播客精选：里德·霍夫曼谈人工智能如何解答我们的重大问题

你为什么关心哲学？

Why do you care about philosophy?

回答这些重大问题为什么重要？

Why are answering these big questions important?

你知道吗，我有时会对商学院说，哲学背景对创业的重要性甚至超过MBA。

You know, one of the things that I sometimes will tell MBA schools, background in philosophy is more important for entrepreneurship than an MBA.

哲学对这些事情至关重要，因为它教会我们如何清晰地思考可能性、当今人类本性的理论，以及这些理论如何因新产品、服务和技术等而改变。

Philosophy is very important to this stuff because it's understanding how to think about very crisply what are possibilities, what are theories of human nature as they are manifest today, and as they may be modified by new products and services, new technology, etcetera.

通常在这个节目中，我们会讨论人们如何实际使用JETHYPT。

Usually in this show, we talk about actionable ways that people use JETHYPT.

但一个更有趣的问题是，人工智能总体上如何改变‘人性’的含义？

But a more interesting question is, how does AI in general, and how might it change what it means to be human?

这些都是非常深刻、重大的哲学问题。

These are really deep, big philosophical questions.

我认为你对这个交叉领域可能有独特的见解。

I thought you might have a unique perspective on this intersection.

雷德，欢迎来到节目。

Reid, welcome to the show.

很高兴能来这里。

It's great to be here.

很高兴有你。

Great to have you.

我相信每位听众或观众都知道，你是一位著名的创业者、风险投资人、作家，最广为人知的身份是领英的联合创始人，以及灰度资本的合伙人。

So I'm sure that everyone listening or watching knows this, but you are a renowned entrepreneur, you're a venture capitalist, you are an author, you're best known as the co founder of LinkedIn, your partner at Greylock.

你还是OpenAI的董事会成员或早期投资人。

You are a board member or a board member, and an early backer at OpenAI.

你还有一个非常出色的播客《规模大师》。

And you also have an incredible podcast, Masters of Scale.

但或许与我们这次对话最相关的是，你曾在斯坦福大学和牛津大学学习哲学，几乎成为一名哲学教授——这一点我在研究这次采访前并不知道。

But perhaps most relevant to this conversation, you also studied philosophy at Stanford and Oxford, you almost became a philosophy professor, which I didn't know before researching this interview.

这真的很酷。

It's really cool.

嗯。

Yeah.

不，这确实是其中一部分，因为我一直对人类的思想和语言感兴趣。

No, it was definitely Part of it was I've always been interested in human thought and language.

我在斯坦福大学主修了一个叫符号系统专业。

Started with Stanford with a major called symbolic systems.

我是第八个选择这个专业的学生。

I was the eighth person to declare that.

作为斯坦福大学的一个专业，我逐渐意识到，我们其实并不完全理解思想和语言究竟是什么。

And as a major at Stanford and then kind of thought, we don't really know what thought and language fully are.

也许哲学家们更懂。

Maybe philosophers do.

于是我跑去斯坦福上了一些课，然后又去了牛津，想看看哲学家们是否对这个问题有更深入的理解。

And so trundled off, took some classes at Stanford, but then also trundled off to Oxford to see if philosophers had a better understanding of it.

我太喜欢了。

I love it.

这很有趣。

It's funny.

我觉得从那以后，符号系统专业已经成为斯坦福大学那些好奇且富有分析力、最终创办初创公司的人的首选专业。

I feel like since then, Symbolics and Systems has become the go to like Stanford major for like curious analytical people who end up doing startups.

所以知道你是最早的一批，真的很有趣。

So that's pretty funny to know that you're one of the first.

所以通常在这档节目中，我们会讨论人们如何实际使用ChekibT。

So usually in this show, we talk about, like, actionable ways that people use ChekibT.

而这正是关键问题。

And and that's that's the big question.

我认为，这就是人们来这里的原因。

That's, I think, what people come here for.

但在表象之下，我认为一个更有趣的问题是：人工智能整体，尤其是Treacy Petey，可能会如何改变‘人性’的含义？

But underneath that, I think what what a more interesting question is is, like, how does AI in general and Treachy Petey in particular, how might it change what it means to be human?

它可能会如何改变我们看待自己和世界的方式？

How might it change how we see ourselves and how we see the world?

它如何提升我们的创造力、智力以及诸如此类的东西？

How might it enhance our creativity, our intelligence, all that kind of stuff?

这些都是非常深刻、宏大的哲学问题。

And these are really deep, big philosophical questions.

作为一名曾系统学习哲学、并且可能至今仍在思考这些问题的人，我认为你对这一交汇点可能有独特的见解。

And as someone who rigorously studied philosophy and probably still thinks about those questions, I thought you might have a unique perspective on, on this intersection.

因为人们往往非此即彼：要么属于哲学阵营，要么属于语言模型阵营，而那些处于中间地带的人反而显得很有趣。

Cause I think people tend to be like, they're either in the philosophy camp or they're in the like language models camp and like people who are sort of in the middle is kind of kind of an interesting one.

我想先从这个问题开始，因为可能有些听众或观众会想：为什么？

And what I wanted to start with, cause I think there are probably people who are listening or watching who are like, why?

我只是想了解里德的实用建议。

I just want Reed's actionable tips.

我想问的是：为什么？你能告诉我你为什么关心哲学吗？

Is is to is to ask, like, why like, tell me more about why you care about philosophy.

我想你在谈论自己如何接触哲学时已经稍微提到过，但还是请你告诉我们，你为什么如此重视哲学？

And I think you got into that a little bit in in talking about how how you got into it, but like, yeah, tell us why is why do you care about philosophy?

为什么回答这些大问题很重要？

Why are answering these big questions important?

所以，当我给商学院做演讲时，有时我会说，哲学背景对创业的重要性甚至超过MBA，这当然令人震惊且反直觉。

So one of the things that sometimes will tell MBA schools when I give talks there is a background in philosophy is more important for entrepreneurship than an MBA, which of course is startling and contrarian.

这部分原因是为了让人们更清晰地思考这些问题，因为作为创业者，你正在思考世界可能成为怎样的样子？

And part of that is to get people to think crisply about this stuff, because part of what you're doing as an entrepreneur is you're thinking about what is the way the world could be?

它可能变成什么样？

What could it possibly be?

如果你用分析哲学的语言来说，就是逻辑可能性之类的东西。

What is, if you wanted to use analytic philosophy language, logical possibility or something like that.

但归根结底，就是在探讨什么是可能的。

But it's kind of what is possible.

