AlphaFold, AGI, and the Scientific Method

00:00

It's hard for us humans to make any kind of clean predictions about highly nonlinear dynamical systems.

我们人类很难对高度非线性的动力系统做出任何清晰的预测。

00:05

But again to your point, we might be very surprised what classical learning systems might be able to do about even fluid. >> Yes, exactly.

但回到你的观点，经典学习系统在流体问题上能做到什么程度，可能会让我们非常惊讶。

00:13

I mean fluid dynamics, Navia Stokes equations, these are traditionally thought of as very very difficult intractable problems to do on classical systems.

>> 是的，完全同意。

00:21

They take enormous amounts of compute, you know, weather prediction systems, you know, these kind of things all involve fluid dynamics

流体动力学、Navier-Stokes方程，这些传统上被认为是非常非常困难、在经典系统上难以处理的问题。

00:28

calculations.

计算。

00:29

But again, if you look at something like VO, our video generation model, it can model liquids quite well, surprisingly well, and materials, specular lighting.

但话说回来，像我们的视频生成模型VO，它处理液体的效果出奇地好，还有材质和镜面光照。

00:38

I love the ones where, you know, there's there's people have generated videos where there's like clear liquids going through hydraulic presses and then being squeezed out.

我最喜欢那些人们生成的视频，比如透明液体被液压机挤压出来的那种。

00:47

I I used to write uh physics engines and graphics engines and in my early days in gaming.

我早年做游戏的时候写过物理引擎和图形引擎，我知道要写出能模拟这些的程序有多痛苦。

00:52

And I know it's just so painstakingly hard to build programs

然而这些系统却能从看YouTube视频中逆向工程出来。

00:56

that can do that.

能做到这些。

00:57

And yet somehow these systems are, you know, reverse engineering from just watching YouTube videos.

但不知为何，这些系统仅仅通过观看YouTube视频就能进行逆向工程。

01:03

So presumably what's happening is it's extracting some underlying structure around how these materials behave.

所以很可能发生的情况是，它正在提取关于这些材料如何表现的某种底层结构。

01:10

So perhaps there is some kind of lower dimensional manifold that can be learned if we actually fully understood what's going on under the hood.

因此，也许存在某种低维流形，如果我们真正理解了背后的原理，就可以学习到它。

01:18

That's maybe, you know, maybe true of most of

这可能也适用于大多数现实世界。

01:22

reality.

现实。

01:22

The following is a conversation with Demis Hassabis, his second time on the podcast.

以下是和Demis Hassabis的对话，这是他第二次来播客。

01:29

He is the leader of Google Deep Mind and is now a Nobel Prize winner.

他是Google Deep Mind的负责人，现在也是诺贝尔奖得主。

01:35

Demis is one of the most brilliant and fascinating minds in the world today, working on understanding and building intelligence and exploring the big

Demis是当今世界上最聪明、最迷人的头脑之一，致力于理解和构建智能，并探索宇宙的重大奥秘。

01:47

mysteries of our universe.

宇宙的奥秘。

01:49

This was truly an honor and a pleasure for me.

这对我来说真的是一种荣幸和快乐。

01:53

This is the Lex Freedman podcast.

这里是Lex Freedman播客。

01:56

To support it, please check out our sponsors in the description and consider subscribing to this channel.

如果你愿意支持我们，请查看描述中的赞助商，并考虑订阅这个频道。

02:04

And now, dear friends, here's Deus Hassavas.

现在，亲爱的朋友们，有请Deus Hassavas。

02:07

In your Nobel Prize lecture, you propose what I think is a super interesting

在你的诺贝尔奖演讲中，你提出了一个我觉得非常有趣的猜想，原话是：“任何可以在自然界中生成或发现的模式，都可以被经典学习算法高效地发现和建模。

02:13

conjecture that quote any pattern that can be generated or found in nature can be efficiently discovered and modeled by a classical learning algorithm.

猜想是：任何能在自然界中生成或发现的模式，都可以被经典学习算法高效地发现并建模。

02:22

What kind of patterns of systems might be included in that?

那这包括哪些系统或模式呢？

02:26

Biology, chemistry, physics, maybe cosmology, >> neuroscience.

生物学、化学、物理学，也许还有宇宙学，>> 神经科学。

02:30

What what are we talking about? >> Sure.

我们到底在说什么？

02:32

Well, look, I I felt that it's sort of a tradition, I think, of Nobel Prize lectures that you're supposed to

>> 好。

02:39

be a little bit provocative and I wanted to follow that tradition.

稍微带点挑衅性，我也想延续这个传统。

02:42

What I was talking about there is if you take a step back and you look at um all the work that we've done especially with the alpha x projects so I'm thinking alpho of course alpha fold what they really are is we're building models of very combinatorily highdimensional spaces that you know if you tried to brute force a solution find the best move and go or find the the exact shape of a protein and if you enumerated all the

我当时说的意思是，如果你退一步看，看看我们做的所有工作，尤其是alpha x系列项目，比如alpha go、alpha fold，本质上我们是在构建那些组合维度极高的空间模型。

03:05

possibilities you there wouldn't be enough time in the in the you know the time of the universe.

可能性是，就算用尽宇宙的时间都不够。

03:11

So you have to do something much smarter and what we did in both cases was build models of those environments.

所以你必须要做更聪明的事情，而我们在两种情况下做的都是构建这些环境的模型。

03:16

Um and that guided the search in a in a smart way and that makes it tractable.

嗯，这能以更聪明的方式引导搜索，让它变得可行。

03:21

So if you think about protein folding which is obviously a natural system you know why should that be possible?

所以如果你想想蛋白质折叠，这显然是一个自然系统，你知道为什么这可能吗？

03:27

How does physics do that?

物理是怎么做到的？

03:28

You know proteins fold in milliseconds in our bodies.

蛋白质在我们体内几毫秒内就折叠好了。

03:31

So somehow

所以某种程度上，粗略地说，任何能被进化的东西都能被高效建模。

03:32

physics solves this problem that we've now also solved computationally.

物理学解决了这个问题，我们现在也用计算的方式解决了。

03:36

And I think the reason that's possible is that in nature, natural systems have structure because they were subject to evolutionary processes that that shape them.

我觉得这之所以可能，是因为在自然界中，自然系统是有结构的，因为它们经历了塑造它们的进化过程。

03:47

And if that's true, then you can maybe learn uh uh what that structure is.

如果真是这样，那你或许就能学习到那种结构是什么。

03:51

So this perspective, I think, is really interesting one.

所以这个视角，我觉得真的很有意思。

03:55

You've hinted it

你之前也暗示过这一点，粗略来说就是，任何能被进化出来的东西都可以被高效建模。

03:56

at it, which is almost like uh crudely stated.

这事儿吧，粗略来说就是——任何能被演化的事物，都可以被高效建模。

03:59

Anything that can be evolved can be efficiently modeled.

我觉得这话有点道理。

04:02

Think there's some truth to that.

对，我有时候管它叫“最稳定者生存”之类的，因为你看，生命当然有演化，但你要是想想地质时间尺度，山的形状也是被风化过程塑造出来的，对吧，几千年下来。

04:04

Yeah, I sometimes call it survival of the stablest or something like that because you know it's it's of course there's evolution for life uh living things but there's also you know if you think about geological time so the shape of mountains that's been shaped by weathering processes right over thousands of years but then you can even

但甚至这也不是随机的模式。

04:23

take it cosmological the orbits of planets the um shapes of asteroids these have all been survived kind of processes that have acted on them many many times so if that's true then there should some sort of pattern um that you can kind of reverse learn and uh a kind of manifold really that helps you uh uh search to the right solution to the right shape um and actually allow you to predict things about it uh in an efficient way because

从宇宙尺度来看，行星的轨道、小行星的形状，这些都经历了无数次作用过程的筛选。

04:50

it's not a random pattern right so um it may not be possible for for man-made things or abstract things like factorizing large numbers because unless there's patterns in the number space which there might be but if there's not and it's uniform then there's no pattern to learn there's no model to learn that will help you search.

这不是随机模式，对吧。

05:06

So you have to do brute force.

所以，对于人造的东西或者抽象的东西，比如分解大数，可能就不太可行，因为除非数字空间里有某种模式——也许确实有——但如果没有，而且是均匀分布的话，那就没有可学习的模式，也没有能帮你搜索的模型可以学。

05:07

So in that case you you know you maybe need a quantum computer something like this.

那你只能暴力破解。

05:11

But in most things in nature that we're interested in uh are not like that.

这种情况下，你可能就需要量子计算机之类的东西。

05:15

They have structure

但在我们感兴趣的大多数自然事物中，并不是这样的。

05:16

um that evolved for a reason and survived over time.

嗯，那个东西是进化出来的，有它的道理，而且一直存活了下来。

05:19

And if that's true I think that's potentially learnable by a neural network. >> It's like nature is doing a search process and it's so fascinating that it's in that search process is creating systems that could be efficiently modeled.

如果真是这样，我觉得神经网络是有可能学会它的。

05:32

That's right.

>> 就像大自然在进行一种搜索过程，而且特别神奇的是，这个搜索过程本身创造出来的系统，竟然可以被高效地建模。

05:33

Yeah. >> So interesting. >> So they can be efficiently rediscovered or recovered um because nature is not random, right?

没错。

05:39

These everything that we see around us, including like the

是啊。

05:43

elements that are more stable, all of those things, they're subject to um some kind of selection process pressure.

那些更稳定的元素，它们都会受到某种选择过程的压力。

05:50

Do you think because you're also a fan of theoretical computer science and complexity, do you think we can come up with a kind of complexity class like a complexity zoo type of class where maybe it's the set of learnable systems, the set of learnable natural systems, lns. >> Yeah,

你觉得——因为你也喜欢理论计算机科学和复杂性——我们能不能提出一种复杂性类，就像复杂性动物园那种，比如“可学习系统集”，或者“可学习自然系统集”，LNS？

06:10

>> this is a deis new class of systems that could be actually learnable by classical systems in this kind of way.

>> 这是一个全新的系统类别，可以用经典系统以这种方式真正学会。

06:16

Natural systems that can be uh modeled efficiently.

那些可以被高效建模的自然系统。

06:19

Yeah, I mean I' I've always been fascinated by the P= MP question and what is modelable by classical systems I non-quantum systems you know cheuring machines in effect and that's exactly what I'm working on actually in kind of my few moments of spare time

是啊，我一直对P=MP这个问题很着迷，也一直好奇经典系统——非量子系统，其实就是图灵机——到底能建模什么。

06:35

with a few colleagues about is should there be you know maybe a new class of problem that is solvable by this type of neural network process and kind of mapped on to these natural systems so you know the things that exist in physics and have structure.

和几个同事讨论的是，是不是应该有一种新的问题类别，可以通过这种神经网络过程来解决，并且某种程度上映射到这些自然系统上，就是那些存在于物理学中、有结构的东西。

06:48

So I think that could be a very interesting uh new way of thinking about it.

所以我觉得这可能是一个非常有趣的思考方式。

06:52

And it sort of fits with the way I think about physics in general which is that you know I think information is primary.

而且这跟我对物理学的整体看法也挺吻合的，就是我认为信息是首要的。

06:58

Information is the most sort of fundamental unit of the universe more

信息是宇宙中最基本的单位，甚至更基础。

07:02

fundamental than energy and matter.

比能量和物质更根本。

07:04

I think they can all be converted into each other but I think of the universe as a kind of informationational system. >> So when you think of the universe as anformational system then the P= NP question is a is a physics question. >> That's right.

我认为它们可以相互转化，但我把宇宙看作一种信息系统。

07:17

And it's a question that can help us actually solve the entirety of this whole thing going on. >> Yeah, I think it's one of the most uh fundamental questions actually if you think of physics asformational uh and and the answer to that I think is

>> 所以当你把宇宙看作信息系统时，P=NP问题就成了一个物理学问题。

07:30

going to be you know very enlightening. more specific to the PNNP question.

会非常有启发性。

07:34

This again, some of the stuff we're saying is kind of crazy right now.

更具体到PNNP这个问题。

07:38

Just like the Christian Edinson Nobel Prize speech controversial thing that he said sounded crazy and then you went and got a Nobel Prize for this with John Jumper solved the problem.

再说一次，我们刚才说的有些东西听起来挺疯狂的。

07:49

So, let me let me just stick to the P equals MP.

就像Christian Edinson那个诺贝尔奖演讲里说的有争议的话，当时听起来也很疯狂，结果后来你和John Jumper真的拿了诺贝尔奖，解决了那个问题。

07:52

Do you think there's something

所以，我还是先回到P等于MP这个问题上。

07:54

in this thing we're talking about that could be shown if you can do something like uh polomial time or constant time compute ahead of time and construct this gigantic model then you can solve some of these extremely difficult problems in a theoretical computer science kind of way. >> Yeah, I think that there are actually a huge class of problems that could be couched in this way. the way we did

在我们讨论的这件事里，如果能提前做某种多项式时间或常数时间的计算，然后构建出一个巨大的模型，那就能用理论计算机科学的方式解决一些极其困难的问题。

08:19

alpha go and the way we did alpha fold where you know you you model what the dynamics of the system is the the the the properties of that system the environment that you're trying to understand and then that makes the search for the solution or the prediction of the next step efficient basically polomial time so tractable by a uh classical system uh which a neural network is it runs on normal computers

Alpha Go和Alpha Fold的做法，就是你先建模系统的动态、系统的属性、你想理解的那个环境，然后这样就能高效地搜索解决方案或预测下一步，基本上就是多项式时间，所以经典系统就能处理——神经网络就是跑在普通计算机上的。

08:46

right classical computers uh chewing machines in effect and um I think it's one of the most interesting questions there is is how far can that paradigm go?

经典计算机，其实就是图灵机，而且我认为这是最有趣的问题之一：这个范式到底能走多远？

08:54

You know, I think we've proven uh and the AI community in general that classical systems, cheuring machines can go a lot further than we previously thought.

你知道，我觉得我们已经证明了——整个AI社区也证明了——经典系统，也就是图灵机，能做的事情远远超出了我们以前的想象。

09:03

You know, they can do things like model the structures of proteins and play go to better than world champion level.

它们可以建模蛋白质结构，可以在围棋上达到超越世界冠军的水平。

09:10

And uh you know a lot of people would have thought maybe 10 20

而很多人可能在十年前、二十年前都不会想到

09:13

years ago that was decades away or maybe you would need some sort of quantum machines to to quantum systems to be able to do things like protein folding.

对，经典计算机其实就是图灵机。

09:22

And so I think we haven't really uh even sort of scratched the surface yet of what uh classical systems socalled uh uh could do.

我觉得最有趣的问题之一就是，这个范式到底能走多远？

09:30

And of course AGI being built on a on a neural network system on top of a neural network system on top of a classical computer would be the ultimate

你看，我们已经证明了，整个AI社区也证明了，经典系统、图灵机能做的比我们以前想象的要多得多。

09:39

expression of that.

在我们讨论的这个东西里，如果能提前做多项式时间或常数时间的计算，构建出这个巨大的模型，那就可以用理论计算机科学的方式解决一些极其困难的问题。

09:40

And I think the limit you know the the what what the bounds of that kind of system what it can do it's a very interesting question and and and directly speaks to the P equals MP question.

>> 对，我觉得实际上有一大类问题都可以用这种方式来表述。

09:50

What do you think again hypothetical might be outside of this maybe emergent phenomena like if you look at cellular automa some of the you have extremely simple systems and then some complexity emerges yes >> maybe that would be outside or even would you guess even that might be

就像我们做的那样。

10:06

amendable >> to efficient modeling by a classical machine >> yeah I think those systems would be right on the boundary right so um I think most emergent systems cellular automter things that could be modelable by a classical system.

对，我觉得那些系统正好就在边界上。

10:20

You just sort of do a forward simulation of it and it probably be efficient enough.

大多数涌现系统，比如细胞自动机这类东西，应该是可以被经典系统建模的。

10:25

Um, of course there's the question of things like chaotic systems where the initial conditions really matter and then you get to some, you know, uncorrelated end

你只要做个正向模拟，可能效率就足够了。

10:34

state those could be difficult to model.

这些可能很难建模。

10:37

So I think these are kind of the open questions.

所以我觉得这些算是开放性问题。

10:39

But I think when you step back and look at what we've done with the systems and the and the problems that we've solved and then you look at things like V3 on like video generation sort of rendering physics and lighting and things like that, you know, really core fundamental things in physics.

但当你退一步，看看我们用这些系统做了什么，解决了哪些问题，再看看像V3这样的视频生成，它能够渲染物理、光照这些东西，你知道，物理学里非常核心的基础内容。

10:55

Um it's pretty interesting.

嗯，这挺有意思的。

10:56

I think it's telling us something quite fundamental

在我看来，这其实在告诉我们一些关于宇宙结构非常根本的东西。

10:59

about how the universe is structured in my opinion.

对，我觉得我们可能会不断被经典计算机能建模的东西所惊讶。

11:03

Um so you know in a way that's what I want to build AGI for is to help uh us uh as scientists answer these questions uh like p=mp. >> Yeah I think we might be continuously surprised about what is modelable by classical computers.

我是说，AlphaFold 3在相互作用方面的进展，能在那个方向上取得任何突破都挺让人惊讶的。

11:18

I mean alpha fold 3 on the interaction side is surprising that you can make any kind of progress on that direction.

Alpha Genome能把基因代码映射到功能，玩弄那些涌现现象，也很惊人。

11:26

Alpha genome is

你会觉得有那么多组合可能性，结果呢，还真能找到那个能被高效建模的核心。

11:27

surprising that you can map the genetic code to the function kind of playing with the emergent kind of phenomena.

是的。

11:32

You think there's so many combinatorial options that and then here you go.

因为存在某种结构，某种景观，比如能量景观或者别的什么，你可以沿着某个梯度走。

11:36

You can find the kernel that is efficiently modeled. >> Yes.

而神经网络最擅长的就是沿着梯度走。

11:38

Because there's some structure there's some landscape you know in the energy landscape or whatever it is that you can follow some gradient you can follow.

所以只要有一条梯度可以跟随，并且你能正确设定目标函数，你就不必处理所有那些复杂性——我觉得我们过去几十年可能天真地以为，那些问题如果穷举所有可能性，看起来完全不可解，而且有很多很多这样的问题。

11:46

And of course what neural networks are very good at is following gradients.

然后你会想，蛋白质结构有10的300次方种可能，围棋有10的70次方种可能，这些都远超宇宙中的原子数，怎么可能找到正确的解或者预测下一步呢？

11:49

And so if there's one to follow and object and you can specify

但事实证明这是可能的，当然现实世界、自然界也确实做到了，蛋白质确实会折叠，这给了你信心：如果我们理解了物理学是怎么做到的，从某种意义上说，然后我们能够模仿那个过程、建模那个过程，那么在经典系统上应该也是可行的——这基本上就是那个猜想的核心。

11:52

the objective function correctly you know you don't have to deal with all that complexity which I think is how we maybe have naively thought about it for decades those problems if you just enumerate all the possibilities it looks totally intractable and there's many many problems like that and then you think well it's like 10^ the 300 possible protein structures uh it's 10^ theund you know 70 possible go positions all of these are way more than atoms in

当然还有非线性动力系统，高度非线性的动力系统，所有涉及流体的东西。

12:18

the universe so how could one possibly find the the right solution or predict the next step and and it but it turns out that it is possible and of course reality nature does do it right proteins do fold so that that gives you confidence that there must be if we understood how physics was doing that uh in a sense uh then and we could mimic that process I model that process uh it should be possible on our classical systems is is is basically what the

对。

12:44

conjecture is about >> and of course there's nonlinear dynamical systems, highly nonlinear dynamical systems, everything involving fluid. >> Yes. >> Right. >> You know, I recently had a conversation with Terrence Ta who mathematically uh it contends with a very difficult aspect of systems that have some singularities in them that break the mathematics and it's just hard for us humans to make any kind of clean predictions about highly nonlinear dynamical systems.

对吧。

13:12

But again to your

但回到你刚才说的，流体动力学、纳维-斯托克斯方程，这些在传统上被认为是经典系统上非常非常棘手的难题。它们需要巨大的算力，你知道，天气预报系统这类东西都涉及流体动力学计算。不过话说回来，如果你看看像VO这样的视频——也许你能展望一下未来5到10年的电子游戏？什么？你刚才说了很多非常有趣的内容。第一，你描述的那种开放世界本身就包含了深度个性化。所以这不只是说开放世界意味着你可以打开任何一扇门，门后总会有东西。

13:13

point we might be very surprised what classical learning systems might be able to do about even fluid.

你知道吗，我最近和Terrence Tao聊过，他在数学上处理一个非常困难的方面，就是系统中存在一些奇点，会破坏数学的完整性，我们人类很难对高度非线性的动力系统做出任何清晰的预测。

13:18

Yes, exactly.

但再说回你的观点，我们可能会非常惊讶，经典学习系统甚至能在流体问题上做到什么。

13:19

I mean fluid dynamics, Navia Stokes equations, these are traditionally thought of as very very difficult intractable kind of problems to do on classical systems.

我的意思是流体动力学，Navia Stokes方程，这些传统上被认为是经典系统上非常非常难处理的棘手问题。

13:28

They take enormous amounts of compute, you know, weather prediction systems, you know, these kind of things all involve fluid dynamics calculations.

它们需要巨大的算力，你知道，天气预报系统之类的，这些都涉及流体动力学计算。

13:36

And um but again, if you look at something like VO, our video

但是呢，如果你看看像VO这样的视频——

13:39

generation model, it can model liquids quite well, surprisingly well, and materials, specular lighting.

对，没错。

13:45

I love the ones where, you know, there's there's people have generated videos where there's like clear liquids going through hydraulic presses and then being squeezed out.

我是说流体动力学、Navier-Stokes方程，这些传统上被认为是非常非常困难、在经典系统上难以处理的问题。

13:54

I I used to write uh physics engines and graphics engines and in my early days in gaming and I know it's just so painstakingly hard to build programs that can do that and yet somehow these systems are, you know,

它们需要巨大的计算量，比如天气预报系统，这些东西都涉及流体动力学计算。

14:06

reverse engineering from just watching YouTube videos.

从只看YouTube视频就能做逆向工程。

14:10

So presumably what's happening is it's extracting some underlying structure around how these materials behave.

所以很可能它是在提取这些材料行为背后的某种底层结构。

14:16

So perhaps there is some kind of lower dimensional manifold that can be learned if we actually fully understood what's going on under the hood.

也许存在某种低维流形，如果我们真正搞懂了底层机制，就能学会它。

14:25

That's maybe you know maybe true of most of reality. >> Yeah.

这可能适用于大部分现实。

14:29

I've been continuously precisely by this aspect of V3.

>> 对。

14:32

I think a lot of

我一直被V3的这个方面深深吸引。

14:34

people highlight different aspects, including the comedic and the meme and all that kind of stuff.

人们会关注不同的方面，包括喜剧效果和梗图之类的东西。

14:39

And then the ultra realistic ability to capture humans in a really nice way that's compelling and feels close to reality and then combine that with native audio.

还有那种超逼真的能力，能以很吸引人的方式捕捉人类，感觉非常接近现实，再结合原生音频。

14:48

All of those are marvelous things about V3.

这些都是V3的厉害之处。

14:51

But the exactly the thing you're mentioning, which is the physics. >> Yeah, >> it's not perfect, but it's pretty damn good.

但你提到的那个点，就是物理效果。

14:58

And then the really interesting

>> 对，>> 它并不完美，但已经相当不错了。

15:00

scientific question is what is it understanding about our world >> in order to be able to do that because of the cynical take with diffusion models there's no way it understands anything >> but it seem I mean I don't think you can generate that kind of video without understanding and then our own philosophical notion what it means to understand then is like brought to the surface like do to what degree do you think V3 understands our world?

大家会强调不同的方面，比如喜剧效果、meme文化之类的。

15:27

I think

还有那种超写实的能力，能把人类捕捉得非常棒，很有感染力，感觉接近真实，再加上原生音频。

15:27

to the extent that it can predict the next frames you know in a coherent way that's some that is a form you know of understanding right not in the anthropomorphic version of you know it's not some kind of deep philosophical understanding of what's going on I don't think these systems have that but they they certainly have uh modeled enough of the dynamics you know put it that way that they can pretty accurately generate whatever it is 8 seconds of consistent

只要它能以连贯的方式预测接下来的帧，这本身就是一种理解的形式，对吧，不是那种拟人化的理解，也不是什么深层的哲学理解——我不觉得这些系统有那种东西，但它们确实建模了足够多的动态，可以这么说，以至于它们能相当准确地生成任何东西，比如8秒的连贯内容。

15:54

video that by eye at least you know at a glance is quite hard to distinguish what the issues are and imagine that in two or three more years time.

这段视频，光用肉眼看，你一眼就能看出问题在哪，很难分辨，再想象一下两三年后。

16:02

That's the thing I'm thinking about and how incredible that there will look uh given where we've come from, you know, the early versions of that uh one or two years ago.

这就是我一直在想的，想想我们是从哪里来的，一两年前的早期版本，这进步速度太不可思议了。

16:11

And so, um the rate of progress is incredible.

我觉得，我和你一样，很多人都喜欢那些脱口秀演员，他们确实捕捉到了很多人类动态和肢体语言，但真正让我印象最深、最着迷的是物理行为、光照、材质和液体。

16:13

And I think um I'm like you is like a lot of people love all of the the the the standup comedians and the the that actually

它能做到这些，真的很惊人。

16:20

captures a lot of human dynamics very well and and body language, but actually the thing I'm most impressed with and fascinated by is the physics behavior, the lighting and materials and liquids.

它很好地捕捉了大量人类动态和肢体语言，但实际上最让我印象深刻和着迷的是物理行为、光照、材质和液体。

16:33

And it's pretty amazing that it can do that.

它能做到这点真的很惊人。

16:36

And I think that shows that it has some notion of at least intuitive physics, right? um how things are supposed to work uh intuitively maybe the way that uh a human child would

我觉得这说明它至少有一些直观物理的概念，对吧？

16:47

understand physics right as opposed to a you know a PhD student really uh being able to unpack all the equations it's more of an intuitive physics understanding >> well that intuitive physics understanding that's the base layer that's the thing people sometimes call like common sense like it it really understands something I think that really surprised a lot of people it blows my mind that >> I just didn't think it would be possible to generate that level of realism without understanding.

一种表达方式。

17:14

>> You there's this notion that you can only understand the physical world by having an embodied AI system, a robot that interacts with that world.

有一种观点认为，只有通过具身AI系统——也就是与物理世界互动的机器人——才能理解物理世界。

17:22

That's the only way to construct an understanding of that world.

那是构建世界理解的唯一方式。

17:25

But V3 is directly challenging that it feels like >> yes and it's very interesting you know even if we if you were to ask me 5 10 years ago I would have said even though I was immersed in all of this I would have said well yeah you probably need to understand intuitive physics you know

但V3感觉直接挑战了这一点。

17:40

like if I push this off the table this glass it will maybe shatter you know um and and the liquid will spill out right so we know all of these things but I thought that you know and there's a lot of theories in neuroscience it's called action in perception where you know you you need to act in the world to really truly perceive it in a deep way.

如果我把它推下桌子，这个玻璃杯可能会碎掉，液体也会洒出来，对吧？

17:58

And there was a lot of theories about you need embodied intelligence or robotics or something or maybe at least simulated action uh so that you would understand

我们都知道这些事。

18:06

things like intuitive physics.

高效的太阳能，嗯，如果能源基本上是免费、可再生且清洁的，那就能解决一大堆其他问题。

18:08

But it seems like um you can understand it through passive observation which is pretty surprising to me and and again I think hints at something underlying about the nature of uh reality in in my opinion beyond um just the you know the cool videos that it generates.

比如，水资源获取问题就消失了，因为你直接用海水淡化就行。

18:24

Um and and of course there's next stages is maybe even making those videos interactive.

我们有技术，只是太贵了。

18:29

So uh one can actually step into them and move around them.

