大多数人对进化的误解：人工生命、开放式演化与超越地球的智能

我想澄清一下，进化绝不是随机的。所以，人工生命真正关乎的是理解，正如你所说，生命可能的样子，而不是生命现在的样子。我们需要理解所有可能智能的空间，所有可能生命形式的空间。

I would just want to clarify that evolution is anything but random. So, artificial life is really about understanding, like you said, life as it could be rather than life as it is. We need to understand the space of all possible intelligences, the space of all possible life forms that can be.

康威的生命游戏是约翰·康威的一个非常著名的模拟，你有一个细胞网格。这是一个二维的细胞网格，它们可以是开或关，活或死。就这样。从这个超级简单的世界，可能是你能想到的最简单的东西，你运行这个，你会得到疯狂的复杂性，直到你实际运行它，没人会想到它能做到。

Conway's Game of Life is a super famous simulation by John Conway where you have this grid of cells. There's a 2D grid of cells and they can either be on or off, alive or dead. That's it. And from this super simple world, it's like possibly the simplest possible thing you can think of, you run this, you get insane amounts of complexity that no one would have thought it could do until you actually run it.

你可以在其他宇宙中进行反事实分析。在哪些世界中，我们得到这种开放式的复杂性和涌现的创造？在哪些世界中，我们没有得到？为什么这个物理，当你运行它时，它会产生这个惊人的现象？而当你运行这个物理时，它不会那样。它只产生随机噪声。

You can do counterfactual analysis in other universes. In what worlds do we get this open-ended creation of complexity and emergence? And in what worlds do we not get that? Why does this physics, when you run it, it produces this amazing phenomenon? And when you run this physics, it doesn't do that. It produces just random noise.

与其硬编码一个单一的模拟，你参数化一个模拟空间，这要容易得多。只需模拟它。看看会发生什么，将其插入基础模型，看看基础模型对发生的事情有何看法。

Instead of hardcoding a single simulation, you parameterize a space of simulations, which is much easier. Just simulate it. See what happens, plug it into a foundation model, and see what the foundation model thinks about what happened.

有 260,000 条规则。我们绘制了所有规则。所以，我们发现实际上有一个大岛的非开放式解决方案，还有一个小岛，所有最酷的模拟都在那里。其中一些最终创造了看起来像细胞的小模式，四处移动。这个物体，这个类似细胞的物体，正在以一种方式利用物理，在这个世界中创造一个持久物体，对吧？

There's 260,000 rules. We plotted all of them. So, we found that there's literally like a big island of solutions which are not open-ended, and there's a small island of solutions where all of the coolest simulations lie. Some of them actually end up creating little patterns which move around that look like cells. This object, this cell-like object, is exploiting the physics in a way to create a persisting object in this world, right?

如果你训练它们微积分，它们会学习微积分。你不必先训练它们算术。我们训练这些模型的方式是如此奇怪和异类。我真的喜欢称它们为统计智能，它们在统计上基本上是完美的，但这个 FER 想法和 PicBreeder 系列工作，基本上是我们试图研究的一种不同的智能范式。所以，这是统计智能和基于规律性的智能之间的区别。

If you train them on calculus, they'll learn calculus. You don't have to train them on arithmetic first. And it's such a weird alien way that we train these models. I really like to call them a statistical intelligence and they're basically perfect statistically, but this FER idea and this PicBreeder line of work, it's basically a different paradigm of intelligence that we're trying to study. So, it's a difference between a statistical intelligence and a regularity-based intelligence.

你可以在 PicBreeder 分数中看到这一点。它在我们找到的神经回路等解决方案中。规律性被烘焙到回路中。

And you can see this in the PicBreeder scores. It's in the solutions like the neural circuits that we find. The regularities are baked into the circuits.

我们想看看如果你使用 LLM 来进化基本上相互竞争的小型汇编程序会发生什么。所以，有一个 1980 年代发明的编程游戏，人们过去常玩。它叫做 Core War。我的程序需要做的是，它基本上需要在你的程序执行线程前注入一个无效操作。它会通过崩溃来停止你的程序。你的程序会崩溃。我们能否从中创造这种开放式的进化军备竞赛？

We wanted to see what would happen if you used LLMs to evolve basically small assembly programs that compete against each other. So, there's this programming game invented in the 1980s that people used to play. It's called Core War. What my program needs to do in order to do that is it needs to basically inject an invalid operation in front of your program's execution thread. It's going to halt your program by crashing. Your program will crash. And can we create this open-ended evolutionary arms race of programs from this?

这就像《权力的游戏》巡回版。

It's like Game of Thrones touring edition.