而部分原因在于，这些都是人类的活动，你需要有对人性的底层理论——理解人类现在是什么样子、在某种意义上永恒不变的是什么，以及当环境变化时（比如技术、政治权力、制度等）人类会如何变化。

And then partially because these are human activities, what's your underlying theories of human nature about how human beings are now, how they are kind of quasi eternally, and how they are as circumstances change, as the environment in which the ecosystems we live in change, which is technology and political power and institutions and a bunch of other things as ways of doing that.

哲学对这些内容至关重要，因为它帮助我们清晰地思考：什么是可能性？什么是关于人性的理论？什么是人类本性在当下以及在新技术、新服务等影响下可能被改变的方式？

And philosophy is very important to this stuff because it's understanding how to think about very crisply what are possibilities, what are theories of human nature, what are theories of human nature as they are manifest today and as they may be modified by new products and services, new technologies, etcetera.

因此，人们通常会说，这是一个哲学问题，因为它是一个无法回答的问题，比如真理的本质，或者虽然我们都使用并理解语言，但我们并不真正知道语言是如何运作的。

And so obviously people tend to say, oh, that's a philosophical question because it's an unanswerable question, Nature of truth, or while we all speak and understand languages, we don't really know how that works.

这也是为什么哲学中出现了语言学转向，维特根斯坦等人因此而闻名，即这些哲学问题可能是语言问题。

And as part of the reason why there was the linguistic turn in philosophy that Wittgenstein and others were so known for, which is, well, maybe these problems in philosophy are problems in language.

如果我们理解了语言，我们就能理解哲学。

And if we understand language, we'll understand philosophy.

而围绕这些看似无法回答的问题，实际上，科学本身也充满了大量无法回答的问题。

And this question around these unanswerable questions, but actually, in fact, like science itself is full of a lot of unanswerable questions.

科学是通过不断演进的动态理论来运作的，而这正是人类境况的一部分。

And it's the working theory as we dynamically improve, and that's part of what the human condition is.

而这正是深度哲学的真正内涵。

And that's part of what actually the in-depth philosophy is.

它并不是说今天哲学中的某些问题——与柏拉图、亚里士多德，甚至前苏格拉底哲学家们所探讨的真理、知识等问题——是完全相同的。

It isn't to say that the same questions today, some of the same questions today in philosophy, the same questions that Plato and Aristotle and even the pre Socratics and other folks are grappling with truth, knowledge, etcetera.

但有些问题也是全新的，问题本身也在不断演变。

But some of the questions are also new questions and the questions evolve.

科学从哲学中演变出来的一部分，正是当我们发展出更具体的理论并提出新问题时，这些问题都是由此衍生出来的。

And part of how science has evolved from philosophy was this question as we get to our more specific theories and kind of developing the new questions that we get to, those are outgrowths.

在构建技术、产品和服务，以及创业过程中，情况也是如此。

And the same thing is true in building technology, in building products and services, in entrepreneurship.

这就是为什么哲学在应对严肃问题时实际上既强大又重要，而不是像我当年在牛津写论文时所研究的那样——思想实验的用途与滥用。

And that's why philosophy is actually, in fact, robust and important as applied to serious questions versus the, you know, one of the things I wrote my thesis on in Oxford was the uses and abuses of thought experiments.

最经典的例子就是电车难题。

And, the most classic one is trolley problems.

在电车难题的方法论中，既有合理的应用，也有滥用的情况。

And there are both uses and abuses within the methodology of trolley problems.

最有趣的一个例子是，如果人们还没看过的话，有一部电视剧叫《良善之地》，它在某一集中以一种极其幽默的方式展现了电车难题。

The most entertaining of which, if people haven't watched it, is there's a TV series called The Good Place, which embodied the trolley problem on a TV episode in an absolutely hilarious way.

这真的很有趣。

That's really interesting.

是啊，人们通常是怎么误用这个的？

Yeah, like, what is the way that people tend to misuse that?

因为我觉得电车难题在有效利他主义的讨论中非常常见，人们在网上经常遇到这个问题。

Because I feel like trolley problems are so common in EA discourse and people run into that a lot online.

根本问题在于，他们试图通过它来唤起某种直觉、原则等。

The fundamental problem is they try to frame it to an intuition, to drive an intuition, a principle, etcetera.

他们试图构建一个人为不同的环境。

They try to frame an artificially different environment.

所以就像是，没错，这是一辆电车，它要么撞上五个罪犯，要么撞上一个婴儿，而默认设置是撞向那个婴儿。

So it's like, no, no, it's a trolley, and the trolley will either hit the five criminals or the one human baby, and it's default set to hit the human baby.

你会不会扳动开关？

And do you throw the switch or not?

当你开始分析这个问题时，你会说：我怎么知道我不能把电车弄坏呢？

And then when you start attacking the problem, you say, well, how do I know that I can't break the trolley?

我可以只是让它停下来。

I could just not make it continue to run.

但你知道这一点。

It's like, well, you know that.

你会想：哦，所以你在思想实验中假设我拥有完美的知识，知道无法破坏电车。

You're like, oh, so you're positing in your thought experiment that I have perfect knowledge that breaking the trolley is impossible.

因此，为了让你的思想实验成立，你假设了我们在现实中从未遇到过、通常会认为是疯狂的情况，对吧？

So in your posit to make your thought experiment work, you're positing something we never Or when we encounter, we generally think people are crazy, right?

比如你拥有完美的知识。

Like you have perfect knowledge.

为什么我会知道，自己拥有完美的知识，知道无法破坏这辆电车？

Like why the fact do I know that I have perfect knowledge that I can't break the trolley?

而因为你接下来会问：面对这个电车难题，人类正确的反应是什么？

And because you're going to say, what is the right human response to this trolley problem?

我的反应是试图破坏电车，让它不会撞到任何人，

Is I'm going to try to break the trolley so it doesn't hit either of them,

这真的很有趣。

That's really interesting.

对。

Right.

你甚至可能会说，问题在于，你说你拥有完美的知识，知道你无法破坏它。

And you might even say that the problem is, is that to say, you say, even you say, well, you have perfect knowledge that you can't break it.

你可能会说，好吧。

You're like, well, okay.

你知道，A 并不具备完美的知识。

You know, A, don't have perfect knowledge.

而且 B，即使你真的有，也许这仍然是正确的反应。

And B, even if you did, maybe it's still the right response.

你试图让我承认：我是该什么都不做，任由它碾过婴儿？

You're trying to get me to say, do I do nothing and run over the baby?

还是该采取行动，碾过那五个罪犯？

Or do I do something and run over the five criminals?

这只有我仅有的两个选择。

Those are my only two options.

而你却说：不，不是这样的。

And you're like, Well, no.