所以只有像新加坡、以色列这样相当富裕的国家才真正在用。

18:32

Um which

但如果它便宜了，那所有有海岸线的国家都能用。

18:33

would be really mind-blowing especially given my games background.

有个叫Jimmy Apples的人发了条推文：“让我用我的V3视频玩个游戏吧。

18:37

So you can imagine and then and then I think you know you're we're starting to get towards what I would call a world model a model of how the world works the mechanics of the world the physics of the world and the things in that world.

谷歌做得真不错，可玩的世界模型，拼写是we n问号。

18:49

And of course that's what you would need for a true AGI system. >> I have to talk to you about video games.

”然后你引用转推说：“那岂不是很有意思？

18:55

So, you were being a bit trolly.

”那么，用AI构建游戏世界有多难？

18:56

I I think you're you're having more and more fun on Twitter on X, which is great to

也许你能展望一下未来五到十年的电子游戏？

19:01

see.

只要它能以连贯的方式预测下一帧，那本身就是一种理解的形式，对吧？

19:01

So, guy named Jimmy Apples tweeted, "Let me play a video game of my V3 videos already.

不是拟人化的那种，不是某种深刻的哲学理解，我不认为这些系统有那种东西。

19:07

Uh, Google cooked so good playable world models when spelled we n question mark." Uh, and then you quote tweeted that with, "Now, wouldn't that be something?" So, how how hard is it to build game worlds with AI?

但它们确实建模了足够多的动态，可以这么说，所以它们能相当准确地生成，比如说，8秒的连贯内容。

19:21

Maybe can you look out into the future uh of video games 5 10 years out?

也许你能展望一下未来五到十年的电子游戏？

19:26

What

什么？

19:27

do you think that looks like? >> Well, games were my first love really and doing AI for games was the first thing I did professionally in my teenage years and and was the first major AI systems that I built and uh I always want to have I want to scratch that itch one day and come back to that. though, you know, and I will do, I think, and um I think I'd sort of dream about, you know, what would I have done back in the '90s if I'd had access to the kind of AI systems we have today?

你觉得那看起来像什么？

19:53

And I think you

>> 嗯，游戏其实是我的初恋，为游戏做AI是我十几岁时第一份专业工作，也是我构建的第一批大型AI系统。

19:54

could build absolutely mind-blowing games.

能做出让人完全震撼的游戏。

19:56

Um, and I think the next stage is I always used to love making all the games I've made are openw world games.

嗯，我觉得下一步就是——我一直特别喜欢做开放世界游戏，我做过的所有游戏都是开放世界类型的。

20:02

So, they're games where there's a simulation and then there's AI characters and then the player uh interacts with that simulation and the simulation adapts to the way the player plays.

就是那种有模拟系统、有AI角色，然后玩家跟这个模拟系统互动，模拟系统会根据玩家的玩法去调整。

20:12

And I always thought they were the coolest games because uh so games like theme park that I worked on where everybody's game experience would be unique to them, right?

我一直觉得这种游戏最酷，比如我参与过的《主题公园》，每个人的游戏体验都是独一无二的，对吧？

20:21

Because you're

因为你其实是在跟系统一起共创这个游戏。

20:22

kind of co-creating the game, right?

比如直觉物理。

20:24

Uh we set up the parameters, we set up initial conditions, and then you as the player immersed in it, and then you are co-creating it with the with the simulation.

但似乎你可以通过被动观察来理解它，这让我挺惊讶的，而且我觉得这又暗示了关于现实本质的某种底层东西，在我看来，不仅仅是那些它生成的酷炫视频。

20:32

But of course, it's very hard to program open world games. you know, you've got to be able to create uh content whichever direction the player goes in and you want it to be compelling no matter what the player chooses.

当然，下一步可能是让这些视频变得可交互。

20:43

Um, and so it was always quite difficult to build uh things like cellular automter actually type of those kind of classical

这样你就可以真正走进去，在它们周围移动。

20:49

systems which created some emergent behavior.

游戏。

20:51

Um, but they're always a little bit fragile, a little bit limited.

有趣的是，那正是研究的前沿所在，而且它跟艺术——图形、音乐，还有整个新的叙事媒介——有一种不可思议的融合，我特别喜欢。

20:55

Now we're maybe on the cusp in the next few years, 5 10 years of having AI systems that can truly create around your imagination. um can nar sort of dynamically change the story and storytell the narrative around uh and make it dramatic no matter what you end up choosing.

对我来说，这种跨学科的努力，又是我一生都热爱的事情。

21:10

So it's like the ultimate choose your own adventure sort of game.

我得问你，我差点忘了最近一件非常了不起、但好像还没得到足够关注的事情——AlphaEvolve。

21:14

And uh you know I think maybe we're

我们之前聊过一点进化，但这是Google DeepMind那个能进化算法的系统。

21:16

within reach if you think of a kind of interactive version of VO uh and then wind that forward 5 to 10 years and you know imagine how good it's going to be. >> Yeah.

如果想象一种互动版的《魔兽世界》，再快进5到10年，想象它会有多好，那其实触手可及。

21:26

So you said a lot of super interesting stuff there.

你刚才说了很多超级有趣的内容。

21:29

So one the open world built into that is a deep personalization the way you've described it. >> So it's not just that it's open world like you can open any door and there'll be something there.

第一，你描述的那种开放世界，本质上是一种深度个性化。

21:41

It's that the choice of which door you open

关键在于你选择打开哪扇门，

21:44

>> in an unconstrained way defines the worlds you see.

>> 对。

21:47

So some games try to do that to give you choice.

你刚才说了很多超级有趣的东西。

21:50

Yes.

首先，你描述的那种开放世界本身就包含了深度个性化。

21:50

But it's really just an illusion of choice because >> the only uh like like Stanley Parable game I recently played.

但这其实只是选择的幻觉，因为 >> 唯一像《Stanley Parable》的游戏我最近刚玩过。

21:57

It's it's it's really there's a couple of doors and it really just takes you down a narrative.

它真的就是有几扇门，然后把你带进一个叙事里。

22:03

Stanley Parable is a great video game I recommend people play that kind of uh in a meta way uh mocks the illusion of

《Stanley Parable》是个很棒的电子游戏，我推荐大家玩，它以一种元叙事的方式 >> 嘲弄了这种幻觉。

22:11

choice and there's philosophical notions of free will and so on.

>> 所以这不只是开放世界，比如你可以打开任何一扇门，门后都有东西。

22:15

But uh I do like one of my favorite games of Elder Scrolls is Daggerfall.

而是你选择打开哪扇门

22:19

I believe that they really played with a like random generation of the dungeons. >> Yeah. >> Of you can step in and they give you this feeling of an open world and there you mentioned interactivity.

我觉得他们确实玩过那种随机生成地牢的机制。 >> 对。 >> 你走进去，他们给你一种开放世界的感觉，还有你提到的互动性。

22:32

You don't need to interact.

你不需要互动。

22:34

That's a first step cuz you don't need to interact that

那是第一步，因为你不需要互动，

22:37

much.

>> 以一种不受限的方式，决定了你看到的世界。

22:38

You just when you open the door, whatever you see is randomly generated for you. >> Yeah.

有些游戏试图给你选择，但那只是一种选择的幻觉，因为

22:43

And that's already an incredible experience because you might be the only person to ever see that. >> Yeah.

而那本身就已经是一种不可思议的体验，因为你可能是唯一见过这个场景的人。 >> 没错。

22:49

Exactly.

没错。

22:49

And and so but what you'd like is a little bit better than just sort of a random generation, right?

但说到底，你想要的肯定不只是随机生成那种水平，对吧？

22:55

So you'd like uh and and also better than a simple AB hardcoded choice, right?

而且呢，也比简单的AB固定选项要好，对吧？

22:59

That's not really uh open world, right?

那其实算不上开放世界，对吧？

23:01

As you say, it's just giving you the

就像你说的，它只是给你一个——

23:03

illusion of choice.

>> 就像我最近玩的《史丹利的寓言》，它其实只有几扇门，最终只是把你带进一个叙事里。

23:04

What you want to be able to do is is potentially anything in that game environment.

《史丹利的寓言》是个很棒的游戏，我推荐大家玩，它以一种元叙事的方式嘲弄了

23:09

Um, and I think the only way you can do that is to have uh generated systems, systems that uh will generate that on the fly.

嗯，我觉得唯一能做到这点的办法，就是拥有生成式系统，那些能实时生成内容的系统。

23:16

Of course, you can't create infinite amounts of game assets, right?

当然，你不可能无限量地制作游戏资源，对吧？

23:20

It's expensive enough already how AAA games are made today.

现在的3A大作制作成本已经够高了。

23:23

And that was obvious to to us back in the '9s when I was working on all these games.

这在90年代我做那些游戏的时候就已经很明显了。

23:27

I think maybe Black and White uh was the game that I worked

我觉得可能《Black and White》是我参与制作的一款游戏。

23:31

on, early stages of that that had the still probably the best AI learning AI in it.

选择的幻觉，还涉及自由意志的哲学概念。

23:35

It was an early reinforcement learning system that you, you know, you were, you were looking after this mythical creature and growing it and nurturing it and depending how you treated it, it would treat the villagers in that world in the same way.

但我最喜欢的《上古卷轴》系列之一是《匕首雨》。

23:47

So if you were mean to it, it would be mean.

我觉得他们真的在随机生成地牢上做了文章。

23:50

If you were good, it would be protective.

如果你玩得好，它就会变得有保护性。

23:52

And so it was really a reflection of the way you played it.

所以它其实是你游戏方式的真实反映。

23:55

So actually all of the uh I've been working on sort of

实际上，我一直在研究类似的东西。

23:58

simulations and AI through the medium of games at the beginning of my career and and really the whole of what I do today is still a follow on from uh those early more hardcoded ways of doing the AI to now you know fully general learning systems that that are trying to achieve the same thing. >> Yeah, it's been uh interesting, hilarious, and uh fun to watch you and Elon obviously itching to create games because you're both gamers.

>> 对。

24:24

And one of

还有一点——

24:24

the sad aspects of your uh incredible success in so many domains of science like serious adult stuff. >> Yeah. >> That you might not have time to really create a game.

>> 你可以走进去，它们给你一种开放世界的感觉。

24:34

You might end up creating the tooling that others would create the game.

你提到了互动性。

24:38

You have to watch >> other others create the thing you've always dreamed of.

其实你不需要太多互动，这是第一步。

24:43

Do you think it's possible you can somehow in your extremely busy schedule actually find

当你打开门时，看到的一切都是为你随机生成的。

24:48

time to create something like black and white? some some an actual video game where like you could make the childhood dream come become reality. >> You know, there's two things way to think about that is maybe with vibe coding as it gets better and there's a possibility that I could, you know, one could do that actually in in your spare time.

是时候做出像《黑与白》那样的东西了吗？

25:07

So, I'm quite excited about that as a as that would be my project if if I got the time to do some vibe coding.

就是那种能让你童年梦想成真的电子游戏。

25:13

Um I'm actually itching to do that.

嗯，我其实特别想试试这个。

25:15

And

所以它不仅仅是开放世界，比如你可以打开任何一扇门，门后都会有东西。

25:15

then the other thing is, you know, maybe it's a sbatical after agi has been safely stewarded into the world and delivered into the world.

你知道，有两种方式可以思考这个问题。

25:23

You know, that and then working on my physics theory as we talked about at the beginning.

一种是随着vibe coding越来越强，说不定真能利用业余时间做到。

25:28

Those would be the two my my two post AGI projects.

我对此挺兴奋的，因为如果我有时间搞点vibe coding，这就会是我的项目。

25:31

Let's call it that way. >> I I would love to see which game post AGI which you choose.

说实话，我手痒得很。

25:36

Solving uh the the problem that some of the smartest people in human history contended with, you know, P equals MP

解决嗯，人类历史上最聪明的一批人争论过的问题，比如P等于NP。

25:43

or creating a cool video.

另一种方式是，也许等AGI被安全地引导并交付到这个世界之后，我会休个假。

25:45

Yeah.

然后就像我们开头聊的，去研究我的物理学理论。

25:45

Well, but they might but in my world they'd be related because it would be an openw world simulated game uh as realistic as possible.

这两个就是我所谓的“后AGI项目”。

25:53

So, you know what what is what is the universe?

所以，你知道，宇宙到底是什么？

25:56

That's that's that's speaking to the same question, right?

这其实是在问同一个问题，对吧？

25:59

NPL MP.

NP和P。

25:59

I think all these things are related, at least in my mind.

我觉得这些事都是相关的，至少在我脑子里是这样。

26:03

I mean in a really serious way like video games sometimes are looked down upon as just

我是说，从很严肃的角度看，电子游戏有时候被人瞧不起，觉得只是——

26:08

this fun side activity but especially as AI does more and more of the difficult uh boring tasks something we in in modern world call work.

我真的很想看看你后AGI会选哪个项目——是解决人类历史上最聪明的人都没搞定的问题，比如P等于NP，还是做一款酷炫的游戏。

26:18

You know video games is the thing in which we may find meaning in which we may find like what to do with our time.

你知道，电子游戏其实可能是我们找到意义、找到该怎么度过时间的东西。

26:27

You could create incredibly rich, meaningful experiences.

你能创造出极其丰富、有意义的体验。

26:31

Like that's what

就像这样，其实面对面直接体验他人——就像我们现在坐在这里聊天——也有巨大的价值。

26:32

human life is.

不过在我眼里，这两件事其实是相关的，因为那会是一个尽可能逼真的开放世界模拟游戏。

26:33

And then in video games, you can create more sophisticated, more diverse ways of living, >> right? >> I think so.

你想想，宇宙是什么？

26:42

I mean, those of us who love games, and I still do, is is is um you know, it's almost can let your imagination run wild, right?

这跟P等于NP其实是同一个问题。

26:51

Like I I used to love games um and working on games. so much because it's the fusion

至少在我看来，这些事是紧密相连的。

26:58

especially in the '9s and early 2000s the sort of golden era maybe the 80s of of of game of the games industry and it was all being discovered new genres were being discovered we weren't just making games we felt we were we were creating a new entertainment medium that never existed before especially with these open world games and simulation games where you were co-create you as the player were co-creating the story there's no other media uh entertainment media where you do that where you as the audience actually co-create the the

我觉得是这样。

27:25

story and of course Now with multiplayer games as well, it can be a very social activity and can explore all kinds of interesting worlds in that.

但另一方面，享受和体验物理世界也非常重要。

27:33

But on the other hand, you know, it's very important to um also enjoy and experience uh the physical world.

问题是，我觉得我们得再次直面那个根本问题：现实的本质到底是什么？

27:39

But the question is then, you know, I think we're going to have to kind of confront the question again of what is the fundamental nature of reality? uh what is the going to be the difference between these increasingly realistic

这些越来越逼真的模拟、多人游戏、涌现现象，和我们在现实世界里做的事，区别究竟在哪？

27:52

simulations and uh multiplayer ones and emergent um and what we do in the real world. >> Yeah, there's clearly a huge amount of value to experiencing the real world nature.

没错，体验真实自然世界显然有巨大价值，像我们现在这样直接面对面地感受他人也有巨大价值。

28:02

There's also a huge amount of value in experiencing other humans directly in person the way we're sitting here today. >> But we need to really scientifically rigorously answer the question why. >> Yeah.

>> 但我们真的需要用科学严谨的方法来回答“为什么”。

28:14

And which aspect of that can be mapped into the virtual world.

>> 是啊。

28:18

>> Exactly. >> It's not it's not enough to say, "Yeah, you should go touch grass and hang out in nature." It's like, why exactly is that valuable? >> Yes.

>> 没错。

28:26

And I guess that's maybe the thing that's been uh haunting me, obsessing me from the beginning of my career.

>> 光说“你应该去接触大自然、多出去走走”是不够的。

28:31

If you think about all the different things I've done, that's they're all related in that way.

关键是要说清楚，为什么这件事有价值？

28:36

This simulation, nature of reality, and what is the bounds of, you know, what can be modeled.

>> 对。

28:41

Sorry for the ridiculous question, but so far, what is the greatest video game of all time?

我觉得这可能就是我从职业生涯一开始就一直困扰我、让我着迷的东西。

28:45

What's up there?

你看我做过那么多不同的事情，其实它们都围绕同一个核心：模拟、现实的本质，以及什么是可以被建模的边界。

28:47

>> Well, my favorite one of all time is Civilization.

>> 嗯，我从小到大最喜欢的游戏是《文明》。

28:50

I have to say that that was the the the Civilization 1 and Civilization 2.

必须得说，《文明1》和《文明2》是我有史以来最爱的游戏。

28:54

My favorite games of all time.

>> 我猜你肯定没碰最新那代，不然你估计直接消失去玩它了。

28:56

Um >> I can only assume you've avoided the most recent one because it would probably you would that would be your sobatical that you would disappear. >> Yes, exactly.

>> 没错，这些《文明》游戏太耗时间了，我得小心点。

29:06

They take a lot of time these Civilization games.

而且，其中的哪些方面可以被映射到虚拟世界里呢。

29:09

So, I got to be careful with them. >> Fun question.

这些《文明》系列游戏真的很花时间。

29:12

You and Elon seem to be

所以我得小心点玩。

29:14

somehow solid gamers.

>> 有意思的问题。

29:15

Uh is there a connection between being great at gaming and and uh being great leaders of AI companies? >> I don't know.

你和Elon好像都是挺硬核的玩家。

29:22

I It's an interesting one.

擅长玩游戏和成为优秀的AI公司领导者之间，有什么联系吗？

29:23

I mean uh we both love games and uh it's interesting he wrote games as well to start off with.

>> 我不知道。

29:28

It's probably especially in the era I grew up in where home computers were just became a thing, you know, in the late ' 80s and '9s, especially in the UK.

这个问题挺有意思的。

29:37

I had a Spectrum and then a Commodore Omega 500 which is

我们俩确实都喜欢游戏，而且有意思的是，他一开始也是写游戏的。

29:40

my my favorite computer ever and that's why I learned all my programming and of course it's a very fun thing uh to program is to program games.

>> 我得问你一个差点忘了的问题，最近有一件非常了不起的事情，但好像还没得到足够的关注，就是AlphaEvolve。

29:49

So I think it's a great way to learn programming probably still is and um and then of course I immediately took it in directions of AI and simulations which so I may was able to express my interest in in games and my sort of wider scientific interests alto together.

>> 我们之前聊过一点进化，这是Google DeepMind的一个系统，用来进化算法。

30:05

And then the final thing I think that's

>> 对。

30:07

great about games is it fuses um artistic design, you know, art with the the the most cutting edge programming.

>> 是的，没错。

30:15

Um so again, in the '90s, all of the most interesting uh technical advances were happening in gaming, whether that was AI, graphics, physics engines, uh hardware, even GPUs of course were designed for gaming originally.

LLM会提出一些可能的解决方案，然后你在上面用进化计算来探索搜索空间中新的部分。

30:29

Um so everything that was pushing computing forward in the in the '9s was due to

所以我认为这是一个非常有前景的方向，把LLM或基础模型和其他计算技术结合起来。

30:35

gaming.

有趣的是，那正是研究的前沿，而且它和艺术、图形、音乐，还有整个叙事的新媒体融合得不可思议，我特别喜欢这一点。

30:35

So interestingly that was where the forefront of research was going on and it was this incredible fusion with with art um you know graphics but also music and just the whole new media of storytelling and I love that.

对我来说，这种跨学科的努力正是我一生都热爱的东西。

30:48

For me it's this sort of multi-disiplinary kind of effort is again something I've enjoyed my whole my whole life.

我得问你，我差点忘了，最近最让我惊叹的事情之一就是……>> 对，没错。

30:55

I have to ask you, I almost forgot about one of the many and I would say one of the most incredible things recently uh that

所以LLMs会提出一些可能的解决方案，然后你在上面用进化计算来探索搜索空间中新颖的部分。

31:02

somehow didn't yet get enough attention is alpha evolve. >> We talked about evolution a little bit but it's the Google deep mind system that evolves algorithms. >> Yeah. >> Are these kinds of evolution-like techniques promising as a component of future super intelligence system?

我们有很多低垂的果实，就是理解如何建模、如何模拟进化，然后利用我们对这种受自然启发的机制的理解，去做得越来越好、越来越好。

31:18

So for people who don't know, it's kind of um I don't know if it's fair to say it's LLM guided evolution search. >> Yeah. >> So evolutionary algorithms are doing the search and LLMs are telling you where.

对。

31:30

>> Yes.

对，完全正确。

31:31

Exactly.

所以LLMs会提出一些可能的解决方案，然后你在上面用进化计算去探索搜索空间中那些新颖的部分。

31:31

So LLMs are kind of proposing some possible solutions and then you do you use evolutionary computing on top to to to find some novel part of the of the search space.

实际上我认为这是一个很有前景的方向，就是把LLMs或者基础模型跟其他计算技术结合起来。

31:42

So actually I think it's an example of very promising directions where you combine LLMs or foundation models with other computational techniques.

进化方法只是其中一种，你还可以想象蒙特卡洛搜索，基本上就是多种类型的搜索系统。

31:51

Evolutionary methods is one but you could also imagine Monte Carlo research basically many types of search

嗯，如果你想发现一些前所未见的新东西，那你就需要某种搜索过程来把你带到搜索空间中新颖的区域。

31:58

algorithms or reasoning algorithms sort of on top of or using the foundation models as a basis.

算法或者说推理算法，基本上是构建在基础模型之上，或者以基础模型为基础来使用的。

32:03

So, I actually think there's quite a lot of interesting uh things to be discovered probably with these sort of hybrid systems, let's call them. >> But not to romanticize evolution.

所以我其实觉得，在这些所谓的混合系统里，可能还有很多有趣的东西值得去发现。

32:14

Yeah, >> I'm only human.

>> 但别把进化浪漫化了。

32:15

But you think there's some value in whatever that mechanism is because we already talked about natural systems.

是啊，>> 我只是个凡人。

32:21

Do you think where there's a

但你觉得，不管那个机制是什么，它本身是有价值的，因为我们之前聊过自然系统。

32:23

lot of lowhanging fruit of us understanding being being able to model uh being able to simulate evolution and then using that whatever we understand about that nature inspired mechanism to to then do surge better and better and better. >> Yes.

我们有很多唾手可得的成果，比如理解并模拟进化，然后用我们对这种自然启发机制的理解，来不断改进、优化。

32:38

So if you think about uh again breaking down the sort of systems we've built uh to their really fundamental core, you've got like the model of the of the underlying dynamics of the

>> 对，所以如果你再想想，把我们构建的系统拆解到最根本的核心，你会得到底层动态的模型。

32:50

system.

嗯，如果你想发现一些全新的、前所未见的东西，那你就需要某种搜索过程在上面，把你带到搜索空间里一个全新的区域。

32:50

Uh and then if you want to discover something new, something novel that hasn't been seen before, um then you need some kind of search process on top to take you to a novel region of the of the of the search space.

这个可以通过很多方式实现，进化计算是其中一种。

33:04

And um you can do that in a number of ways.

在AlphaGo里，我们用的是蒙特卡洛搜索，正是它发现了第37手，那种从未见过的全新策略。

33:07

Evolutionary computing is one. um with Alph Go we just use Monte Carlo research right and that's what found move 37 the new kind of never seen before strategy

>> 有趣的问题。

33:17

in go and so that's how you can go beyond potentially what is already known so the model can model everything that you currently know about right all the data that you currently have but then how do you go beyond that so that starts to speak about the ideas of creativity how can these systems create something new discover something new obviously this is super relevant for scientific discovery or pushing met science and medicine forward, which we want to do with these systems.

在Go语言里，这就是你如何超越已知范畴的方式——模型可以建模你目前所知的一切，也就是你已有的所有数据，但问题是如何超越这些，这就开始涉及创造力的概念：这些系统如何创造新事物、发现新事物？

33:42

And you can actually

显然，这对科学发现、推动医学和科学进步至关重要，这也是我们希望这些系统能做到的。

33:43

bolt on some uh fairly simple search systems on top of these models and get you into a new region of space.

在这些模型上加上一些相当简单的搜索系统，就能把你带到空间的新区域。

33:48

Of course, you also have to um make sure that you're not searching that space totally randomly.

当然，你还得确保不是完全随机地搜索那个空间，那样太大了。

33:53

It would be too big.

所以你得有个目标函数来优化，并朝着它爬山，这能引导搜索。

33:54

So, you have to have some objective function that you're trying to optimize and hill climb towards and that guides that search.

但在程序空间里，有些进化机制可能很有意思。

34:01

But there's some mechanism of evolution that are interesting maybe in the space of programs.

程序空间极其重要，因为你大概能泛化到所有东西，比如变异。

34:05

But then the space of programs is an extremely

所以这不只是蒙特卡洛树搜索那种搜索。

34:08

important space because you can probably generalize to to everything you know for example mutation.

有点像共同创造游戏，对吧？

34:15

So it's not just Monte Carlo tree search where it's like a search. >> You could every once in a while >> combine things.

我们设定参数，设定初始条件，然后你作为玩家沉浸其中，和这个模拟一起共同创造。

34:23

Yeah. >> Combine things alter like sub like a components of a thing.

但当然，编程开放世界游戏非常难。

34:28

Yes.

你得能生成内容，无论玩家往哪个方向走，而且无论玩家选择什么，内容都要有吸引力。

34:28

So then you know what evolution is really good at is not just the natural selection.

所以一直以来，构建像元胞自动机这类经典系统都挺难的。

34:34

It's combining things and building increasingly complex hierarchical systems. >> So that component is super interesting especially like with alpha evolve in the space of programs. >> Yeah.

将事物组合起来，构建越来越复杂的层级系统。

34:45

Exactly.

>> 所以这个部分非常有趣，尤其是在程序领域里像AlphaEvolve这样的东西。

34:45

So there's a you can get a bit of an extra property out of evolutionary systems which is some new emergent capability may come about but of course like happened with life.

>> 对，没错。

34:55

Interestingly, with naive uh sort of traditional evolutionary computing methods without LLMs and the modern AI,

所以你可以从进化系统中获得一点额外的特性，那就是可能会出现一些新的涌现能力，当然就像生命演化中发生的那样。

35:01

the problem with them, there was they were very well studied in the 90s and and and and early 2000s and some promising results, but the problem was they could never work out how to evolve new properties, new emergent properties.

在我们讨论的这件事里，如果你能提前做某种多项式时间或常数时间的计算，然后构建出这个巨大的模型，那就有可能用理论计算机科学的方式解决一些极其困难的问题。

35:14

You always had a sort of subset of the properties that you put into the system.

>> 对，我觉得实际上有一大类问题都可以用这种方式来表述。

35:18

But maybe if we combine them with these foundation models, perhaps we can overcome that limitation.

我们当时做的方式是……

35:24

Obviously uh natural evolution clearly did because it it did evolve new capabilities right so

你和Elon似乎……

35:29

bacteria to where we are now.

从细菌到我们现在的位置。

35:31

So clearly that it must be possible with evolutionary systems to generate uh new patterns you know going back to the first thing we talked about and uh new capabilities and emergent properties and maybe we're on the cusp of discovering how to do that. >> Yeah listen uh alpha evolve is one of the coolest things I've ever seen.

所以很明显，通过进化系统，必然有可能生成新的模式——回到我们聊的第一个话题——以及新的能力和涌现特性，也许我们正处在发现如何做到这一点的边缘。

35:51

I've I've on my desk at home, you know, most of my time is spent behind that

>> 是啊，听着，Alpha Evolve 是我见过最酷的东西之一。

35:56

computers just programming.

然后如果你想发现一些新的、前所未见的东西，你就需要某种搜索过程在上面，把你带到搜索空间中一个新颖的区域。

35:58

And next to the the three screens is a skull of a tectalic, which is one of the early organisms that crawled out of the water onto land.

你可以用多种方式做到这一点。

36:08

And I just kind of watch that little guy.

进化计算是一种。

36:11

It's like you whatever the computation mechanism of evolution is is quite incredible.

在AlphaGo里，我们用了蒙特卡洛搜索，对吧，那就是发现第37手——那种前所未见的新策略的方式。

36:18

It's truly truly incredible.