是的。是的，没错。这是一个很好的说法。LLM 不擅长 Redcode。如果你只是尝试零样本学习这种语言，它很糟糕。但如果你进行这种进化，并且你有这个验证器，它结果非常好。进化可能是我们所知的解决离散组合问题的最强大算法。如果你解决了一个谜题的一部分，并坚持这一点，然后搜索其他部分，你可以将指数搜索问题转化为线性搜索问题。

Yeah. Yeah, exactly. That's a good way to put it. The LLM is not good at Redcode. If you just try to zero-shot this language, it's terrible. But if you do this evolution and you have this verifier, it turns out to be really good. Evolution is probably the most powerful algorithm that we know to solve discrete combinatorial problems. If you solve even one part of a puzzle and you hang on to that and you search for the other ones, you can turn your exponential search problem into a linear search problem.

我们基本上是在寻找细胞自动机，比如神经细胞自动机，它们从自我复制的分子到漂浮在原始汤中的细胞，再到捕食者-猎物动态，再到漫游的外星动物。你可以想象，如果你用一个大粒子模拟，比如数万亿个粒子，并且你有正确的基础模型，你搜索这个事件序列，那么会发生什么？也许你最终会为这个大粒子系统找到一些物理规则，它真的给了你整个序列，你得到动物，就像我们宇宙中发生的那样，对吧？人工生命领域，我真的认为是对智能的一个非常长期的赌注。

We're basically trying to search for cellular automata like neural cellular automata which go from self-replicating molecules to cells floating in primordial soup to predator-prey dynamics to alien animals roaming around. You can imagine that if you did this with a big particle simulation with like quadrillions of particles and you have the right foundation model and you searched for this sequence of events, then what would happen? Maybe you end up with some physics rule for this big particle system, which really does give you this entire sequence and you get animals at them just like what happened in our universe, right? The field of artificial life, I really think is like a very long-term bet on intelligence.

这篇论文是《ASAL：用基础模型自动化搜索人工生命》。Akarsh 和他的合著者使用视觉语言模型在生命游戏、Lenia、Boids 等中寻找有趣的模拟。在论文早期，你引用说人工生命不仅是我们所知道的，而且是它可能的样子。你是什么意思？

The paper is ASAL: Automating the Search for Artificial Life with Foundation Models. Akarsh and his co-authors use vision language models to find interesting simulations across Game of Life, Lenia, Boids, and more. Early on in the paper you quoted artificial life is not just as we know it, but also as it can be. What do you mean by that?

所以，人工生命真正关乎的是理解，正如你所说，生命可能的样子，而不是生命现在的样子。如果我们真的只想理解地球上的生命以及它现在的样子，我们会研究生物学，对吧？这和我们不仅仅研究神经科学来理解智能的原因是一样的。我们创建 AI 是因为我们想要更机械化的理解，不仅如此，就我个人而言，我想借用杰夫·克鲁恩的一句话，我非常喜欢。就像我们真的想理解，如果我们想声称我们理解智能或理解生命，我们不能只谈论这种特定的生命形式。我们需要理解所有可能智能的空间，所有可能生命形式的空间，只有这样你才能谈论支配智能或生命的通用原则，对吧？你不能只看一辆车就说所有车都是这样的，对吧？同样，你不能只看 GPT-5 就说所有智能都是这样的。你不能只看人脑就说所有智能都是这样的。有一个事物的空间，我们需要研究整个空间。

So, artificial life is really about understanding like you said, life as it could be rather than life that is as it is. If we really just wanted to understand life on Earth and how it is right now, we'd be studying biology, right? It's the same reason that we don't just study neuroscience to understand intelligence. We create AIs because we want a more mechanistic understanding and not only that, but me personally, I like to I'm going to borrow this quote from Jeff Clune that I liked a lot. Like we really want to understand if we want to claim that we understand intelligence or we understand life, we can't just talk about this specific form of life. We need to understand the space of all possible intelligences, the space of all possible life forms that can be and only then can you talk about general principles for what governs intelligence or what governs life, right? You can't just look at one car and be like this is how all cars are, right? That's the same way you can't just look at GPT-5 and be like all intelligences are like this. You can't just look at the human brain and be like all intelligences are like this. There's a space of things and we need to study the whole space.