我可以说，即使我认为自己无法阻止电车，我仍会尝试这么做，因为这是道德上该做的事。

I could say, Even if I think I can't break the trolley, that's what I'm going to try to do because that's the moral thing to do.

我听过一个

I've heard a

很多电车难题，但我从未听过有人停下来考虑第三种选择。

lot of trolley problems, and I've never heard anyone pause at the third option.

我喜欢这一点。

I love that.

这很棒。

That's great.

而且我也觉得，某种意义上，这些思想实验会劫持你的直觉，让你无法彻底推敲所有这些隐藏的假设，这让我联想到某些末日论或争论。

And I also like there's something about that where it's like, yeah, certain thought experiments sort of like hijack your instincts and and you don't quite reason through all these all these hidden assumptions that I think honestly reminds me of, like, certain doom or arguments.

我不想深入展开整个话题，但我认为这是一种非常有趣的方式来思考它。

And I don't I don't wanna, like, go into go into the full thing, but I think it's a it's a really interesting, way to think about it.

如果让我总结你刚才说的话，对你而言，哲学的价值在于清晰地思考可能性，思考人性与现实。

If I had to, like, summarize what what you just said, like, the value to you of philosophy is like, thinking crisply, thinking crisply about possibilities, thinking about, human nature and reality.

所有这些对商人来说都极其重要。

All those things are like really, really, really important for business people.

我想再进一步，有些哲学家、哲学学生或哲学爱好者经常磨练我们技能的那些问题。

I wanna kind of, like, take it take another step, which is, like, some of those some of those questions that philosophers like, or philosophy students or philosophy nerds just, like, sharpen our skills on.

有一些重大的、永恒的问题，比如什么是真理、什么是现实、我们能知道什么，诸如此类的问题。

There are some of these some of these big questions, some of the big perennial questions like what is truth, what is reality, what what can we know, all that kind of stuff.

当我开始谈论人工智能时，我很好奇，你是否觉得人工智能和大语言模型会为我们理解这些问题提供一些新的视角？

I'm kind of curious if you have a sense as we start to get into talking about AI stuff, what are those questions where, AI large language models are are going to give us a little bit of a new lens on on on on some of those questions?

或者，我们会发现一些比以前更好的新问题，即使它们可能并不能真正解答这些问题？

Or what are what are questions where we'll we'll find new ones to ask that are better than previous ones, even if they maybe don't answer them?

你对此有什么看法吗？

Do you have a sense for that?

嗯，从历史上看，比如那些催生了许多科学学科的问题，对吧？

Well, I mean, historically, it's like, for example, questions that have led to, you know, a bunch of the very science disciplines, right?

从物理世界到生物世界的一切，比如细菌理论等等。

It's everything from things in the physical world to things in the biological world, like germ theory and all the rest.

我认为这甚至更真实。

I think it's actually even true.

这正是哲学成为许多其他学科根源的原因之一。

It's one of the reasons why kind of philosophy is the root discipline for many other disciplines.

当你思考经济学和博弈论这类问题时。

When you get to questions around like, okay, how do you think about economics and game theory?

或者如何思考政治科学、现实政治以及国家与利益之间的冲突。

Or how do you think about of political science and realpolitik and kind of the conflict of nations and interests.

这也是我对大学未能重新创新最深刻的批评之一：学科壁垒的极端化。

And it's also one of the reasons why, probably one of my deepest critiques of the non reinvention of the university is the intensity of disciplinarianism.

也就是说，仅仅局限于政治科学这一学科，或者甚至仅仅局限于哲学这一学科，而不是采取跨学科的方式。

So it's just the discipline of just political science or just the discipline of even philosophy, as opposed to multidisciplinary.

我通常认为一个有趣的现象是，学术学科越来越趋于专业化，而不是每隔二十五年左右，我们就该彻底打破它们并以各种方式重新构建。

Part of the thing that I tend to think is kind of an interesting thing is how much the academic disciplines tend to be more and more disciplinary versus the, hey, maybe every twenty five years, we should think about blowing them all up and reconstituting them in various ways.

而这才是更好的思维方式，也是为什么学术界中最有趣的人往往是那些真正跨越学科界限的人。

And that would be actually a better way of thinking and why some of the most interesting people are the people who are actually blending across disciplines within academia.

我认为这部分非常重要。

And I think that part of it is, I think, extremely important.

而问题和哲学的一部分，就是探讨我们如何演变‘我们所知’这一问题。

And part of the question and philosophy is the kind of the question of like, well, how do we evolve the question of what do we know?

显然，我们通过例如科学史中的大量仪器和新型测量设备来演变我们所知的内容，这些设备有助于理论的构建。

And obviously you evolve the question what you know through, like, for example, a lot of the history of science is instrumentation, new measurement devices that help with kind of provisioning of theories.

这也是为什么人们常常没有充分思考技术如何帮助我们改变‘人类’的定义，因为我们有一种笛卡尔式的想象，认为我们是纯粹的思维生物。

And that's one of the reasons why people frequently don't think enough about how technology helps us change what is the definition of a human because we have this kind of imagination, like the Descartesian imagination that we are this kind of pure thinking creature.

但如果你仔细想想，我们学到的任何东西都表明，事实并非如此，对吧？

And you're like, well, if we've learned anything, that's not really the way it works, right?

这并不意味着我们不会以这种方式思考，不会通过抽象来生成逻辑和对世界的理论等等。

That doesn't mean that we don't think that way to have abstractions to generate logic and theories of the world and all the rest.

但如果你让一位哲学家服用一些LSD，你就会得到不同的输出。

But put your philosopher on some LSD and you'll get some different outputs.

这说得通。

That makes sense.

所以我想，沿着这个思路，如果我退后一步，眯着眼看，可以把哲学史大致分为本质主义和唯名论，至少在哲学的某个部分是这样。

So I guess like, along those lines, if you want if I step step back and squint, I can kind of like you can kind of divide the history of philosophy into essentialism and nominalism for for a certain part of philosophy.

对吧？

Right?

本质主义者认为，是否存在一个可认知的、根本的客观现实，是否存在一种能够‘按自然的关节’来切割世界的方式？

Like, and essentialists are like, do you believe that there are like fundamental there's a fundamental objective reality out there that's knowable and that there's a way to kind of like carve nature at its joints?

而唯名论者，比如我们所熟知的维特根斯坦——我知道你深入研究过他，还有实用主义者，他们认为真理更多是相对的，或基于社会惯例，或取决于什么有效，对此有各种不同的表述。

And nominalists, where we would include Wittgenstein, which I know you studied pretty deeply, and pragmatists, that more or less truth is more or less relative or it's about social convention or it's about what works or there's lot of different formulations of it.

在这两者之间，一直存在着持续的争论。

And there's this sort of like ongoing debate between people who think one thing one thing or the other.

