强大的资本市场，并投资他们所需的基础设施，我们就会没事。
And as long as we understand that the way we make progress is we invest in the right people, in the right businesses, we have a a strong capital market, we invest in the infrastructure that they need, We'll be fine.
这才是我真实的观点。
That is my actual opinion.
我们能否回到AI话题？埃里克，我认为你能帮我们达成——姑且称之为——对这些问题的跨党派共识，因为你思考得非常清晰。要知道2022年ChatGPT推出后，2023到2024年的讨论几乎被AGI概念及其迫在眉睫性主导。
Can we go back to AI So, for a Eric, I think you can help us get to a a let's call it a a bipartisan understanding of these issues because I think you you you think really clearly about You know, in in the wake of ChatGPT launching at the 2022, I think the discourse was really dominated in 2023 and '24 by this idea of AGI and that AGI was imminent.
我认为这几乎在华盛顿政策制定者中制造了恐慌氛围，你会看到诸如'必须限制开源否则中国会得到'的论调——这还是在DeepSea发布前，后来我们发现他们在开源领域其实领先我们。
And I think it created almost like a panicky atmosphere in Washington among policymakers, and you saw things like we gotta restrict open source because, you know, then China will get it, and this is before DeepSea launched, then we saw that actually they're ahead of us in open source.
但感觉关于AGI的讨论热度有所减退，我认为这其实是件好事。
But it feels like there's been a pullback a little bit from the AGI narrative, which I think I think is actually a good thing.
我认为这更有利于冷静理性的政策制定。
I think it's more conducive to calm, rational policy making.
你目前对AGI有何看法？
What's your perception of AGI right now?
我们在这整件事上进展到哪一步了？
Where where are we on that whole train?
首先，总统大约一个月前关于AI战略的演讲——虽然你可能不会这么说，但那篇演讲稿基本上是你为他起草的——完全正确。
So so so first place, the speech that the president delivered about a month ago about AI strategy, which I think you probably wouldn't say it but you kinda wrote it for him, was exactly right.
对吧？
Right?
所以感谢与这位我们都无比敬仰的杰出领导者合作，
So thank collaborated with an amazing leader who we all respect and admire so much,
埃里克。
Eric.
是的。
Yes.
不过，
So, nevertheless
说是我写的就太夸张了。
Saying I wrote it was was way too strong.
我是说，实际上如果
I mean, actually If
如果不是你写的，那一定是你的双胞胎兄弟写的。
didn't if you didn't write it, then it must have been your twin.
但无论如何，你把握住了重点——对研究的投资，对我们所做工作的投资，确实至关重要。
But in any case, the You you got you got the emphasis right, which was that investment in research, investment in the kinds of stuff that we do is really really important.
在AGI这件事上我不同意你的观点，因为有群被我称为'旧金山叙事派'的人——他们都住在旧金山。
I don't agree with you on this on this AGI thing because there's this group which I call the San Francisco narrative because they all live in San Francisco.
他们的论调大致是这样的：
And their narrative goes something like this.
如今我们正在研发智能体。
Today, we're doing agents.
智能体革命将改变商业形态——这点我认同。
The agentic revolution will change businesses, which I agree with.
最终系统将实现递归式的自我智能进化。
That what happens is the systems will become recursively self self intelligent.
所谓递归式自我改进，就是面对无规模限制的问题（比如编程或数学这种可以无限深入的问题），只要投入足够硬件或软件资源，就能获得指数级的高速进展。
With recursive self improvement, as it's called, if you have a scale free problem, and a scale free problem, for example, is programming or math where you can just keep doing it, you get these enormous fast gains if you buy enough hardware or do enough software and so forth and so on.
这个进程仍在继续。
That is still underway.
他们集体认为在未来三年左右，我们将实现某种超级智能。
The collective of that says that in the next three ish years, they believe that we will get forms of super intelligence.
他们将其定义为专才型智能——化学专才、物理专才、数学专才。
And the way they define it is basically a savant, a chemist savant, a physics savant, a mathematician savant.
我不认同三年这个时间点，但我认为大概需要六到七年。
I don't agree with the three years, but I do agree that it'll be maybe six or seven years.
但如果它只是在某个特定领域的天才，那能算通用智能吗？
But if it's a savant in, you know, a particular area, is that general intelligence?
这还不是通用智能。
It's not general intelligence yet.
通用智能是指能够自主设定目标函数。
General intelligence is when it can set its own objective function.
对。
Right.
但目前没有证据
But there's no evidence
表明这一点。
of that.
目前没有证据显示其具备自主设定目标函数的能力。
There's no evidence right now of the ability to set your own objective function.
这个思考——我正在写一篇相关论文，所以一直在研究——技术难题在于数学证明的非平稳性。
The the the thinking, and I'm writing a paper on this, so I've been studying it, is that the the the technical problem is non stationarity of mathematical proofs.
你试图针对某个函数求解，但目标函数不断变化，这正是人类的运作方式。
And what you're doing is you're trying to solve against a function but the objective function keeps changing which is how humans operate.
人类的目标每天都在变，而计算机难以应对这种情况。
Your goal changes every day whereas computers have trouble with that.
作为数学问题，我们尚未开发出能让大语言模型实现这一点的算法。
As a math problem, we don't have an algorithm yet for LLMs that can do that.
人们正在研究这个问题。
People are working on it.
数学 而测试将是，你能否基本上利用1902年可用的信息，推导出爱因斯坦通过狭义相对论和广义相对论得出的相同结论？
Math And the the test will be, can you basically using the information available in nineteen o two, can you derive the same thing that Einstein did with special relativity followed by general relativity?