如果你观察人类和动物如何学习，他们有一个自然的课程。他们会说：“我要先学这个，然后这个，然后这个。”没有其他方式。但是，如果你看看我们的模型是如何训练的，你可以训练它们任何东西，它就会工作。问题是，很多人认为这没问题，也许比所有人类学习的方式更好。这是一种可能性。它可能只是更好，对吧？但是，有理由相信存在根本性的错误。宇宙不是那样运作的。宇宙不是你可以随时学习任何东西。我认为，人类按顺序学习以及进化以这种路径依赖的方式运作的事实，是它在所使用的表示中创造出如此良好的规律结构的原因。这就是它如此健壮、适应性强、可泛化的原因。我认为正是这些原因。

If you look at how humans learn and animals learn, they have a natural curriculum. They're like, "I'm going to learn this and then this and then this." There's no other way. But, if you look at how our models are trained, you can train them on anything and it'll just work. And the question is a lot of people think that that's fine and that maybe it's better than all humans learn. And that's a possibility. It could be just better, right? But, there's reasons to believe that there is something fundamentally wrong. And the universe doesn't work like that. The universe doesn't work that you can just learn anything at any given time. And that's I think the fact that humans learn in the sequential order and evolution works in this path-dependent way is the reason that it creates such nice regular structures in the representations that it uses. And that's the reason it's so robust, adaptable, generalizable. It's for those reasons, I think.

我们看到这体现在课程学习的形式中。这就是我们一步一步学习，并且以越来越复杂的方式学习。对吧。

We see this cashed out in the form of curriculum learning. Which is this idea that we learn step by step and we learn in kind of with an increasing complexity. Right.

你刚才提到，当你一步一步地学习、逐步增加复杂度时，你学到的表征会更好。但你会如何用一种直白的方式来描述这一点？直觉是什么？

And you were just pointing to when you learn step by step, increasing the complexity, the representations you learn are better. But how would you describe that in a kind of no-nonsense way? What's the intuition?

课程不是唯一重要的东西，因为人们在强化学习中一直在做课程学习。而我曾痴迷于强化学习中的课程学习。我认为关键在于，这回到了为什么伟大不能被计划。你需要一个偶然的课程，而不是预先计划好的——比如我要先做这个再做这个再做这个。它必须更加偶然，拥抱我们本性、宇宙的混沌，对吧？它必须拥抱这一点。所以你必须接受这样一个想法：不仅仅是沿着这个课程走，而是任何可能的课程都可能带我去某个有趣的地方。所以这只是提一下，你需要偶然的课程。而要回答课程为什么最终重要的问题，我认为那是我真正有兴趣回答的根本问题。我不知道简短的答案。但这就像我们宇宙的一个基本属性：如果你想捕捉、像你喜欢说的那样，沿着关节切开宇宙并理解它，对吧？就是沿着关节切开它。我认为为了做到这一点，你需要有这种复杂化的过程，你以这种顺序方式在其他规律之上建立规律。好吧？我不完全确定为什么必须这样，但我知道自然进化就是这样运作的。我知道很多理论，比如组装理论，它们就是这样运作的。就像生物化学中的进化理论等等。人类学习也是这样运作的。它以顺序方式运作。所以我不完全确定为什么这样运作，但它确实似乎是某种我们宇宙的基本属性。我们真的需要更好地理解这一点。

The curriculum is not the only thing that matters, because people have been doing curriculum learning in RL. And I was obsessed with curriculum learning in RL. I think the key thing is that this goes back to why greatness cannot be planned. And you need a serendipitous curriculum, which is not a pre-planned like I'm going to do this then this and this. It has to be much more serendipitous and embrace the chaos of our nature, of the universe, right? It has to embrace it. So you have to embrace the idea that not just like going down this curriculum, but any possible curriculum which might lead me to somewhere interesting. So that's just a shout out to like you need serendipitous curriculums. And to answer the question of why curriculums actually end up mattering, I think that's a fundamental question that I'm really interested in answering. And I don't know the short-form answer for that. But it's like a fundamental property of our universe that if you want to capture, as you like to say, carve the universe and understand it by the joints, right? That is carve it up by the joints. I think in order to do that, you need to have this process of complexification where you build regularities on top of other regularities in this sequential way. All right? And I'm not exactly sure why it has to be like that, but I know that natural evolution works like that. And I know that a lot of theories like assembly theory, they work like that. Like that's like an evolutionary theory in biochemistry and stuff. And human learning also works like that. It works in a sequential way. So I'm not exactly sure why it works like that, but it definitely seems something like it's something fundamental to our universe. And we really need to understand that better.

对吧？

Right?

如果有人能回答那个问题，为什么这很重要，或者证明它通常不是必要的，那将是一个巨大的发现，对吧？是的。

If someone can answer that question, why that is important or prove that it's not in general necessary, that would be a huge discovery, right? Yeah.

