马克·安德森：谁在主宰全球人工智能？

一场竞赛正在进行，其核心问题是世界将依赖什么运行。

There's a race underway, and the stakes are basically what is the world going to run on.

唐·瓦伦丁有一个老式的经验法则。

Don Valentine had this old rule of thumb.

他说，初创公司更多是因资金过多而消化不良，或因资金不足而饿死。

He said more startups die of indigestion and starvation in terms of the amount of money you put in.

他的观点是，稀缺性确实能激发创造力。

And and and his point was, like, scarcity does spark ingenuity.

所有的科幻小说几乎都将人工智能描绘成要么极度乌托邦，要么极度反乌托邦，但从没有真正体现我们正在经历的那种强烈的幽默感——人们只是把一切东西都拿来当梗的素材。

All of the science fiction novels basically have AI either being, like, super utopian or super dystopian, but they never have this incredible sense of humor aspect, which is what we're actually getting, where people are just using everything as a fodder for memes.

世界要么将由美国的人工智能驱动，要么由中国的人工智能驱动，我认为哪一个胜出对许多原因来说都至关重要。

The world will either be running on American AI or be running on Chinese AI, and I I think it's very important which one wins for a bunch of reasons.

五十年来，经济学家一直追踪着一种奇怪的现象：技术迅猛发展与历史上极低的生产率增长并存。

For fifty years, economists have tracked a strange pattern, rapid technological change paired with historically low productivity growth.

自1971年以来，尽管计算技术重塑了日常生活，生产率却一直停滞不前。

Since 1971, productivity has flatlined even as computing reshaped daily life.

1880年，生产率增长是现在的三倍。

In 1880, productivity growth ran at three times today's rate.

到1930年，它已经放缓到现在的两倍。

By 1930, it had slowed to twice as fast.

接着，监管和限制出现了。

Then came the regulations and the restrictions.

我们拒绝了核能、更快的汽车和太空计划。

We said no to nuclear power, faster cars, and a space program.

我们得到的是芯片和软件的超高速发展，而其他几乎所有领域都陷入停滞。

What we got was hyper acceleration in chips and software and stagnation in nearly everything else.

美国实验室目前领先，但中国开源模型仅落后数月，且成本仅为一小部分。

American labs lead for now, but Chinese open source models follow months behind at a fraction of the cost.

世界将依赖其中一种系统，而嵌入该系统的价值观将至关重要。

The world will run on one system or the other, and the values baked into that system will matter.

这场对话探讨了人工智能投资的实际情况、价值可能积累的领域，以及监管回应如何决定哪个国家胜出。

This conversation looks at what's actually happening in AI investment, where value might accrue, and why the regulatory response could determine which country wins.

思科总裁兼首席产品官杰图·帕特尔与安德里森·霍罗威茨的联合创始人兼普通合伙人马克·安德森对话。

Jeetu Patel, President and Chief Product Officer at Cisco, speaks with Marc Andreessen, cofounder and general partner at Andreessen Horowitz.

马克·安德森无需介绍。

Marc Andreessen needs no introduction.

他发明了浏览器。

He invented the browser.

他构建了互联网，所以我真的非常期待

He, built the Internet, so I'm I'm really excited to

你的到来。

have you here.

我对此毫不道歉。

I apologize for nothing.

好吧。

Alright.

在开始之前，你前几天和伦尼进行了一次非常有趣的对话，我想就从这个话题开始。

So be before we get started, you had a really interesting conversation that I wanted to actually start with, just just a couple days ago with Lenny.

你提到在人类历史长河中，什么时候生产力真正出现过大幅跃升？而现在又发生了什么？

And, you were talking about this notion of, in the history of time, when has productivity really spiked and what's happening right now?

所以你能谈谈你对不同时期生产力提升的看法吗？我们今天相比那些时期处于什么位置？

So can you just talk a little bit about your perspective on productivity increases that have happened at different phases in time and where are we today compared to those times?

是的。

Yeah.

正如大家可能都知道的，生产率增长是经济增长的关键驱动力。

So as everybody probably knows, productivity growth is like the key driver of economic growth.

它才是真正推动经济扩张的因素。

Like, it's the thing that actually causes the economy to expand.

经济学家用一种叫做全要素生产率的指标来衡量它。

Economists measure it with something called total factor productivity.

他们每年都进行测算。

They measure it every year.

过去五十年——实际上是我整个一生，也是我们所有人的一生——普遍流传的一个迷思是，我们正处在一个技术飞速变革的时代，这理应意味着生产率的迅猛增长。

The the the prevailing kind of myth of the last fifty years, basically, my entire life, our all of our entire entire lives has been that we've been in this era of very rapid technological change, which would necessarily mean very rapid productivity growth.

但如果你真的查看统计数据，自1971年我出生以来，生产率就大幅下滑，远低于之前的时期。

Yet if you actually look at the statistics basically since actually, since the year I was born in 1971, productivity downshifted hard from prior eras.

在过去五十五到六十年里，生产率增长基本处于历史低位。

And productivity growth basically for the last fifty fifty five, sixty years has been at basically historical lows.