我们今天做不到这一点。
We cannot do that today.
大多数人认为，解决这个问题的方法是通过类比。
And most people believe that the way this will be solved is through analogy.
因此，伟大天才的理论在于他们极其精通某一领域，他们的智慧使得后人能够将其思想应用于完全不同的领域。
So the theory of great geniuses is that they understand one area extremely well and they're so brilliant that lay later man can then take their ideas and apply it to a completely different domain.
如果我们能解决这个问题，那么我认为一切就结束了。
If we can solve that problem, then I think it's over.
然后我们将实现人工通用智能（AGI），届时将是一个全新的世界。
Then we get to AGI and then it's a whole different world.
我认为难以取代人类的原因之一——我和JK经常争论这点——在于人类是端到端的。
I think one of the reasons why it's hard to replace a human and, you know, JK and I debate this, is that humans are end to end.
我们能够完成整个工作。
You know, we can do the whole job.
你拥有某种程度的全面理解。
You have sort of a complete understanding.
你可以非常灵活地转换方向。
You can pivot very easily.
人工智能，至少就目前我们所知，并非端到端的。
AI, at least as we know it today, is not end to end.
它需要被提示才能工作。
It has to be prompted.
你得到了一个答案。
You get an answer.
这个答案需要被验证。
That answer has to be validated.
然后你必须提出新问题，因为它永远不会给你真正想要的东西。
Then you have to ask a new question because it never gives you exactly what you want.
你需要提供更多上下文。
You have to apply more context.
你必须经历一个迭代循环。
You have to go through an iterative loop.
最终你会得到一个具有商业价值的答案。
Finally, you get to an answer that has business value.
生物学对此的解释是：人工智能并非端到端的。
The way biology puts it is that AI is not end to end.
它是中间到中间的。
It's middle to middle.
人类才是端到端的。
Humans are end to end.
因此结果不是AI取代我们所有人，而是AI将与人类高度协同——因为我们可以定义目标函数，进行提示工程，与AI协作迭代，而AI负责中间大部分工作。
And so as a result of that, instead of AI replacing all of us, AI will be very synergistic with humans because we can define the objective function, we do the prompting, and we work with it to iterate, and it does a lot of the work in the middle.
在我看来这是个非常乐观、不那么悲观的看法。
That seems to me like a very optimistic, less doomeristic take on it.
你刚才说的正是未来几年将要发生的事。
What you What you just said is exactly what's going to happen for the next few years.
我们每个人都将拥有这样的助手，只需一声令下或简单提示，它就能为我们解决任何个人问题，你知道的，那些私人的困扰。
That each of us will have assistance which on our command and our prompting will be incredibly helpful to whatever problem we have, you know, personal.
现在有人已经在用这类工具获取恋爱建议，甚至用来与孩子沟通交流。
You have people who are using these things for relationship advice, for you know, talking to their kids.
我是说，这些都太疯狂了。
I mean, it's all crazy stuff.
但事实就是如此。
But the fact of the matter is that's it.
对我来说，真正的问题是：什么时候它会跨越界限，拥有自主意志和主动获取信息解决新问题的能力？
To me, the real question is when does it cross over to having its own volition, its own ability to seek information and solve new problems?
那将是完全不同的概念。
That's a different animal.
但我们目前看到任何递归自我改进的证据了吗？
But have we seen any evidence of recursive self improvement yet?
还没有。
Not yet.
我...我投资过几家声称即将实现这一点的初创公司。
I'm I'm I've funded a number of startups which claim to be close to it.
不过这些都是初创企业，谁说得准呢？按我的估算大概还要五到十年。
But of course, these are startups and you never know, which tells me it's five, ten years cutting numbers.
你觉得谷歌在这方面有什么动作？
What do think Google's doing on this front?
这个嘛，我已经不在谷歌工作了。
Well, I'm not at Google anymore.
Gemini的每一期都位居排行榜榜首。
Every issue of Gemini is top of the leaderboard.
所以2.5版本刚刚超越所有人，我确信下一个版本即将到来。
So two point five just overcame everybody and I'm sure there's another one coming.
Demis正在全力研究这个关于科学发现的问题。
Demis is working really hard on this question about scientific discovery.
所以这是一条通往AGI的道路。
So that's a that is a path to getting to AGI.
Eric，我们很感谢你所做的工作。
Eric, we appreciate the work you're doing.
我们很感谢你能和我们在一起。
We appreciate you being here with us.
我们感谢你所做的一切，以及你对硅谷社会产生的影响
We appreciate what you've done, the impact you've had on Silicon Valley as a society
这真的很棒。
It's really nice.
是的。
Yeah.
不。
No.
但这确实已经
But it's it's really been
我非常高兴能参与其中。
I am so happy to be part of this.
你创建了这个了不起的社区，所有这些聪明人都把时间花在听你说话上。
You created this incredible community, and there's all of these smart people that spend all their time listening to you.
是啊。
Yeah.
是啊。
Yeah.
非常令人担忧。
Very concerning.
哇。
Wow.
真奇怪，埃里克。
Strange, Eric.
是啊。
Yeah.
谢谢。
Thank you.
谢谢你，埃里克。
Thank you, Eric.
谢谢你，埃里克。
Thank you, Eric.
谢了，
Thanks,
感激你。
Appreciate you.
干杯。
Cheers.
好吧。
Alright.
关于 Bayt 播客
Bayt 提供中文+原文双语音频和字幕，帮助你打破语言障碍，轻松听懂全球优质播客。