确实会。我能想到的唯一类比是斯蒂芬·沃尔夫勒姆。所以我们在这里谈论的映射到他的想法上，就是我们需要对表征采取建构主义的方法，这意味着我们一步一步地构建它们。他指出有一种现象叫做计算不可约性，基本上就是没有捷径。我想在我脑海中兑现的方式是，它说的是如果你想学习一个复杂的表征，你想获得一个复杂的表征，你实际上不能跳跃前进。你实际上需要合成这些构建块，你需要走这条路。

It would. And the only analogy I can think of is Stephen Wolfram. So what we're talking about here mapped onto his ideas is that we need to have a constructivist approach to representations, which means that we build them step-by-step. And he noted that there's this phenomenon called computational irreducibility, which is that basically there are no shortcuts. And I guess the way that cashes out in my mind is that it's saying that if you want to learn a complex representation, you want to acquire a complex representation, you can't actually jump ahead. You actually need to synthesize these building blocks and you need to take the path.

是的，我喜欢肯解释他的哲学时的一点是，每当人们，每当他说“哦，我们需要新颖性搜索。我们需要这个东西。”而人们说“哦，但那就像在黑暗中搜索。那就像随机。”它实际上绝不是随机的，因为随机就像只是做随机动作。新颖性，你知道你之前做了什么，你想避免它。所以我想你说的是，如果你想构建，如果你想尊重你之前来自的路径，无论是通过新颖性避免它，还是因为你尊重你在表征中沿途学到的对称性，你确实需要尊重它，这给了你一个想法，它约束了你能去哪里，那个问题就像它约束了你多少？很有可能它很大程度上约束了你，并且它以正确的方式约束你走向正确的路径，对吧？这就是为什么每当我思考 FER 假设和这一系列工作时，我真的很喜欢通过正则化的视角来看待它。就像人们谈论深度学习中的正则化，比如权重衰减或其他某种正则化机制，我喜欢那个。我认为一切都是正则化。数据是一种正则化形式。它正则化了你需要建模世界的方式？那是一种形式，就像基本上以你需要建模世界的方式塑造粘土。L2 正则化如权重衰减是另一种形式。基本上它是说你想找到低权重范数的解，对吧？然后我认为这种课程学习基本上你想尊重对称性只是另一种正则化形式，但我认为这是你让这个模型接受的正则化，也许这是我们真正需要的正确正则化。

Yeah, and one of the things I like whenever Ken explains his philosophy is that whenever people, whenever he says like, "Oh, we need novelty search. We need this thing." And people say, "Oh, but that's just like searching in the dark. That's like random." It's actually anything but random because randomness is like just doing random actions. Novelty, you know where what you did before and you want to avoid that. So I guess what you're saying is that if you want to build, if you want to respect the path of where you previously came from, whether it's like by avoiding it via novelty or because you're respecting the symmetries that you learned along the way in a representation, you do need to respect it and that gives you an idea that it constrains where you can go and that question is like how much does that constrain you? There's a good chance it constrains you a lot and it constrains you towards in a right way towards the right path, right? That's why whenever I think about the FER hypothesis and this line of work, I really like looking at it through the lens of regularization. Like people talk about regularization in deep learning with like weight decay or some sort of other regularization mechanism and I love that. I think everything is just regularization. The data is a form of regularization. It regularizes what like do you need to model the world? That's a form of like it's like basically molding the clay in a way that you need to model the world. L2 regularization like weight decay is another form of that. Basically it's saying you want to find solutions with low weight norm, right? And then I think that this curriculum learning basically you want to respect the symmetries is just another form of regularization, but I think it's regularization that you're subjecting this model to and maybe it's the right regularization that we really really need.

这里快速回顾一下。我们去年发布的《断裂与纠缠的表征》纪录片解释了为什么在神经网络表征方面，路径比目的地更重要。这是肯尼斯·斯坦利。这种分层锁定告诉我们一些关于表征如何出现的重要事情。我们现在认为这是关于找到正确的构建块，但奇怪的是，这是关于让未来的发现更。

And here's a quick refresher. Our Fractured and Entangled Representations documentary that we published last year explained why the path is more important than the destination when it comes to representations in neural networks. This is Kenneth Stanley. This hierarchical locking tells us something really important about how representations emerge. We think it's about finding the right building blocks now, but weirdly it's about making future discoveries more.

重要的不仅仅是你到达了哪里，而是你如何到达那里。你知道，这是现在缺失的东西，因为我们倾向于只关心你到达了哪里。就像我们看领域中的基准分数，那就像是结果。它做得非常好。它刚刚通过了数学奥林匹克。但如果重要的是你如何到达那里呢？

It matters not just where you get, but how you got there. You know, and that's like something missing right now, you know, cuz we tend to just care where you get. Like we look at the benchmark score you know, in the field and it's like that's the result. It's doing really well. It just passed the math Olympiad. But what if it matters how you got there?

但我认为那种结构化的进化，与用统计方式做是完全不同的范式。

But that kind of structured evolution, I think, is a completely different paradigm to doing it the statistical way.