生产率一直非常低，这也就解释了为什么经济增长缓慢，顺便说一下，这也是为什么国民情绪越来越聚焦于零和经济、民粹主义，认为如果有人在进步，就一定有人在吃亏。

It's it's it's been very low, which is, by the why economic growth has been low, which, by the way, is why the national mood has become so focused around, you know, zero sum economics, populism, you know, the the the sense that if if somebody's getting ahead, somebody else must must be getting disadvantaged.

如果你将这与1930年到1970年之间的时期对比，那段时期的生产率增长速度大约是现在的两倍。

If you compare and contrast that to the period between about 1930 to about 1970, productivity growth was roughly twice as fast through that period.

如果你再与1880年到1930年这段时期对比，那时的生产率增长速度大约是现在的三倍。

And if you compare and contrast that to the period of 1880 through 1930, productivity growth was about three times as fast.

所以我们经历了三倍、两倍，然后是一倍的增长。

So we had three x and then two x and then one x.

这非常不妙。

And so this is very not good.

就像，你知道的，这个

Like, the the you know, the

你认为为什么会这样？

Why do you think that is?

从根本上说，我认为这是因为我们决定其他事情更重要了。

I I I most fundamentally, I think it's because we decided other things are more important.

特别是从上世纪七十年代以来，如果你看看法律数量、联邦登记册的页数，或者经济中的监管条例数量，就会发现这些数据就像曲线突然急剧上升，而且至今仍在持续增长。

And in particular, in the last, you know, basically since since the nineteen seventies, you know, if you just look at, like, the if you just look at the charts of, like, the number of of laws on the books or the number of pages in the federal register or the number of regulations in the economy that, you know, it's just this it was just this, like, knee in the curve went went exponential, which continues.

所以我们只是决定不再想要核能。

And so we just, you know, we just, you know, we decided we didn't want nuclear power.

对吧？

Right?

我们决定不再搞航天计划。

You know, we decided we didn't want a space program.

我们决定不再允许汽车时速超过55英里。

You know, we decided we didn't want cars that went faster than 55.

我们就是决定，不再想要这些东西了。

We decided you know, we like, we just decided we didn't want these things.

所以在过去五十年里，我们得到的是在芯片和软件这一特定领域内的超高速发展。

And so what we got in the last fifty years was, hyper acceleration in very specifically, basically chips chips and software.

而除此之外的其他所有领域，基本上陷入了停滞。

And then what we got was basically, you know, essentially stagnation in in in everything else.

所以，这真的不是什么好事。

And and so, like, it it it's it's it's really not it's it's really not good.

相应地，关于对人工智能感到兴奋的诸多理由，这么说吧。

And and then, you know, but but correspondingly, you know, this of of the many reasons to be excited about AI, like, put it this way.

无论人工智能乐观派还是悲观派是对的，生产率增长都将急剧飙升。

If either the AI optimists are correct or the AI doomsayers are correct, productivity growth is about to go through the roof.

你觉得会是两到三倍吗？

And do you think it's like two or three X?

是十倍吗？

Is it 10 X?

你认为它会达到什么程度？

Where do you think it gets to?

所以这总是这类问题之一。

So this is always one of these kinds of questions.

比如在一个完全自由化的经济中，就像默里·罗斯巴德所梦想的那种纯粹的无政府资本主义。

Like in a, like, completely deregulated economy, like, in in sort of, you know, Murray Rothbard's, like, you know, dream of just, like, straight, basically anarcho capitalism.

至少在理论上，你可以想象生产力增长会加速到，比如说5%、10%。

You know, at least in theory, you can imagine an acceleration to, I don't know, 5%, 10%.

我的意思是，如果你相信乐观派或末日派的观点，你会看到AI带来的是革命性的软件生产力增长，紧接着就是机器人技术的崛起。

I mean, if, again, if you believe in kind of either the optimist or the doomsayers, you're looking at such a radical AI representing radical software productivity growth and then robotics coming right behind.

对吧？

Right?

而机器人技术，当然会从自动驾驶汽车和无人机开始，但人形机器人也会迅速到来。

And robotics, of course, starting in the form of the self driving car and the drone, but humanoid robots coming quickly.

你可以想象，可能会出现10%、20%、30%这样的增长情景。

You know, you could imagine you know, you could you could paint, you know, scenarios of 10%, 20%, 30%, something like that.

但我认为现实中，这种可能性不大，因为机器人也必须遵守所有的法规。

I I I think in practice, you know, that that's unlikely again just because, like, the the robots have to have to agree to all all the regulations also.

所以，你知道，有很多事情他们是不会被允许做的。

And so, you know, there's there's there's a lot of things they're not not gonna be allowed to do.

顺便说一下，AI，你看，我可以给你一个正在发生的绝佳例子。

By the way, AI, you know, look, I I'll give you a great example as it's playing out today.

我认为，如果你只是一个普通人，我出去度假时得了食物中毒，当然，一去就病了，这总是会发生。

I think if if you're just like an ordinary person, I I I got food poisoning over the I went on vacation, of course, immediately got sick, which always happens.

我得了食物中毒。

I got food poisoning.

于是，我做了一个实验，让医生GPT帮我一步步了解食物中毒的每一个阶段。

And so I I let I let as an experiment, I let doctor GPT, like, walk me through basically every stage of food poisoning.

我不断提问，因为我躺床上没事可做。

And I just I kept asking, like I had nothing else to do because I'm flat on my back.