这真的太不可思议了。

36:20

Now whether that's exactly the thing we need to do to do our search but never dismiss the power of nature what it did here. >> Yeah.

它们的问题在于，90年代和2000年代初期被研究得很透彻，也有一些有希望的结果，但问题在于它们始终搞不清楚如何进化出新的属性、新的涌现属性。

36:28

And it's amazing um which is a relatively simple algorithm right effectively and it can generate all of this immense complexity emerges obviously running over you know 4 billion years of time but but it's it's it's you know you can think about that as again a pro a search process that ran over the physics substrate of the

你总是只能得到你放进系统的那部分属性的一个子集。

36:48

universe for a long amount of computational time but then it generated all this incredible uh rich diversity. >> So uh so many questions I want to ask you.

宇宙在极长的计算时间里运行，然后它生成了所有这些令人难以置信的丰富多样性。

36:58

But one, you do have a dream.

>> 所以，呃，我有好多问题想问你。

37:00

One of the natural systems you want to uh try to model is a is a cell. >> Yes, >> that's a beautiful dream.

但其中一个，你确实有一个梦想。

37:07

Uh I could ask you about that.

你想尝试建模的一个自然系统是一个细胞。

37:09

I also just for that purpose on the AI scientist front just

>> 是的，>> 那是个很美的梦想。

37:13

broadly.

大体上来说。

37:14

So there's a essay uh from Daniel Cocatalio, Scott Alexander, and others that outlines steps along the way to get to ASI and has a lot of interesting ideas in it. one of which is uh including a superhuman coder and a superhuman AI researcher and in that there's a term of research taste that's really interesting.

有一篇由Daniel Cocatalio、Scott Alexander等人写的文章，概述了通往ASI的各个阶段，里面有很多有趣的想法。

37:36

So in everything you've seen, do you think

其中一个包括超级人类程序员和超级人类AI研究员，里面有个叫研究品味的概念特别有意思。

37:40

it's possible for AI systems to have research taste to help you in the way that AI co-scientist does to help steer human um human brilliant scientists and then potentially by itself to figure out what are the directions where you want to generate truly novel ideas because that seems to be like a really important component how to do great science.

AI系统有可能具备研究品味，像AI co-scientist那样帮助你，引导人类、呃，人类杰出的科学家，然后可能自主地找出那些你想产生真正新颖想法的方向，因为这似乎是做好科学的一个非常重要的组成部分。

38:04

Yeah, I think that's

对，我觉得那

38:06

going to be one of the hardest things to to uh mimic or model is is this this idea of taste or or judgment.

这会是模仿或建模最难的事情之一，就是这种品味或判断力。

38:12

I think that's what separates the you know the the great scientists from the good scientists like all all professional scientists are good technically right otherwise they wouldn't have made it that far in in academia and things like that but then do you have the taste to sort of sniff out what the right direction is what the right experiment is what the right question is.

我觉得这就是伟大科学家和优秀科学家的区别，所有专业科学家技术上都很强，对吧，不然他们也不可能在学术界走到那么远，但你能不能有那种品味，去嗅出正确的方向是什么，正确的实验是什么，正确的问题是什么。

38:31

So the

所以，挑选正确的问题是科学里最难的部分。

38:32

it's the it's picking the right question is is the hardest part of science.

就是，选对问题是科学中最难的部分。

38:36

Um and and making the right hypothesis and um that's what you know today's systems definitely they can't do.

呃，并且提出正确的假设，呃，而这正是如今的系统肯定做不到的。

38:42

So you know I often say it's harder to come up with a conjecture a really good conjecture than it is to solve it.

所以我常说，提出一个真正好的猜想比解决它更难。

38:49

So we may have systems soon that can solve pretty hard conjectures. um you know I I um mass Olympiad problems where we we you know alpha proof last year our system got you

所以我们可能很快就会有能解决相当难猜想的系统。

38:59

know silver medal in that really hard problems maybe eventually we'll be able to solve a millennium prize kind of problem but could a system have come up with a conjecture worthy of study that someone like Terren Tower would have gone you know what that's a really deep question about the nature of maths or the nature of numbers or the nature of physics and that is far harder type of creativity and we don't really Oh, systems clearly can't do that and we're not quite sure what that mechanism would

在那些真正困难的问题上拿到银牌，也许最终我们能解决一个千禧年大奖难题。

39:26

be.

这种想象力的飞跃，就像爱因斯坦在提出狭义相对论和广义相对论时，基于他当时所掌握的知识所做出的那种飞跃。

39:27

This kind of leap of imagination like like Einstein had when he came up with, you know, special relativity and then general relativity with the knowledge he had at the time. >> As for conjecture, the you want to come up with a thing that's interesting and amenable to proof. >> Yes. >> So like it's easy to come up with a thing that's extremely difficult. >> Yeah. >> It's easy to come up with a thing that's extremely easy. at that at that very edge, >> that sweet spot, right, of of basically advancing the science and splitting the

这种想象力的飞跃，就像爱因斯坦当年凭借当时的认知提出狭义相对论、再到广义相对论时的那种感觉。

39:54

hypothesis space into two ideally, right?

>> 至于猜想，你想提出一个既有趣又适合证明的东西。

39:57

Whether if it's true or not true, you you've learned something really useful and um and and that's hard and and and and making something that's also uh you know falsifiable and within sort of the technologies that you have you currently have available.

>> 说到猜想，你想提出一个既有趣又可以被证明的东西。

40:13

So it's a very creative process actually highly creative process that um I think just a kind of naive search on top of a model

>> 对。

40:21

won't be enough for that. >> Okay.

>> 至于猜想，你想提出一个既有趣又可以被证明的东西。

40:23

The idea of splitting the hypothesis space in two is super interesting.

>> 所以，提出一个极其困难的东西很容易。

40:27

So uh I've heard you say that there's basically no failure in or failure is extremely valuable if it's done if you construct the questions right if you construct the experiments right if you design them right that failure success are both useful.

>> 是啊。

40:41

So perhaps because it splits the hypothesis basically two, it's like a binary search. >> That's right.

>> 提出一个极其简单的东西也很容易。

40:46

So when you do like, you

在那个临界点上，

40:48

know, real blue sky research, there's no such thing as failure really as long as you're picking experiments and hypotheses that that that that meaningfully spit the hypothesis space.

>> 是的。

40:58

So you know, and you learn something, you can learn something kind of equally valuable from an experiment that doesn't work.

你看，从失败的实验里也能学到同样有价值的东西。

41:04

That should tell you, if you've designed the experiment well and your hypothesis are interesting, it should tell you a lot about where to go next. and um and then it's you're effectively

只要实验设计得好，假设有意思，它就能告诉你下一步该往哪儿走。然后呢，基本上就是——

41:14

doing a search process um and using that information in in you know very helpful ways.

>> 是的。

41:19

So to go to your dream of uh modeling a cell uh what are the big challenges that lay ahead for us to make that happen?

所以回到你那个建模细胞的梦想，要实现它，前面最大的挑战是什么？

41:27

We should maybe highlight that alpha I mean there's just so many leaps. >> Yeah. >> So AlphaFold solved if it's fair to say protein folding and there's so many incredible things we could talk about

我们得先提一下Alpha，这中间真的跨越了太多步。

41:39

there including the open sourcing uh the everything you've released.

>> 所以，提出一个极其困难的东西很容易。

41:44

Alpha Fold 3 is doing protein, RNA, DNA interactions, >> which is super complicated and and fascinating.

嗯。

41:51

That's amendable to modeling.

可以说AlphaFold解决了蛋白质折叠问题，这里面有太多值得聊的惊人成果了。

41:53

Alpha genome uh predicts uh how small genetic changes like if we think about single mutations, how they link to actual uh function.

AlphaFold 3现在在做蛋白质、RNA、DNA之间的相互作用，

42:02

So um those are it seems like it's creeping along to

这超级复杂，也特别迷人。

42:06

sophistic to to much more complicated u things like a cell but a cell has a lot of really complicated components. >> Yeah.

>> 所以，提出一个极其困难的东西很容易。

42:13

So what I've tried to do throughout my career is I have these really grand dreams and then I try to as you've noticed and then I try to break but I try to break them down any you know it's easy to have a kind of a crazy ambitious dream but the the the trick is how do you break it down into manageable achievable uh interim steps that are

我职业生涯里一直尝试的是，先有那些特别宏大的梦想，然后就像你注意到的，我会试着把它们拆解——你知道，拥有一个疯狂又野心勃勃的梦想很容易，但关键在于怎么把它分解成可控、可实现、阶段性的步骤。

42:32

meaningful and useful in their own right and so virtual cell which is what I call the project of modeling a cell I've had this idea you know of wanting to do that for maybe more like 25 is and I used to talk with Paul Nurse who is a bit of a mentor of mine in biology.

它们本身就有意义且有用。

42:48

He runs the the you know founded the Craig Institute and and won the Nobel Prize in in 2001. uh is is we've been talking about it since you know before the you know in the '90s and um and I come used to come

所谓的“虚拟细胞”——也就是我所说的细胞建模项目——这个想法我已经想了大概25年了。

42:59

back to every 5 years is like what would you need to model the full internals of a cell so that you could do experiments on the virtual cell and what those experiment you know in silicone and those predictions would be useful for you to save you a lot of time in the wet lab right that would be the dream maybe you could 100x speed up experiments by doing most of it in silicone the search in silicico and then you do the validation step in the wet lab.

每五年回看一次，就像是在想：你需要建模一个细胞的完整内部结构，这样就能在虚拟细胞上做实验，那些在硅片上的实验，预测结果能帮你在湿实验室里省下大量时间，对吧？

43:23

That would be that's the that's the dream.

那会是梦想成真。

43:25

And so u but maybe now finally uh so I

也许你能把实验速度提升100倍，大部分在硅片上搜索和模拟，最后在湿实验室里做验证。

43:27

was trying to build these components alpha fold being one that that would allow you eventually to model the full interaction a full simulation of a cell and I'd probably start with a yeast cell and partly that's what Paul nurse studied because a yeast cell is like a full organism that's a single cell right so it's the kind of simplest single cell organism and so it's not just a cell it's a full organism and um and yeast is

>> 对。

43:54

very well understood And so that would be a good candidate for uh a a kind of full simulated model.

>> 提出一个极其简单的东西也很容易。

43:59

Now alpha fold is the is the solution to the kind of static picture of what does a what does a protein look 3D structure protein look like a static picture of it.

现在AlphaFold解决的是静态图像的问题，也就是蛋白质的三维结构长什么样，一个静态的呈现。

44:08

But we know that biology all the interesting things happen with the dynamics the interactions and that's what alpha 3 is is the first step towards is modeling those interactions.

但我们知道生物学里所有有趣的事情都发生在动态和相互作用中，而Alpha 3正是迈向模拟这些相互作用的第一步。

44:18

So first of all

首先，

44:19

pairwise you know proteins with proteins proteins with RNA and DNA but then um the next step after that would be modeling maybe a whole pathway maybe like the to pathway that's involved in cancer or something like this and then eventually you might be able to model you know a whole cell >> also there's another complexity here that stuff in a cell happens at different time scales is that tricky like there you know protein uh folding is you know super fast >> yes

在那个非常边缘的地带，

44:46

>> um I don't know all the bi ological mechanisms, but some of them take a long time.

>> 那个最佳点，对吧，就是推动科学进步并分割……

44:50

And so is that that's an level.

所以这其实是一个层级问题。

44:52

So the levels of interaction has a different temporal scale that you have to be able to model. >> So that would be hard.

相互作用的层级有不同的时间尺度，你必须能够建模这些尺度。

44:58

So you'd probably need several simulated systems that can interact at these different temporal dynamics or at least maybe it's like a hierarchical system.

那确实很难。

45:07

So um you can jump up and down the the different temporal stages.

所以你可能需要几个模拟系统，它们能在这些不同的时间动态下相互作用，或者至少像一个层级系统那样运作。

45:10

So can you avoid I mean one of

所以，能不能避免——我是说，你知道，我首先把自己看作一个人。>> 对。>> 其次，我认同自己是人类。

45:12

the challenges here is not avoid simulating for example the the the quantum mechanical aspects of any of this right you want to not overm model you can skip ahead to just model the really highlevel things that get you a really good estimate of what's going to happen >> so you you got to make a decision when you're modeling any natural system what is the cutoff level of the granularity that you're going to model it to that then captures the dynamics that you're

这里的挑战不是要避免模拟，比如说，量子力学层面的东西，对吧？

45:39

interested in.

感兴趣的。

45:40

So probably for a cell I would hope that would be the protein level uh and that one wouldn't have to go down to the atomic level.

所以对于细胞，我希望那会是蛋白质层面，呃，而且不需要深入到原子层面。

45:48

Um so you know of course that's where alpha volt stock kicks in.

呃，当然，那就是alpha fold发挥作用的地方。

45:52

So that would be kind of the basis and then you'd build these um uh higher level simulations that um take those as building blocks and then you get the emergent behavior.

那会是一个基础，然后你构建这些，呃，更高层次的模拟，把它们当作积木，然后你就能得到涌现行为。

46:02

Apologize for

抱歉

46:03

the pthead questions ahead of time, but uh will do you think uh we'll be able to simulate and model the origin of life.

我提前看了那些问题，但你觉得我们能不能模拟和建模生命的起源？

46:11

So being able to simulate the first from from non-living organisms the the birth of a living organism. >> I think that's a one of the of course one of the deepest and most fascinating questions.

就是模拟从非生命体到第一个生命体的诞生。

46:24

Um I love that area of biology. you know, uh, people like there's a great book by Nick Lane, one

>> 我觉得这当然是最深刻、最迷人的问题之一。

46:31

of the top top experts in this area called the the 10 great inventions of of of evolution.

这个领域的顶尖专家之一，叫做进化史上的十大伟大发明。

46:35

I think it's fantastic and it also speaks to what the great filters might be, you know, prior or are they ahead of us.

我觉得这太棒了，它也说明了那些大过滤器可能是什么，你知道，是在我们之前还是之后。

46:41

I think I think they're most likely in the past if you read that book of how unlikely to go, you know, have any life at all and then single cell to multisell seems an unbelievably big jump that took like a billion years, I think, on Earth to do, right?

我觉得它们很可能在过去，如果你读过那本书，就知道从无生命到有生命有多不可能，然后从单细胞到多细胞似乎是一个难以置信的巨大跳跃，在地球上大概花了十亿年，对吧？

46:54

So it shows you how hard it was, right? >> Bacteria were super happy for a very long time,

所以这说明它有多难，对吧？

46:58

>> a very long time before they captured mitochondria somehow, right?

>> 在它们不知怎么捕获线粒体之前，过了很久，对吧？

47:02

I don't see why not why AI couldn't help with that some kind of simulation.

我不觉得AI为什么不能帮上忙，通过某种模拟。

47:06

Again, it's again, it's a bit of a search process through a combinatorial space.

再说一次，这有点像在一个组合空间里搜索的过程。

47:10

Here's like all the chem, you know, the chemical soup that that you start with, the primordial soup that, you know, maybe was on Earth near these hot vents.

这里有所有的化学物质，你一开始的化学汤，那个原始汤，可能在地球上那些热液喷口附近。

47:19

Here's some initial conditions.

这里有一些初始条件。

47:21

Can you uh generate something that looks like a cell?

你能不能生成一个看起来像细胞的东西？

47:23

So perhaps that would be a next

所以这可能是虚拟细胞项目之后的下一步，就是你怎么能从化学汤中真正涌现出那样的东西？

47:25

stage after the virtual cell project is well how how could you actually um something like that emerge from the chemical soup? >> Well, I would love it if there was a move 37 for the origin of life.

虚拟细胞项目之后的阶段是，嗯，怎么才能让这种东西从化学汤里涌现出来呢？

47:36

Yeah, >> I think that's one of the sort of great mysteries.

>> 我倒希望生命起源也有一个“第37手”。

47:39

I think ultimately what we will figure out is their continuum.

是啊，>> 我觉得这算是那些巨大的谜团之一。

47:42

There's no such thing as a line between non-living and living.

最终我们会发现，其实是一个连续体。

47:45

But if we can make that rigorous Yes. >> that that the very thing from the be big bang to today has been the same process.

非生命和生命之间并没有一条明确的界线。

47:51

If we can break down that wall that

但如果我们能让这个说法变得严谨就好了。

47:53

we've constructed in our minds of the actual origin of from non-living to living and it's not a line that it's a continuum that connects physics and chemistry and biology.

我们在脑海中构建的从非生命到生命实际起源的图景，它并不是一条线，而是一个连接物理、化学和生物的连续体。

48:03

There's no line. >> I mean this is my whole reason why I've worked on AI and AGI my whole life because I think it can be the ultimate tool to help us answer these kind of questions.

没有那条线。

48:14

And I don't really understand why um you know the average person doesn't think like worry about this

>> 这就是我一生致力于AI和AGI的全部原因，因为我认为它可以成为帮助我们回答这类问题的终极工具。

48:20

stuff more like how how can we not have a good definition of life and not and not living and non-living and the nature of time and let alone consciousness and gravity and all these things.

事情，比如我们怎么能没有一个关于生命、非生命、时间本质的好定义，更不用说意识、引力以及所有这些东西了。

48:30

It's it's just and quantum mechanics weirdness.

还有量子力学的怪异。

48:33

It's just to me it's I've always had this sort of screaming at me in my face the whole and that it's getting louder you It's like how what is going on here?

对我来说，这就像我一直有这种感觉在对我尖叫，而且越来越响。

48:41

You know, in in I mean that in the deepest sense like in the you know the

就像，这到底是怎么回事？

48:46

nature of reality which has to be the ultimate question uh that would answer all of these things.

现实的本质，那一定是能回答所有这一切的终极问题。

48:51

It's sort of crazy if you think about we can stare at each other and all these living things all the time.

想想看，这有点疯狂，我们可以互相凝视，时刻看着所有这些活物。

48:56

We can inspect it with microscopes and take it apart uh almost down to the atomic level and yet we still can't answer that clearly in a simple way that question of how do you define living? >> Yeah, >> it's kind of amazing.

我们可以用显微镜观察它，把它拆解，呃，几乎到原子层面，然而我们仍然无法简单明了地回答那个问题：如何定义生命？

49:08

Yeah, living you can kind of talk your way out of thinking about but like consciousness like we have this very obviously

>> 是啊，>> 这挺神奇的。

49:14

subjective conscious experience like we're at the center of our own world and it it feels like something and then h how how are you not screaming >> at the mystery of it all I mean but really humans have been contending with the mystery of the world around them for long there's a lot of mysteries like what's up with the sun and and the rain, >> like what's that about?

主观意识体验就像我们自己是自己世界的中心，那种感觉是真实的，然后你怎么能不尖叫着面对这一切的奥秘呢？

49:38

And then like last year we had a lot of rain and this

但说真的，人类一直在应对周围世界的奥秘，比如太阳和雨是怎么回事？

49:41

year we don't have rain.

那年我们没下雨。

49:42

Like what did we do wrong?

就像，我们做错了什么？

49:44

Humans have been asking that question for a long time. >> Exactly.

人类问这个问题已经很久了。

49:47

So we're quite I guess we've developed a lot of mechanisms to cope with this these deep mysteries that we can't fully we can see but we can't fully understand and we have to have to just get on with daily life and and and we get we keep ourselves busy right in a way.

>> 没错。

50:02

Do we keep ourselves distracted?

所以我们大概，我觉得我们已经发展出很多机制来应对这些深层的谜团——我们能看见，但无法完全理解，而且我们还得继续过日子，对吧？

50:04

I mean weather is one of the most important questions of human history.

某种程度上，我们让自己忙起来。

50:08

We

我们是在让自己分心吗？

50:08

still that's that's the go-to small talk direction of of the weather >> especially in England >> and then it's which is you know famously is an extremely difficult system to model and uh even that system uh Google deep mind has made progress on.

现在，天气还是那种最常用的闲聊话题。

50:23

Yes, we've yeah, we've created the the best weather prediction systems in the world and they're better than traditional fluid dynamics sort of systems that

>> 尤其是在英国。

50:33

usually calculated on massive supercomputers takes days to calculate it.

要在巨型超级计算机上算好几天。

50:37

And we've managed to model a lot of the weather dynamics with neural network systems with our weather next system.

而我们用神经网络系统，用我们的GraphCast系统，成功建模了很多天气动态。

50:43

And again, it's interesting that those kinds of dynamics can be modeled even though they're very complicated, almost bordering on chaotic systems in some cases.

而且，有意思的是，这些动态虽然非常复杂，有些甚至接近混沌系统，但依然可以被建模。

50:52

A lot of the interesting aspects of that um can be modeled by these neural network systems, including very recently we had, you know, cyclone

这些神经网络系统能建模很多有趣的方面，包括最近我们做的，比如气旋

51:00

prediction of where, you know, paths of hurricanes might go. of course super useful super important for the world and and and it's super important to do that very timely and very quickly and as well as accurately and uh I think it's very promising direction again of you know simulating and uh uh so that you can run forward predictions and simulations of very complicated real world systems. >> I should mention that uh I've got a chance in uh Texas to meet a community of folks called the stormchasers.

预测，比如飓风的路径可能往哪走。

51:27

>> Yes.

>> 是的。

51:27

And what's really incredible about them, I need to talk to them more, is they're extremely tech-savvy because what they have to do is they have to use models to predict where the storm is.

他们真正了不起的地方——我还得多跟他们聊聊——是他们非常懂技术，因为他们必须用模型来预测风暴的位置。

51:37

So they're it's just it's it's this beautiful mix of like crazy enough to like go into the eye of the storm and like >> in order to protect your life and predict where the extreme events are going to be, they have to have increasingly sophisticated models of of weather. >> Yeah.

所以这就像是一种美妙的结合，既有疯狂到敢冲进风暴眼的人，又有，嗯，为了保命和预测极端事件会发生在哪，他们得用越来越复杂的天气模型。

51:52

>> Yeah.

>> 是啊。

51:52

It's it's a a beautiful balance of like being in it as living organisms and the the cutting edge of science.

这是一种美妙的平衡，一边是作为活生生的人身处其中，一边是科学的前沿。

51:58

So they actually might be using uh deep mind system.

所以他们可能真的在用DeepMind的系统。

52:01

So that's >> Yeah, they hopefully they are and I I'd love to join them on one of those chases.

所以那——>> 对，希望他们在用，我也很想加入他们去追一次风暴。

52:06

They look amazing, right?

那些看起来太棒了，对吧？

52:07

To actually experience it one time. >> Exactly.

真正体验一次。

52:10

And then also to experience the correct prediction where something will come and how it's going to evolve.

>> 没错。

52:15

It's incredible. >> Yeah. >> You've estimated that we'll have AGI by

然后还能体验一次正确的预测——知道某样东西会在哪出现、会怎么演变。

52:19

2030.

当然，现在人们还在争论这个。

52:20

Um so there's interesting questions around that.

而我的标准一直挺高的，就是能不能匹配大脑的认知功能，对吧？

52:23

How will we actually know that we got there?

我们知道，我们的大脑大致上是通用的图灵机。

52:27

Uh and uh what maybe the move quote move 37 of AGI. >> My estimate is sort of 50% chance by in the next 5 years.

而且，当然，我们用我们的心智创造了不可思议的现代文明。

52:35

So you know by 2030 let's say and uh so I think there's a good chance that that could happen.

这也说明了大脑有多通用。

52:42

Part of it is what what is your definition of

而要让我们知道自己有了真正的AGI，我们得确保——

52:46

AGI?

AGI？

52:46

Of course, people are arguing about that now and and uh mine's quite a high bar and always has been of like can we match the cognitive functions that the brain has, right?

当然，现在大家都在争论这个，而我对它的标准一直很高，就是能不能匹配大脑的认知功能，对吧？

52:56

So, we know our brains are pretty much general cheuring machines approximate.

我们知道，我们的大脑大致上就是通用的计算设备。

53:01

And of course, we've created incredible modern civilization with our minds.

而且，当然，我们用我们的心智创造了不可思议的现代文明。

53:06

So, that also speaks to how general the brain is.

这也说明了大脑有多通用。

53:08

And um for us to know we have a true AGI, we would have to like make sure

要确认我们有了真正的AGI，我们得确保——

53:13

that it has all those capabilities. it isn't kind of a jagged intelligence where some things it's really good at like today's systems but other things it's really uh flawed at and and that's what we currently have with today's systems they're not consistent so you'd want that consistency of intelligence across the board and then we have some missing I think capabilities like sort of uh the true invention capabilities and creativity that we were talking about earlier so you'd want to see those how you test that um I think you just

它具备所有这些能力。

53:41

test it one way to do it would be a kind of brute force test of tens of thousand thousand of cognitive tasks that um you know we know that humans can do uh and maybe also make the system available to uh a few hundred of the world's top experts uh the terren towers of each each subject area and see if they can find you know give them give them a month or two and see if they can find an obvious flaw in the system and if they can't then I think you're you're pretty

一种测试方法是，对成千上万种人类已知的认知任务进行某种暴力测试，嗯，你知道，人类能做的那些，然后可能也让系统开放给全球几百位顶尖专家，每个领域的顶尖人物，看看他们能不能，给他们一两个月时间，看看他们能不能找出系统明显的缺陷。

54:08

uh you know pretty you can be pretty confident we have a a fully general system >> maybe to push back a little bit it seems like humans are really incredible as the the intelligence improves across all domains to take it for granted. >> Mhm. >> Uh like you mentioned Terrence Tao uh these brilliant experts they might quickly in a span of weeks take for granted all the incredible things it can do and then focus in well haha right

>> 也许稍微反驳一下，人类其实也很厉害，随着智能在所有领域的提升，我们容易把这一切视为理所当然。

54:36

there.

>> 嗯。

54:36

You know I I consider myself uh first of all human. >> Yeah. >> Uh second I identify as human.

嗯，你知道，有些人听我说话时会觉得：“这家伙不太会说话，结结巴巴的。”所以，即使人类在跨领域方面也有明显的局限，甚至不只是这些。

54:44

Um I you know some people listen to me talk and they're like that guy is not good at talking the stuttering the you know so like even humans have obvious across domains limits even just outside of

所以我认为，有一种全面的测试，只是为了确保你有一致性，但我也觉得有一些“灯塔式”的测试。

55:00

mathematics and physics and so on it I I I wonder if it will take something like a move 37 so on the positive side versus like >> a barrage of 10,000 cognitive tasks where it would be one or two where it's like yes, holy this is >> I think exactly.

>> 就像你提到的Terrence Tao，这些天才专家可能几周内就会习惯系统做的所有惊人事情，然后开始挑刺，对吧。

55:17

So I think there's the sort of blanket testing to just make sure you've got the consistency, but I think there are the sort of lighthouse

其中一个就是提出新的猜想或假设，比如爱因斯坦对物理学做的那样。

55:26

moments like the move 37 that I would be looking for.

>> 没错。

55:29

So one would be inventing a new conjecture or new hypothesis about physics like Einstein did.

所以，你甚至可以非常严格地做回溯测试：设定一个知识截止点，比如1900年，然后把所有1900年之前的知识给系统，看看它能不能像爱因斯坦那样提出狭义相对论和广义相对论。

55:34

So maybe you could even run the back test of that very rigorously like have a cut off of knowledge cutff of 1900 and then give the system everything that was you know that was written up to 1900 and then and then see if it could come up with special relativity and general relativity right like Einstein did that

然后，为了完善这一点，你还要检查一致性，确保那个系统里也没有漏洞。>> 对。

55:51

that would be an interesting test another one would be can it invent a game like go not just come up with move 37 a new strategy but can it invent a game that's as deep as aesthetically beautiful as elegant as go and those are the sorts of things I would be looking out for.

那会是一个有趣的测试，另一个测试是它能不能发明像围棋这样的游戏——不只是想出第37手这种新策略，而是能不能发明一个像围棋一样深邃、一样美学上优美、一样优雅的游戏。

56:07

Uh and probably a system being able to do uh uh several of those things, right, for it to be very general.

这些就是我会关注的事情。

56:13

Um not just one domain.