确切，确切。我认为你说对了词。这就像一种理解智能的范式。但这个 FER 想法和这个 pick breeder 系列工作，基本上就像我们试图研究的另一种智能范式。我们不是从统计学的角度研究它，我们是从规律的角度研究它，对吧？所以这是统计智能和基于规律的智能之间的区别。我真的很喜欢统计智能的想法。我认为我们在这条路上做得很好。我们应该继续走这条路，但我们也应该不要把所有的鸡蛋放在一个篮子里，真正尝试理解另一个东西，因为有很多证据表明另一个东西在这里有很多潜力，对吧？它真的与所有对深度学习的批评相关。它们基本上是在谈论规律。就像每当人们谈论我们需要符号处理，每当我们需要几何深度学习，我们需要更多的归纳偏置。人们有所有这些不同的领域。

Exactly, exactly. And I think you said the right word. It's like a paradigm to understand intelligence. But this FER idea and this pick breeder line of work, it's basically like a different paradigm of intelligence that we're trying to study. We're not studying it from the standpoint of statistics, we're studying it from the standpoint of regularities, right? So it's a difference between like a statistical intelligence and a regularity-based intelligence. And I really like the idea of statistical intelligence. I think we did a good job like going down this path. We should still go down this path, but we should also not put all our eggs in one basket and really try to understand this other thing because there's lots of evidence that this other thing has lots of potential here, right? And it really ties into what all the criticisms of deep learning. They're basically talking about regularities. Like whenever people talk about like we need symbolic processing, whenever we the way we need geometric deep learning, we need more inductive biases. And people have like all these different fields.

关于我们需要什么，有完全不同的领域，对吧？它们都是向系统添加规律性和对称性的形式。我非常喜欢这种 FER 方法的地方在于——我喜欢把它看作一种自动对称学习算法，对吧？因为我喜欢举这个例子：人们喜欢 CNN 是因为它们具有平移不变性，对吧？人们喜欢 Transformer 是因为它们具有置换不变性，并且编码了某些对称性。这很棒。我的意思是，我们需要置换不变性来进行集合操作，对吧？但你能把光照不变性编码到你的模型中吗？比如，如果你希望你的 CNN 或图像模型对白天还是夜晚具有不变性。世界上可能没有方程能让你做到这一点。也许未来会有，但我无法想象你能提出一个数学归纳偏置来保证这一点，对吧？所以你真正需要的是某种数据驱动的对称学习机制，它能自动尊重这种不变性，并形成天然内置了那种归纳偏置的神经回路，对吧？你可以在 pick breeder 头骨中看到这一点。并不是说归纳偏置在我们搜索的假设空间中，而是在我们找到的解——神经回路中。规律性被烘焙进了回路里。

There are whole different fields on what we need, right? And they're all forms of adding regularities and symmetries to the system. What I really like about this FER approach is—I like to think about it as an automatic symmetry learning algorithm, right? Because I love to give this example: people like CNNs because they're translation invariant, right? People like transformers because they're permutation invariant and they encode certain symmetries. And that's great. I mean, we need permutation invariance for set operations, right? But can you code in lighting invariance into your model? Like if you want your CNN or image model to be invariant to whether it's nighttime or daytime. There's no equation in the world that will probably get you there. Maybe there is in the future, but it's impossible for me to imagine that you could come up with a mathematical inductive bias that's going to guarantee that, right? So what you really need is some sort of data-driven symmetry learning thing that automatically respects it, and it forms neural circuits which naturally have that inductive bias built in, right? And you can see this in the pick breeder skulls. It's not that the inductive bias is in the hypothesis space of things we're searching for. It's in the solutions—the neural circuits that we find. The regularities are baked into the circuits.

嗯，我想另一种看待方式是用人类智能。我们的知识获取过程是扎根于世界的。当我们构建这些人工生命模拟时，它们并不是以同样的方式扎根于世界，对吧？所以它们只是——但它们确实具有我们刚才谈论的这些属性。所以它们实际上在构建这些表征，而且当你观察这些事物如何演化的现象时，会发生有趣的事情。你会得到某些东西，它们仅仅是因为你正在进行这种迭代计算而自然产生，即使是以一种无方向的方式。

Well, I suppose another way to look at it is with human intelligence. Our knowledge acquisition process is grounded in the world. When we build these artificial life simulations, they're not grounded in the same way, right? So they just—but they do have these properties that we're just talking about. So they're actually building these representations and interesting things happen when you look at the phenomena of how these things evolve. And you get certain things that just fall out of the fact that you're doing this iterative computation even in an undirected way.