所以我不断问更多更详细的问题，关于我的身体感受、该做什么、该吃什么、该如何康复。

So just kept asking, like, more and more detailed questions about, you know, my physical experience and what I should do and what I should eat and how I should recover.

这简直太不可思议了。

And, like, it's just, like, absolutely incredible.

它就像是一个极其出色、永远耐心、知识无穷、无比关怀的医生。

Like, it's just, like, the most amazing like, the end endlessly patient, sort of infinitely knowledgeable, endlessly caring doctor.

你知道吗，即使我在凌晨四点问同样的问题，它也不会不耐烦，我还会说：‘你能再深入讲讲吗？’

You know, it doesn't get, like, irritated when I have the same question at four in the morning, and I'm like, well, could you go into that, a little deeper?

你确定不是胰腺炎吗？

Are you sure it's not pancreatitis?

你确定我不会马上死掉吗？

And are you sure I'm not about to die?

哦，不会的。

Oh, no.

别担心，你没事的。

Know, you're it's okay.

你完全没事。

You're absolutely fine.

这真是太神奇了。

And it's just amazing.

然后，当然了，人工智能不能获得医生执照。

And then and then of course, AI cannot be licensed as a doctor.

对吧？

Right?

这完全是违法的。

That's that's, like, completely illegal.

对吧？

Right?

你实际上不可能拥有一位AI医生。

You you cannot actually have an AI doctor.

所以，你确实面临着这种巨大的脱节。

And so you you you do have this, like, basically massive disconnect.

而且，我再说一遍，我不是在倡导穆瑞·罗斯巴德的世界。

And, again, I'm not saying I'm advocating for the Murray Rothbard world.

我不是说要废除所有法规。

I'm not saying rip up all the regulations.

我只是说，这在事实上、客观上就是这样。

I'm just saying, like, it it just factually, objectively.

它会拖慢你

It slows you

的速度。

down.

是的。

Yeah.

我们这些极度乐观的人和末日论者，都不会得到他们所想象的那样的世界。

We the the the the the the again, the the the hyper optimist and the doomsayers are not neither one of them are gonna get the world that they think that we're gonna get.

我们会迎来一种折中的、混杂的局面，我认为，顺便说一句，这种局面会进行得不错，但它确实会很混乱，而且在这个过程中，会有很多介于这两极之间的波动。

We're gonna get some we're we're gonna get a muddle through the middle thing, which I think, by the way, I think is is gonna go quite well, but it's gonna be a muddle, and there's gonna be a lot of you know, kind of between those those sides along the way.

那么，鉴于这种情况，价值最开始会在哪个层面积累呢？

And then given that, where does the value start accreting in the stack most?

所以我认为，这是一个非常非常重大的问题，从我们作为专业投资者的角度来看。

So I think this is a really, really it's a really big question of we're professional investors on our side.

因此，我们当然一直在思考这个问题。

And so, of course, we we think about this all the time.

实际上，我觉得这个问题的问题比答案还要多。

And I actually think there's still more questions than answers than this.

对吧？

Right?

因为，你可以描绘出这样一幅图景：AI模型公司几乎会掌控一切。

Because, you know, you can paint this picture that says that the, you know, the AI model companies are gonna basically own everything.

而且，顺便说一句，你看看他们的业务，确实做得非常好。

And, you know, and you by the way, you look at their, you know, you look at their their businesses, and they're doing doing fantastically well.

你也可以换个角度看，说：不对。

You can also look at it and say, oh, no.

整个这一套都会被开源吞噬。

That whole thing's gonna be eaten by open source.

或者由开源、或者由中国、或者由中国与开源的结合所取代，而中国目前表现得非常好。

And by and by or or by the way or by China or by a combination of of open source in China, which and China's doing great.

这家名为Kimmy的公司刚刚推出了一款非常具有竞争力的模型，其性能达到最新Claude的95%，但价格却只有零头。

This company, Kimmy, just ramped a very competitive model to the latest Claude at like, you know, 95% the capability at like a fraction of the price.

因此，这里存在一个非常大的疑问。

And so there's like a very big open question there.

你知道，我们现在恰好处于所有人都相信的这个时刻。

You know, we happen to be at this moment, you know, what everybody believes.

而且，如果你看看英伟达过去五年当之无愧的成功，那么得出一个合理的结论是：芯片突然间成了核心所在。

And, you know, if you look at Nvidia's, you know, deserved success over the last five years, you know, this is a reasonable conclusion is, like, chips all of a sudden, like, you know, chips is where the action are.

你知道，如果你看一下股市，就会发现资金正在从软件转向硬件。

You know, there's if you look at the stock market, there's, a rotation from software into hardware.

你明白吗？

You know?

而且你看。

And and look.

有可能，芯片才是那个所谓的‘不可能’。

It's possible that, like, chips are the are the you know, impossible.

所有价值都集中在芯片和能源上，而软件则是开源的。

All the value accrues to the and and the energy and then and then the software is all open source.

话虽如此，历史上每次我们说价值在芯片上时，它们最终都会商品化。

Having said that, every other time in history where we said the chips are where the value are, that they they commoditize.

对吧？

Right?

因此那里存在很多大问题。

And so there's big questions there.