你认为语言模型是否改变了，或者为这场辩论的任一方增添了新的视角？

Do you think language models like change or add any way to either side of that debate?

我认为它们提供了新的视角和色彩。

I think they add perspective and color.

但它们并没有解决这场辩论。

I don't think they resolve the debate.

而且确实存在一个问题，因为语言模型的功能更类似于后期维特根斯坦，或者说更偏向唯名论，你会说：它们的运作方式是否倾向于支持唯名论一方呢？

The And there's certainly some question about since they function more like later Wittgenstein or more, you know, kind of nominalist, you know, you say, well, does that weigh in on the side of nominalists because of actually, in fact, the way they function?

事实上，当你观察我们如何试图开发大型语言模型时，我们会努力让它们体现更多的本质主义特征。

And actually, in fact, you say, well, if you look at how we're trying to develop the large language models, we're actually trying to get them to embody more essentialist characteristics as they do it.

比如，如何使它们更贴近真理，减少幻觉等等。

Like how do you ground in truth, have less hallucination, you know, etcetera.

为了提及一位更早的德国哲学家，黑格尔，我认为人类境况的一部分正是正题、反题与合题的辩证过程。

And you know, to gesture at a different earlier German philosopher, you know, Hegel, one of the things I think is kind of part of a I think it was kind of the human condition is the thesis, antithesis, synthesis.

你可以说，我们有一个本质主义的正题，一个唯名论的反题，而合题则是我们以各种方式将二者结合起来的方式。

Like you could say, hey, we have an essentialist thesis, we have a nominalist antithesis, And the synthesis is how we're putting them together in various ways.

因为你可以说，我们——而且我认为后期维特根斯坦也不会说世界仅仅是语言，就像解构主义者和德里达所走向的那样。

Because you say, look, we and I don't even think later Wittgenstein would have said that the world is only language, you know, kind of what the deconstructionist and Derrida went to.

他们认为语言只是遮蔽世界的面纱，你完全无法接触现实，因此完全不扎根于世界。

Was like, you know, it is only the veil of language and you have no contact with the world, so you're not grounded in the world at all.

我认为他会觉得这种观点相当荒谬，对吧？

I think he would think that's kind of absurd, right?

但他的观点是，我们在生活中所体现的生活形式，其运作方式并非简单的指称，他明白这不仅仅是指称‘猫在垫子上’或‘猫在垫子上的可能性’，而是宇宙可能的各种配置。

But his point was, is to say that there is also in how we live his forms of life, the way that it operates is not a simple kind of denote of and he understood it wasn't just denoting the cat on the mat or the possibilities the cat is on the mat and the possibility of the cat is on that, but actually possible configurations of the universe.

这种所谓的‘可能性语言’所描述的是一种概念上的逻辑可能性，认为这种可能性语言是本质主义的，实际上是对我们如何发现真理、如何实践真理的误解。

And that was this kind of notional logical possibility that was described as language of possibility was to say that kind of essentialist about a language of possibility is actually incorrect to actually how we discover truth and how we operationalize truth.

你仍然拥有一个强大的真理理论，这与解构主义者所主张的截然不同。

And you still have a robust theory of truth, which is not essentially what the deconstructionists do.

但这种强大的真理理论部分建立在‘语言游戏’和生物性生活形式这一观念之上。

But the robust theory of truth is partially grounded in this notion of language games and a biological form of life of how you do that.

然后，你显然会深入探讨：那么，数学作为经典的真理语言，是如何成为一种语言游戏的呢？这是一种试图理解这一点的方式。

And then obviously you go into this deeply with saying, well, okay, how is mathematics language game as a classic language of truth is a way of trying to understand that.

这也正是哲学家们所称的‘克里普克斯坦’的由来，即索尔·克里普克对维特根斯坦思想的卓越解读视角。

And that's part of where you get what philosophers refer to as Kripkenstein, the Saul Kripke excellent you know, lens on reading a part of what Wittgenstein was about.

接着，你将这一切应用起来，大家都会问：这与大语言模型有什么关系？

And you kind of then apply all that, you know, everyone's going, where is this going to large language models?

而你会说，事实上，语言正是这种语言游戏的展开。

And you say, well, actually, in fact, you know, language is this play out of this language game.

大型语言模型正在以各种方式演绎这种语言游戏。

Large language models are playing out this language game in various ways.

但揭示出来的一部分是，我们不能简单地认为真理就是语言中所表达的内容。

But part of what is revealed is we don't just go, truth is what is expressed in language.

真理是一个动态的过程，是一种人类对话。

Truth is a dynamic process and kind of human discourse.

它可以是综合，即正题、反题、合题，或者其他形式。

Could be synthesis, synthesis, know, thesis, antithesis, synthesis, or other things.

正是这种源自对话时代的、人类的对话——真理的发现、逻辑与推理过程——无论是归纳、演绎还是溯因推理，这些推理过程使我们形成了那些始终处于不断演进中的真理理论。

It's this human discourse that's coming out of this dialogic period, this truth discovery, this logical, this reasoning, whether it's induction, is reasoning, whether it's abduction, whether it's deduction, and these reasoning processes that get us to what we think are these kind of theories of truth that are always, to some degree, works in progress.

这真的非常有趣。

That's really fascinating.

我想试着总结一下，因为说实话，这部分可能有点难理解。

I I wanna try to summarize that in case in case it was a little bit difficult to follow, to be honest.

我觉得里面有个观点我可能没跟上，你告诉我我漏掉了什么。

Like, there's a there's a point in there that I think I missed something, so you tell me what I what I missed.

但我认为，我听到的一些内容中，有一部分特别有趣，那就是当我们思考如何构建人工智能时——它是通过预测下一个词元来实现的，这与后期维特根斯坦的观点或实用主义观点非常契合，即重点在于句子中不同词语之间的关系。

But I think one of the like, some of the things that I heard in there that that I thought I thought was really interesting is when you think about how we built AI, which is predicting the next token, that's a very sort of late Wittgenstein compatible idea or pragmatic, like, compatible idea where it's really about the relationship between different words in a sentence.

我们并不是在发现关于世界的新事物。

And it's not we're not finding anything out about the world.

还有其他一些人工智能方法，比如七八十年代的那种，试图逐个列出世界上所有的物体。

Like there are other AI approaches, I don't know, in the eighties or seventies where it was like literally like, let's list out every single object in the world.

但那些方法并没有真正成功。

And those didn't really work.

那会是一种更偏向本质主义的人工智能路径。

And that would be like something along the lines of a more essential approach to AI.