是的。所以我想谈谈你提到的一个关键点：你说它不是扎根于我们的世界，对吧？这是我最喜欢人工生命的地方之一——你可以把它扎根于任何你想要的人工世界中，对吧？所以人们常常认为人工生命只谈论生命。我喜欢它的原因是人工生命中有一个子领域叫做人工化学。它只是改变——不是使用正常的氢、氦，你可以改变它们相互作用的方式，看看会发生什么，对吧？所以这就是人工生命的力量。你可以在其他宇宙中进行反事实分析，对吧？我真的很喜欢这个想法，因为人工生命不仅关乎生命，还关乎研究人工物理、人工化学中一般会发生什么。我们能否研究每个不同宇宙中的涌现属性？在哪个宇宙中——这些是我真正想围绕其展开研究议程的问题：在哪些世界中我们会得到这种开放式的复杂性和涌现的创造？在哪些世界中我们得不到？为什么——这是关键问题——为什么这个参数，为什么这个物理，当你运行它时，它会产生这种惊人的现象，而当你运行这个物理时，它不会那样做，它只产生随机噪声之类的，对吧？这就是我实际从事人工生命自动搜索——这项 ASEL 工作——的关键原因，因为我想直接回答那个确切的问题。但 ASEL 试图基本上参数化一个智能空间，找到好的，找到坏的，然后做一些简单的事情，比如在它们之间插值，看看会发生什么，对吧？在它们之间插值参数空间，对每个中间点运行模拟，看看会发生什么——就像简单的科学分析那样。

Yeah. So I want to touch on one key point that you said: you said it's not grounded in our world, right? And that's one of my favorite things about artificial life—that you can ground it however you want to in whatever artificial world you want to, right? So people often think about artificial life as only talking about life. The reason I like it is because there's a subfield in artificial life called artificial chemistry. It's about just changing—instead of using the normal hydrogen, helium, you can change how they interact and see what happens, right? So that's the power of artificial life. You can do counterfactual analysis in other universes, right? And I really like this idea because artificial life is not only about life, but it's also about studying what happens in general in artificial physics, artificial chemistries. Can we study the emergent properties in each of these different universes? In what universe—these are the questions that I really want to focus my research agenda around: in what worlds do we get this open-ended creation of complexity and emergence? And in what worlds do we not get that? And why is it—and this is the key question—why is it that this parameter, why does this physics, when you run it, it produces this amazing phenomenon, and when you run this physics it doesn't do that, it produces just random noise or something, right? And that was the key reason I actually worked on automated search for artificial life—this ASEL work—because I want to answer that exact question directly. But ASEL was trying to basically parameterize a space of intelligences, find the good ones, find the bad ones, and just do simple stuff like interpolate between them and just see what happens, right? And interpolate the parameter space between them and run the simulation for each intermediate point and just see what happens—just simple scientific analysis like that.

是的，另一件让我印象深刻的事情是，你在规则空间中得到了这些收敛的模式。所以即使规则——我们使用“涌现”这个术语，也许我们也应该谈谈它——我们不一定是指 David Krakauer 意义上的强涌现，但你知道，弱涌现就是当你顺序地应用这些规则很多很多次时，你会看到这些非常令人惊讶、不可预测的现象，而这些现象具有似乎在规则空间中共有的特征，比如某些类型的滑翔机、物体和模式。当然，Wolfram 认为宇宙本身是计算性的，但这几乎就像我们看到了影子。

Yeah, and another thing that jumps out to me is that you get these convergent patterns in rule space. So even though the rules—and we use this term emergence, and maybe we should talk about that as well—which is, we don't necessarily mean emergence in the strong David Krakauer sense, but you know, weak emergence is just that when you sequentially apply these rules many, many times, you see these very surprising, unpredictable phenomena, and these phenomena have characteristics which seem to be shared in rule space, like certain types of glider and object and pattern. And of course, Wolfram believes that the universe itself is computational, but it's almost like we're seeing shadows.

你这么说是什么意思？就像，你看到了——我猜——宇宙运行所基于的基质的影子？

What do you mean by that? Like, you're seeing shadows of—I guess—the substrate that the universe is running on?

嗯，我的意思是这只是一个假设，所以就像我们在规则空间的不同部分看到这些收敛的模式，现在怀疑的人可能会说，好吧，那些只是——它们就像设计偏差，它们只是你进行这种计算的方式的特征，它们是生命游戏结构的特征，也许是因为你使用网格，或者你使用某种类型的程序解释器来做这件事，那些东西影响了这些模式。但它们也可能只是其他东西的暗示，你知道，它们可能是影子。

Well, I mean it's only a hypothesis, so like we see these convergent patterns in different parts of rule space, and now the skeptical person might just say, okay, well those are just—they're like design bias, they are just features of the way that you are doing this computation, they are features of the structure of the Game of Life maybe because you're using a grid or you're using a certain type of program interpreter to do it, that those things are influencing these patterns. But they might also be hints of just something else, you know, they could be shadows.