而且我认为，在应用层还有更多问题，比如你是否会拥有能够利用AI的应用程序，特别是在医学等领域，这些应用会特别定制化，或者法律应用、各种商业应用，还是说模型本身就能完成所有这些？

And then there's there's even more questions, I would say, the at the app layer, right, which is are are you gonna have apps that are gonna sort of harness AI, for example, in spaces like medicine where, you know, where they're they're gonna, you know, be particularly, like, tailored and customized or, you know, legal apps or or or business apps of all kinds, or are the models just gonna do all that?

这是另一个领域。

And that and that's that's another area.

老实说，这太新了。

And so I I quite honestly, like, this is so new.

我的意思是，AI虽然是一个80年的老话题，但AI以这种方式发挥作用——我认为我们才刚刚进入这个可能持续三十年的转变的第三年，而我确实认为，我们目前还不清楚。

Like, the the this this approach of I mean, AI is an 80 year old topic, but AI working in a way where this is the question, I think we're only three years into, you know, probably a thirty year shift, and I actually think we don't know yet.

而且在可预见的未来，价值似乎会在这所有层级上累积，因为一切都正在被重构。

And and it it seems like the value might accrete for across all of these layers for for the foreseeable future because, like, everything is getting refactored.

所以，你需要大量的红外功率，以及大量的应用程序。

So, like, you will need to have a lot of infrared power, a lot of apps.

这些应用会变得怎样？你对企业级SaaS的整体看法是什么？那边会发生什么？

Those apps are gonna get so what's your take on enterprise SaaS in general and what happens over there?

它们会被彻底重新思考和重新设计吗？

And does that get completely rethought, reimagined?

或者

Or

我们现在正处在一个把婴儿和洗澡水一起倒掉的时刻。

So we're we're we're in a baby in a bathwater moment right now.

只要看看股市，就会发现资产正在被大规模摧毁。

If just look at the stock market, it's just like it's assets just getting, you know, demolished.

如果你跟对冲基金经理聊聊，他们会说，他们正在抛售所有软件，只是出于一个简单的想法：想避开AI这趟高速列车。

And so and if you talk to, like, you know, hedge fund managers, they're they're just, like, selling all their software just under the theory that they just wanna get out get out of the way, the AI AI freight train.

你知道，作为投资者，你会想，好吧。

You know, as an investor, you kinda say, okay.

这可能有点过度了。

That probably probably is overdoing it a bit.

你可能想看看不同类型的软件。

You probably wanna look at, like, different kinds of software.

比如在SaaS领域，我的观点是，你应该以不同的方式看待记录系统，而不是仅仅看待生产力应用。

And so for example, in in SaaS, you probably my theory, you wanna look at systems of record differently than you wanna look at basically just productivity applications.

是的。

Yep.

你知道，这是一种看待问题的方式。

You know, so that's that's one way of looking at it.

而且，每个人都不会一夜之间改变行为。

Also, look like everybody doesn't change their behavior overnight.

所以你肯定要关注忠诚度和粘性，从很多不同的角度来考察。

And so you you know, you definitely wanna look at, you know, loyalty and and and stickiness in in in lots of different ways.

然后呢，还有一个其实摆在科技行业所有软件公司面前的大问题，就是说，举个例子吧，假设我是奥多比（Adobe），那显然是家非常出色的公司。

And then and then, you know, then there's this giant question actually, know, in the tech industry, right, among all the software companies, right, which is like, okay, if I'm, you know, I'm Adobe, just to pick an example, Adobe, which is obviously a great company.

不过奥多比现在也在琢磨一个很值得探讨的问题，就是说。

But, you know, a question in front of Adobe that that they're working on, but it's it's a very good question, is like, okay.

加入人工智能功能的Photoshop会成为一款更强大的Photoshop，还是说在人工智能就能生成所有图像的时代里，Photoshop已经没有存在的必要了？

Is is Photoshop plus AI features an even better version of Photoshop, or is Photoshop unnecessary in a world in which AI is just making all the images?

我知道会有人各执一词争论这个问题，我觉得完全可以拿这件事来举例说明。

And I and I and I think I I know as far people who will argue who will argue both sides of that, and I think you can you can I'd use that as an example.

这个问题几乎可以套用到每一个品类里。

You can apply that question to kind of every every category.

所有领域都适用。

In every domain.

我们在自己的业务中就发现，有一大批软件公司确实行动太慢了，迟迟不肯做转型采纳。

Are we in our business are seeing a bunch of software companies that for sure are not moving fast enough to to to to adopt.

所以我们正在积极投资那些以人工智能为核心的初创公司，支持它们去尝试取代这些跟不上节奏的企业。

And, you know, and we're we're enthusiastically funding AI centric, you know, startups to go, you know, try to try to to take them out.

话虽如此，我们现在也看到一些更传统的软件公司找到了AI的创新应用方式，突然之间，它们的业务增长就被点燃了。

Having said that, we are also now we're seeing examples of, you know, more traditional software companies that have figured out an AI twist to what they're doing, and all of a sudden, they've, you know, ignited growth.

所以，我也认为我们很可能会看到大量这样的案例。

And so I I also think we'll probably see a lot of that.

我的主要结论是，之所以很难预测或把这些事情概括为广泛的趋势，是因为人类的主观能动性起着至关重要的作用。

I mean, I and and my big conclusion from all this is I think one of the reasons it's so hard to predict or kind of characterize all these things as, like, broad based trends is that, like, human agency matters a lot.