而真正有效的方法则是一种更实用、更贴近后期维特根斯坦思想的方法。

And the one that works is is a more pragmatic and more late Wittgensteinian one.

但有趣的是，我们现在有了这种实用主义的基础，并正在努力使其更加扎根于现实，或更精确地还原为能够谈论本质真理的能力。

But what's quite interesting is now that we have that pragmatic base that we've bootstrapped, we're in this process of trying to make it more grounded, more grounded in reality or more reduced down to being able to talk about the essential ground truth.

我认为维特根斯坦真正有趣的一点是，他 famously 说过：‘我的语言的界限就是我的世界的界限。’

And I think what's really interesting about Wittgenstein is he's sort of famous for saying, like, the limits of my language are the limits my world.

我不知道。

I don't know.

我不记得这是晚期还是早期的了。

I don't remember if that's late or early.

但大致来说，我认为你的意思是，维特根斯坦并不认为语言之外一无所有，但他确实认为，我们谈论世界的方式或使用语言的方式，是这种社会对话的一部分，在其中我们所有人不断互动，共同发明语言、结构和语言游戏。

But but more or less, like, I think what you're saying is that Wittgenstein doesn't think that, like, there's nothing outside of language, but he does think that the way we talk about the world or the way that we use language is part of this sort of like social discourse where we're all kind of like going back and forth to like co invent language and structures and language games together.

而在语言模型中，你确实能看到这种现象发生，比如当你进行RLHF时，这就像是我们在与语言模型玩一种语言游戏，告诉它：不，不，你不应该那样。

And and you kind of see that happening with language models where like when you do something like RLHF, that's sort of us playing with a language model, like playing a language game to be like, No, no, you don't like that.

这大致就是你想表达的意思吗？

Is that generally what you're getting at?

是的。

Yes.

你所说的全部都是对的。

So everything you said.

但除此之外，后期的维特根斯坦其实一直在以各种方式探索一个额外的问题，因为他并不想完全把真理视为一种社会建构。

But then the additional thing, which later Wittgenstein was really trying to explore in various ways because he wasn't trying to do a kind of a completely just social construction of truth.

你知道吗，我其实认为，只有成为一位维特根斯坦学者，才能真正理解早期和晚期维特根斯坦其实是同一项事业的一部分。

You know, I'm actually a fan of you have to be a Wittgenstein scholar to actually understand how both early and late Wittgenstein are actually part of the same project.

早期的维特根斯坦是个傻瓜，现在让我看看，我已经虔诚地转向了这种不同的观点。

And early Wittgenstein was an idiot, and now let me look, I've religiously converted to this different point of view.

但有一个特别的问题，那就是：你如何获得对‘真理’的理解？

But there is a particular thing, which is how do you get to the notion of understanding truth?

而真理是通过语言并借助某种方式发现的动态过程，它必须具有一些明确的外部条件，而不是‘我的真理’或‘你的真理’。

And truth is the dynamic of discovery through language and through kind of, it has to have some explicit external conditions that isn't my truth, your truth.

只在某种程度上，才存在‘我们的真理’，或以各种方式呈现的真理。

There is only to some degree our truth or the truth in various ways.

而你如何通过你的所作所为，建立起这样的真理条件？

And how do you get to that as what you're doing and having truth conditions.

在早期维特根斯坦那里，真理条件体现为一种可能性与现实性的状态，存在于这个可能性的逻辑空间中，包括物理空间在内，但又超越了它。

And in kind of early Wittgenstein, the truth condition was it cashes out into a state of possibilities and actualities in this logical space of possibilities, include physical space as part of it, but broader than that.

而晚期维特根斯坦则说：事实上，这种对逻辑可能性的建模根本不是事物运作的真实方式，对吧？

And then later Wittgenstein said, well, actually, in fact, this modeling of logical possibility is actually not the fact the way this works, right?

而我们实际上并不是以这种方式来奠定基础的。

And we're not actually, in fact, grounding it that way.

我们奠定基础的方式在于语言游戏和语言中的行动这一概念。

The way that we're grounding it is in the notion of how we play language games, make moves in language.

这种基础在某种程度上依赖于共享某种生物性的、所谓的生活形式，通过它我们认识到某种语言游戏中的行动是有效的。

And the way that's grounded is to some degree sharing a certain biological, you know, kind of form of life by which we recognize that's a valid move in the language game.

这种行为在语言游戏中是无效的。

This is not a valid move in the language game.

当涉及到大型语言模型时，有趣的是，你会想：大型语言模型是否与我们拥有相同的生物性生活形式，还是不同？

Now, is what's interesting when it gets to large language models, because you go, well, large language models, are they the same biological form of life as us, or are they different?

这会如何体现出来？

And how does that play out?

我认为维特根斯坦会认为这个问题极其迷人，并会深入探究，试图弄清楚它。

And I think Wittgenstein would have found that question utterly fascinating and really would have gone very deep on it, trying to figure that out.

顺便说一句，答案可能是部分是、部分不是，既不是100%是，也不是100%否。

And by the way, the answer might be some and some, not 100% or 100% no.

100% 是，100% 否。

100% yes, 100% no.

因为支持这一观点的理由是，大型语言模型是在人类知识、语言及其他所有内容的语料库上训练的，它们在这些基础上进行语言模式的运作。

Because the argument in favor is the large language models are trained on the corpus of human knowledge and language and everything else, and they're doing language patterns on that.

有些人甚至认为，它们的一些模式与人类学习和大脑的模式非常相似。

Some might even argue that some of their patterns are very similar to the patterns of human learning and brains.

另一些人则认为并非如此。

Others would argue that it's not.

但接着你会说，它也不是一个生物实体。

But then you'd say, well, but it's also not a biological entity.

而且它的学习方式实际上与人类的学习方式大不相同。

And it learns actually very differently than human beings learn.

因此，尽管它的语言游戏看起来像是人类的语言游戏，但实际上在重要方面是不同的。

And so maybe its language game, which looks like it's the human language game, is actually different in significant ways.

因此，其真值功能实际上也非常不同。

And so therefore, the truth functions are actually very different.

从某种意义上说，当我们改进和推进LLM的构建方式时，我们试图让它们在真实性基础上更加可靠。

And in a sense, what we're trying to do when we are modifying and making progress with how we build these LLMs is to make them much more reliable on a truth basis.

我们喜欢它的创造性和生成性，但在大量真正有助于增强人类能力的场景中，我们希望它能具备更好的真实感，对吧？

Like we love the creativity and the generativity, but we want it to almost, for a huge amount of the really useful cases in terms of amplifying humanity, we want it to have a better truth sense, right?

我的意思是，当前GPT中的悖论在于，当你用一些关于质数的简单问题去试探它时，就能发现这些问题。

I mean, like the paradoxes in current GBT are when you can kind of tease it out with like very simple questions around prime numbers.