我有点同意你的看法，因为在我的 ASEL 工作中，我跨越了所有可能的二维元胞自动机的空间，对吧？在我的案例中，我有很多多样性。在一个世界里，滑翔机看起来像康威生命游戏的滑翔机。在另一个世界里，最小的滑翔机看起来像沿着轨道行驶的火车，对吧？所以它留下了一条长长的路径。所以在不同的世界里，它可以是不同的，但我猜你所说的是，当它们在所有不同的世界中都相同时，那就指向了某种根本性的东西——不仅是你所在的世界，不仅是你操作的规则，而是你工作的整个基质。也许那个基质中所有可能的规则都有这种特性。

I kind of agree with you because in my ASEL work, I spanned the space of all possible 2D cellular automata, for example, right? And in my case, I had a lot of diversity. In one world, a glider looked like a Conway's Game of Life glider. In another world, the minimal glider looked like a train going down the tracks, right? So it left like a long path. So in different worlds, it can be different, but I guess what you're talking about is that when they're all the same in all the different worlds, that points to something fundamental about—not only the world that you're in, not only the rule that you're operating on, but the entire substrate that you're working in. Maybe all possible rules in that substrate have this type of thing.

也许我们应该稍微谈谈其中一些模拟器，以便观众理解我们在说什么。为了让它更生动，你能过一遍其中一些模拟器并解释它们如何工作吗？

Maybe we should just talk a little bit about some of these simulators just so that folks in the audience understand what we're talking about. Just to bring this to life, can you go through some of these simulators and just explain how they work?

是的，我可以过一遍一些模拟器。所以我可以从线性的那个开始。我将从最简单的开始，康威生命游戏。康威生命游戏是 John Conway 的一个超级著名的模拟，你有一个细胞网格，一个二维细胞网格，它们可以是开或关，活或死。就这样。

Yeah, I can go through some of the simulators. So I can start off with the linear one. I'll start with the simplest one, Conway's Game of Life. So Conway's Game of Life is a super famous simulation by John Conway, where you have this grid of cells, this 2D grid of cells, and they can either be on or off, alive or dead. That's it.

基本上，它有着超级简单的规则，比如如果我死了，而我的八个邻居中有三个是活的，那么我就会复活，对吧？就像复活一样。这样的规则大概有两三条。规则超级简单，就这些。你只需要知道这些。然后你从一个随机死或活的细胞网格开始，或者从人类创造的某种模式开始，然后你运行这个模拟很多很多时间步。从这个超级简单的世界——可能是你能想到的最简单的东西——你运行它，就会得到疯狂数量的复杂性，直到你实际运行它之前，没人会想到它能做到。现在人们就像——有整个 Discord 服务器。我实际上就在其中一个里，人们试图浏览并分享他们发现的东西。他们会说，‘哦，我找到了这个很酷的飞船，也是一个每 64 个时间步重复一次、以这个速度移动的重复模式。’我真的很喜欢这个社区，因为他们称呼——比如滑翔机，它移动得多快。我想它是以某个速度移动，对吧？他们意识到系统中有一个因果速度，就像每个时间步一个细胞。那就是信息传播的速度。所以他们称之为光速 C。然后他们用他们的滑翔机说，‘这个以 C/16 的速度移动。’对吧？就像在真实物理中人们谈论这类东西一样。所以它真的像一个人造宇宙。很多人似乎认为它像个玩具系统。但我想它的一个要点是，它足够玩具化，我们可以研究它。如果它像我们的真实世界一样超级复杂，那我们就无法研究它了，对吧？但我确实认为这是一个非常有趣的探索。这是我的探索之一：创造更引人入胜的宇宙，这些宇宙能捕捉——像康威的生命游戏捕捉了我们宇宙的一些属性，比如涌现，以及你可以从局部相互作用构建全局模式的想法。但还有其他东西我们也想捕捉，对吧？比如自组织和开放式进化的想法。所以我的很多研究都是试图找到具有我们关心的所有那些属性的模拟。这就是 A Cell 的要点，是的。