而且，可能一个系统需要能做到好几件这样的事，对吧，才能算得上非常通用。

56:15

And so I think that would be the signs at least

不只是单一领域。

56:18

that I would be looking for that we've got a system that's a GI level.

你知道，我首先觉得自己是人。

56:21

And then maybe to fill that out, you would also check the consistency, you know, make sure there's no holes in that system either. >> Yeah.

类似于一种新的猜想或科学发现。

56:29

Something like a new conjecture or scientific discovery.

有点像一种新的猜想或科学发现。

56:32

That would be a cool feeling.

那感觉会很酷。

56:33

Yeah, that would be amazing.

是啊，那会特别棒。

56:35

So, it's not not just helping us do that, but actually coming up with something brand new >> and you would be in the room for that.

所以它不只是帮我们做这件事，而是能真正创造出全新的东西，而且你还能亲眼见证。

56:41

So, it would be like probably 2 or 3 months before announcing it.

所以大概会在公布前的两三个月吧。

56:45

>> Mhm. >> And you would just be sitting there trying not to tweet >> something like that.

嗯。

56:51

Exactly.

>> 你就只能坐在那儿，拼命忍住不发推文 >> 类似那种东西。

56:51

It's like what is this amazing new you know physics idea?

对。

56:55

And then we would probably check it with world experts in that domain, right? and validate it and kind of go through its workings and it I guess it would be explaining its workings too.

就像在说，这是什么惊人的新物理想法？

57:06

Um yeah be an amazing moment. >> Do you worry that we as humans even expert humans like you might miss it

然后我们大概会找那个领域的全球顶尖专家来验证，对吧？

57:12

might miss >> it may be pretty complicated.

可能会错过 >> 它可能相当复杂。

57:15

So it could be the analogy I give there is I don't think it will be um uh uh totally mysterious to the to the best human scientists but it may be a bit like for example in chess if I was to talk to Gary Kasparov or Magnus Carlson and play a game with them and they make a brilliant move I might not be able to come up with that move but they could explain why afterwards that move made sense and we were to understand it to

所以我给的类比是，我不认为它会完全神秘到让最优秀的人类科学家无法理解，但可能有点像，比如在国际象棋中，如果我和加里·卡斯帕罗夫或马格努斯·卡尔森聊天，和他们下一盘棋，他们走了一步妙招，我可能想不出那步棋，但他们之后可以解释为什么那步棋合理，我们也能在一定程度上理解，虽然达不到他们的水平，但如果他们擅长解释——这其实也是智能的一部分，就是能用简单的方式表达你在想什么。

57:41

some degree not to the level they do but in you know if they were good at explaining which is actually part of intellg igence too is being able to explain in a simple way that what you're thinking about.

这会是个有趣的测试。

57:51

Um uh I I think that that would be very possible for the best human scientists. >> But I wonder maybe you can you can educate me on the side of go.

另一个测试是，它能不能发明一种像围棋一样的游戏，不只是想出第37手那种新策略，而是能不能发明一种像围棋一样深邃、一样美学上优雅、一样精巧的游戏。

57:59

I wonder if there's moves for Agnes or Gary where they at first will dismiss it as a bad move. >> Yeah, sure.

这些就是我会关注的事情。

58:05

It could be.

嗯，很可能一个系统能做到其中好几样，对吧，才能说明它非常通用。

58:05

But then afterwards they'll figure out with their

嗯，不只是单一领域。

58:08

intuition that that this why this works.

直觉上这就是为什么它有效。

58:10

And then and then and then empirically the nice thing about games is one of the great things about games is you can it's a sort of scientific test.

然后，然后，然后从实证角度看，游戏的好处之一就是它是一种科学测试。

58:18

Does it do you win the game or not win?

它赢了还是没赢？

58:20

And then um that tells you okay that move in the end was good.

然后那会告诉你，那步棋最终是好的，那个策略是好的。

58:23

That strategy was good.

然后你可以回去分析，甚至对自己多解释一下为什么，并围绕它进行探索。

58:24

And then you can go back and analyze that and and and and explain even to yourself a little bit more why explore around it.

这就是国际象棋分析之类的工作方式。

58:30

And that's how chess analysis and things like that works.

所以也许这就是为什么——

58:33

So perhaps that's why

可能这就是原因。

58:34

my brain works like that cuz I I've been doing that since I was four and you're train you know it's sort of hardcore training in that way.

某种程度上，不一定达到人类的水平，但你知道，如果它们擅长解释——这其实也是智能的一部分，就是能用简单的方式解释你在想什么。

58:43

But even even now like when I generate code there there is this kind of nuanced fascinating con contention that's happening where I might at first identify as a set of generated code is incorrect in in some interesting nuanced ways but then I'm always have to ask the

嗯，我觉得对于最优秀的人类科学家来说，这是完全可能的。

59:01

question is there a deeper insight here that that I'm the one who's incorrect >> and that's going to as the systems get more and more intelligent you're going to have to contend with that.

问题是，这里有没有更深层的洞察，其实是我错了？

59:11

It's like what what what do you is this a bug or a feature of what you just came up with? >> Yeah.

>> 随着系统变得越来越智能，你将不得不面对这个问题。

59:16

And they're going to be pretty complicated to do.

就像，你刚想出来的东西，这是个bug还是个feature？

59:18

But of course it will be you can imagine also AI systems that are producing that code or whatever that is and then human programmers looking at

>> 是啊。

59:26

it but also not unaded with the help of AI tools as well.

我的大脑是这样运作的，因为我从四岁就开始做这个，你知道，这是一种高强度的训练方式。

59:29

So it's going to be kind of an interesting you know maybe different AI tools to the ones that the more you know kind of monitoring tools to the ones that generated it.

但即使是现在，当我生成代码时，也会有一种微妙而迷人的争论在发生：我可能一开始觉得某段生成的代码在某些有趣而微妙的方式上是错误的，但我总是要问自己一个问题：这里有没有更深层的见解，其实是我自己错了？

59:40

So if we look at a AGI system, sorry to bring it back up, but alpha evolve, super cool.

所以如果我们看一个AGI系统——抱歉又提回来——但Alpha Evolve，超酷的。

59:46

So Alpha Evolve enables on the programming side something like recursive self-improvement uh

Alpha Evolve在编程层面实现了类似递归自我改进的能力。

59:52

potentially like what who can imagine what that AGI system maybe not the first version but a few versions beyond that.

可能像，谁能想象那个AGI系统——也许不是第一个版本，而是再往后几个版本——实际上会是什么样子？

59:59

What does that actually look like?

你觉得它会很简单吗？

60:01

Do you think it would be simple?

你觉得它会像是一个简单的自我改进程序吗？

60:02

You think it'll be something like a self-improving program in a simple one? >> I mean, potentially that's possible.

>> 我的意思是，那是有可能的。

60:09

I would say um I'm not sure it's even desirable because that's a kind of like hard takeoff scenario.

但我不确定那是否可取，因为那是一种硬起飞的情景。

60:14

But but you you these current systems like Alpha Evolve,

但像AlphaEvolve这样的现有系统——

60:18

they have, you know, human in the loop deciding on various things.

他们有人类在循环中决定各种事情，是独立的混合系统在交互。

60:21

They're separate hybrid systems that interact.

嗯，可以想象最终端到端地做到这一点，我不觉得这不可能，但目前嘛，我觉得系统还不够好，无法在代码架构层面做到这一点。

60:24

Uh one could imagine eventually doing that end to end.

嗯，这又有点跟提出新猜想假设的想法相关。

60:27

I don't see why that wouldn't be possible but right now um you know I think the systems are not good enough to do that in terms of coming up with the architecture of the code.

就是说，如果你给它们非常具体的指令，告诉它们要做什么，它们表现不错。

60:36

Um and again it's a little bit connected to this idea of coming up with a new conjectural hypothesis.

嗯，但如果你给一个非常模糊的高层指令，目前就行不通。

60:42

How like they're good if you give them very

比如，嗯，我觉得这跟“发明一个像围棋一样好的游戏”这个想法有关，对吧？

60:44

specific instructions about what you're trying to do.

关于你具体想做什么的明确指令。

60:47

Um, but if you give them a very vague high level instruction, that wouldn't work currently.

嗯，但如果只给一个非常模糊的高层指令，目前是行不通的。

60:51

Like, uh, and I think that's related to this idea of like invent a game as good as go, right?

比如，我觉得这跟“发明一个像围棋一样好的游戏”这种想法有关，对吧？

60:56

Imagine that was the prompt.

想象一下这就是提示词。

60:57

That's that's pretty underspecified.

那太不具体了。

60:59

And so the current systems wouldn't know, I think, what to do with that, how to narrow that down to something tractable.

所以目前的系统，我觉得，不知道该怎么处理，怎么把它缩小到可操作的范围。

61:05

And I think there's similar like, look, just make a better version of yourself that's too that's too unconstrained.

而且我觉得类似的情况还有，比如“做个更好的自己”，那也太没有约束了。

61:10

But we've done

但我们已经做到了。

61:11

it in, you know, and as you know with Alpha Evolve, like things like faster matrix multiplication.

就像在Alpha Evolve里，你知道的，比如更快的矩阵乘法这些事。

61:17

So when you when you hone it down to very specific thing you want um it's very good at incrementally improving that but at the moment these are more like incremental improvements sort of small iterations whereas if you know if you wanted a big leap in uh understanding you need a you need a much larger uh advance. >> Yeah.

当你把它缩小到非常具体的目标时，它很擅长逐步改进，但目前这些更像是渐进式的改进，小步迭代。

61:36

But it could also be sort of to

而如果你想要理解上的大飞跃，你需要一个更大的突破。

61:37

push back against hard takeoff scenario.

对。

61:40

It could be just a sequence of um incremental improvements like matrix multiplication like it has to sit there for days thinking how to incrementally improve a thing and that it does so recursively and as you do more and more improvement it'll slow down so there'll be like a like uh the path to AGI won't be like a it'll be a gradual improvement over time.

但这也可以用来反驳硬起飞场景，它可能只是一连串渐进式的改进，比如矩阵乘法，它得花好几天时间思考怎么逐步改进一个东西，然后递归地这么做，随着你改进得越来越多，速度会慢下来，所以通往AGI的路不会是一条直线，而是随时间逐渐提升。

62:04

>> Yes.

对。

62:05

If it was just incremental improvements that's how it would look.

如果只是渐进式改进，那看起来就是这样。

62:08

So the question is could it come up with a new leap like the transformers architecture right could it have done that back in 2017 when you know we did it and brain did it and it's it's not clear that that these systems something like Alpha wouldn't be able to do make such a big leap so for sure these systems are good we have systems I think that can do incremental hill climbing and that's a kind of bigger question about is that all that's needed from

所以问题是，它能不能带来像transformer架构那样的新飞跃？

62:32

here or do we actually need one or two more um uh big breakthroughs >> and can the same kind of systems provide the breakthroughs also.

而且同样的系统能不能也带来这些突破？

62:39

So make it a bunch of scurves like incremental improvement but also every once in a while leaps. >> Yeah.

所以让它变成一堆S曲线，既有渐进式改进，偶尔也有飞跃。

62:44

I don't think anyone has systems that can have shown unequivocally those big leaps that the the right.

我不觉得有哪个系统已经明确展示过那种质的飞跃，对吧。

62:50

We have a lot of systems that do the hill climbing of the S-curve that you're currently on. >> Yeah.

我们有很多系统在做你当前所在S曲线的爬山过程。>> 对。

62:55

And that would be the move 37 is a

而那个就像是第37步，嗯，类似那样的东西。

62:57

leap. >> Yeah.

对。

62:58

I think would be a leap.

我觉得没人有系统能明确展示出那些大飞跃，我们有很多系统能在你当前所在的S曲线上爬山。

63:00

Um something like that.

呃，你觉得scaling laws在pre-training、post-training、test time、compute这些方面还很强吗？

63:01

Uh do you think the scaling laws are holding strong on the pre-training, post- training, test time, compute?

呃，反过来看，你预计AI进展会遇到瓶颈吗？>> 我们当然觉得在scaling方面还有很大空间。

63:09

Uh do you uh on the flip side of that anticipate AI progress hitting a wall? >> We certainly feel there's a lot more room just in the scaling.

所以嗯，实际上所有步骤——pre-training、post-training和inference time都是。

63:19

So um actually all steps pre-training, post-training and inference time.

呃，大概就是那样。

63:24

So uh there's sort

嗯，就是那种……

63:25

of three scalings that are happening concurrently.

对。

63:28

Um and we again there it's about how innovative you can be and we you know we pride ourselves on having the broadest and um deepest research bench. uh we have amazing you know incredible uh researchers and uh people like Nam Shazir who you know came up with transformers and and Dave Silva you know who led the Alph Go project and so

那就像move 37，是一个飞跃。

63:50

on and um it's it's it's that research base means that if some new new breakthrough is required like an Alph Go or Transformers uh I would back us to be the place that does that.

嗯，那个研究基础意味着，如果未来需要某种新的突破，比如AlphaGo或Transformer，我相信我们会是做出这些突破的地方。

64:01

So I'm actually quite like it when the terrain gets harder, right?

所以我其实还挺喜欢局面变得更难的，对吧？

64:04

Because then it veers more from just engineering to to true research and you know re or research plus engineering and that's our sweet spot.

因为那样它就更少是纯粹的工程问题，而更多是真正的研究，或者说研究加工程，而这正是我们的强项。

64:12

And I I think that's harder

而且我觉得这更难——发明新东西比快速跟进要难得多。

64:14

it's harder to invent things than to than to um you know fast follow.

对。

64:18

And um so you know we don't know I would say it's a it's kind of 50/50 whether new things are needed or whether the scaling the existing stuff is going to be enough.

我觉得那算是一个飞跃。

64:29

And so in true kind of empirical fashion, we're pushing both of those as hard as possible.

类似那样的东西。

64:34

The new blue sky ideas and you know maybe about half our resources are on that and then and then

你觉得scaling laws在pre-training、post-training、test time compute上还很强吗？

64:40

uh scaling to the max the the current the current capabilities and um we're still seeing some you know fantastic progress on uh each different version of Gemini.

我们确实觉得在scaling上还有很多空间。

64:52

That's interesting the way you put it in terms of the deep bench that if uh progress towards AGI is more than just scaling compute so the engineering side of the problem and is more on the scientific side where there's

实际上，pre-training、post-training和inference time这三个步骤都有三种scaling在同时进行。

65:07

breakthroughs needed then you feel confident deep mind as well Google deep mind is well positioned to kick kick ass in that domain >> well I mean if you look at the history of the last decade or 15 years um it's been I you know maybe I don't know 80 90% of the breakthroughs that that underpins modern AI field today was from you know originally Google brain Google research and deep mind so yeah I would back that to continue hopefully >> uh so on the data side are you concerned

需要突破性进展，你才有信心 DeepMind 以及 Google DeepMind 在这方面很有优势，能大杀四方 >> 嗯，我觉得如果你看看过去十年或十五年的历史，大概 80% 到 90% 支撑现代 AI 领域的突破，最初都来自 Google Brain、Google Research 和 DeepMind，所以我希望这种情况能继续下去 >> 那在数据方面，你担心会耗尽高质量数据吗，尤其是高质量的人类数据 >> 我不太担心，部分原因是我觉得数据足够多，而且已经证明能让系统表现得相当不错，这又回到了模拟的问题上——如果你有足够的数据来做模拟，就能生成更多来自正确分布的合成数据。

65:34

about running out of highquality data especially high quality human data >> I'm not very worried about that partly because I think there's enough data uh or and it's been proven to get the systems to be pretty good and this goes back to simulations again if you do you have enough data to make simulations or so that you can create more synthetic data that are from the right distribution.

关于高质量数据，尤其是高质量人类数据即将耗尽的问题 >> 我不太担心这个，部分原因是我觉得数据足够多，而且已经证明可以让系统变得相当好。

65:56

Obviously, that's the key.

这又回到了模拟：如果你有足够的数据来创建模拟，那么你就可以生成更多来自正确分布的合成数据。

65:58

So, you need enough real world data in

显然，这是关键。

66:00

order to be able to uh uh create those kinds of generator data generators and um I think that we're at that step at the moment. >> Yeah, you've done a lot of incredible stuff on the side of science and biology doing a lot with not so much data. >> Yeah. >> I mean, it's still a lot of data, but I guess enough >> take off that going.

关于高质量数据，尤其是高质量人类数据即将耗尽的问题 >> 我对此不太担心，部分原因是我认为数据足够多，嗯，而且已经证明可以让系统变得相当不错，这又回到了模拟的问题上。

66:19

Exactly.

如果你有足够的数据来构建模拟，或者可以生成更多来自正确分布的合成数据。

66:20

Yeah. >> So exactly >> uh how crucial is the scaling of compute to building AGI?

显然，这是关键。

66:24

This is a question that's an engineering question.

所以，你需要足够的真实世界数据，才能嗯嗯创建那种数据生成器，而且我认为我们现在已经走到了这一步。

66:27

It's a

>> 是啊，你在科学和生物学领域做了很多了不起的事情，用不算太多的数据就取得了很大成就。

66:28

almost geopolitical question >> because it also integrated into that is the supply chains and energy a thing that you care a lot about which is um potentially fusion.

这几乎是一个地缘政治问题 >> 因为其中还整合了供应链和能源——这是你非常关心的一件事，也就是聚变。

66:37

So innovating on the side of energy also.

所以在能源方面也要创新。

66:39

Do you think we're going to keep scaling compute? >> I think so for several reasons.

你觉得我们会继续扩展算力吗？

66:44

I think compute there's there's the amount of compute you have for training often it needs to be colloccated so actually even like you know uh bandwidth constraints

>> 我认为会，有几个原因。

66:54

between data centers can affect that so it's it's it's there's additional constraints even there and that that's important for training obviously the largest models you can but there's also because now AI systems are in products and being used by billions of people around the world you need a ton of inference compute now um and then on top of that there's the thinking systems, the new paradigm uh of the last year that uh where they get smarter the

数据中心之间的连接会影响这一点，所以它……它……它甚至在那里还有额外的限制，而这对训练显然很重要——你可以训练最大的模型——但同时也因为现在AI系统已经产品化，被全球数十亿人使用，你需要大量的inference算力。

67:19

longer amount of inference time you give them at test time.

你给它们的推理时间越长，测试时表现就越好。

67:22

So all of those things need a lot of compute and I don't really see that slowing down.

所以所有这些都需要大量算力，而且我看不出这个需求会放缓。

67:27

Um and as AI systems become better, they'll become more useful and there'll be more demand for them.

嗯，随着AI系统变得更好，它们会更有用，需求也会更大。

67:33

So both from the training side, the training side actually is is only just one part of that.

所以从训练侧来看，训练侧其实只是其中一部分，甚至可能成为整体所需算力中较小的一部分。

67:38

It may even become the smaller part of of what's needed um uh in the overall compute that that's required.

对，这有点像那种 meme 式的东西，就是 V3 的成功和惊人之处，人们会调侃说它越成功，服务器就越吃力。

67:44

Yeah, that's one sort of almost memey kind of thing which is

>> 对，完全就是那个区别 >> 是啊。

67:47

like the success and the incredible aspects of V3 there people kind of make fun of like the more successful it becomes the you know the servers are sweating. >> Yes, exactly the difference in >> Yeah.

V3的成功和那些令人惊叹的地方，大家其实会调侃说它越成功，服务器就越吃不消。

67:58

Yeah.

没错，区别就在于。

67:58

Exactly.

对对对，我们之前还拍了个小视频，服务器在煎鸡蛋什么的，确实是这样。

67:58

We did a little video of of the servers frying eggs and things and um that's right and and and we're going to have to figure out how to do that.

我们得想办法解决这个问题。

68:06

Um there's a lot of interesting hardware innovations that we do as you know we have our own TPU line and we're looking at like inference only things

我们在硬件创新上做了很多有趣的事，比如我们有自己的TPU产品线，也在研究专门做inference的芯片，以及怎么让它们更高效。

68:14

inference only chips and how we can make those more efficient.

数据中心之间的连接会影响这一点，所以即使在那里也有额外的限制，这对训练显然很重要，尤其是训练最大的模型。

68:18

We're also very interested in building AI systems and we have done the help with energy usage so help um data center energy like for the cooling systems be efficient um grid optimization um and then eventually things like helping with plasma containment fusion reactors.

但同时也因为现在AI系统已经产品化，被全球数十亿人使用，你需要大量的推理计算。

68:34

We've done lots of work on that with Commonwealth Fusion and also uh one could imagine reactor design.

此外，还有去年出现的新范式——思考系统，它们变得更聪明了。

68:40

Um

呃，

68:40

and then material design I think is one of the most exciting new types of solar material solar panel material super room temperature superconductors has always been on my list of dream breakthroughs and um optimal batteries and I think a solution to any you know one of those things would be absolutely revolutionary for you know climate and energy usage and we're probably close you know again in the next 5 years to having AI systems that can materially help with those problems.

至于那是不是我们做搜索需要做的具体方式，但永远不要低估自然的力量，它在这里做到的这些。

69:06

If you were to bet, sorry for

>> 是啊。

69:08

the ridiculous question, but what what is the main source of energy in like 20, 30, 40 years, do you think it's going to be nuclear fusion? >> I think fusion and solar are the two that I I would bet on.

那个问题有点离谱，但你觉得未来20、30、40年，主要的能源来源会是什么？

69:20

Um solar, I mean, you know, it's the fusion reactor in the sky, of course, and I think really the problem there is is is batteries and transmission.

会是核聚变吗？

69:28

So you know as well as more efficient more and more efficient solar material perhaps eventually you

>> 我觉得聚变和太阳能是我会押注的两样东西。

69:34

know in space you know these kind of Dyson sphere type ideas and fusion I think is definitely doable seems uh if we have the right design of reactor and we can control the plasma and uh fast enough and so on and I think both of those things will actually get solved so we'll probably have at least those will probably be the two primary sources of renewable clean almost free or perhaps

在太空中，你知道那种戴森球之类的想法，还有核聚变，我觉得绝对可行，只要反应堆设计对路，能控制住等离子体，速度够快之类的。

69:59

free energy What a time to be alive.

推理专用芯片，以及我们如何让它们更高效。

70:02

If I uh traveled into the future with you 100 years from now, how much would you be surprised if we've passed a type one card scale civilization?

我们也对构建AI系统非常感兴趣，并且在能源使用方面提供了帮助，比如帮助数据中心冷却系统更高效，优化电网，最终还有像帮助等离子体约束聚变反应堆之类的事情。

70:12

I would not be that surprised if there was a like a 100redyear time scale from here.

我们与Commonwealth Fusion在这方面做了很多工作，也可以想象反应堆设计。

70:17

I mean, I think it's pretty clear if we crack the energy problems in one of the ways we've just discussed, fusion or or very

我觉得，如果我们用刚才讨论的某种方式解决了能源问题，比如核聚变或者非常……

70:26

efficient solar, um, then if energy is kind of free and renewable and clean, um, then that solves a whole bunch of other problems.

高效太阳能，嗯，然后如果能源基本上是免费、可再生且清洁的，嗯，那就能解决一大堆其他问题。

70:33

So, for example, the water access problem goes away because you can just use desalination.

比如说，水资源获取问题就不存在了，因为你可以直接用海水淡化。

70:37

We have the technology, it's just too expensive.

我们有这个技术，只是太贵了。

70:40

So, only, you know, fairly wealthy countries like Singapore and Israel and so on like actually use it.

所以只有，你知道，像新加坡和以色列这样相当富裕的国家才真正在用。

70:46

But but if it was uh cheap then every then you know all countries that have a coast could

但如果它便宜的话，那么，你知道，所有有海岸线的国家都能用上。

70:50

but also you'd have unlimited rocket fuel.

而且你还会拥有无限的火箭燃料。

70:53

You could just separate sea water out into hydrogen and oxygen using energy and that's rocket fuel.

你可以直接用能量把海水分解成氢气和氧气，那就是火箭燃料。

70:59

So uh combined with you know Elon's amazing self landing rockets then it could be like you sort of like a bus service to to space.

所以，结合Elon那不可思议的自动着陆火箭，这就像是一种通往太空的公交服务。

71:07

So that opens up you know incredible new resources and domains. uh asteroid mining I think will become a thing and maximum human flourishing to the stars.

这样一来，就能打开无数新的资源和领域。

71:16

That's what I uh dream about

我觉得小行星采矿会成为现实，人类也能最大程度地繁荣发展，冲向星辰。

71:18

as well is like Carl Sean's sort of idea of bringing consciousness to the universe, waking up the universe.

材料设计我认为是最令人兴奋的新方向之一，比如新型太阳能材料、太阳能板材料、超室温超导体，这些一直是我梦想中的突破。

71:24

And I I think human civilization will do that in the full sense of time if we get AI right and uh and and and crack some of these problems with it. >> Yeah.

还有最优电池，我认为解决其中任何一个问题，都会对气候和能源使用产生革命性的影响。

71:32

I wonder what it would look like if you just a tourist flying through space.

我们可能在未来5年内就能拥有可以切实帮助解决这些问题的AI系统。

71:37

You would probably notice Earth because if you solve the energy problem, you would see a lot of space rockets probably.

如果你要打赌，抱歉。

71:43

So it would be like traffic

那就会像交通一样。

71:45

here in London. >> But in space, >> just a lot of rockets >> and then you would probably see floating in space some kind of source of energy like solar. >> Yeah. >> Potentially.

就像Carl Sagan那种把意识带到宇宙、唤醒宇宙的想法。

71:56

So earth would just look more on the surface more um technological and then then you would use the power of that energy then to preserve the natural >> yes >> like the rainforest and all that kind of stuff >> because for the first time in in human

我也认为，如果我们把AI搞对了，并用它解决一些难题，人类文明终将在时间的长河中实现这一点。

72:12

history we wouldn't be uh resource constrainted and I think that could be amazing new era for humanity where it's not zero sum right I have this land you don't have it or if we take you know if the tigers have their forest just then the the local villagers can't what are they going to use?

>> 是啊。

72:30

I I I think that this will help a lot.

我在想，如果你只是个在太空飞行的游客，会看到什么景象。

72:32

No, it won't solve all problems because there's still other human foibless that will will will still

你可能会注意到地球，因为如果能源问题解决了，你大概会看到很多太空火箭。

72:38

exist, but it will at least remove one I think one of the big vectors which is scarcity of resources, you know, including land and more materials and energy and um you know, we should be I sometimes call it like and others call it about this kind of radical abundance era where um there's plenty of resources to go around.

>> 但在太空里，

72:55

But of course the next big question is making sure that that's fairly you know shared fairly uh and everyone in society benefits from that.

但当然，下一个大问题是确保这能被公平地分享，嗯，公平地，并且让社会中的每个人都能受益。

73:02

So there is something about human nature where I go you know it's like borat like my neighbor like I like you start trouble we we we do start conflicts and that's why games throughout as I'm learning actually more and more even in ancient history serve the purpose of pushing people away from war actually hot war so maybe we can figure out

>> 就是一大堆火箭

73:28

increasingly sophisticated video games that pull us they that give us that uh scratch the itch of like conflict, whatever that is, about us, the human nature, and then avoid the actual hot wars that would come with increasingly sophisticated technologies because we're now long past the stage where the weapons we're able to create can actually just destroy all of human

>> 然后你可能会看到太空中漂浮着某种能源，比如太阳能。

73:55

civilization.

>> 对。

73:56

So, it's no longer um that's no longer a great way to to uh start with your neighbor.

所以，这不再是，嗯，不再是跟邻居开始较量的好方式了。

74:01

It's better to play a game of chess >> or football or football.

更好的办法是下一盘棋 >> 或者踢足球，踢足球。

74:05

Yeah. >> And I think I mean I think that's what my modern sport is.

对。 >> 我觉得，嗯，我觉得这就是现代体育的意义。

74:09

So, and I love football watching it and and I just feel like uh and I used to play it a lot as well and it's it's it's it's very visceral and it's tribal and I think it does channel a lot of those energies

所以，我喜欢看足球，也踢过很多，它非常直观，很有部落感，我觉得它确实疏导了很多那种能量。

74:22

into a which I think is a kind of human need to belong to some some group and um but into a into a into a fun way, a healthy way and and a not a not destructive way kind of constructive uh thing.