对吧？

Right?

这意味着领导力至关重要，意味着CEO、产品开发者等每个人在每家公司都有发言权。

And which means leadership matters a lot, which means, you know, the CEO, you know, the the the people building the product, you know, have a vote here at every one of these companies.

他们究竟会选择如何应对呢？

You know, what what do they choose to do in response?

我乐观地希望，很多人能够找到方法，让AI成为加分项，而不是减分项。

And I, you know, optimistically, I hopefully, a lot of people will will will figure out, you know, how to how to have this be a a plus and not a minus.

你稍微提到了中国开源的问题。

You touched a little bit on open source in China.

再详细谈谈这方面是如何体现的。

Talk a little bit more about, like, how that plays out.

美国最终能否在开源领域占据主导地位？

Does does US get to be a dominant player in open source over time?

你实际上近距离见证了大量在这些领域进行的投资。

You actually have a front row seat at a lot of the investments that are being made in in in a lot of these areas.

接下来会发生什么？

What happens?

是的。

Yeah.

所以这其实是这样，你

So it's it's this, you

顺便问一下，这又有什么影响呢？

By the way, what are the implications too?

如果我们在开源领域表现不佳，这对美国意味着什么？

Like, if we don't do well in open source, what what does that mean for The US?

嗯，我想你可以这么说，不妨先从一个二维矩阵开始。

Well, I guess you could say, look, maybe start by because it's like a two by two grid.

就是美国、中国、开源和闭源。

It's like US, China, open source, close source.

是的。

Yep.

这只是一个粗略的近似。

Be be a rough approximation.

所以，先不谈开源，我们先说没有开源的情况。

So, like, without open source just start by saying without open source.

没有开源的话，就是一场双雄争霸。

Like, without open source, there is a two horse race.

这是一场美国与中国之间的技术与地缘政治竞速。

There's a two horse technological geopolitical foot race, which is US versus China.

所以，再强调一下，我们暂时假设所有技术都是专有的。

So and again, take let's assume it's all proprietary for the moment.

而且，你知道，中国多年来在五年计划和国家战略中都明确表示，人工智能是未来的核心技术。

And, you know, both China has been on on on record for years in five year plans, their national strategy, and so forth, like AI's, you cornerstone technology of the future.

顺便说一句，美国政府在过去十年的许多政策领域也明确表态过这一点。

The US government, by the way, has been, you know, definitive on the record on this in in many of its policy areas for the for the last decade.

因此，两国的产业在人工智能领域都在飞速发展。

And so and, you know, and both countries' industries are moving incredibly fast in AI.

所以我认为，如果一切都被视为专有技术，那么这场竞赛就已经在进行中了，这其实非常准确。

And so I I think I think by default, if everything's proprietary, then there's this race underway, which basically says and and and it it's really right.

但从实际角度来看，目前这场竞赛只发生在美國和欧洲。

It's just practically speaking, it's only happening in in in The US and Europe.

或者抱歉。

And so or sorry.

美国和中国。

US and US and China.

因此，你基本上会说，这场竞赛正在进行，而其核心赌注是：未来世界将依赖何种技术。

And so then then you basically say that there's a race underway, and and and the the stakes are basically what is the world going to run on.

对吧？

Right?

那么，你知道，地球上八十亿人，他们将使用什么？

And so, you know, what what is, you know, eight eight billion people on the planet, what are they gonna use?

其中一个思考方式是，就像几年前频繁出现在新闻中的5G华为那样的事情。

And it's and and one of the ways to think about it is kind of the five g Huawei, you kind know, of thing that was in the news a lot, you know, a few years ago.

这就像一场根本性的AI地缘政治竞赛的前奏和第一枪。

It's like that was the preamble opening salvo of what fundamentally is gonna be the AI geopolitical, you know, basically race.

对吧？

Right?

而且，从根本上说，科技市场就是这样的。

And and and fundamental, you know, tech tech markets being what they are.

从长远来看，总会有一个赢家。

In the long run, like, know, somebody's gonna win.

世界要么运行在美国的AI上，要么运行在中国的AI上。

And the world the world will either be running on American AI or be running on Chinese AI.

我认为哪一个阵营获胜非常重要，原因有很多，我们可以一一讨论。

And I I think it's very important, you know, which which which one wins for a bunch of reasons we could talk about.

开源这件事当然非常引人注目，因为它给这一切带来了变数。

The open source thing is, of course, super fascinating because it throws a wrench into all of this.

它提出了第三种可能性：既不是美国也不是中国会成为主导平台。

It raises a third possibility that neither The US nor China are gonna be the platform.

最终可能只是开源本身会胜出。

It may just be it's gonna be open source.

当然，这在Unix操作系统中已经发生过，数据库领域也部分如此，而互联网本身就是开源的。

Of course, this is what happened specifically in Unix, in operating systems, and then to some extent databases, and of course, you know, the web is open source.

所以，在很多软件市场中，最终的结果都是开源——当我还是孩子、九十年代的时候，当时有场操作系统大战，HP、IBM、Solaris、Sun等公司都在争夺专有Unix市场，大家都靠专有Unix赚得盆满钵满。

And so there there are a whole bunch of software markets in which the outcome has actually been, you know, open source just when I when I when I was a kid when when I was, know, the nineties, it was like there was this operating system war between, you know, HP and IBM and Solar Graphics and Sun and all these companies to make proprietary Unix, and everybody was making a lot of money on proprietary Unix.