然后你会说：‘你这个答案错了。’

And you go, well, you got that answer wrong.

它会说：‘哦，对，我答错了。’

It's, oh yeah, I got it wrong.

这是正确答案。

Here's the answer.

但这个答案也是错的。

Well, answer is wrong too.

哦，我这个也答错了。

Oh, I got that one wrong too.

这是答案。

Here's the answer.

而一个理解这些事情的人类，我只是在不断犯错。

And a human being understanding these things, I'm just getting these things wrong.

明白了。

Like, got it.

我明白了。

I got it.

我错了。

I'm wrong.

而不是说，对不起，你对了。

As opposed to, oh, I'm sorry, you're right.

我答错了。

I got it wrong.

这是另一个错误的答案。

And here's another wrong answer.

我们在做这件事的同时，正试图将这种对真理的感知融入其中。

We're trying to get that truth sense into it as we're doing it.

因为我们确实有一些概念，哦，对了，这才是典型的特征。

Because we do have some notion of, Oh, right, this is what's characteristic.

比如，数学让我们进入了某些语言游戏非常纯粹的定义中。

Like mathematics gets us into very pure definitions of certain kinds of language games.

这也是为什么几个世纪前，人们认为数学可能是宇宙的语言、上帝的语言或其他类似的语言的原因之一。

It's one of the reasons why, you know, centuries ago, people thought math was maybe the language of the universe or language of God or language of etc.

因为你会觉得，好吧，那些最纯粹的真理之一，比如二加二等于四，就深植于其中。

Because you're like, okay, there is the one where the purest truths, some of the purest truths that we know, two plus two equals four, is kind of embedded in.

当我们探索如何创造这些语言工具、语言装置时，我们仍在不断厘清这一点。

And we're still working that out as we play with how we create these language tools, these language devices.

这也是我认为这个问题非常有趣的原因之一，因为你实际上可以将其建模为我们在开发下一代技术时所试图实现的物理机制。

And it's part of the reason why I think this question is really interesting because you can actually model it to some of the actual, as it were, the technological physics that we're trying to create when we're doing the next version.

比如，我们如何将这些特性融入优秀的推理机器，而不仅仅是优秀的生成机器？

Like, how we get these things into good reasoning machines, not just good generativity machines?

它们从生成能力中具备一些推理能力，但经典的问题在于，当它们的推理在我们作为人类、作为我们理想中的最佳自我所珍视和追求的方式上失效时，我们就暴露了这一点。

And they have some reasoning from their generativity, but like part of classic showing where they break is showing where their reasoning stops working in ways that we value and aspire to in terms of what we try to do as human beings as in our best selves.

有一件事很让人疲惫，就是导出导入的繁琐流程。

Here's something exhausting, the export import dance.

如果这听起来很熟悉，告诉我一声。

Let me know if this sounds familiar.

你在 Figma 中设计一个东西，导出后粘贴到其他地方，然后祈祷别出问题。

You design something in Figma, you export it, you paste it somewhere else, and you pray that nothing breaks.

但通常都会出问题。

And usually something does.

都快 2026 年了，这说明是时候停止这种做法了。

Almost 2026, which means that it's definitely time to stop doing this.

Framer 已经打造了发布精美、生产级网站的最快方式，现在它正在重新定义我们设计网页的方式。

Framer already built the fastest way to publish beautiful, production ready websites, and now it's redefining how we design for the web.

随着最近推出的基于画布的免费工具 Design Pages，Framer 已经不再只是一个网站构建器。

With the recent launch of Design Pages, a free canvas based tool, Framer is more than a site builder.

这是一个真正的全栈设计平台。

It's a true all in one design platform.

从社交媒体素材、活动视觉设计到矢量图和图标，直至上线的网站，一应俱全。

From social assets to campaign visuals to vectors and icons, all the way to a live site.

Framer 是创意从开始到最终上线的完整场所。

Framer is where ideas go live from start to finish.

Framer 的设计工具与你在其他播客中看到的那些传统网站构建器截然不同。

Framer's design tool is different from old school website builders you might see advertising on other podcasts.

它免费提供矢量编辑、3D 变换、渐变和动画功能。

It offers vector editing, three d transforms, gradients, animations, all for free.

它支持无限项目、无限页面和无限协作者。

It has unlimited projects, unlimited pages, and unlimited collaborators.

但真正改变我对 Framer 认知的是，无需进行交付。

But what really changed how I think about Framer is that there's no handoff.

你设计的即是网站本身。

What you design is the website.

无需开发人员解读。

No developer interpretation.

无需再讨论如何让实现效果匹配设计稿。

No can you make it match the mock up conversation.

你设计它，发布它，它就上线了。

You design it, you publish it, and it's live.

你准备好在一个工具中完成设计、迭代和发布了吗？

Are you ready to design, iterate, and publish all in one tool?

立即免费开始创作，请访问 framer.com/design，并使用代码 DAN 获取一个月的 Framer Pro 服务。

Start creating for free at framer.com/design, and use the code DAN for a month of Framer Pro.

访问 framer.com/design，优惠代码 Dan。

That's framer.com/designpromo code Dan.

规则和限制可能适用。

Rules and restrictions may apply.

现在，回到本集节目。

And now back to the episode.

这真的非常有趣。

That's really fascinating.

你刚才说了很多。

You said a lot there.

我一会儿真的很想深入探讨一下推理的问题，但我想先回过头来谈谈你提到的后期维特根斯坦和早期维特根斯坦的区别，因为我以前从来没听过这种说法。

I really wanna get into the reasoning thing in a second, but I wanna go back to the the sort of the way that you talked about late Wittgenstein versus early Wittgenstein because I haven't really heard it said that way.

通常人们会说，他年老时只是完全否定了自己以前的一切观点之类的。

And the usual, like, thing people say is, like, he just disagreed with everything when he was older or whatever.

而我现在听到你的意思是，这两种情况下，他实际上在说一些相似的内容，或者持有某些相同的观点。

And what I hear you saying now is more or less in both cases, he's saying some of the of the same things or he has some of the same views.

但真正的区别在于，他是如何诠释‘什么是真’的。

But like the real difference is how he cashes out what it what it means to be true.

某件事是否为真。

Something is whether something is true.

在他的早期阶段，他用一种逻辑可能性空间来谈论真理，这种空间可以分解为他称之为‘原子事实’的最小单位。

And in the first, in his, like, sort of first period, he's, talking about truth in terms of, a logical space of possibilities that can be broken down into these, like, little, he calls atomic facts.