我觉得这其实是一种人类对归属感的需求，想要属于某个群体，但用一种有趣、健康的方式，而不是破坏性的方式，而是建设性的东西。

74:32

And I think going back to games again is I think they're originally why they're so great as well for kids to play things like chess is they're great little microcosm simulations of the world.

再回到游戏这个话题，我觉得游戏最初之所以对孩子们来说这么棒，比如下棋，是因为它们是世界的绝佳微缩模拟。

74:43

They are simulations of the world too.

它们也是世界的模拟，是现实世界某些情境的简化版本，不管是扑克、围棋、象棋的不同方面，还是外交中的不同方面。

74:45

They're simplified versions of some real world situation, whether it's poker

而且还能让你在其中练习。

74:49

or or go or chess, different aspects or diplomacy, different aspects of of the real world.

数据中心之间的物理距离会影响这一点，所以即使在那里也有额外的限制，这对训练最大的模型显然很重要。

74:54

And allows you to practice at them, too.

但同时也因为现在AI系统已经产品化，被全球数十亿人使用，你需要大量的推理算力。

74:56

And and cuz, you know, how many times do you get to practice a massive decision moment in your life, you know, what job to take, what university to go to, you know, you get maybe, I don't know, a dozen or so key decisions one has to make, and you got to make those as best as you can.

然后在此基础上，还有thinking systems，这是过去一年出现的新范式，它们会变得越来越聪明。

75:11

Um, and games is a kind of safe environment, repeatable environment where you can get

嗯，游戏是一种安全的环境，可重复的环境，你可以在其中获得……

75:16

better at your decision- making process.

嗯，我觉得练习输和赢也非常重要，对吧？

75:18

Um, and it maybe has this additional benefit of channeling some energies into uh into more creative and constructive pursuits. >> Well, I think it's also really important to practice um losing and winning, >> right? >> Like losing is a really, you know, that's why I love games.

输其实是很重要的，这就是为什么我喜欢游戏，甚至喜欢像巴西柔术这样的东西。

75:34

That's why I love even um things like uh Brazilian jiu-jitsu. >> Yeah. >> Where you can get your ass kicked in a safe environment over and over.

对。

75:42

It reminds you about

在那里你可以在一个安全的环境里反复被揍，这能提醒你关于物理规律、世界运作的方式，有时候你会输，有时候你会赢，但大家还是可以做朋友。

75:43

>> the way about physics, about the way the world works, about sometimes you lose, sometimes you win.

关于物理，关于世界运作的方式，关于有时候你会输，有时候你会赢。

75:49

You can still be friends with everybody.

你仍然可以和所有人做朋友。

75:51

But that that feeling of losing, I mean, it's a weird one for us humans to like really like make sense of like that's just part of life.

但那种失败的感觉，对我们人类来说，真的很难去真正理解，就像那只是生活的一部分。

75:58

That is a fundamental part of life is losing. >> Yeah.

失败是生活的基本组成部分。

76:01

And I think in martial arts as I understand it, but also in things like light chess is a at least the way I took it, it's a lot to do with self-improvement, self-nowledge.

>> 是的。

76:11

You

而且我认为在武术中，据我所知，还有像国际象棋这样的东西，至少我是这么理解的，很大程度上关乎自我提升和自我认知。

76:11

know that, okay, so I did this thing.

知道吗，好吧，我确实做过这件事。

76:13

It's not about really being the other person.

这其实不是要成为另一个人，而是最大化你自己的潜力。

76:15

It's about maximizing your own potential.

如果你用健康的方式去做，你会学会如何利用胜利和失败。

76:17

If you do it in a healthy way, you learn to use victory and losses in a way.

不要被胜利冲昏头脑，觉得自己就是全世界最牛的；而失败能让你保持谦逊，时刻知道总有更多东西要学，总有更厉害的专家可以指导你。

76:21

Don't get carried away with victory and and think you're the just the best in the world.

我觉得你从中学到的就是这个，我很确定。

76:26

Keep and and and the losses keep you humble and always knowing there's always something more to learn. there's always a bigger expert that you can mentor you, you know, I think you learn that I'm pretty sure in

坚持和坚持再坚持，那些损失会让你保持谦逊，让你始终知道总有更多东西要学。总会有更厉害的专家可以指导你，我觉得你肯定能学到这一点。

76:36

martial arts and and and I think that's also uh the way that at least I was trained in chess.

武术和和和，我觉得这也是我学棋的方式。

76:41

And so in the same way and it can be very hardcore and very important and of course you want to win, but you also need to learn how to deal with setbacks in a in a healthy way that and and and and wire that that feeling that you have when you lose something into a constructive thing of next time I'm going to improve this, right?

所以同样地，它可以非常硬核、非常重要，当然你想赢，但你也需要学会如何健康地面对挫折，把输掉时的那种感觉，转化为建设性的东西——下次我要改进这个，对吧？

76:58

Or get better at this.

或者在这方面做得更好。

76:59

There is something that's a source of happiness, a source of meaning, that improvement step.

那种进步本身，就是快乐和意义的来源。

77:03

It's

它不在于输赢。

77:04

not about the winning or losing. >> Yes.

>> 对。

77:06

The mastery.

是 mastery。

77:06

There's nothing more satisfying in a way is like, "Oh, wow.

没有什么比“哇，我以前做不到的事，现在能做到了”更让人满足的了。

77:09

This thing I couldn't do before, now I can." And and and again, games and physical sports and and mental sports, they're way they're ways of measuring.

而且，游戏、体育运动、脑力运动，它们都是衡量的方式。

77:17

They're beautiful because you can measure that that progress. >> Yeah.

它们很美，因为你能看到自己的进步。

77:21

I mean there's something about this is why I love role playing games like the uh number go up of like my on the skill tree like literally that is a source of meaning for us humans whatever our

我是说，这就是为什么我喜欢角色扮演游戏，比如技能树上的数字往上涨——对我们人类来说，这本身就是一种意义来源，无论我们做什么。

77:31

>> yeah we're quite we're quite addicted to this sort of yeah these numbers going up and uh and and and and maybe that's why we made games like that because obviously that is something we're we're hill climbing systems ourselves right >> yeah it would be quite sad if we didn't have any mechanism by >> color belts all of the we do we do this everywhere right where we just have this thing that >> it's and I don't want to dismiss that that there is a source of deep meaning for us as humans.

>> 是啊。

77:56

U so one of the

这就是为什么我喜欢角色扮演游戏，比如技能树上的数字往上涨——对我们人类来说，那本身就是一种意义来源，不管我们……

77:57

incredible stories on the business on the leadership side is um what Google has done over the past year.

在商业和领导力方面，过去一年Google的故事真的很不可思议。

78:03

So I uh I think it's fair to say that Google was losing on the LLM product side uh a year ago with Gemini 15 and now it's winning with Gemini 25 and you took the helm and you led this effort.

我觉得可以说，一年前Google在LLM产品方面是落后的，当时有Gemini 1.5，但现在凭借Gemini 2.5它正在赢，而你接手并领导了这项工作。

78:15

What did it take to go from, let's say, quote unquote losing to quote unquote winning in the in in the span of a year? >> Yeah.

从所谓的“落后”到所谓的“赢”，这一年里到底做了什么？

78:23

Well, firstly, it's absolutely

>> 嗯，首先，这绝对是我们拥有的一个不可思议的团队，由Cory、Jeff Dean和Oral领导，还有我们Gemini团队里那些出色的人。

78:25

incredible team that we have, you know, led by Cory and Jeff Dean and and Oral and the amazing team we have on Gemini.

>> 对，我们确实对这种数字上涨上瘾，嗯，也许这就是为什么我们做出这样的游戏，因为我们自己就是 hill climbing 系统，对吧？

78:32

Absolutely world class.

绝对是世界级的。

78:34

So, you can't do it without the best talent.

所以，没有最优秀的人才，你根本做不到。

78:36

Um, and of course, you have, you know, we have a lot of great compute as well.

嗯，当然，我们也有大量的算力。

78:41

But then it's the research culture we've created, right? and basically coming together both different groups in in Google you know there was Google brain world-class

但关键还是我们创造的研究文化，对吧？就是把谷歌内部不同的团队聚集在一起，比如Google Brain就是世界级的。

78:51

team and and then the old deep mind and pulling together all the best people and the best ideas and gathering around to make the absolute greatest system we could hard um but we're all very competitive uh and we you know love research this is so fun to do um and we you know it's great to see our trajectory wasn't a given but we're very pleased with um the the where we are in the rate of progress

>> 是啊，如果我们没有任何机制……

79:17

is the most important thing.

>> 色带、段位，我们到处都在做这种事，就是有这么一个东西……

79:19

So if you look at where we've come two from 2 years ago to one year ago to now you know I think our we call it relentless progress along with relentless shipping of that progress is um being very successful and you know um it's unbelievably competitive uh the whole space the whole AI space with some of the greatest entrepreneurs and leaders uh and companies in the world all competing now because everyone's

所以，如果你看看我们从两年前到一年前再到现在的进展，我觉得我们称之为“ relentless progress”（不懈进步），同时不断把这种进步落地，这非常成功。而且，整个AI领域竞争极其激烈，全球最顶尖的企业家、领导者和公司都在参与竞争，因为每个人都在……

79:44

realized how important AI is um and it's very you know been pleasing for us to see that progress you know, Google's a gigantic company.

>> 我不想否认，这对我们人类来说是一种深层意义的来源。

79:51

Uh can you speak to the natural things that happen in that case is the bureaucracy that emerges like you want to be careful like you know like that the natural kind of there's there's meetings and there's managers and that like what what are some of the challenges from a leadership perspective breaking through that in order to like you said ship like the the number of

所以，我们有一个非常棒的团队，由 Cory、Jeff Dean、Oral 领导，还有 Gemini 上那些出色的伙伴。

80:10

products >> Gemini related products that's been shipped over the past year is just insane >> right it is yeah exactly that's that's what relentlessness looks like um I think it's it's a question of like any big company you know ends up having uh a lot of layers of management and things like that is sort of the nature of how it works.

我们都很 competitive，热爱研究，这太有趣了。

80:31

Um but I still operate and I was always operating with old Deep Mind as a as a startup still large one but

看到我们的轨迹——虽然不是必然的——但我们很满意现在的位置和进步的速度。

80:37

still as a startup and that's what we still act like today as with Google Deep Mind and acting with decisiveness and the energy that you get from the best smaller organizations and we try to get the best of both worlds where we have this incredible billions of users surfaces uh incredible products that we can power up with our AI and our and our research.

作为一家初创公司，我们至今仍然保持着这种作风，就像现在的Google DeepMind一样，以最佳小型组织特有的果断和活力来行动。

80:57

Um, and that's amazing.

我们试图兼得两者之长：一方面拥有数十亿用户的平台和优秀产品，可以用我们的AI和研究来赋能；另一方面，这种组合非常罕见——全球没几个地方能让你一边做世界级的前沿研究，第二天就能把成果落地，改善数十亿人的生活。

80:59

And you can, you know, that's very few places in the world you can get that do

我们一直在努力砍掉官僚主义，让研究文化和快速交付文化能够蓬勃发展。

81:03

incredible world-class research on the one hand and then plug it in and improve billions of people's lives the next day.

高效的太阳能，嗯，如果能源基本上是免费、可再生且清洁的，嗯，那就能解决一大堆其他问题。

81:10

Uh, that's a pretty amazing combination.

比如说，水资源获取问题就消失了，因为你可以直接用海水淡化。

81:12

And we're continually fighting and cutting away bureaucracy to allow the research culture and the relentless shipping culture to flourish.

我们有这个技术，只是太贵了。

81:19

And I think we've got a pretty good balance whilst being responsible with it, you know, as you have to be as a large company and also uh with a number of, you know, uh huge product surfaces that

所以只有像新加坡和以色列这样相当富裕的国家才真正在用。

81:30

we have. >> Uh so a funny thing you mentioned about like the the surface of the billion.

嗯，有个挺有意思的事，你刚才提到“冰山表面”这个说法。

81:36

I I had a conversation with a guy named um brilliant guy uh here at the British Museum called Irvin Fininkle.

我之前在大英博物馆跟一个叫Irvin Fininkle的人聊过，他是个天才，是世界级的楔形文字专家，就是那种古代泥板上的文字。

81:44

He's a world expert at Kuneaforms, which is a ancient writing on tablets. and he doesn't know about Chad GBT or Gemini.

但他完全不知道ChatGPT或者Gemini，对AI也一无所知。

81:52

He doesn't even know anything about AI.

他第一次接触AI

81:55

But his first encounter with this AI

但他第一次接触这个AI的时候……

81:58

>> is AI mode on Google.

>> 是Google的AI模式。

81:59

Yes. >> He's like, "Is that what you're talking about?

对。

82:02

This AI mode and you know, it's just it's just a reminder that there's a large part of the world that doesn't know about this AI thing." >> Yeah.

>> 他就问，“这就是你说的那个吗？

82:09

I know.

这个AI模式？

82:09

It's funny cuz if you live on uh X and Twitter and I mean it's sort of at least my feed it's all AI and and there's certain places where you know in the valley and certain pockets where everyone's just all they're thinking about is AI but a lot of the normal world hasn't hasn't come across it yet

” 你知道吗，这提醒了我们，世界上还有很大一部分人根本不知道AI这回事。

82:24

but >> that's a great responsibility to the their first interaction >> on the the the grand scale of the rural India or anywhere across the world like you get to >> right and we want it to be as good as possible and in a lot of cases it's just under the hood powering making something like maps or search work better and um and it's ideally for a lot of those people should just be seamless.

但 >> 他们的第一次接触，责任重大啊。

82:46

It's just new technology that makes their lives more, you know, productive and and and helps them. >> A bunch of folks on the Gemini product

>> 在印度农村那种大范围的地方，或者世界任何角落，你都得

82:54

and engineering teams spoken extremely highly of you on another dimension that I almost didn't even expect cuz I kind of think of you as the like deep scientist and caring about these big research scientific questions.

>> 对，我们希望它尽可能好用。

83:06

But they also said you're a great product guy like how to create a thing that a lot of people would use and enjoy using.

很多时候，AI只是藏在底层，让地图或者搜索这类东西更好用。

83:13

So can you maybe speak to what it takes to create a a AI based product that a lot of people would enjoy using?

对很多人来说，理想状态就是无缝体验。

83:20

>> Yeah.

都对你评价极高，而且是在一个我几乎没想到的维度上。

83:21

Well, I mean again that comes back from my game design days where I used to design games for millions of gamers.

因为我一直觉得你是那种深度科学家，关心那些宏大的科研问题。

83:27

People would forget about that.

但他们也说，你是个很棒的产品人，知道怎么做出大家愿意用、喜欢用的东西。

83:29

I've had experience with cutting edge technology in product.

所以，你能不能聊聊，要做出一个大家喜欢用的AI产品，需要些什么？

83:33

That that that that is how games was in the '90s.

九十年代的游戏就是那个样子。

83:36

And so I love actually the combination of cutting edge research and then being applied in a product and to power a new experience.

所以我其实很喜欢这种结合——前沿研究直接落地到产品里，驱动全新的体验。

83:43

And so um I think it's the same skill really of of you know imagining what it

我觉得这本质上就是同一种能力，就是去想象它……

83:48

would be like to use it viscerally um and having good taste.

>> 嗯，这又得回到我做游戏设计的那些日子了，那时候我为几百万玩家设计游戏。

83:51

Coming back to earlier the same thing that's useful in science um I think is is can also be useful in in product design.

可能很多人都忘了这茬。

83:58

And um I I've just had a very you know always been a sort of multi-disiplinary person.

我在产品里用过最前沿的技术，90年代的游戏行业就是这样。

84:03

So I don't see uh the boundaries really between you know arts and sciences or product and research.

所以我其实很喜欢把前沿研究和产品应用结合起来，去驱动一种新体验。

84:08

It's it's a continuum for me.

我觉得这跟想象

84:10

I mean I only work on I like working on products that are cutting edge.

我只做那种，我喜欢做前沿的产品。

84:14

I wouldn't be able to you

我根本没法……

84:15

know have cutting edge technology under the hood.

用户用起来是什么感觉、要有好的品味，是同一个技能。

84:18

I wouldn't be excited about them if they were just run-of-the-mill products.

回到之前说的，科学里有用的东西，在产品设计里也有用。

84:23

Um so it requires this invention creativity capability.

我一直是个跨领域的人，所以对我来说，艺术和科学、产品和研究之间没有界限，它是一个连续体。

84:26

What are some specific things you kind of learned about when you um even on the LLM side, you're interacting with Gemini, you're like this doesn't feel like the layout, the the interface, >> maybe the trade-off between the latency, like how >> how to present to the user how long to

我只喜欢做前沿的产品。

84:42

wait >> and how that waiting is shown or the reasoning capabilities.

>> 可能是在延迟和用户体验之间的权衡，比如怎么向用户展示，要等多久，

84:46

There's some interesting things cuz like you said, it's the very cutting edge.

有些挺有意思的地方，就像你说的，这玩意儿太前沿了。

84:50

We don't know >> how to present it, how to present it correctly.

我们根本不知道该怎么呈现它，怎么正确地呈现它。

84:54

So is there some specific things you've you've learned? >> I mean it's such a fast evolving space.

那你有没有学到什么具体的门道？

84:59

We're evaluating this all the time, but where we are today is that you want to continually simplify things.

我们一直在评估这个，但就目前而言，核心思路就是不断把东西做简单。

85:05

Um the whether that's the interface or all the inter what you build on top of the

嗯，不管是界面还是你在上面构建的各种东西。

85:10

model.

模型。

85:10

You kind of want to get out of the way of the model.

你其实应该给模型让路。

85:13

The model train is coming down the track and it's improving unbelievably fast.

模型这列火车正沿着轨道飞驰而来，进步速度快得惊人。

85:17

This relentless progress we talked about earlier.

就是我们之前聊到的那种 relentless progress。

85:19

You know, you look at 2.5 versus 1.5 and it's just a gigantic improvement.

你看 2.5 和 1.5 的对比，提升简直巨大。

85:23

And we expect that again for the future versions.

我们预期未来版本也会是这样。

85:26

And so the models are becoming more capable.

所以模型的能力越来越强。

85:28

So you've got the interesting thing about the design space in in in today's world these AI first products is you got to design not for what the thing can do today the

现在设计这些 AI-first 产品，有意思的地方在于，你不能只盯着今天技术能做什么，而要考虑一年后它能做什么。

85:37

technology can do today but in a year's time.

今天技术能做到什么，一年后又能做到什么。

85:39

So you actually have to be a very technical product person because uh you got to kind of have a good intuition for and feel for okay that thing that I'm dreaming about now can't be done today but is the research track on schedule to basically intercept that in 6 months or a year's time.

所以你实际上必须是一个非常懂技术的产品人，因为你需要有一种良好的直觉去判断：我现在梦想的东西今天还做不了，但研究路线图是否能在6个月或一年后赶上这个节点。

85:56

So you kind of got to intercept where this highly changing technology is going as well as that um uh uh new capabilities are coming online

你既要把握这个快速变化的技术走向，也要跟上新能力不断涌现的节奏。

86:04

all the time that you didn't realize before that can allow like D research to work or now we got video generation what do we do with that um this multimodal stuff you know is it one question I have is is it really going to be the current UI that we have today these textbox chats seems very unlikely given once you think about these super multimodal uh uh systems Shouldn't it be something more like Minority Report where you're you're sort of vibing with it in a in a in a

>> 我觉得被低估的艺术形式是界面设计。

86:32

kind of collaborative way?

有点协作的方式，对吧？

86:33

Right?

现在看起来限制很多。

86:34

It seems very restricted today.

我觉得再过几年，我们回头看今天的界面、产品和系统，会觉得它们相当原始。

86:36

I think we'll look back on today's interfaces and products and systems as quite archaic in maybe in just a couple of years.

所以我认为在产品端和研究端其实都有很大的创新空间。

86:43

So I think there's a lot of space actually for innovation to happen on the product side as well as the the research side.

然后我们私下聊到，键盘这个开放问题是：我们会以多快、多大程度转向以音频为主要交互方式？

86:50

And then we're offline talking about this keyboard is the the open question is how when and how much will we move to audio as the primary way of

而AI会根据任务来生成内容。

86:58

interacting with the machines around us versus typing stuff.

>> 你会觉得“让路”本身不是一种真正的艺术形式。

87:02

Yeah, I mean typing is a very low bandwidth way of doing even if you're very fast, you know, typer and I think we're going to have to start utilizing other devices whether that's smart glasses, you know, audio, earbuds, um, and eventually maybe some sorts of neural devices where we can increase the the input and the output bandwidth to something uh, you know, maybe 100x of what is today.

对，我是说，打字其实是一种带宽很低的交互方式，就算你打字速度很快也一样。我觉得我们得开始用其他设备了，比如智能眼镜、音频耳机，嗯，最终可能是一些神经接口设备，能把输入输出的带宽提升到现在的100倍。

87:26

>> I think that you know underappreciated art form is the interface design.

以一种协作的方式？

87:30

But I think you can not unlock the power of the intelligence of a system if you don't have the right interface.

对吧？

87:36

The interface is really the way you unlock its power.

今天看起来还非常受限。

87:39

It's such an interesting question of how to do that.

我觉得可能再过几年，我们回头看今天的界面、产品和系统，会觉得相当原始。

87:42

So h how >> you would think like getting out of the way isn't real art form. >> Yes.

所以我认为在产品端和研究端其实都有很大的创新空间。

87:47

You know, it's the sort of thing that I guess Steve Jobs always talked about, right?

我们私下聊过，这个开放问题是：我们会以何种方式、在何时、以多大程度转向以音频为主要交互方式？

87:52

It's simplicity, beauty,

简单就是美。

87:53

and elegance that we want, right?

>> 对。

87:55

And we're not there.

你知道，这大概就是 Steve Jobs 总在说的东西吧？

87:56

Nobody's there yet in my opinion.

我们想要的是 simplicity、beauty 和 elegance。

87:58

And that's what I would like us to get to.

但我们还没做到。

88:00

Again, it sort of speaks to like Go again, right?

在我看来，没人做到。

88:03

As a game, the most elegant, beautiful game.

这就是我希望我们达到的状态。

88:05

Can you, you know, that can you make an interface as beautiful as that?

这又让我想到围棋，作为一款游戏，它是最优雅、最美丽的。

88:09

And actually, I think we're going to enter an era of AI generated interfaces that are probably personalized to you so it fits the way that you your aesthetic, your feel, the way that your brain works.

你能做出像那样美丽的界面吗？

88:19

And um and

而且我觉得我们即将进入一个 AI 生成界面的时代，这些界面可能会为你个性化定制，贴合你的审美、你的感觉、你大脑的运作方式。

88:20

and and the AI kind of generates that depending on the task.

>> 对。

88:23

You know, that feels like that's probably the direction we'll end up in. >> Yeah.

因为有些人是重度用户，他们希望屏幕上每个参数都显示出来，所有东西都基于——比如我这种用键盘导航的人，我喜欢给所有操作设快捷键。

88:28

Because some people are power users and they want every single parameter on screen, everything, everything based like perhaps me with a key keyboard based navigation.

而有些人喜欢极简风格，把那些复杂性全藏起来。

88:36

I like to have shortcuts for everything.

我喜欢给所有东西都设快捷键。

88:38

And some people like the minimalism >> just hide all of that complexity.

有些人喜欢极简主义——就把所有复杂的东西都藏起来。

88:42

Exactly. >> Yeah.

没错。

88:43

Uh well, I'm glad you have a Steve Jobs mode in you as well.

嗯，很高兴你身上也有个乔布斯模式。

88:46

This is great.

这很棒。

88:47

Einstein mode, Steve Jobs mode.

爱因斯坦模式，史蒂夫·乔布斯模式。

88:49

Um all right, let me try to trick you into answering a question.

嗯，好吧，让我试着骗你回答一个问题。

88:53

When when will Gemini 3 come out?

Gemini 3 什么时候出？

88:55

Is it before or after GTA 6?

是在 GTA 6 之前还是之后？

88:57

The world waits for both.

全世界都在等这两个。

88:59

And what does it take to go from 25 to 3 0?

还有，从 2.5 到 3.0 需要什么？

89:02

Because it seems like there's been a lot of releases of 25 which are already leaps in performance. >> So what what does it even mean to go to a new version?

因为看起来已经出了很多 2.5 版本，性能上已经是飞跃了。

89:12

Is it about performance?

>> 那么，推出新版本到底意味着什么？

89:14

Is this about a completely different flavor of an experience? >> Yeah.

还是说这是一种完全不同的体验风格？

89:18

Well, so the way it works with our different uh version numbers is we you know we try to collect so maybe it takes you know roughly 6 months or something to to do a new kind of full run and the full productization of a new version and during that time lots of new interesting research iterations and ideas come up and we sort of collect them all together

>> 对。

89:40

that you know you could imagine the last 6 months worth of interesting ideas on the architecture front uh maybe it's on on the data front.

你可以想象，过去六个月里，架构方面有趣的想法，嗯，可能还有数据方面的。

89:48

It's like many different possible things and we collect package that all up, test which ones are likely to be useful for the next iteration and then bundle that all together and then we start the new you know giant hero training run right and and then uh and then of course that gets monitored uh and then at the end then there's the of the pre-training then there's all the post- training there's

有很多不同的可能性，我们收集打包所有东西，测试哪些可能对下一次迭代有用，然后整合在一起，接着就开始新的巨型英雄训练运行，对吧。

90:09

many different ways of doing that different ways of patching it so there's a whole experimental phase there which you can also get a lot of gains out and that's where you see the version numbers usually referring to the base model, the pre-trained model.

有很多不同的方式来做后训练，也有不同的修补方法，所以整个实验阶段也能带来很多收益。

90:22

And then the interim versions of 2.5, you know, and the different sizes and the different little additions, they're often uh patches or post-training ideas that can be done afterwards off the same basic architecture.

这就是版本号通常指基础模型、预训练模型的地方。

90:34

And then of course on top of that, we also have different sizes,

然后中间的 2.5 版本，你知道，还有不同尺寸和不同的小补充，它们通常是基于相同基础架构的补丁或后训练想法。

90:38

pro and flash and flashlight that are often distilled from the biggest ones, you know, the flash model from the Pro model.

Pro、Flash 和 Flashlight，它们通常是从最大的模型蒸馏出来的，比如 Flash 模型来自 Pro 模型。

90:45

And that means we have a range of different choices if you are the developer of do you want to prioritize performance or speed right and cost.

这意味着，如果你是开发者，你有多种选择，是优先考虑性能还是速度，还有成本。

90:54

And we like to think of this parto frontier of of you know on the one hand uh the y- axis is you know like performance and then the the the x-axis is you know cost

我们喜欢把这个看作帕累托前沿，一方面，Y 轴是性能，X 轴是成本

91:04

or latency and and speed uh basically and we we have models that completely define the frontier.

或者延迟和速度，基本上就是这样。

91:11

So whatever your trade-off is that you want as an individual user or as a as a developer, you should find one of our models satisfies that constraint. >> So behind the version changes, there is a big hero run. >> Yes. >> And then there's uh just an insane complexity of productization.

我们有完全定义这个前沿的模型。

91:31

Then there's the distillation of the different sizes along that predator front.

接着是沿着那个帕累托前沿对不同尺寸模型进行蒸馏。

91:35

And then as with each step you take, you realize there might be a cool product.

然后每走一步，你都会发现可能有个很酷的产品。

91:40

There's side quests. >> Yes. >> Exactly. >> But and then you also don't want to take too many side quests because then you have a million versions of million products.

还有支线任务。

91:49

It's very unclear, but you also get super excited because it's super cool.

>> 对。

91:53

Like how does even you look at VO how does it fit into the bigger thing? >> Exactly.

>> 没错。

91:58

Exactly.

>> 但你也不想接太多支线任务，因为那样会有上百万个版本和上百万个产品。

91:58

And then you're

非常混乱，但你也会超级兴奋，因为太酷了。

91:59

constantly this process of converging upstream we call it you know ideas from the from the product surfaces or or or from the post training and and even further downstream than that you you kind of upstream that into the the core model training for the next run.