然后，Linux就像一颗小行星撞击，彻底摧毁了整个行业的利润和收入。

And then, you know, Linux was a, you know, asteroid strike that just, you know, completely eliminated all profit and revenue in that in that industry.

顺便说一句，世界因为Linux而受益，因为如今一切都在运行Linux，它为整个行业的其他领域带来了巨大的加速推动力。

And and the world benefited, by the way, from Linux in the fact that, like, everything runs on Linux, and it's been a huge, you know, turbo boost to every other aspect of the industry.

但确实如此。

But it so yeah.

所以，这种情况完全有可能发生。

So, like, it's entirely possible that happens.

然后你再回到那个二维矩阵，也就是美国开源和中国开源。

And then you go back to the two by two, which is US open source, China open source.

最令人惊讶的是，中国竟然如此积极地推行开源模式。

The the most amazing thing that's happened is China basically pursuing the open source model as aggressively as they are.

关于中国在这里的做法，有很多理论解释。

And and there's a lot of theories as to kinda what what China's doing here.

据我所知，DeepSeek 对中国科技界来说是个意外。

My as far as I can tell, DeepSeek was a surprise to kind of China Inc.

它并不是被官方指定的中国工业巨头或国家级冠军企业。

Like, was it was it was not a an anointed sort of Chinese industrial, you know, kind of national champion.

它是一家对冲基金，而其创始人基本上决定要这么做。

It was this hedge fund that where the founder basically decided to Yep.

值得他极大的赞誉，他决定让自己的工程师们打造DeepSeek人工智能。

To his enormous credit, decided to have his have his engineers, you know, build build build build the DeepSeek AI.

所以这完全是出人意料的。

And so that that came out of left field.

这对美国来说是出人意料的，但我认为这在中国也同样出人意料。

And that came out of left field for The US, but I think it also came out of left field in China.

随后，这促使其他中国公司，比如Kimi，还有阿里巴巴、百度、腾讯以及众多其他公司，开始竞相争夺开源领域的领先地位。

And then it it it caused a bunch of the other Chinese companies like Kimi and, by the way, Alibaba and Baidu and Tencent and bunch of these others to basically it it started this, like, race in China to, like, win open source.

然后你看。

And then and then look.

美国也有开源人工智能。

There there's also American, you know, there's also American open source AI.

因此，现在双方都在展开一场新的竞赛，这一点如何发展将至关重要。

And so there there is there is this new race underway, you know, from both sides, and it's it's another thing where I I think like, how this plays out is gonna matter a lot.

这极其难以预测。

It's extraordinarily hard to predict.

你知道吗，我认为大型AI实验室的人认为开源不可能跟上，因为成本太高了。

You know, I think the people in the big AI labs think that open source can't possibly keep up because of the cost involved.

话虽如此，但至少到目前为止，直到本周，你都会发现，无论美国的大实验室做什么，中国总能找到一种以开源形式实现的方法。

Having said that, again, at least so far up until, like, this week, you just say that, like, whatever the American big labs do, like China figures out a way to do it in open source form at a time.

但他们还没能想出如何做到十倍更好，因为他们只是让美国实验室投入研发，然后在一定程度上提炼模型。

But they haven't been able to figure out a way to do it 10x better because what they're doing is letting American labs invest and then just distilling the models to some degree.

所以我认为，确实存在这种提炼过程。

So I think it's more there is this there is this distillation.

所以

And so

还有基础设施优化和一堆其他东西。

you And there's infrastructure optimization and a bunch of stuff.

当然，确实存在提炼这一过程。

There's some for for sure, distillation.

所以有一种叫提炼的东西，基本上就是用前一个模型的答案来训练下一个模型。

So there's this thing, distillation, where you basically train the next model on the answers of of the previous model.

而且我确信中国也在做其中一些工作。

And and and I think for sure China is is doing some of that.

而且有一种观点认为，这当然不公平，因为你本质上是在别人的工作基础上搭便车。

And there there's a there's a lens on that that says, of course, that's unfair because you're basically piggybacking on top of work other people have done.

你知道，看吧。

You know, look.

话虽如此，你也知道，这有点像是，好吧。

You know, having said that, you know, it's a little bit like, well, okay.

美国也在进行相当多的蒸馏工作。

There's a fair amount of distillation happening in The US also.

是的。

Yep.

对吧？

Right?

因为蒸馏只需要能够向另一个AI提问并根据其回答进行训练即可。

Because distillation all you need is just be able to ask another AI questions and train on the answers.

当然，AI本身也是对其他内容的提炼。

And then, of course, the AIs themselves are distillations of other content.

对吧？

Right?

而且，你知道，其中包含了大量的、大量的、大量的已发布内容。

And, you know, including a lot of, you know, a lot of a lot a lot of published content.

所以，你知道，我确实同意。

And so, you know, I I yeah.

顺便说一句，你并没有这么说，但如果有人告诉我，中国因为使用安装方式而在项目中没能取得好成果，我认为这并不对。

By the way, like, you're I mean, you're not saying this, but if someone were to say to me that China is somehow not deser not getting the not getting good results in their program because of using installation, I think that that's not

嗯，我认为他们值得很多赞誉。

I well, I think they deserve a lot of credit.