这些从未被真正定义过，但你可以从那里逐步构建出真理，将这些可能性映射到现实世界中的实际事物。

And those are never really defined, but, like, you can kind of build up truth from there, mapping those those possibilities into actualities, like what's actually in the world.

而在后期的维特根斯坦那里，一切都围绕着这些所谓的语言游戏、社会关系，以及词语或短语在人们语境中的使用。

And in later Wittgenstein, it's all about these sort of like language games, the social relationships, like the the use of that word or that phrase in the context of people.

我真的很想问你的是，那种早期的维特根斯坦观点——即那种可能性的逻辑空间。

And one of the things that I I really wanted to ask you about is like that first that first version of Wittgenstein where it's sort of that logical space of possibilities.

这让我想起了嵌入向量，嵌入向量是催生人工智能的关键底层技术之一。

What that reminds me of is embeddings, where embeddings are one of the key underlying technologies that gave rise to AI.

在传统的自然语言处理中，它们允许你将词语或标记表示在高维空间中。

In traditional NLP, they're allowing you to represent words or tokens in a high dimensional space.

而语言模型的创新之处在于，它不只是关注词语本身，而是关注词语在特定语境中的使用。

And then the language model, like innovation is kind of like, it's not just words, it's words in their particular context.

每个词在特定语境中都占据空间中的独特位置。

Each each word in that particular context has its own part of the space.

所以在一个语言模型中，如果‘king’这个词被这样分词，你知道，国际象棋里也有一个‘国王’。

So like, in a in a language model, the word king, if if it's if it's tokenized that way, you know, there's a king in chess.

有一个国王。

There's a king.

有一个真正的国王。

There's an actual king.

有一个英格兰的国王。

There's like a king of England.

有一个李尔王。

There's a king Lear.

它们都像是国王，但属于不同的空间。

And they're all kind of like kings, but they're like different spaces.

语言模型能够表示所有这些不同的含义，当我们说‘国王’时，我们指的是许多不同的事物，而模型能够表达这一切。

And language models are able to represent all of those different like, when when we say king, we mean many different things that are able to represent all of that.

这让我想起了原子事实，或者维特根斯坦早期的作品？

And that just actually reminds me a lot of of, like, atomic facts or or or the the first, like, Wittgenstein's early early work?

我只是很好奇，因为我觉得你提到语言模型由于下一个词预测的机制，更像是后期维特根斯坦的思想，但我很好奇你是如何考虑嵌入机制的，毕竟它是这一技术的核心部分。

And I'm just kind of curious, like because I think you said that language models sort of because of the next token prediction, they they're they're sort of late Wittgensteinian, but I wonder how you, like, factor in the fact that embeddings work and they're sort of a core part of this.

嗯，事实上，这正是晚期维特根斯坦与早期维特根斯坦不同的部分，早期维特根斯坦是个傻瓜。

Well, and actually, this is part of the fact that late Wittgenstein is not early Wittgenstein was an idiot.

因为，是的，我认为‘因果’这个概念本质上是一种概率性推测，即哪些不同的词元可能适用。

Because, yes, I do think that the notion of cause that were a probabilistic bet for what are the set of different tokens that apply are kind of there.

我之所以更倾向于将当前实践归为晚期维特根斯坦而非早期，是因为早期维特根斯坦认为，一旦你掌握了其中的逻辑，只要正确地表达，就几乎不可能犯真理性的错误，因为逻辑已经内嵌其中。

Now, the reason why I would kind of slant more as current practice late Wittgenstein than early Wittgenstein, because early Wittgenstein thought that once you had the grasp on the logic of it, you then almost by speaking correctly couldn't make truth mistakes because the logic was embedded in it.

尽管词元嵌入确实属于一种非常广泛的符号性、准符号性网络的一部分。

And even though the token embeddings are kind of part of a very broad symbolic, quasi symbolic, I would say, kind of network.

之所以说是‘准符号性’，是因为它仍然是基于激活状态等机制，而不是纯粹围绕‘国王’这个词元、或15个不同‘国王’词元、或23个部分‘国王’词元进行推理，而更像是从海量语言使用中映射出的概念空间。

And the reason it's quasi symbolic is because it's still kind of activations and so forth and isn't purely the reasoning around a token of king or 15 different tokens of king or 23 different partial tokens king, as much as there's kind of conceptual spaces in that tokenization as mapped from a very large use of language.

但语言的一部分不仅在于历史用法，还在于对它的重新应用。

But part of language isn't just the historical language, but is the reapplication of it.

比如，你说‘这是播客之王’，对吧？

Like if you say, This is the king of podcasts, right?

或者，‘这是麦克风之王。'

Or, this is the king of microphones.

还没有，但也许吧。

Not yet, but maybe.

是的，但仅仅作为实例而言，这正是后期维特根斯坦所关注的：我们如何玩这些语言游戏，以及如何重新应用它们。

Yes, but just kind of as instances, that's part of why kind of later Wittgenstein went to, well, it's how we're playing these language games and how we're reapplying them.

当我们说，比如在这个播客中，这可能会成为播客之王。

When we say, like for example, we say on this podcast, this could become the king of podcasts.

我们所有人都能感受到我们在做什么。

We all have a sense of what we're doing.

那么，哪些情况下这句话为真，哪些情况下为假呢？

It's like, well, would be the cases where that would be true and what would the cases where it be false?

它做出了什么样的预测？

And what prediction is that making?

这种说法为什么是有用的？

And how is it that that's a useful thing?

我肯定有人之前说过‘播客之王’，但我以前从未听过，对吧？

I'm sure someone said king of podcasts before, but I've never heard it before, right?

而且这是一种不同的分词方式，尤其是在讨论中不断发展和细化之后。

And it's a different tokenization, especially as it gets developed and elaborated a lot in discussion.

然后，如果你突然获得了关于国王、王国等讨论的另一个太字节的信息，那么它所学习的词元空间就会发生变化，对吧？

And then actually, if you suddenly had another terabyte of information about discussions of kings and kingdoms and all the rest, and all of a sudden that token space that it's learning from would change, right?

而基于它的泛化也会随之改变。

And then the generalizations off it would change.

这正是我认为它更接近后期维特根斯坦观点的原因，尽管并未完全脱离早期的嵌入方式。

And that's part of the reason I would say it's kind of more later Wittgenstein, even though not completely disconnected from those embeddings early.

这也是为什么实际上，后期维特根斯坦认为真理并非仅仅是语言所说的那样。

And it's one of the reasons why, actually, in fact, later Wittgenstein is not truth is just what language says.

不，语言是通过我们作为生物体如何与世界互动而嵌入其中的。

It's no, there's ways in which it's embedded in the world by how we navigate as biological beings.