一直在做这个我们称之为“上游汇聚”的过程，从产品表面或后训练，甚至更下游的地方来的想法，你会把它们上游到下一次运行的核心模型训练中。

92:15

Right.

对吧。

92:15

So then the main model the main Gemini track becomes more and more general and eventually you know AGI >> one hero run at a time. >> Yes.

所以主模型，也就是 Gemini 主线，变得越来越通用，最终，AGI >> 一次一个英雄运行。

92:23

Exactly.

>> 对，没错。

92:24

A few hero runs later.

再跑几次英雄运行之后。

92:25

>> Uh yeah.

嗯，是的。

92:26

So sometimes when you release these new versions or every version really are benchmarks um productive or counterproductive for showing the performance of a model you need them and and but it's important that you don't overfitit to them right so there shouldn't be the end the be all and end all so there's there's lmina or it used to be calledis that's one of them that turned out sort of organically to be one of the the main ways people like to test

所以有时候当你发布这些新版本，或者说每个版本其实都这样，基准测试对于展示模型性能来说，要么有用要么适得其反。

92:52

these systems at least the chat bots um obviously there's loads of academic benchmarks on from from that test mathematics and coding ability, general language ability, science ability and so on.

>> 我觉得被低估的艺术形式是界面设计。

93:02

And then we have our own internal benchmarks that we care about.

但如果你没有合适的界面，就无法释放系统的智能潜力。

93:06

It's a kind of multi-objective, you know, optimization problem, right?

界面真的是解锁其力量的关键。

93:09

You you don't want to be good at just one thing.

如何做到这一点，是一个很有趣的问题。

93:12

We're trying to build general systems that are good across the board and you try and make no regret uh

所以，如何 >> 你会觉得“让界面不碍事”本身就是一门真正的艺术形式。

93:18

improvements. though where you're improving like you know coding uh but it doesn't reduce your performance in other areas right so that's the hard part cuz you you can of course you could put more coding data in or you could put more um I don't know gaming data in but then does it make worse your language uh system or or uh in your translation systems and other things that you care about.

>> 没错。

93:41

So it's you've got to kind of continually monitor this increasingly

所以你得不断持续监控这个，越来越深入。

93:45

larger and larger suite of of benchmarks.

越来越大的基准测试套件。

93:47

And also there's uh when you stick them into products these models you also care about the direct usage and the direct stats and the signals that you're getting from the end users whether they're coders or or or the average person using using the chat interfaces. >> Yeah.

而且，当你把这些模型放进产品里时，你也会关心直接的使用情况、直接的统计数据，以及从终端用户那里得到的信号——不管他们是程序员，还是用聊天界面的普通人。

94:02

Because ultimately you want to measure the usefulness but it's so hard to convert that into a number right.

因为最终你想衡量的是实用性，但要把这个转化成数字实在太难了，对吧。

94:08

It's it's really vibe based benchmarks across a large number of users and it's

这基本上就是基于大量用户的“vibe”式 benchmark。

94:12

hard to know and I it would be just terrifying to me to you know you have a much smarter model but it's just something vibe based.

>> 对，因为最终你想衡量的是实用性，但要把这个转化成数字实在太难了，对吧。

94:20

It's not not not quite working.

这基本上就是基于大量用户的“vibe”式基准测试，而且很难判断。

94:22

That's such a scary cuz and everything you just said it has to be smart and useful across so many domains.

对我来说，想象一下你有一个更聪明的模型，但它的表现却只是靠“感觉”来判断，这太可怕了。

94:29

So you you get super excited because it's all of a sudden solving programming problems you've never been able to solve before. >> But now it's crappy poetry or something

如果它没完全奏效，那真的很吓人，因为你刚才说的所有东西——它必须在这么多领域里既聪明又有用。

94:40

and it's just I don't know that's a stressful that's so difficult >> um to balance and because you can't really trust the benchmarks you really have to trust the end users. >> Yeah.

>> 但现在它写出来的诗却很烂，或者别的什么，我就不知道了，这压力很大，也很难平衡，因为你不能完全相信基准测试，你只能相信终端用户。

94:50

And then other things that are even more esoteric come into play like um you know the style of the persona of the the the system you know how it you know is it verbose is it succinct is it humorous you know and and different people like different things so um you

然后还有一些更玄乎的因素会掺进来，比如系统的风格、人设——它是啰嗦还是简洁？有没有幽默感？不同的人喜欢不同的东西，所以……

95:06

know it's very interesting it's almost like cutting edge part of psychology research or person personality research you know I used to do that in my PhD like five factor personality what do we actually want our assistance to be like and different people will like different things as well.

>> 对。

95:21

So, these are all just sort of new problems in product space that I don't think have ever really been tackled before, but um we're going to sort of rapidly have to deal with now.

然后还有一些更玄乎的东西会掺和进来，比如系统的风格、个性——它话多还是话少，幽默还是不幽默，而且不同的人喜欢不同的东西。

95:30

I think is a super fascinating space developing the character of the thing

所以这很有趣，几乎像是心理学或人格研究的前沿领域。

95:34

and in so doing it puts a mirror to ourselves what are the kind of things um that we like cuz prompt engineering allows you to control a lot of those elements but can the product uh make it easier for you to uh control the different flavors of those experiences the different characters that you interact with. >> Yeah, exactly.

而在这个过程中，它也像一面镜子照出我们自己：我们到底喜欢什么样的东西？

95:53

So >> So what's the probability of Google Deep Mai winning?

因为prompt engineering能让你控制很多这些元素，但产品能不能让你更容易地控制这些体验的不同风格、你与之互动的不同角色？

95:57

Well, I don't see it as sort of winning.

嗯，我不觉得这算是“赢”。

95:59

I mean, I think we need

我是说，我们需要……

96:01

to think winning is the wrong way to look at it given how important and consequential what it is we're building.

把这件事看作“赢”其实是错的，考虑到我们正在构建的东西有多么重要和深远。

96:07

So, funnily enough, I don't I try not to view it like a game or competition, even though that's a lot of my mindset.

所以，有趣的是，我尽量不把它当成一场游戏或竞争，尽管那确实是我很多思维模式的一部分。

96:13

It's it's about in my view, all of us have those of us at the leading edge have a responsibility to um steward this unbelievable technology that could be used for incredible good, but also has risks. um steward it safely into the

在我看来，我们所有人——那些处于前沿的人——都有责任去引导这项不可思议的技术，它既能带来巨大的好处，也伴随着风险。

96:26

world for the benefit of humanity.

我同意你的看法。

96:28

That's always um what I've um uh uh I dreamed about and what we've always tried to do and I hope that's what eventually the community maybe the international community will rally around when it becomes obvious that as we get closer and closer to to AGI that um that's what's needed. >> I agree with you.

说得非常好。

96:47

I think that's beautifully put.

你说过，随着竞争加剧，你和一些实验室的负责人关系不错，也保持沟通。

96:48

You've said that um you talk to and are on good terms with the

维持这些关系有多难？

96:52

leads of some of these uh labs as the competition heats up.

到目前为止还好。

96:56

Um how hard is it to maintain sort of those relationships?

我努力让自己成为一个善于合作的人。

96:59

It's been okay so far.

我本来就是合作型的人。

97:00

I try to pride myself in being uh collaborative.

研究本身就是合作的事业，科学也是合作的事业，对吧？

97:03

I'm a collaborative person.

最终，如果你能治愈那些可怕的疾病，带来神奇的疗法，这对全人类都是净收益。

97:04

Research is a collaborative endeavor.

能源也是一样。

97:06

Science is a collaborative endeavor.

所有我想用AI帮助解决的问题都是如此。

97:08

Right?

所以我只是希望这项技术存在于世界上，被用在正确的地方，它的好处——比如生产力提升——能被所有人共享。

97:09

It's all good for humanity in the end if you cure incredible, you know, terrible diseases and you come with an incredible cure. this is net win for humanity and the

所以我努力和所有主要实验室的人保持良好关系。

97:18

same with energy.

而且我认为，当事情变得比现在更严峻时，保持这些沟通渠道会非常重要。

97:19

All of the things that I'm interested in in in helping solve with AI.

这也会促进合作或协作——尤其是在安全这类问题上。

97:23

So I just want that technology to exist in the world and be used for the right things and and and the the kind of the benefits of that the productivity benefits of that being shared for every the benefit of everyone.

我只是希望这项技术能存在于世界上，并被用在正确的地方，而它带来的生产力效益，能被所有人共享。

97:34

So I try to maintain good relations with all the leading lab uh people.

所以我尽量跟各大实验室的人保持良好关系。

97:38

They have very interesting characters many of them as you might expect.

他们很多人性格都挺有意思的，估计你也猜得到。

97:42

Um, but yeah, I'm on good terms.

嗯，不过呢，我跟他们关系都还不错。

97:43

I I hope with pretty much all of them.

我希望基本上跟所有人都处得来。

97:46

And uh I I think that's going to be important when when things get even more serious than they are now.

是的，我希望在一些风险较低的事情上也能有合作，这样也能作为维持友谊和关系的一种方式。

97:51

Uh that there are those communication channels and uh that's what will facilitate uh cooperation or collaboration if that's what is required especially on things like safety. >> Yeah, I hope there's some collaboration on stuff that's uh sort of less high stakes and in so doing serves as a mechanism for maintaining friendships and relationships.

比如，我觉得如果互联网能看到你和Elon合作做一款电子游戏，大家会很高兴。

98:10

So, for example, I think the internet would love it if you

这种事情能建立友谊和良好关系，而且你们俩都是真正的玩家。

98:14

and Elon somehow collaborated on creating a video game.

创作本身也很有趣。

98:17

That kind of thing that I think that enables camaraderie in good terms and also you two are legit gamers.

我觉得这种关系能带来 camaraderie（同事情谊）和良好互动，而且你们俩也是正经玩家。

98:22

So, it's just fun to Yeah. >> fun to create. >> Yeah, that would be awesome.

所以，这本身就挺有意思的。 >> 有意思的是去创造。 >> 对，那会很棒。

98:26

And we've talked about that in the past and it may be a cool thing that that you know we can do.

我们以前也聊过这个，这说不定是件我们能做的酷事。

98:32

And I agree with you.

我同意你的看法。

98:33

It'd be nice to have um kind of side projects in a way where where one can just lean into the collaboration aspect of it and it's

有个小项目其实挺好的，就是那种可以完全投入到协作里的感觉。

98:40

a sort of uh win-win for both sides and it's um and it kind of builds up that that that uh collaborative muscle. >> I see the scientific endeavor as that kind of side project for humanity and I I think deep Google deep mind has been really pushing that.

对，那会很棒。

98:53

I would love it if to see other labs do more scientific stuff and then collaborate cuz it just seems like easier to collaborate on the big scientific questions.

我们以前也聊过这个，可能确实是一件我们可以做的酷事。

99:02

I agree and I would love to see a lot of people a lot of the other labs talk about science but

我同意你的看法，如果能有一些副项目，让大家可以专注于合作的部分，那会很好。

99:07

I think we're really the only ones using it for science and doing that and that's why projects like Alpha Fold are so important to me and I think to our mission is to show uh how AI can vis you know be clearly used in a very concrete way for the benefit of humanity and and also we spun out companies like isomorphic off the back of Alphafold to do drug discovery and it's going really well and build sort of you know you can think of build additional alpha fold

我把科学事业看作人类的一个大型副项目。

99:32

type type systems to go into chemistry space to help accelerate drug design and the examples I think we need to show uh and society needs to understand what AI can bring these huge benefits. >> Well, from the bottom of my heart, thank you for pushing the scientific efforts forward wi with rigor, with fun, with humility, all of it.

类型系统进入化学空间，帮助加速药物设计，我觉得我们需要展示这些例子，社会也需要理解AI能带来这些巨大好处。

99:52

I just love to see and still talking about P equals NP.

>> 嗯，从心底里说，感谢你们以严谨、有趣和谦逊的态度推动科学进步。

99:56

I mean, it's just incredible.

我真的很喜欢看到你们还在讨论P是否等于NP，这太不可思议了。

99:57

So, I love

所以，我很喜欢。

99:58

it.

>> 对，没错。

99:58

Uh there there's been uh seemingly a war for talent.

嗯，感觉人才争夺战一直就没停过。

100:01

Some of it is meme, I don't know.

有些说法其实挺扯的，我也不确定。

100:03

Um, what do you think about Meta buying up talent with huge salaries and and the heating up of this battle for talent?

你怎么看Meta用高薪挖人，还有这场人才争夺战越来越激烈这件事？

100:11

And I I should say that I think a lot of people see Deep Mind is a really great place to do uh cutting edge work for the reasons that you've outlined is like there's this vibrant scientific culture.

我得说，很多人觉得Deep Mind是个做前沿研究的好地方，原因就像你刚才说的，那里有很活跃的科研氛围。

100:23

>> Yeah.

>> 那么Google DeepMind赢的概率有多大？

100:23

Well, look, of course, um, you know, there's a strategy that that Meta is taking right now.

嗯，我不觉得这是个“赢”的问题。

100:28

I think that um from my perspective at least I think the people that are real uh believers in the mission of AGI and what it can do and understand the real consequences both good and bad from that and what's what that responsibility entails I think they're mostly doing it to be like myself to be on the frontier of that research so you know they can help influence the way that goes and steward

我是说，我们需要它。

100:50

that technology safely into the world and you know meta right now are not at the frontier maybe they'll they'll manage to get back on there and um you know it's probably rational what they're doing from their perspective because they're behind and they need to do something but I think um there's more important things than than just money.

>> 对，当然。

101:06

Of course one has to pay you know people their market rates and all of these things and that continues to go up.

嗯，你知道，Meta现在采取了一种策略。

101:12

Um but as pro and and and I was expecting this because more and more people are

从我的角度来看，我认为那些真正相信AGI使命、理解它可能带来的好与坏后果、并明白其中责任的人，他们这么做主要是为了像我一样——站在研究的前沿，这样他们就能影响方向，并安全地把这项技术引入世界。

101:16

finally realizing leaders of companies what I've always known for 30 plus years now which is that AGI is the most important technology probably that's ever going to be invented.

终于，公司领导们意识到了我三十多年来一直明白的事——AGI很可能是人类有史以来最重要的技术。

101:25

So in some senses it's it's rational to be doing that.

从某种意义上说，他们这么做是理性的。

101:27

But I also think there's a much bigger question.

但我也觉得有个更大的问题。

101:30

I mean people in AI these days are very well paid.

现在搞AI的人收入都很高。

101:32

You know I I remember when we were starting out back in 2010 you know I didn't even pay myself for a couple of years because wasn't enough money.

你还记得吗，2010年我们刚起步的时候，我连着好几年没给自己发工资，因为根本没钱，也融不到资。

101:40

We couldn't raise any money.

而现在，实习生拿的薪水都快赶上我们当年第一轮种子轮融资的总金额了，想想还挺好笑的。

101:41

And these days interns are

我还记得以前我得免费干活，甚至自己贴钱去实习。

101:43

being paid you know the amount that we raised as our first entire seed round.

>> 嗯，我很高兴你也有个“史蒂夫·乔布斯模式”，这挺好的。

101:46

So it's pretty funny.

所以还挺好笑的。

101:48

And I remember the days where we used I used to have to to work for free and and almost pay my own way to do an internship. right now it's all the other way around but that's just how it is. it's the new world and um but I think that you know we've been discussing like what happens post AGI and energy systems are solved and so on what is even money going to mean so I think uh you know and the economy and and we're going to have much bigger

我还记得以前我得免费干活，甚至自己掏钱才能去做个实习。现在完全反过来了，但就是这样。这是个新世界。不过我觉得，我们一直在讨论AGI之后、能源问题解决之后会发生什么，到时候钱还有什么意义，经济又会变成什么样，我们会有更大的……

102:10

issues to work through and how does the economy function in that world and companies so I think you know it's a little bit of a side issue about uh uh salaries and things of like that today >> yeah when you're facing such gigantic consequences and and gigantic fascinating scientific questions >> which maybe only a few years away.

是啊，面对这么巨大的后果和这么迷人的科学问题，可能也就几年之遥了。

102:30

So, >> so on the practical pragmatic sense, if we zoom in on jobs, we can look at programmers because it seems like AI

所以，从实际务实的角度来说，如果我们聚焦到具体工作上，可以看看程序员，因为AI似乎……

102:37

systems are currently doing incredibly well in programming and increasingly so.

改进方面，比如你在提升编程能力，但不会降低其他领域的表现，对吧？

102:41

So, a lot of people that uh program for a living, love programming, are worried they will lose their jobs.

这就是难点，因为你可以加入更多编程数据，或者更多游戏数据，但这样会不会让你的语言系统、翻译系统或者其他你在意的方面变差？

102:48

How worried should they be, do you think? and what's the right way to uh sort of adjust to the new reality and ensure that you survive and thrive as a human in the programming world. >> Well, it's interesting that programming and it's again counterintuitive to what

所以你得持续监控这个越来越复杂的平衡。

103:03

we thought years ago maybe that some of the skills that we think of as harder skills are turned out maybe to be the easier ones for various reasons but you know coding and math because you can create a lot of synthetic data and verify if that data is correct.

我们几年前以为，那些我们觉得更难的技能，反而可能因为各种原因变得更容易了，比如编程和数学，因为你可以生成大量合成数据，并验证这些数据是否正确。

103:17

So because of that nature of that it's easier to make things like synthetic data to train from.

所以基于这个特性，用合成数据来训练就更容易了。

103:22

Um it's also an area of course we're all interested in because as programmers right to help us and get faster at it and more productive.

当然，这也是我们都感兴趣的领域，因为作为程序员，我们希望它能帮我们更快、更高效。

103:29

So I think the for the next

所以我认为，在下一个时代，比如未来5到10年，我们会发现那些拥抱这些技术的人，几乎会与它们融为一体，无论是在创意行业还是技术行业，他们都会变得超级高效。

103:30

era like the next 5 10 years I think what we're going to find is people who are kind of embrace these technologies become almost at one with them um whether that's in the creative industries or the technical industries will become sort of superhumanly productive I think.

所以从实际务实的角度来说，如果我们聚焦到具体工作上，可以看看程序员。

103:45

So the great programmers will be even better but they'll be even 10x even what they are today and because there you'll be able to use their skills to utilize the the tools to the maximum uh you exploit them

因为现在AI系统在编程方面表现极其出色，而且越来越强。

103:57

to the maximum and um so I think that's what we're going to see in the next domain um so that's going to cause quite a lot of change right and so that's coming a lot of people benefit from that so I think one example of that is if coding becomes easier um it becomes available to many more creatives to do more uh and uh but I think the top programmers will still have huge advantages as terms of specifying going back to specifying what the architecture

你拿到的报酬，差不多是我们整个种子轮融到的金额。

104:23

should be the question should be how to guide these um uh coding assistants in a way that's useful and you know check whether the code they produce is good so I think there's plenty of um uh headroom there for the foreseeable you know next few years >> so I think there's there's several interesting things there one is there's a lot of imperative to just get better and better consistently of using these tools.

嗯，编程这事儿挺有意思的，而且又有点反直觉。

104:45

So they they're riding the wave of the improvement improving models.

我觉得未来5到10年，我们会发现那些拥抱这些技术的人，几乎会跟技术融为一体——不管是在创意行业还是技术行业——他们会变得超级高效。

104:49

>> Yes. >> Versus like competing against them. >> But sadly, but that's the the nature of of life on earth.

系统目前在编程方面表现非常出色，而且越来越强。

104:58

Um there could be a huge amount of value to certain kinds of programming at the cutting edge and less value to other kinds.

所以很多以编程为生、热爱编程的人担心会失业。

105:08

For example, it could be like, you know, front end >> web design might uh be more amendable to

你觉得他们应该多担心？

105:17

to to as as you mentioned to generation >> uh by AI systems and maybe for example game engine design or something like this or backhand design or or guiding systems in high performance situations, high performance programming type of design decisions that might be extremely valuable.

就像你提到的，由AI系统生成的这种代际转变，比如游戏引擎设计、反手设计，或者高性能场景下的引导系统、高性能编程这类设计决策，可能会非常有价值。

105:33

But it it will shift where the humans are needed most and that's scary for people to adjust.

但这会改变人类最需要被投入的地方，这对人们来说调整起来很可怕。

105:39

I can I think that's right that the any time where

我觉得没错，每当出现大量颠覆和变化的时候——而且这不只是这一次，人类历史上已经经历过很多次了，比如互联网、移动设备，再往前还有工业革命——这将是其中一个充满巨变的时代。

105:42

there's a lot of disruption and change you know and we've had this it's not just this time we've had this in many times in human history with the internet u mobile but before that obviously industrial revolution um and it's going to be one of those eras where there will be a lot of change I think there'll be new jobs we can't even imagine today just like the internet created and then those people with the right skill sets to ride that wave will become incredibly uh valuable right those skills but maybe

>> 嗯，有意思的是，编程这件事——又跟直觉相反——

106:09

people will have to relearn or adapt a bit uh their current skills.

生产力提升带来的好处会被分享和分配给社会，可能以基础服务的形式出现。

106:13

And it's the the thing that's going to be harder to deal with this time around is that I think what we're going to see is something like probably 10 times the impact the industrial revolution had and but 10 times faster as well.

如果你想要更多，你还是得去掌握一些极其稀缺的技能，让自己变得独特。

106:25

Right?

但会有一个基本保障被提供出来。

106:25

So instead of 100 years, it takes 10 years.

如果你把政府看作一种技术，那就不只是经济学，还有政治学上也出现了有趣的问题：你怎么设计一个系统——

106:28

And so that's going to make it, you know, it's like a 100x uh the impact and the speed combined.

所以这会让它——你知道，相当于把影响力和速度结合起来，效果放大一百倍。

106:33

So that's what's I

这就是我的意思。

106:34

think going to make it more difficult for society to to to deal with and it's there's a lot to think through and I think we need to be discussing that right now and I I you know I encourage top economists in the world and philosophers to start thinking about um uh how should is society going to be affected by this and what should we do including things like um you know universal basic provision or something like that where a lot of the um

我觉得这会让社会更难应对，有很多东西需要想清楚，而且我认为我们现在就该讨论这些。

107:01

increased productivity uh gets shared out and distributed uh to society um and maybe in the form of surface services and other things where if you want more than that you still go and get some incredibly rare skills and things like that um and and make yourself unique.

生产力提升的成果会被分配和共享给社会，可能以基础服务的形式出现，如果你想要更多，那还是得去掌握一些极其稀缺的技能，让自己变得独特。

107:17

Um but uh uh but there's a basic provision that is provided >> and if you think of government as a technology there's also interesting questions not just in economics but just politics.

但基本的生活保障是提供的。

107:27

How do you design a system

>> 如果你把政府看作一种技术，那就不只是经济学的问题，还有政治学的问题。

107:29

that's responding to the rapidly changing times such that you can represent the different pain that people feel from the different groups?

>> 如果你把政府看作一种 technology，那就不只是经济学，政治学里也有有趣的问题。

107:39

And how do you reallocate resources in a way that um addresses that pain and represents the hope and the pain and the fears of different people uh in a way that doesn't lead to division because politicians are often really good at

你怎么设计一个系统，能应对快速变化的时代，同时代表不同群体感受到的痛苦？

107:57

sort of fueling the division and using that to get elected the other defining the other and then saying that's bad and sort of based on that >> I think that's often counterproductive ive to leveraging a rapidly changing technology how to help the world flourish.

在煽动分裂、利用这一点来当选、把对方定义为“他者”、然后说那很糟糕，并以此为基础来运作？

108:14

So we almost need to improve our political systems as well rapidly if you think of them as a technology >> definitely and I think I think we'll

我认为这往往对利用快速变化的技术来帮助世界繁荣是适得其反的。

108:23

need new governance structures institutions probably to help with this transition.

>> 我觉得这往往对利用快速变化的 technology 来帮助世界繁荣是适得其反的。

108:28

So I think political philosophy and political science is going to be key uh to that.

所以我们几乎也需要快速改进我们的政治系统，如果你把它们也看作一种 technology。

108:33

But I think the number one thing first of all is to create more abundance of resources right then there's the so that's the number one thing increase productivity get more resources maybe eventually get out of the zero sum situation then the second question is how to use uh those

但我认为首先最重要的事情是创造更多的资源富余，对吧？这是第一位的——提高生产力，获取更多资源，也许最终能摆脱零和博弈的局面。然后第二个问题是如何利用这些资源。

108:50

resources and distribute those resources but yeah you can't do that without having that abundance first.

一方面是顶尖的世界级研究，然后第二天就能把它应用到产品里，改善几十亿人的生活。

108:57

Uh you mentioned to me uh the book the maniac uh by Benjamin Levitut a book on uh first of all about you there's a bio about you um strange yeah >> it's unclear yeah sure it's unclear how much is fiction how much is reality um but I think the central figure that is John vonman I would say it's a haunting

嗯，这个组合真的很了不起。

109:18

and beautiful exploration of madness and genius and let's say the double-edged uh sword of discovery And you know for um people who don't know John vonman is a kind of legendary mind.

>> 确实如此，我认为我们需要新的治理结构和制度，可能来帮助这个过渡。

109:30

He contributed to quantum mechanics.

所以政治哲学和政治科学会非常关键。

109:33

He was on the Manhattan project.

但首先第一件事是创造更多的资源丰裕，对吧？

109:35

He is widely considered to be the father of or pioneer the modern computer and AI and so on.

这是第一位的：提高生产力，获得更多资源，也许最终走出零和博弈的局面。

109:41

So as many people say he's like one of the smartest

然后第二个问题才是如何使用和分配这些资源。

109:45

humans ever.

需要资源和分配这些资源，但如果没有先实现富足，你是做不到这一点的。

109:46

So it's just fascinating.

你之前跟我提到过那本书《The Maniac》，作者是Benjamin Labatut，这本书首先是关于你的，有一篇你的传记，有点奇怪。

109:48

And what's also fascinating is as a person who saw nuclear science and physics become the atomic bomb.

对，不太清楚有多少是虚构、多少是现实。

109:55

So you you got to see ideas become a thing that has a huge amount of impact on the world.

但我认为核心人物是John von Neumann，我觉得这是一本令人难忘的书——

110:02

He also foresaw the same thing for computing. >> Yeah. >> He he and that's the a little bit again

他也预见到了计算领域的同样情况。 >> 对。 >> 他——这又有点回到之前的话题了。

110:09

beautiful and haunting aspect of the book. um than taking a leap forward and looking at this at least at all alpha zero alpha go alpha zero big moment that maybe John vonman's thinking was brought to to to to reality.

这本书里既美又令人深思的部分。

110:25

So I I I guess the question is um what do you think if you got to hang out with John Noman now?

嗯，比直接跳到未来看Alpha Zero、AlphaGo这些大事件更有意思，这些可能让冯·诺依曼的想法变成了现实。

110:32

What what would he say about what's going on?

所以我想问的是，如果你现在能和冯·诺依曼坐下来聊聊，你觉得他会怎么评价现在发生的一切？

110:36

>> Well, that would be an amazing experience. you know, he's a fantastic mind and and I also love the where he he spent a lot of his time at Princeton at the Institute of Advanced Studies, a very special place for thinking and um it's amazing how much of a polymath he was in the the spread of things he helped invent including of course the vonoyman architecture that all the modern computers are based on.

>> 嗯，那会是一次很棒的经历。

111:01

And um he had amazing foresight.

你知道，他有一个非凡的头脑，我也很喜欢他花了很多时间在普林斯顿高等研究院，那是一个非常特别的思考场所，而且他真是个博学家，他帮助发明的东西范围之广令人惊叹，当然包括所有现代计算机所基于的 von Neumann 架构。

111:02

I think he would

而且他有惊人的远见。

111:04

have loved where we are today and he would have um I think he would have really enjoyed Alph Go being you know games he also did game theory.

>> 那肯定是一次不可思议的经历。

111:11

I think he foresaw a lot of what would happen with learning machines systems that that that are kind of grown I think he called it rather than programmed.