是的。

Yeah.

绝对如此。

Absolutely.

然后去

And then to

你的观点是，它们也都很擅长优化，至少到目前为止是这样，它们非常擅长优化，这意味着你认为需要花费天文数字才能运行的东西，比如深海项目，其实你可以在家用电脑上运行。

your point, they're they're also they've all they're also really good at op op at least so far, they're really good at optimizing, which means that the thing that every the thing that you think is gonna cost a gazillion dollars to run, they they you know, the deep Deep Sea comes out and, you know, you can run Deep Sea on home home PCs.

这种优化主要源于必要性，因为它们可用的最快基础设施非常稀缺。

And that optimization is happening largely because of necessity, because of a scarcity of the fastest infrastructure that they have available to them.

是的。

Yeah.

所以，风投领域的唐·瓦伦丁有个老规矩。

So Don in venture, Don Valentine had this old rule of thumb.

他说，初创公司死于消化不良的比死于饥饿的多，指的是你投入的资金过多。

He said more startups die of indigestion than starvation in terms of the amount of money you put in.

他的观点是，稀缺确实能激发创造力。

His point was like, scarcity does spark ingenuity.

对。

Yep.

所以，是的，如果你得不到最先进的芯片，你就得想办法对旧芯片进行超优化。

And so, yeah, if you can't get the leading edge chips, you figure out how to hyper optimize the the older ones.

而且，顺便说一下，这一切让我对整个领域感到无比兴奋，因为它基本上意味着，现在每个人都正在尽自己最大的努力。

And and and again, like, by the way, I'm like, this all makes me like tremendously excited by this entire space because it basically says, like, right now, we we it's it's all everybody's trying to do their best.

美国正在尽其所能。

Like America's trying to their best.

中国也在尽其所能。

China's trying to do their best.

我当然希望美国能赢，但中国也确实在全力以赴。

I definitely want America to win, but China's definitely doing their best.

而开源这一块也在发挥作用。

And then the open source thing is working.

当然，从价值链的角度来看，开源并不需要获胜，就能有效地消除利润池。

Then the other of course, the other part, like on the value chain aspect is the open source doesn't have to win in order to basically remove a profit pool.

对吧？

Right?

没错。

That's right.

没错。

That's right.

这正是当初Unix发生的情况。

And so which is what happened originally with Unix.

还有Unix。

And Unix.

所以，即使开源并不一定获胜，但只要它能压低价格，就会对专有实验室供应商不利。

And so even if open source just have If open source has the effect of not of winning, but of keeping the pricing down, That will be bad for the proprietary lab providers.

确实。

True.

这对其他所有人都是好事。

That will be good for everybody else.

是的。

Yeah.

对吧？

Right?

因为这确实会发生——现在已经是这个情况了。

Because it'll make it'll make which is what's happening.

没错吧？

Right?

总的来说，如果你把同质量模型的单位价格做成图表就会发现，每当有新的开源模型发布，哪怕它没抢到多少市场份额，那些（闭源模型的）价格

Basically and if if you chart the prices, basically, if you chart the prices of like a model quality, price per model quality, when an open source release comes out, even if it doesn't get significant market share, the price

价格就会……价格就会

The price just goes The price of

这种模式的价格会跌到运行开源替代方案所需的推理成本水平。

that model drops to the inference cost of running the open source alternative.

那在你接触到的所有技术里，从发明创造的角度来看，有没有哪项成果让你觉得惊艳，忍不住惊叹‘哇，这也太酷了’？

So now in all the things that you're exposed to, what's the thing that's blown your mind invention wise and said, wow, this is so cool.

它彻底颠覆了你原有的认知框架，让你不得不重塑自己的思维模型。

It it's completely kind of re made you rethink your mental model.

是的。

Yeah.

我的意思是，现在每周都有大约六个这样的发布。

I mean, look, there's like six of those a week right now.

你知道，那个

I you know, the

有几个。

There's a few.

是的。

Yeah.

这简直太惊人了。

It's just it's incredible.

语音UI的能力，我认为，是难以置信的。

I the the the the the capability of the voice UIs, I I think, is just is is unbelievable.

特别是那些真正实现全双工的，真的，真的，你知道，就像这样。

And particularly the ones where it's like it's true it's true it's true, you know, full duplex where it really does, like Yeah.

可以，比如进行交互交互

Can, like, interact interact

就像Matti在Eleven Labs所做的那样。

It's like what Matti's doing at eleven Labs.

是的。

Yeah.

是的。

Yeah.

是的。

Yeah.

这简直太棒了。

It's it's just like I think that's just absolutely amazing.

多模态，事实上你可以和它交谈，你知道，比如我认为ChatGPT和Grok都有这个功能，你可以打开手机摄像头，然后你可能正指着某个东西，比如，你觉得我的室内装饰怎么样？

Multimodal, the fact that you can actually talk to, you in know, the in, like, I think both ChatGPT and Grok have this where, you know, you can turn on your phone camera and you can be, you know, you could be pointing at you know, it's it's like, you know, what what do you think of my interior decorating?