而这正是世界如何影响语言的一部分机制。

And that's part of how the world kind of comes and impacts it.

因此，语言并非像笛卡尔式意识那样自由漂浮，而是以某种方式嵌入在现实中的。

And therefore it's not just language by itself free floating like the Cartesian consciousness, but it's embedded in some ways.

他试图做的部分工作是，从哲学的角度出发，理解这些嵌入表示，并基于这种生物性嵌入来构建我们的语言真值讨论。

And part of what he was trying to do is figure out, well, from a philosophy standpoint, how do we understand those embeddings and how do we drive our truth discourse in language based upon that biological embedding.

这说得通。

That makes sense.

所以我认为我听到你的意思是，尽管嵌入表示将词语映射到一个高维空间，这似乎像是将词语映射到某种原子事实或逻辑可能性空间中。

So I think what I what I hear you saying is despite the fact that embeddings are in this sort of they're they're mapping words into this high dimensional space, which sort of seems like kind of mapping words into this, like, sort of atomic facts or, like, logical possibility space.

但这个空间的构建方式，以及决定某个词进入空间哪个部分的原因，更接近后期维特根斯坦的观点，因为它非常关注词语在实际使用中的情况，以及是否对人类在世界中的实践有用，而不是基于某种深层的逻辑秩序——那种秩序一旦建立，你就不能说任何错误的话，因为你只使用了该空间中的词语。

The way that that space is constructed and and what makes something go into one part of the space or another is more late Vic and Shenyang because it's very much about how it's used in practice and whether it's useful for humans in the world rather than, like, it's about some deep underlying logical ordering where if you've created that ordering, you can't say anything wrong because you're only using words from that space.

我的理解对吗？

Is that kind of on target?

是的，完全正确。

Yes, exactly.

而且部分原因是，我们知道有些情况下，语言的连贯使用仍然可能是虚假的。

And part of it is we know that there's truths where the coherent use of language still is a falsity.

因此，我们试图弄清楚的是，如何让更多的真理、真实表达和推理——因为推理就是寻找真理——融入到这些大语言模型的工作方式中。

And so do, like part of what we're trying to figure out is how do we get more of those truths and truth telling and reasoning, because reasoning is about finding truth, into how do these LLMs work.

为了更深入地探讨这一点，你觉得在将推理能力融入这些语言模型方面，最有前景的方法是什么？

And just to move into that point a little bit, what is most promising to you in terms of ways that we are getting reasoning into these language models?

你认为哲学领域，无论是维特根斯坦还是其他人的思想，是否与这一目标相关？

And do you think that there are any ideas from philosophy, whether Wittgenstein or otherwise, that are relevant to that project?

答案肯定是肯定的，相关的思想确实存在。

Well, answer is certainly yes on the relevant ideas.

目前，我认为我们正在做几件事。

Currently, I think we're doing a couple of things.

所以，我认为我们正在尝试将人类的知识纳入训练内容之中。

So I think we're taking kind of, call it, human knowledge and figuring out how to get that as part of what's trained.

最早的发现之一实际上是，如果你用代码进行训练，这些模型就能学到远超计算机代码本身的更广泛的推理模式。

So the earliest discoveries were actually, in fact, if you trained on code, computer code, then these models learn patterns of reasoning much broader than just computer code.

因此，所有进行此类工作的模型现在都会训练于计算机代码，即使它们的目标并不是成为微软Copilot、代码生成等工具。

And so all of the models that are doing this are now also training on computer code, even if they don't have a target of being a Microsoft copilot, code generation, etcetera.

即使它们并不专注于此，因为代码中蕴含着类似数学那样清晰的推理建模模式。

Even if they're not doing that, because there's pattern just like math of crisp kind of modeling of reasoning.

另一个目前正在发生的是：你们是如何使用教科书的？

Another one is that's currently happening is, well, what are you doing with textbooks?

其理念是，如果你采用人类通过教科书所接受的那种训练方式，就可以构建出更小但依然非常有效的模型，以教科书作为实现途径。

And the notion is if you take the same kind of training discipline that we use for human beings encapsulated in textbooks, you can, for example, build much smaller, but still very effective models based on textbooks as ways of doing it.

因此，教科书是另一个方法。

And so textbooks is another one.

现在，当你开始思考时，可能会有一些有趣的计算哲学问题。

Now, as you begin to, like there's probably like some interesting, as it were, computational philosophy.

如果你开始探讨，我们如何将科学理论——比如各种不同的科学哲学理论——转化为模型，例如将拉卡托斯对波普尔的修正，以及库恩的科学范式理论融入其中？

If you begin to say, well, how do we cash out kind of theories of, whether it's kind of, call it theories of science, the kind of different theories of science, and you're kind of building those models into how do you get, you know, it's kind of like Lakotosh as a development on Popper, given thinking about Kunian kind of models, a scientific paradigm?

你如何基于这些理论做出预测？

How do you kind of make predictions on those kinds of bases?

还有一些深入的逻辑研究，比如贝叶斯逻辑，可能有助于从这个角度进行分析。

And some of the in-depth work in logic, maybe Bayesian logic, as ways of possibly looking at this.

我非常确信，还有许多非常有用的内容值得进一步拓展。

I'm quite certain that there probably are some very useful things to elaborate beyond it.

现在，当然，这些事物的一个概念是它们是学习机器。

Now, currently, of course, part of the notion of these things are they're learning machines.

因此，你需要为它们提供一个相当丰富的数据集来学习。

So you have to give a fairly substantive corpus of data from them to learn from.

当然，还有合成数据，而且，哲学可能在于：我们如何创建那些基于现有数据、仍然有用的合成数据模式？

Now, of course, there's synthetic data, and look, there may be like philosophy is in what patterns do we create synthetic data that is still useful to learn from off the current data?

你知道，无论如何，这里有很多不同的研究方向，但我确信它们是存在的，尽管我并没有提出具体的理论来说明这些是如何实现的，而只是做了一些手势。

You know, might be anyway, so there's a bunch of different kind of gestural areas, but I'm certain those are there, even though I'm not bringing up I'm making gestures rather than specific theories as to how that there, there cashes out.

这真的很有趣。

That's really interesting.

所以，似乎我们试图将推理引入模型的方式，就是寻找那些具有清晰推理过程的数据源，从而让模型从中学到推理。

So it seems like basically the way that we're trying to get reasoning into models is to find sources of data that just has really crisp reasoning, and so they'll learn the reasoning from that.

是的。

Yep.

我有点好奇，如果是这样的话，难道逻辑中能做的‘动作’不就只有那么几种吗？

I'm sort of curious if if that's the case, like, aren't there are only a certain number of, like, moves you can make in logic.