你知道，他是个天才，而且我也很喜欢他当年在普林斯顿高等研究院度过的那段时光，那是个非常适合思考的地方。

111:19

I'm not sure how even maybe he wouldn't even be that surprised this the fruition of what I think he already foresaw in the 1950s. >> I wonder what advice he would give.

他真是个全才，涉猎极广，帮助发明了那么多东西，当然也包括所有现代计算机都基于的von Neumann架构。

111:28

You got to see the building of the atomic

他有着惊人的远见。

111:30

bomb with the Manhattan project.

曼哈顿计划那样的炸弹。

111:32

I'm sure there's >> interesting stuff that maybe is not talked about enough.

我相信肯定有一些>> 有趣的内容，可能没有被充分讨论。

111:36

Maybe some bureaucratic aspect, maybe the influence of politicians, maybe >> maybe not enough of picking up the phone and talking to people that are called enemies by the said politicians.

也许是一些官僚层面的东西，也许是政客的影响，也许>> 也许是不够主动拿起电话，和那些被政客称为敌人的人交谈。

111:45

There might be some like deep wisdom that we just may have lost from that time actually. >> Yeah, I'm sure.

可能有一些深刻的智慧，实际上是我们从那个时代丢失的。

111:51

I'm sure there is.

>> 是啊，我确信。

111:52

I mean, I've we we you know studied I read a lot of books for that time as well. chronicle time um and some brilliant

我确信有。

111:58

people involved.

>> 我好奇他会给出什么建议。

111:59

I I agree with you.

他亲眼见证了曼哈顿计划中原子弹的建造过程。

112:00

I think maybe there needs to be more dialogue and understanding.

我敢肯定，有些有趣的事情可能还没被充分讨论过。

112:04

Um I hope we can learn from those those times.

也许是一些官僚层面的问题，也许是政客们的影响，也许是不够主动拿起电话去和那些被政客称为“敌人”的人沟通。

112:06

I think the difference here is that the AI has so many it's a multi-use technology obviously we're trying to do things like that like solve you know all diseases um uh help with energy uh and scarcity these incredible things.

那个时代可能有一些深刻的智慧，我们其实已经丢失了。

112:20

This is why all of us and and myself, you know, I worked

这就是为什么我们所有人——包括我自己，你知道，我工作过——

112:23

started on this journey 30 plus years ago.

资源，然后分配这些资源，但没错，没有先有这种充裕，你根本做不到。

112:26

And um but of course there are risks too and probably vonoman my guess is he foraw both and um and I think he sort of said I think is to his wife that that that it would be this is computers would be even more impactful in the world and as we just discussed you know I think that's right.

你跟我提过一本书，《The Maniac》，作者是Benjamin Labatut，这本书首先是关于你的，有一本你的传记，嗯，挺奇怪的。

112:44

I think it's going to be 10 times at least of the industrial revolution.

对，确实不清楚有多少是虚构，多少是现实。

112:48

So I think he's

但我觉得核心人物是John von Neumann，我得说，这本书挺让人难忘的。

112:49

right.

>> 是啊，我确信。

112:49

So I think he would have been I imagine fascinated by uh uh uh where we are now. >> And I think one of the maybe you can correct me but one of the takeaways from the book is that reason as uh said in the book mad dreams of reason.

我读过很多那个时代的书，记录那个时代的，里面有一些非常杰出的人物。

113:05

It's not enough for guiding humanity as we build these super powerful technology that there's something else.

我同意你的看法。

113:12

I mean there's also like a religious component.

我觉得可能需要更多的对话和理解。

113:16

Whatever God, whatever religion gives it G, it pulls at us something in the human spirit that raw cold reason doesn't give us. >> And I I agree with that.

在30多年前就开始了这段旅程。

113:24

I think we need to approach it with whatever you want to call it, the a spiritual dimension or humanist dimension.

但当然也有风险，我猜冯·诺依曼可能两者都预见到了，我觉得他大概对他妻子说过，计算机会对世界产生更大的影响，就像我们刚才讨论的，我认为没错。

113:30

Doesn't have to be to do with religion, right?

我觉得它的影响至少会是工业革命的10倍。

113:33

But this idea of of a soul, what makes us human, this spark that we have perhaps is to do with consciousness when we finally understand that.

所以我认为他——

113:40

Um, I think that has to be at the heart of the endeavor.

嗯，我认为这必须成为这项事业的核心。

113:43

Um, and

嗯，然后——

113:44

technology, I've always seen technology as the enabler, right?

技术，我一直把技术看作是一种赋能工具，对吧？

113:47

The tools that that enable us to to flourish and to understand more about the the world.

那些能让我们蓬勃发展、更深入理解世界的工具。

113:52

And I I'm sort of with Fman on this, and he used to always talk about science and art being companions, right?

我某种程度上同意Fman的观点，他过去总说科学和艺术是相伴相生的，对吧？

113:58

You can understand it from both sides, the beauty of a flower, how beautiful it is, and also understand why the colors of the flower evolved like that, right?

你可以从两个角度去理解——一朵花有多美，以及它的颜色为什么这样演化，对吧？

114:06

That just makes it more beautiful, the the the just the intrinsic beauty of the

这只会让花更美，那种花本身内在的美。

114:11

flower.

花。

114:11

And and I've always sort of seen it like that.

我一直是这么看待它的。

114:14

And maybe you know in the renaissance times the great discoverers then like people like Da Vinci you know they were I don't think he saw any difference between science and art uh and perhaps religion right they were everything was it's just part of being human and um being inspired about the world around us and that's what I the philosophy I tried to take and u one of my favorite philosophers is Spininoza and I think he combined that all very

也许在文艺复兴时期，那些伟大的探索者，比如达·芬奇，我觉得他并不认为科学与艺术之间有什么界限，或许还有宗教，对吧？

114:38

well you know this idea of trying to understand the universe and understanding our place in it and that was his kind of way of understanding religion and I think that's quite beautiful and for me every all of these things are related interrelated the technology and um what it means to be human and uh I think it's very important though that we remember that as when we're immersed in the technology and the the research I think a lot of

>> 对。

115:05

researchers that I see in in our field are a little bit too narrow and only understand the technology and I think also that's why it's important important for this to be debated by society at large and I'm very supportive of things like this the AI summits that will happen and governments understanding it and I think that's one good thing about the chatbot era and the product era of AI is that everyday person can actually feel and and interact with cutting edge AI and and and feel feel it for themselves.

无论是什么神，什么宗教赋予它那种东西，它触动了我们人性中某种东西，是冰冷纯粹的理性无法给予的。

115:31

>> Yeah.

>> 你觉得会不会出现类似曼哈顿计划的东西？

115:32

Because they they force the technologist to have the human conversation.

这种技术的力量会升级，国家用它们老一套的思维试图把它用作武器技术，然后出现这种升级竞赛？

115:37

Yeah, for sure.

对，确实如此。

115:38

That's the hopeful aspect of it.

这是比较乐观的一面。

115:40

Like you said, it's a dual use technology that we're forcefully integrating the entire of humanity into it by into the discussion about AI because ultimately AI AGI will be used for the things that states use technologies for which is uh conflict and so on.

就像你说的，这是一项双刃剑技术，我们正强行把全人类都卷入关于AI的讨论中，因为说到底，AI和AGI最终会被国家用来做那些国家用技术做的事，比如冲突之类的。

115:57

And the more we uh integrate

而我们越是这样整合——

115:59

humans into this picture by having chats with them, the more it will guide >> Yeah. be able to adapt. society will be able to adapt to these technologies like we've always done in the past with with uh the incredible technologies we've invented in the past. >> Do you think there will be something like a Manhattan project where um there will be an escalation of the power of this technology and states in their old way of thinking will try to

嗯，那会是一次很棒的体验。

116:25

use it as weapons technologies and there will be this kind of escalation. >> I hope not.

>> 我希望不会。

116:31

Um I think that would be uh very dangerous to do and I think also um you know not the right use of the technology.

我觉得那样做会非常危险，而且也不是技术的正确用途。

116:39

I I hope we'll end up with more something more collaborative if needed like more like a like a CERN project you know where um it's research focused and the best minds in the world

我希望最终能更像一个合作项目，比如CERN那样，以研究为导向，让全世界最聪明的人聚在一起，谨慎地完成最后几步，确保在部署到世界之前是负责任的。

116:52

come together to carefully complete the final steps and make sure it's responsibly done before you know like deploying it to the world.

花朵。

117:00

We'll see.

我一直是这么看的。

117:00

I mean it's difficult with the current geopolitical climate I think uh to to see cooperation but things can change and um I think at least on the scientific level it's important for the researchers to to to to keep in touch and and and keep close to each other on at least on those kinds of topics.

也许在文艺复兴时期，那些伟大的探索者，比如达·芬奇，我觉得他不认为科学与艺术有什么不同，也许宗教也一样，对吧？

117:18

>> Yeah.

>> 嗯，我个人觉得在教育方面和移民方面，如果两边都能互相流动——西方人移民到中国，中国人也去西方——那就太好了。

117:18

And I I personally believe on the education side and um immigration side, it would be great if both directions uh people from the west immigrated to China and China back.

我是说，人们互相融合，这里面有家庭和人性的一面。

117:29

I mean there is some like family human aspect of people just intermixing. >> Yeah. >> And thereby those ties grow strong.

>> 嗯。

117:37

So you can't sort of divide against each other this kind of old school way of thinking.

>> 这样一来，纽带就会变得更牢固。

117:42

And so uh multi- uh multicultural multid-disciplinary

你就没法再用那种老套的思维方式去挑拨对立。

117:46

research teams working on scientific questions.

花。

117:48

That's like the hope.

我一直是这么看待它的。

117:49

Don't don't let the the warm leaders that are warmongers because it divide us.

也许在文艺复兴时期，伟大的发现者像达芬奇这样的人，我认为他不认为科学与艺术之间有任何区别，也许还有宗教，对吧，一切都只是作为人类的一部分，以及对我们周围世界的灵感，这就是我试图秉持的哲学，我最喜欢的哲学家之一是斯宾诺莎，我认为他把这一切都很好地结合了

117:53

I think science is the ultimately really beautiful connector. >> Yeah.

我觉得科学本质上是一个非常美好的连接纽带。 >> 没错。

117:57

Science has always been uh I think quite a a very collaborative endeavor and you know scientists know that it's it's a it's a collective endeavor as well and we can all learn from each other.

科学一直以来都是一项非常需要协作的事业，而且你知道，科学家们也明白这是一项集体努力，我们都能互相学习。

118:06

So perhaps it could be a vector to get a bit of cooperation.

所以也许它可以成为促成一些合作的途径。

118:09

What's your uh ridiculous question?

你那离谱的问题是什么？

118:11

What's your pdoom?

你的pdoom是多少？

118:12

Probability the human

人类概率

118:14

civilization destroys itself. >> Well, look, I I don't have a it's a you know, I don't have a pdoom number.

文明自我毁灭。

118:21

The reason I don't is because I think it's would imply a level of precision that is not there.

>> 嗯，你看，我其实没有，你知道，我没有什么pdoom数字。

118:27

So, like I don't know how people are getting their poom numbers.

我没有的原因是因为我觉得那会暗示一种并不存在的精确度。

118:31

I think it's a kind of a little bit of a ridiculous notion because um what I would say is it's definitely nonzero and it's probably non

所以，我不知道人们是怎么得出他们的pdoom数字的。

118:40

negligible.

微不足道。

118:41

So that in itself is pretty sobering and my my view is it's just hugely uncertain, right?

所以这本身就挺让人清醒的，我的看法是这充满了巨大的不确定性，对吧？

118:46

What these technologies are going to be able to do, how fast are they going to take off, how controllable they going to be.

这些技术到底能做什么、发展速度有多快、可控性有多强，都是未知数。

118:53

Some things may turn out to be and hopefully like way easier than we thought, right?

有些事情可能会比我们想象的简单得多，希望如此吧。

118:58

Um but it may be there's some really hard um uh uh problems that are harder than we guess today and I think uh we don't know that for sure and so in under

但也可能有一些非常棘手的问题，比我们今天的猜测更难解决，我觉得我们对此并不确定，所以本质上...

119:07

those conditions of a lot of uncertainty but huge stakes both ways you know on the one hand we we could solve all diseases energy problems the not the the the scarcity problem and then travel to the stars and conscious of the stars and maximum human flourishing on the other hand is this sort of p doom scenarios so given the uncertainty around it and the importance of It's clear to me the only rational sensible approach is to proceed with cautious optimism.

这些情况充满不确定性，但两边都赌注巨大。

119:33

So we want the

一方面，我们可能解决所有疾病、能源问题、资源稀缺问题，然后飞向星辰，感知宇宙，实现人类最大程度的繁荣；另一方面，又有这些p(doom)的末日场景。

119:34

outcome.

你更担心什么来源？

119:34

We want the um uh the benefits of course uh and uh all of the the amazing things that AI can bring and actually I would be really worried for humanity if I if given the other challenges that we have climate dis you know aging uh resources all of that if I didn't know something like AI was coming down the line right how would we solve all those other problems I think it's hard um so I think we've you know it

是人类造成的，还是AI/AGI造成的？

119:59

could be amazingly transformative for good um but on the other And you know there are these risks that we know are there but we can't quite quantify.

人类滥用那种技术，还是AGI本身通过你提到的那种机制——比如欺骗之类的东西——变得越来越好、越来越隐蔽，然后……

120:08

So the the best thing to do is to use the scientific method to do more research to try and uh more precisely define those risks and of course address them.

所以最好的办法就是用科学方法做更多研究，试着更精确地定义这些风险，当然也要去解决它们。

120:18

Um and I think that's what we're doing.

嗯，我觉得这就是我们现在在做的事。

120:20

I think there probably needs to be uh 10 times more effort on that than there is now as

我认为这方面可能需要投入比现在多10倍的努力才行。

120:26

we're getting closer and closer to the to the to the AGI line. >> What would be the source of worry for you more?

我觉得它们运作在不同的时间尺度上，而且同样重要，都需要应对。

120:32

Would it be human caused or AI AGI caused? >> Humans abusing that technology versus AGI itself through mechanism that you've spoken about which is fascinating deception or this kind of stuff >> getting better and better and better secretly and then >> I think they they operate over different time scales and they're equally important to address.

所以问题就在于……

120:52

So there's just

所以就是

120:53

the the the the common garden of variety of like you know bad actors using new technology uh in this case general purpose technology and repurposing it for harmful ends and that's a huge uh risk and I think that has a lot of complications because generally you know I'm in huge favor of open science and open source and in fact we did it with all our science projects like AlphaFold and all of those things uh for the benefit of of of the scientific

各种各样的恶意行为者利用新技术——这里指的是通用技术——并将其重新用于有害目的，这就像是一个常见的大杂烩。

121:20

community.

社区。

121:20

Um but how does one restrict bad actors access access to these powerful systems whether they're individuals or even rogue states uh and but enable access at the same time to good actors to to maximally build on top of.

但如何限制坏行为者——无论是个人还是流氓国家——获取这些强大系统，同时又能让好行为者最大程度地利用它们进行构建？

121:34

It's pretty tricky problem that there's I've not heard a clear solution to.

这是个相当棘手的问题，我还没听到过清晰的解决方案。

121:39

So there's the bad actor use case problem and then there's obviously uh as the systems become more agentic and and

所以有坏行为者使用的问题，然后显然，随着系统变得更具代理性、更接近AGI、更自主，我们如何确保有护栏，让它们坚持做我们想做的事，并且受我们控制？

121:46

closer to AGI um and more autonomous how do we ensure the guard rails and they stick to what we want them to do uh and under our control.

而这里面，一部分是如何不让破坏性技术落入坏行为者手中，另一部分是从地缘政治技术角度，如何减少……

121:54

Yeah, I tend to maybe on my mind is limited worry more about the humans.

对，我可能心里更倾向于有限地担心人类本身。

121:58

So the bad actors >> and there it could be uh in part how do you not put destructive technology in the hands of bad actors but in another part from again geopolitical technology perspective how do you reduce the number

所以坏 actors >> 一方面是如何不让 destructive technology 落到坏 actors 手里，另一方面从 geopolitical technology 的角度来看，如何减少数量

122:12

of bad actors in the world that's that's also an interesting human problem. >> Yeah it's a hard problem.

世界上确实存在一些恶意行为者，这也是一个有趣的人类问题。

122:19

I mean, look, we we we can um maybe also use the technology itself to help um early warning on some of the bad actor use cases, right?

>> 对，这是个难题。

122:27

Whether that's bio or nuclear or whatever it is, like AI could be potentially helpful there as long as the AI that you're using is itself reliable, right?

我的意思是，我们也许可以用技术本身来帮助预警一些恶意行为者的用例，对吧？

122:37

So it's a sort of

不管是生物、核武器还是别的什么，AI 在这方面可能有帮助，前提是你用的 AI 本身是可靠的，对吧？

122:39

interlocking problem and that's what makes it very tricky and and again it may require some agreement internationally at least between China and the and and the US of of of some uh basic standards right I have to ask you about the the book the maniac there there's this the the hand of God moment Lisa doll's move 78 >> that perhaps the last time a human did a

相互交织的问题，这就是为什么它非常棘手。

123:05

move of sort of pure human genius and beat Alph Go or like broke its brain if sorry to anthropomorphize but it's an interesting moment cuz I think in so many domains it will keep happening. >> Yeah, it's a special moment and you know it was great for Lisa Doll and you know I think it's in a way they were kind of inspiring each other.

我得问问你关于《The Maniac》这本书，那个“上帝之手”时刻——李世石的78手。

123:25

We as a team were inspired by Lisa Doll's brilliance and nobleness and then maybe he got inspired

那也许是人类最后一次做出纯粹人类天才的妙手，击败了AlphaGo，或者说搞乱了它的“大脑”——抱歉我用了拟人化——但那是个有趣的时刻，因为我觉得在很多领域，这种情况会不断重演。

123:30

by you know what AlphaGo was doing to then conjure this incredible inspirational moment. it's all you know captured very well in the in the documentary about it and um I think that'll continue in many domains where there's this at least for the for the again for the foreseeable uh future of like the humans bringing in their ingenuity um and asking the right question let's say uh and then utilizing these tools uh in a way that um then

你知道的，AlphaGo当时做的事情，然后它创造出了那个令人难以置信的鼓舞人心的时刻。

123:58

cracks a problem. >> Yeah.

>> 是啊。

123:59

What as the AI becomes smarter and smarter, one of the interesting questions we can ask ourselves is what makes humans special?

随着AI变得越来越聪明，我们可以问自己一个有趣的问题：人类到底有什么特别之处？

124:08

It does feel I'm perhaps biased that we humans are deeply special.

我可能有点偏见，但我确实觉得人类非常特别。

124:13

I don't know if it's our intelligence.

我不知道这是不是我们的智力，也可能是别的什么东西，那种超出理性疯狂梦境之外的东西。

124:15

It could be something else that that other thing that's outside the mad dreams of reason.

我小时候一直这么想，

124:21

I think that's what I've always imagined uh when I was a kid

我觉得那一直是我小时候想象中的画面。

124:26

and starting on this journey of like um I was of course fascinated by things like consciousness did did a neuroscience PhD to look at how the brain works especially imagination and memory I focused on the hippocampus and it's sort of going to be interesting I always thought the best way of course one can kind of philosophize about it and have thought experiments and maybe even do actual experiments like you do in neuroscience on on real brains but in the end I always imagined that building AI a kind intelligent artifact and then

踏上这段旅程时，我自然对意识之类的东西着迷，读了神经科学博士，研究大脑如何运作，特别是想象力和记忆，我专注于海马体，这还挺有意思的。

124:53

comparing that to the human mind and seeing what the differences were uh would be the best way to uncover what's special about the human mind if indeed there is anything special and I suspect there probably is but it's going to be hard to def you know I think this journey we're on will help us uh understand that and define that and you know there may be a difference between carbon based substrates that we are and silicon ones when they process information you know one of the best definitions I like of of of

把那个和人类思维做比较，看看有什么不同，会是揭示人类思维独特之处的最佳方式——如果它确实有什么特别的话。

125:20

consciousness is it's the information feels when we process it, right?

意识就是我们在处理信息时感受到的那种东西，对吧？

125:24

Um, it could be I mean doesn't it's not a very helpful scientific explanation.

嗯，这可能不是一个很有用的科学解释，但我认为这是一个挺有趣的直觉。

125:29

I think it's kind of interesting intuition in intuitive one and um and so you know on this this this journey this scientific journey we're on will I think um help uncover that mystery. >> Yeah.

所以，在这条科学探索的路上，我觉得会帮助我们揭开这个谜团。

125:40

What I cannot create I do not understand.

我不能创造的东西，我就无法理解。

125:42

That's uh somebody you deeply admire Richard Feman like you

那是你深深敬佩的一个人，Richard Feynman，对吧？

125:46

mentioned. you also reach um for the the Wignner's dreams of universality that he saw in constraint domains but also broadly generally in in mathematics and so on.

>> 是啊。

125:57

So so many aspects on which you're pushing towards >> not to start trouble at the end but uh Roger Penrose. >> Yes. >> Okay.

我不能创造的东西，我就不能理解。

126:05

So uh you know do do you think consciousness there's this hard problem

这是你非常敬佩的一个人——Richard Feynman说的，就像你提到的。

126:10

of consciousness how information feels >> um do you think consciousness first of all is a computation and if it is if it's information processing like you said everything is >> is it something that could be modeled by a classical computer? >> Yeah. >> Or is it a quantum mechanical in nature?

关于意识，信息如何被感知……嗯，你觉得意识首先是一种计算吗？

126:27

Well, look, Pero is an amazing thinker, one of the greatest of the modern era, and he we've had a lot of discussions about this.

如果是的话，如果它像你说的那样是信息处理，一切皆是如此……那它是否可以被经典计算机建模？

126:34

Of course, we cordily

当然，我们完全同意。

126:35

disagree, which is, you know, I I feel like um I mean, he collaborated with a lot of good neuroscientists to see if he could find mechanisms for quantum mechanics behavior in the brain.

是的。

126:46

And they, to my knowledge, they haven't found anything um convincing yet.

而且据我所知，他们目前还没找到什么有说服力的证据。

126:50

So my betting is there is is that that that it's mostly you know it is just classical computing that's going on in the brain which suggests that all the phenomena uh are modelable or mimickable

所以我猜测，大脑里进行的其实主要还是 classical computing，这意味着所有现象都是可以被建模或模仿的。

127:02

by a classical computer but we'll see you know there may be this final mysterious things of the feeling of consciousness the qualia these kinds of things that philosophers debate where it's unique to the substrate we may even come towards understanding that when if we do things like neural link and and have neural interfaces to the AI systems, which I think we probably will eventually um maybe to keep up with the AI systems.

还是说它在本质上是量子力学的？

127:26

Uh we might actually be able

呃，我们或许真的能够做到。

127:28

to feel for ourselves what it's like to compute on silicon, right?

嗯，Pero是一位了不起的思想家，现代最伟大的人物之一，我们对此有过很多讨论。

127:31

So um and maybe that will tell us.

当然，我们礼貌地

127:33

Uh so I think it's it's going to be interesting.

嗯，所以我觉得这会很有意思。

127:36

I had a debate once with the late Daniel Dennett about why do we think each other are conscious?

我曾经和已故的Daniel Dennett争论过一次，关于我们为什么认为彼此是有意识的。

127:41

Okay, so it's for two reasons.

好，这有两个原因。

127:42

One is you're exhibiting the same behavior that I am.

第一，你表现出的行为跟我一样。

127:45

So that's one thing. behaviorally you seem like a conscious being if I am.

所以这是其一——从行为上看，如果你是有意识的，那你也像是有意识的。

127:49

But the second thing which is often overlooked is that we're running on the same substrate.

但第二点，常常被忽略的是，我们运行在同样的底层基质上。

127:54

So

嗯，而且它可能无法知道我们的感受，至少在最初阶段是这样。

127:55

if you're behaving in the same way and we're running on the same substrate, it's most parsimonious to assume you're feeling the same experience that I'm feeling.

如果你行为方式相同，而且我们运行在同样的生物基质上，最简洁的解释就是假设你的感受和我一样。

128:04

But with an AI uh that's on silicon, we won't be able to rely on the second part.

但对于运行在硅基上的AI，我们无法依赖第二部分。

128:08

Even if it exhibits the first part, the behavior looks like a behavior of a conscious being.

即使它表现出第一部分，行为看起来像是有意识的存在，它甚至可能声称自己有意识。

128:13

It might even claim it is.

但我们不知道它实际感受如何，而且至少在初期阶段，它可能也无法理解我们的感受。

128:14

Um but we but but we wouldn't know how it actually felt.

也许当我们达到超级智能，以及由此构建的技术时，我们或许能够弥合这个鸿沟。

128:17

Um and it probably couldn't know we what we felt at least in the first stages.

也许吧。

128:21

Maybe

没错。

128:22

when we get to super intelligence and the technologies that builds perhaps we'll we'll be able to um bridge that. >> No, I mean that's a huge test for radical empathy is to empathize with a different substrate. >> Right.

>> 不，我的意思是，这对“激进共情”来说是一个巨大的考验——去共情一个不同的基质。

128:35

Exactly.

没错。

128:35

We never had to confront that before. >> Yeah.

我们以前从未面对过这种情况。>> 是啊。

128:38

So maybe maybe through brain computer interfaces be able to truly empathize what it feels like to be a computer >> for information to be computed not on a carbon system.

所以也许通过脑机接口，才能真正体会作为计算机是什么感觉 >> 让信息在一个非碳基系统上被计算。

128:47

I mean that's deeply I

我是说，这非常深刻。

128:49

mean some people kind of think about that with plants with other life forms which are different similar substrate but sufficiently far enough on the uh evolutionary tree >> that it's requires a radical empathy but to do that with a computer >> I mean Lou we sort of there are animal studies on this of like of course higher animals like you know killer whales and dolphins and dogs and and monkeys you know they have some and elephants you know they have some aspects certainly of

>> 对，没错。

129:16

consciousness right?

>> 是啊，所以也许通过脑机接口，我们才能真正共情成为一台计算机是什么感觉——信息不是在碳基系统上计算的。

129:17

Even though they're not might not be that that that smart on an IQ sense.

这真的很深奥，有些人会想到植物或其他生命形式，它们的基质不同但相似，但在进化树上足够遥远，需要激进共情才能理解。

129:21

So so we can already empathize with that and maybe even some of our systems one day like we built this thing called dolphin Gemma, you know, which can one a version of our system was trained on dolphin and whale sounds and maybe we'll be able to build a an interpreter or translator at some point which should be pretty cool. >> What gives you hope for the future of human civilization? >> Well, what gives me hope is I think our

但要对计算机做到这一点……

129:44

almost limitless ingenuity first of all.

>> 我的意思是，Lou，关于动物研究，当然高等动物比如虎鲸、海豚、狗、猴子，还有大象，它们确实有意识的一些方面，对吧？

129:46

I think the best of us and the best human minds are incredible.

即使它们在IQ意义上可能没那么聪明。

129:50

Um, and you know, I love, you know, meeting and watching any human that's the top of their game, whether that's sport or science or art.

所以我们已经能对它们共情了，也许有一天我们的系统也能做到。

129:58

You know, it's it's it's just nothing more wonderful than that, seeing them in their element in flow.

比如我们建了一个叫Dolphin Gemma的东西，我们系统的一个版本是用海豚和鲸鱼的声音训练的，也许我们最终能造出一个翻译器或解释器，那应该会很酷。

130:03

Um, I think it's almost limitless.

嗯，我觉得这几乎是无限的。

130:05

You know, our brains are general systems, intelligent systems.

你知道，我们的大脑是通用系统，是智能系统。

130:09

So, I think it's

所以，我认为……

130:10

almost limitless what we can potentially do with them.

>> 是什么让你对人类文明的未来充满希望？

130:14

And then the other thing is our extreme adaptability.

然后另一件事是我们极端的适应能力。

130:17

I think it's going to be okay in terms of there's going to be a lot of change.