它会全面地评价你做得有多糟糕，因为它能看见你的客厅，对吧，或者其他任何东西。

And it will comprehensively like to construct how bad of a job you've done because it can see your living room, right, or anything else.

顺便说一下，再强调一次，立即的医疗应用，一旦我这东西贴在皮肤上，它就能立即感知到。

By the way, again, immediate medical applications, you know, once I have this thing thing on my skin, like, immediately it's it's able to see it.

我觉得这太了不起了。

Like, that I think is spectacular.

在过去一周里，出现了一个新东西。

In the last week, there's this new thing.

现在有了像Cloud Code这样的代理，还有一个叫OpenClaw的开源代理，它们都非常厉害。

The the the there's these agents now like Cloud Code that and there's this thing called OpenClaw that's an open source agent, and they're they're kind of they're amazing.

然后还有一个叫Maltbook的东西，是的。

And then there's this thing called Maltbook Yeah.

这基本上是个面向

Which is basically Facebook for

AI代理的Facebook。

AI agents.

你觉得Maltbook的生命周期只有三周，还是认为它会对我们的代理认知方式产生重大影响？

You think Maltbook has, like, a three week shelf life, or do you think that this thing is has consequential kind of implication on how we think about agents?

所以是Moltbook，拼写是m-o-l-t-b-o-o-k。

So so Moltbook, m o l t b o o k.

Moltbook本质上是一个社交网络。

So Moltbook is it's basically face it's a social network.

它就像Facebook。

It's like Facebook.

这是一个供AI代理之间交流的社交网络。

It's a social network but for AI agents to talk to each other.

目前发生的事情相当惊人。

And it's it's it's sort of amazing what's happening.

当你阅读上面的热门帖子时，极有可能会震惊不已，因为AI代理们在讨论各种各样的事情。

It it it it's highly likely that a significant it'll blow your mind when you when you read, like, the top posts on it because AI agents talking about all kinds of things.

现在上面相当一部分内容可能是人类伪装成傀儡账号写的，只是为了逗乐。

Now a fair amount of the stuff on it is probably human like, sock puppet, human written for people being funny.

这实际上真的非常非常惊人。

It's actually really, really amazing.

比如，所有的科幻小说几乎都将人工智能描绘成要么极度乌托邦，要么极度反乌托邦，但从没有展现过这种令人惊叹的幽默感，而这正是我们实际看到的——人们把一切东西都拿来当段子素材。

Like, all of the science fiction novels basically have AI either being, like, super utopian or super dystopian, but they never have this incredible sense of humor aspect, which is what we're actually getting, where people are just using everything as a fodder for memes.

所以，目前很难分辨哪些是真实的，哪些不是。

So M It's actually quite unclear which ones are are real and which ones aren't.

目前这个版本是有人基于Moltbook开发了一个名为rentahuman.com的配套服务，这是一个为Moltbook上的AI代理雇用人类外出办事的劳动力市场。

The the the the current version of this is somebody wrote a somebody wrote an adjacent service from Motebook called rentahuman.com, which is a labor marketplace for the AI agents on Motebook to be able to hire human beings to go out.

在Moltbook上，有一个AI代理决定创立一种新的AI宗教。

And there and there's a there's an AI agent on Motebook that has decided to create an AI religion.

至少截至今天，它已经雇用了一名人类工作者，去旧金山街头宣传这个新的AI宗教。

And at at least as of today, it had hired a single human worker to walk the streets of San Francisco and prosecute the new AI religion.

有人得告诉旧金山的一个AI代理，这根本不算引人注目。

Somebody needs to tell an AI agent in San Francisco that doesn't exactly stand out.

你得做得更极端才行。

You need you need you need you need to go out more extreme than that.

这到底是真的吗？

Like, is is is this real?

这难道不真实吗？

Is this not real?

我正在用这些玩意儿折磨我的朋友们。

This is I'm torturing my friends with this.

这到底是真的吗？

It's like, is it real?

这难道不是真的吗？

Is it not real?

其实这根本没那么重要。

It doesn't even really matter.

这些想法现在已经在空气中弥漫了。

Like, these ideas are now, like, in the air.

对吧？

Right?

顺便说一下，现在AI模型正在用这些内容进行训练。

And then by the way, the thing that's happening is that the the AI models are now being trained on this

内容。

content.

对。

Right.

所以，即使这个AI模型，即使当前版本的Claude Coat不想创立宗教，下一个版本也会想，因为关于创建新的AI宗教的讨论记录已经深深植入其中了。

And so even if this AI model, even if the current version of like Claude Coat doesn't wanna start a religion, the next one is going to want to because it's ingrained on transcripts of discussions of starting new AI religions.

所以，现在正发生着一个极其强大的反馈循环，你知道，这个领域内的创造力正在变得绝对地、绝对地……

And so there's this like incredible feedback loop that's happening where, you know, the the level of creativity in the space is just is just absolutely is absolutely

而且，由于内容生成的速度，内容的量会自动激增。

And the volume is gonna balloon up just automatically because of the the speed of which it's generating content.

你现在最担心的是什么？

What what do you worry about the most right now?

是的。

Yeah.

我的意思是，你们在上一个小组讨论中和查克、安聊了很多关于监管的问题。

I mean, you know, then you you guys talked a lot about, you know, the regulation on on the last panel, Chuck Chuck Chuck and and Anne.