我觉得会没事的，虽然会有很多变化。

130:22

But but look where we are now with our effectively our hunter gatherer brains.

但你看，我们现在用着原始狩猎采集者的大脑，却要应对现代世界，对吧？

130:26

How is it we can you know we can cope with the modern world, right?

坐飞机、录播客、玩电脑游戏和虚拟模拟——这些不都已经成了现实吗？

130:31

Flying on planes, doing podcasts, you know, playing computer games and virtual simulations.

这就是我们人类彼此之间的相处方式。

130:36

I mean, it's already given

我们互相赞美。 >> 没错。

130:38

that was developed for, you know, hunting buffal societyy's already adapted to this mind-blowing AI technology we have today already.

>> 嗯，首先让我充满希望的是我们几乎无限的创造力。

130:47

It's like, oh, I talk to chat bots.

我认为最优秀的人类和最出色的人类头脑是不可思议的。

130:50

It's totally fine. >> And it's uh very possible that this very podcast activity, which I'm here for, will be completely replaced by AI.

我喜欢见到和观察任何在各自领域顶尖的人，无论是体育、科学还是艺术。

130:59

I'm very replaceable and I'm waiting for >> not to the level that you can do it,

没有什么比看到他们在自己的领域里进入心流状态更美妙了。

131:04

Lex.

>> 而且很有可能，我现在参与的播客活动会被AI完全取代。

131:05

So don't think >> Thank you.

我是很容易被替代的，我正等着呢。

131:06

That's that's what we humans do to each other.

而且，我深深感激我们人类拥有这种无限的好奇心、适应力，就像你说的，还有同情心和爱的能力。 >> 正是如此。 >> 所有这些属于人性的特质 >> 那些真正深刻的人性。 >> 嗯，这真是莫大的荣幸，Demis。

131:09

We compliment. >> All right.

你是这世上真正特别的人类之一。

131:10

And uh I'm uh deeply grateful for us humans to have this uh infinite capacity for curiosity, adaptability, like you said, and also compassion and ability to love. >> Exactly. >> All of those human >> all the things that are deeply human. >> Well, this is a huge honor, Demis.

而且，我深深感激我们人类拥有这种无限的好奇心、适应力，就像你说的，还有同情心和爱的能力。

131:24

You're one of the truly special humans in the world.

>> 没错。

131:27

Uh thank you so much for doing what you do and for talking today. >> Well, thank you very much, Lex.

呃，非常感谢你做的这些事，也谢谢你今天来聊。>> 嗯，也特别感谢你，Lex。

131:33

Thanks for listening to this conversation with Demos.

感谢收听这期与Demos的对话。

131:36

To support this podcast, please check out our sponsors in the description and consider subscribing to this channel.

为了支持这个播客，请查看简介里的赞助商，并考虑订阅本频道。

131:44

And now, let me answer some questions and try to articulate some things I've been thinking about.

现在，我来回答一些问题，并试着梳理一下我最近一直在思考的一些东西。

131:50

If you would like to submit questions, including in audio and video form, go to lexfreman.com/am.

如果你想提交问题，包括音频和视频形式，可以访问lexfreman.com/am。

131:57

I got a lot of amazing questions,

我收到了很多很棒的问题、想法和请求。

131:59

thoughts, and requests from folks.

持不同意见。

132:02

I'll keep trying to pick some uh randomly and comment on it at the end of every episode.

你知道，我觉得……嗯，他和很多优秀的神经科学家合作过，试图在大脑中寻找量子力学行为的机制。

132:09

I got a note on May 21st this year that said, "Hi, Lux. 20 years ago today, David Foster Wallace delivered his famous this is water speech at uh Kenyan College.

据我所知，他们还没有找到任何令人信服的证据。

132:22

What do you think of this speech?"

所以我打赌，这基本上就是……你知道，大脑中发生的只是经典计算，这意味着所有现象都可以被经典计算机建模或模仿。

132:25

Well, first I think this is probably one of the greatest and most unique commencement speeches ever given.

嗯，首先我觉得这可能是史上最伟大、最独特的毕业演讲之一。

132:32

But of course, I have many favorites, including the one by Steve Jobs.

当然，我还有很多其他喜欢的演讲，比如Steve Jobs的那场。

132:37

And David Foster Wallace is one of my favorite writers and one of my favorite humans.

David Foster Wallace是我最喜欢的作家之一，也是我最喜欢的人之一。

132:43

There's a tragic honesty to his work.

他的作品里有一种悲剧性的诚实，总感觉他一直在和自己的内心进行一场持续的搏斗。

132:46

And it always felt as if he was engaging

而他的写作，就像是那场搏斗前线传来的笔记。

132:49

in a a constant battle with his own mind. and the writing, his writing were kind of his notes from the front lines of that battle.

但我们会看到的，你知道，可能还有这些最终的神秘事物，比如意识的感觉、qualia这类哲学家争论的东西，它们可能对基质来说是独特的。

132:59

Now, onto the speech.

我们甚至可能会逐渐理解这一点，如果我们做像Neural Link这样的事情，并且拥有与AI系统的神经接口，我认为我们最终可能会这样做，嗯，也许是为了跟上AI系统。

133:01

Let me quote some parts.

呃，我们实际上可能能够

133:03

There's of course the parable of the fish and the water that goes, "There are these two young fish swimming along, and they happen to meet an older fish

>> 所有这些人类的特质——

133:15

swimming the other way who nods at them and says, "Morning boys.

现在回到这个演讲。

133:19

How's the water?" And the two young fish swim on for a bit and then eventually one of them looks over at the other and goes, "What the hell is water?" In the speech, David Foster Wallace goes on to say, "The point of the fish story is merely that the most obvious important realities are often the ones

让我引用一些部分。

133:41

that are hardest to see and talk about." stated as an English sentence.

亲自感受在硅上计算是什么感觉，对吧？

133:46

Of course, this is just a banal platitude.

所以嗯，也许那会告诉我们答案。

133:48

But the fact is that in the dayto-day trenches of adult existence, bal platitudes can have a life or death importance.

呃，所以我认为这会很有趣。

133:56

Or so I wish to suggest to you in this dry and lovely morning.

我曾与已故的Daniel Dennett辩论过一次，关于为什么我们认为彼此是有意识的。

134:00

I have several takeaways from this parable and the speech that follows.

好吧，有两个原因。

134:05

First, I think we must question everything and in

一是你表现出和我一样的行为。

134:08

particular the most basic assumptions about our reality, our life and the very nature of existence and that this project is a deeply personal one in some fundamental sense.

在演讲中，David Foster Wallace接着说：“这个鱼故事的重点仅仅在于，那些最显而易见的重要现实，往往是最难被看见和谈论的。

134:21

Nobody can really help you in this process of discovery.

”用一句英文句子表述出来，这当然只是一句陈腐的套话。

134:25

The call to action here, I think, from uh David Foster Wallace, as he puts it, is to quote, to be just a little less

但事实是，在成年生活的日常战壕里，陈腐的套话可能具有生死攸关的重要性。

134:34

arrogant, to have just a little more critical awareness about myself and my certainties.

来自大家的想法和请求。

134:41

Because a huge percentage of the stuff that I tend to be automatically certain of is, it turns out, totally wrong and deluded.

我会继续尝试随机挑选一些，并在每集结尾评论。

134:52

All right, back to me.

今年5月21日我收到一条留言，上面写着：“嗨，Lex。

134:53

Lex speaking.

20年前的今天，David Foster Wallace在Kenyan学院发表了著名的《这就是水》演讲。

134:54

Second takeaway is that the central spiritual battles of our life are not

你对这个演讲怎么看？

135:01

fought on a uh mountain top somewhere at a meditation retreat but it is fought in the mundane moments of daily life.

在一场山顶的冥想静修中战斗，但它其实是在日常生活的平凡时刻里战斗。

135:11

Third takeaway is that we too easily give away our time and attention to the multitude of distractions that the world feeds us. the insatiable black holes of attention.

第三个收获是，我们太轻易地把时间和注意力交给世界塞给我们的无数干扰——那些贪得无厌的注意力黑洞。

135:25

David Foster Wallace's call to action in this case is to be deeply aware of the beauty in each moment and to find meaning in the mundane.

David Foster Wallace提出的行动号召，是要深刻意识到每个瞬间的美，并在平凡中找到意义。

135:36

I often quote David Foster Wallace in his advice that the key to life is to be unorable.

我经常引用David Foster Wallace的建议，他说生活的关键在于保持不可预测性。

135:43

And I think this is exactly right.

我觉得这完全正确。

135:46

Every moment, every object, every experience

每一个瞬间、每一个物体、每一次体验——他拿起一朵花说，“看它多美。

135:50

when looked at closely enough contains within it infinite richness to explore.

大卫·福斯特·华莱士在此发出的行动号召是：深刻觉察每个瞬间的美，在平凡中寻找意义。

135:56

And since uh Deus Lasabus of this very podcast episode and I are such fans of Richard Feineman, allow me to uh also quote Mr.

我经常引用大卫·福斯特·华莱士的建议，他说生活的关键在于“不可忽视”。

136:05

Fineman on this topic as well. quote, "I have a friend who's an artist and has sometimes taken a view which I don't agree with very well.

我认为这完全正确。

136:16

He'll hold

每一个瞬间、每一个物体、每一次体验，当你足够仔细地观察时，其中都蕴含着无限的丰富性等待探索。

136:17

up a flower and say, "Look how beautiful it is." And I'll agree.

>> 我不想在最后惹麻烦，但Roger Penrose。

136:22

Then he says, "I, as an artist, can see how beautiful this is, but you as a scientist take this all apart and it becomes a dull thing." And I think that's kind of nutty.

当然有个关于鱼和水的寓言是这样说的：“有两条小鱼在水里游，碰巧遇到一条老鱼……”

136:35

First of all, the beauty that he sees is available to other people and to me too,

>> 所有那些真正属于人性的东西。

136:42

I believe.

既然这期播客的Deus Lasabus和我都是理查德·费曼的粉丝，请允许我也引用费曼先生关于这个话题的发言。

136:43

Although I may not be quite as refined aesthetically as he is, I can appreciate the beauty of a flower.

引用：“我有个艺术家朋友，有时会持一种我不太认同的观点。

136:50

At the same time, I see much more about the flower than he sees.

他会举起一朵花说：‘看它多美。

136:54

I can imagine the cells in there, the complicated actions inside which also have beauty.

’我会同意。

137:00

I mean it's not just beauty at this dimension at 1 cm.

然后他说：‘我作为艺术家能看到这有多美，但你作为科学家把它全拆解了，它就变得乏味了。

137:04

There's also beauty at the

’我觉得这有点荒谬。

137:06

smaller dimensions.

两人正在争论上帝是否存在，带着那种第四杯啤酒下肚后特有的激烈劲儿。

137:07

The inner structure also the processes.

无神论者说：“听着，我不是没有不相信上帝的实际理由。

137:10

The fact that the colors in the flower evolved in order to attract insects to pollinate it is interesting.

我也不是没试过整个上帝和祈祷那套。

137:17

It means that the insects can see the color.

就在上个月，我在营地外被那场可怕的暴风雪困住，完全迷路了，结果呢？

137:20

It adds a question.

只是碰巧有几个爱斯基摩人路过，给我指了回营地的路。

137:21

Does this aesthetic sense also exist in lower forms?

”这一切，我觉得教会我们，万事万物都是视角问题，而智慧可能在我们谦卑地不断调整和扩展对世界的视角时到来。

137:25

Why is it aesthetic? all kinds of interesting questions which the science knowledge only adds to the

谢谢你们让我聊了一会儿David Foster Wallace。

137:32

excitement, the mystery and the awe of a flower.

回到大卫·福斯特·华莱士的演讲。

137:35

It only adds all right back to uh David Foster Wallace's speech.

里面有一个我特别喜欢的故事。

137:41

He has a great story in there that I particularly enjoy.

故事是这样的：“有两个人一起坐在阿拉斯加荒野深处的一个酒吧里。

137:45

It goes, "There are these two guys sitting together in a bar in the remote Alaskan wilderness.

一个人是信徒，另一个是无神论者，两人正以第四杯啤酒下肚后那种特有的激烈争论上帝是否存在。

137:53

One of the guys is religious.

无神论者说：‘听着，我不是没有不相信上帝的实际理由。

137:55

The other is an atheist and

我不是没有尝试过整个上帝和祈祷的事。

137:58

the two are arguing about the existence of God with that special intensity that comes after about the fourth beer.

>> 是的。

138:05

And the atheist says, "Look, it's not like I don't have actual reasons for not believing in God.

然后他说：“我作为艺术家，能看到这有多美，但你作为科学家却把这一切拆解开来，让它变得索然无味。”我觉得这想法挺离谱的。

138:11

It's not like I haven't ever experimented with the whole God and prayer thing.

我也不是没试过那种“上帝和祈祷”那套东西。

138:17

Just last month, I got caught away from the camp in that terrible blizzard and I was totally lost

首先，他看到的那些美，别人也能看到，我也能看到。

138:23

and I couldn't see a thing and it was 50 below.

我当时什么都看不见，气温是零下50度。

138:26

And so I tried it.

我还是试了。

138:28

I fell to my knees in the snow and cried out, "Oh God, if there is a God, I'm lost in this blizzard and I'm going to die if you don't help me." And now back in the bar, the religious guy looks at the atheist all puzzled. "Well, then you must believe now," he says.

我跪倒在雪地里，大喊：“上帝啊，如果真有上帝，我在这暴风雪里迷路了，你不帮我我就死定了。

138:45

After all, there you are alive.

”现在回到酒吧里，那个信教的人一脸困惑地看着这个无神论者。

138:47

The atheist just rolls his eyes.

“那你现在肯定信了吧，”他说，“毕竟你还活着。

138:49

No,

”无神论者只是翻了个白眼。

138:50

man.

>> 好吧。

138:50

All that happened was a couple of Eskimos happened to be wandering by and show me the way back to the camp.

那么，你觉得意识这个难题……

138:58

All this, I think, teaches us that everything is a matter of perspective and that wisdom may arrive if we have the humility to keep shifting and expanding our perspective on the world.

所有这些，我觉得都在告诉我们，万事万物都取决于视角，而智慧也许会在我们保持谦逊、不断转换和拓展对世界的看法时悄然降临。

139:12

Thank you for allowing me to talk a bit

谢谢你让我有机会聊几句

139:15

about David Foster Wallace.

说到大卫·福斯特·华莱士，他是我最喜欢的作家之一，灵魂特别美。

139:17

He's one of my favorite writers and he's a beautiful soul.

如果允许的话，我还想再多说一句。

139:21

If I may, one more thing I wanted to briefly comment on.

我发现自己陷入了一个奇怪的处境——经常在网上被各方攻击，有时是通过断章取义来歪曲事实，但更多时候是彻头彻尾的谎言。

139:25

I found myself to be in this strange position of getting attacked online often from all sides, including being lied about sometimes through selective misrepresentation, but often through downright lies.

我不知道还能怎么形容。

139:40

I don't

说实话，这让我心碎，但我逐渐明白，这就是互联网的常态，也是我选择这条路要付出的代价。

139:40

know how else to put it.

不知道还能怎么说。

139:42

This all breaks my heart, frankly, but I've come to understand that it's the way of the internet and the cost of the path I've chosen.

说实话，这一切都让我心碎，但我也逐渐明白，这就是互联网的运作方式，也是我选择这条路必须付出的代价。

139:51

There's been days when it's been rough on me mentally.

有些日子，我在精神上确实很难熬。

139:55

It's not fun being lied about, especially when it's about things that are usually for a long time have been a source of happiness and joy for me.

被人造谣中伤一点都不好玩，尤其是当这些谣言涉及那些长期以来一直是我快乐和幸福源泉的事情时。

140:04

But again, that's life.

但话说回来，生活就是这样。

140:06

I'll

我会

140:06

continue exploring the world of people and ideas with empathy and rigor, wearing my heart on my sleeve as much as I can.

我不知道还能怎么说。

140:16

For me, that's the only way to live.

坦白讲，这让我心碎，但我已经明白这就是互联网的运作方式，也是我选择这条路必须付出的代价。

140:19

Anyway, a common attack on me is about my time at MIT and Drexel, two great universities I love and have tremendous respect for.

有些日子对我来说精神上很煎熬。

140:28

Since a bunch of lies have accumulated online about me on

被人造谣中伤一点都不好玩，尤其是当这些谣言涉及那些长期以来一直是我快乐和幸福源泉的事情时。

140:33

these topics to a sad and at times hilarious degree, I thought I would once more state the obvious facts about my bio for the small number of you who may care.

这些话题有时候让人难过，有时候又搞笑到不行。

140:45

TLDDR, two things.

我想再跟你们当中可能关心的人，简单说清楚我简历上那些显而易见的事实。

140:46

First, as I say often, including in a recent podcast episode that somehow was listened to by many millions of people, I proudly went to Drexen University for my bachelor's,

长话短说，就两件事。

140:59

masters, and doctor degrees.

研究团队在解决科学问题。

141:01

Second, I am a research scientist at MIT and have been there in a paid research position for the last 10 years.

那是希望所在。

141:09

Allow me to elaborate a bit more on these two things now, but please skip if this is not at all interesting.

不要让那些好战的领导者分裂我们。

141:17

So, like I said, a common attack on me is that I have no real affiliation with MIT.

我认为科学最终是真正美丽的连接器。

141:23

The

>> 是的。

141:24

accusation, I guess, is that I'm falsely claiming an MIT affiliation because I taught a lecture there once.

继续以同理心和严谨的态度探索人与思想的世界，尽可能坦诚地表达自己。

141:34

Nope.

对我来说，这是唯一的生活方式。

141:34

That accusation against me is a complete lie.

总之，对我的一种常见攻击是关于我在MIT和Drexel的经历，这两所我非常热爱并极其尊重的优秀大学。

141:39

I have been at MIT for over 10 years in a paid research position from 2015 to today.

由于网上积累了大量关于我的谎言，涉及我的

141:47

To be extra clear, I'm a

说得更清楚一点，我是一个

141:49

research scientist at MIT working in lids, the laboratory for information and decision systems in the college of computing.

MIT 的研究科学家，在计算学院的 information and decision systems 实验室工作。

141:59

For now, since I'm still at MIT, you can uh see me in the directory and on the various lab pages.

目前我还在 MIT，所以你能在目录和各个实验室页面上找到我。

142:07

I have indeed given many lectures at MIT over the years, a small fraction of which I posted online.

这些年我确实在 MIT 讲过很多课，其中一小部分我上传到了网上。

142:15

Teaching for me always has been just for fun and not part of my research work.

我不知道还能怎么说。

142:21

I personally think I suck at it, but I have always learned and grown from the experience.

坦白说，这让我心碎，但我已经明白，这就是互联网的运作方式，也是我选择这条路所付出的代价。

142:27

It's like Fineman spoke about, if you want to understand something deeply, it's good to try to teach it.

有些日子，我在精神上备受煎熬。

142:34

But like I said, my main focus has always been on research.

被人造谣中伤并不好受，尤其是当这些谣言涉及那些长久以来一直是我快乐和幸福源泉的事情时。

142:39

I published many peer-reviewed papers that you can

但话说回来，这就是生活。

142:43

see in my Google Scholar profile.

对我来说，教书一直只是出于兴趣，不是研究工作的一部分。

142:45

For my first four years at MIT, I worked extremely intensively.

我个人觉得自己教得挺烂的，但我一直从中学到东西、有所成长。

142:49

Most weeks were 80 to 100 hour work weeks.

就像 Fineman 说的，如果你想深入理解某件事，试着去教别人是个好办法。

142:52

After that, in 2019, I still kept my research scientist position, but I split my time taking a leap to pursue projects in AI and robotics outside MIT and to dedicate a lot of focus to the podcast.

但就像我说的，我的主要精力一直在研究上。

143:05

As I've said, I've been continuously surprised

我发表了很多同行评审的论文，你可以在我的 Google Scholar 资料里看到。

143:08

just how many hours preparing for an episode takes.

硕士和博士学位。

143:12

There are many episodes of the podcast for which I have to read, write, and think for 100, 200 or more hours across multiple weeks and months.

第二，我是MIT的研究科学家，过去十年一直在那里担任带薪的研究职位。

143:22

Since 2020, I have not actively published research papers.

现在请允许我稍微详细地说明这两点，但如果完全不感兴趣请跳过。

143:26

Just like the podcast, I think it's something that's a serious full-time effort.

所以，就像我说的，一种常见的攻击是我与MIT没有真正的关联。

143:32

But not publishing and doing full-time research

这个

143:36

has been eating at me because I love research and I love programming and building systems that test out interesting technical ideas, especially in the context of human AI or human robot interaction.

在 MIT 的头四年，我工作强度非常大。

143:49

I hope to change this in the coming months and years.

大多数星期都是 80 到 100 小时的工作时间。

143:53

What I've come to realize about myself is if I don't publish or if I don't launch systems that people use, I

之后，到了 2019 年，我仍然保留了研究科学家的职位，但我分出一部分时间，冒险去探索 MIT 之外的 AI 和 robotics 项目，并投入大量精力做播客。

144:00

definitely feel like a piece of me is missing.

确实感觉像少了点什么。

144:04

It legitimately is a source of happiness for me.

这对我来说真的是一个快乐源泉。

144:07

Anyway, I'm proud of my time at MIT.

总之，我为在MIT的时光感到自豪。

144:10

I was and am constantly surrounded by people much smarter than me, many of whom have become lifelong colleagues and friends.

过去和现在，我身边总是围绕着比我聪明得多的人，其中很多人成了我一生的同事和朋友。

144:19

MIT is a place I go to escape the world, to focus on exploring fascinating questions at the cutting edge of science

MIT是我逃离世界、专注于探索科学前沿那些迷人问题的地方。

144:28

and engineering.

从 2020 年开始，我没有再积极发表研究论文。

144:29

This again makes me truly happy.

就像做播客一样，我觉得那是一件需要全职投入的严肃事情。

144:32

And it does hit pretty hard on a psychological level when I'm getting attacked over this.

但不发表论文、不做全职研究，一直让我心里不踏实，因为我热爱研究，热爱编程，热爱构建系统来测试有趣的技术想法，尤其是在 human AI 或 human robot interaction 的背景下。

144:39

Perhaps I'm doing something wrong.

我希望在未来几个月和几年里改变这一点。

144:42

If I am, I will try to do better.

我逐渐意识到，如果我不发表论文，或者不推出人们使用的系统，我确实会觉得缺了一块。

144:45

In all this discussion of academic work, I hope you know that I don't ever mean to say that I'm an expert at anything.

这对我来说真的是快乐的来源。

144:55

In the podcast and in my private life, I don't claim to be smart.

在播客里和私下生活中，我从不觉得自己聪明。

145:00

In fact, I often call myself an idiot and mean it.

事实上，我经常真心实意地管自己叫傻瓜。

145:04

I try to make fun of myself as much as possible and in general to celebrate others instead.

我尽量多拿自己开涮，总体上更愿意去赞美别人。

145:10

Now, to talk about Drexler University, which I also love, am proud of and am deeply grateful for my time there.

接下来聊聊Drexler University，我也同样热爱它、为它自豪，并且深深感激在那里度过的时光。

145:19

As I

如果你觉得有点意思，请去听听那期节目的结尾，或者看看相关的clip。

145:19

said, I went to Drexil for my bachelor's, masters, and doctor degrees in computer science and electrical engineering.

我去Drexel读了计算机科学和电子工程的本科、硕士和博士。

145:27

I've talked about Drexel many times, including, as I mentioned, at the end of a recent podcast, the Donald Trump episode, funny enough, that was listened to by many millions of people, where I answered a question about graduate school and explained my own journey at Drexel and how grateful I am for it.

我多次提到过Drexel，包括在最近一期播客的结尾——就是那个特朗普那期，挺有意思的，那期被几百万人都听过——我在那里回答了一个关于研究生院的问题，讲了我自己在Drexel的经历，以及我对此有多感激。

145:48

If

如果你感兴趣的话，请去听那期节目的结尾，或者看相关的片段。

145:48

it's at all interesting to you, please go listen to the end of that episode or watch the related clip.

如果你觉得有意思，请一定去听听那期节目的结尾，或者看看相关的视频片段。

145:54

At Drexel, I met and worked with many brilliant researchers and mentors from whom I've learned a lot about engineering, science, and life.

在Drexel，我遇到了很多优秀的研究者和导师，和他们一起工作，从他们身上学到了很多关于工程、科学和人生的道理。

146:03

There are many valuable things I gained from my time at Drexel.

我在Drexel收获了很多宝贵的东西。

146:06

First, I took a large number of very difficult math and theoretical computer science courses.

首先，我上了大量非常难的数学和理论计算机科学课程。

146:12

They taught me how to think deeply and rigorously, and

这些课教会了我如何深入、严谨地思考，并且

146:16

also how to work hard and not give up even if it feels like I'm too dumb to find a solution to a technical problem.

还有就是，即使觉得自己笨到解不开技术难题，也要拼命努力、绝不放弃。

146:24

Second, I programmed a lot during that time, mostly C, C++.

第二，那段时间我写了很多代码，主要是C和C++。

146:28

I programmed robots, optimization algorithms, computer vision systems, wireless network protocols, multimodal machine learning systems, and all kinds of simulations of physical systems.

我编程做过机器人、优化算法、计算机视觉系统、无线网络协议、多模态机器学习系统，还有各种物理系统的仿真。

146:41

This is where I really develop a love for programming, including yes, Emacs and the Kinesis keyboard.

正是在这里，我真正培养了对编程的热爱，包括，没错，Emacs和Kinesis键盘。

146:50

Uh I also during that time read a lot.

呃，那段时间我也读了很多书。

146:53

I played a lot of guitar, wrote a lot of crappy poetry and uh trained a lot of uh injudo and jiu-jitsu which I cannot sing enough praises to.

弹了很多吉他，写了很多烂诗，呃，还练了很多柔道和柔术——这些我真是怎么夸都不够。

147:06

Jiu-jitsu humbled me on a daily basis throughout my 20s and it still does to this very day whenever I get a chance to train.

正是在这个阶段，我真正爱上了编程——包括Emacs和Kinesis键盘。

147:15

Anyway, I hope that the folks who occasionally get swept up in enchanting online crowds that want to tear down others don't lose themselves in it too much.

那段时间我也读了很多书，弹了很多吉他，写了很多烂诗，还练了很多柔道和巴西柔术，这两项运动我真是怎么夸都不够。

147:27

In the end, I still think there's more

说到底，我还是觉得还有更多

147:30

good than bad in people.

人性中好的东西总比坏的多。

147:32

But we're all, each of us, a mixed bag.

但我们每个人，说到底，都是好坏参半的。

147:34

I know I am very much flawed.

我知道自己有很多缺点。

147:36

I speak awkwardly.

我说话很笨拙。

147:38

I sometimes say stupid I can get irrationally emotional.

有时候会说蠢话，会莫名其妙地情绪化。

147:42

I can be too much of a dick when I should be kind.

该温柔的时候，我可能反而太刻薄。

147:45

I can lose myself in a biased rabbit hole before I wake up to the bigger, more

我会陷入偏见的死胡同，直到清醒过来，看到更广阔、更

147:51

accurate picture of reality.

柔术在我二十多岁的时候天天让我受挫，到现在只要有机会训练，它依然能让我保持谦卑。

147:54

I'm human and so are you.

总之，我希望那些偶尔被网上那些煽动人心的、喜欢贬低别人的群体冲昏头脑的人，不要太迷失自己。

147:57

For better or for worse.

说到底，我仍然相信人性中善多于恶。

147:59

And I do still believe we're in this whole beautiful mess together.

而且我依然相信，我们都在这片美丽的混乱里一起前行。

148:06

I love you all.

我爱你们所有人。

Deep Analysis · 深度解析

Human Limitations in Predicting Nonlinear Systems

人类在预测非线性系统方面的局限性

End-to-End Systems Still Not Mature

端到端系统尚未成熟

Unresolved Tricky Problems in the Field

领域中未解决的棘手问题

Key Moments · 关键时刻

Play Queue