And basically, it has super simple rules like if I'm dead and three of my neighbors—I have eight neighbors—if three of them are alive and I'm dead, then I become alive, right? As an act of just coming to life. And there are like two or three rules like that. It's super simple rules, and that's it. That's all you need to know. Then you start off with a grid of random dead or alive cells, or you can start off with some pattern that humans have created, and you just run this simulation over many, many time steps. And from this super, super simple world—it's like possibly the simplest possible thing you can think of—you run this, you get insane amounts of complexity that no one would have thought it could do until you actually run it. And now people are like—there are whole Discord servers. I'm actually in one of them where people try to go through and they just share what they found. They're like, 'Oh, I found this cool spaceship that also—that's a repeating pattern that repeats every 64 time steps and it moves at this speed.' And I really love this community because they call—like the glider, how fast it moves. I think it's at a certain speed, right? And they realized that there's like a speed of causality in the system, which is like one cell per time step. That's like the speed at which information can propagate. So they call that the speed of light C. And then they use their gliders and they're like, 'This goes at C over 16 speed.' Right? Just like in real physics how people talk about this kind of stuff. So it really is like an artificial universe. And a lot of people seem to think it's like a toy system. But I guess a point of it is that it's toy enough that we can study it. If it were super complex like our real world, then we wouldn't be able to study it, right? And I do think it's a very interesting quest. That's one of my quests: to create more compelling universes which capture—like Conway's Game of Life captures some properties of our universe like emergence and this idea that you can build up global patterns from local interactions. But there's other stuff we want to capture, too, right? Like the idea of self-organization and open-ended evolution. So a lot of my research is on trying to find simulations which have all those properties that we care about. And that's the point of A Cell again, yeah.

那么，这个空间的结构化程度如何？比如，我们很多人见过神经网络表示空间的可视化，比如用 t-SNE 之类的。当我们看到岛屿和集群等等。对于康威的生命游戏来说，也是一样的吗？例如，如果你改变一个相邻的细胞，它会——你知道，是像虫洞一样把你传送到空间的完全不同的部分，还是在涌现的高层领域做出一些相应的小变化？

So then how structured is the space? So, um you know, like we think about—many of us have seen visualizations of representation spaces of neural networks like using t-SNE and something like that. And when we see islands and clusters and whatnot. Is it the same for something like Conway's Game of Life? So, for example, if you change one neighboring cell, does it trans—you know, is it like a wormhole where you get transported into a completely different part of the space or does it make some commensurate small change in the emergent like high-level domain?

是的，没错。这是个很好的问题，这也是我真正想研究的事情之一。所以，在生命游戏的初始状态中，是的，它非常混乱。在康威的生命游戏中，让我们具体谈谈，它非常混乱。我有过这样的例子：你改变一个细胞，比如在 64x64 网格中改变一个像素，之前——我实际上在的一次演讲中用过这个——它基本上永远运行。就像——不是永远，我不该说永远，但像数万步，它只是继续移动，对吧？它只是继续做有趣的事情。你改变这一件事，它就停止——基本上在不到 100 步内停止成振荡模式。所以，在初始状态中，是的，它非常混乱。但有趣的是，这不一定适用于所有模拟，而在 ASEL 中，我们真正感兴趣的是搜索物理——就像支配动力系统的实际规则，我称之为物理。我们一直在谈论的是初始状态，对吧？但有两件事：初始状态和物理。如果你在物理中做，生命游戏似乎更稳健一些——比如如果你改变一个——基本上有 18 条可能的规则可以打开或关闭，如果你改变其中一条，根据我的经验，它不会影响太大，但也有一些例外。

Yeah, exactly. That's a great question, and this is one of the things I really want to study. So, in the initial state of Game of Life, yes, it's very chaotic. In Conway's Game of Life, let's talk about that specifically, it's very chaotic. I've had examples where you change one cell like one pixel in like the 64 by 64 grid, and beforehand—I actually use this in one of my talks—it basically runs forever. It's just like—not forever, I shouldn't say forever, but like for tens of thousands of steps, it just keeps on moving, right? And it just keeps on doing interesting stuff. You change this one thing, it stops—it basically halts into an oscillating pattern in like less than 100 steps. So, in the initial state, yeah, it's very chaotic. But, the interesting thing is that this is not necessarily true for all simulations, and in ASEL, we're really interested in searching over the physics—like the actual rules that dictate the dynamical system, what I call the physics. What we've been talking about is the initial state, right? But there's two things: initial state and physics. And if you do it in the physics, Game of Life seems to be a little more robust—as in if you change like one—there's basically 18 possible rules you can toggle on or off, and if you change one of them, it doesn't impact it too much in my experience, but there are some exceptions to that.

非常有趣。那么，也谈谈其他一些模拟器吧。

Very interesting. So, um and talk about some of the other simulators as well.

所以，其中一个——我可以过一遍我们论文中使用的列表。在 Bert Chan 的 Lenia 中，他基本上创造了生命游戏的连续泛化，它泛化到三个通道，你可以将它们对应到 RGB。它们基本上就像一个连续微分方程更新规则，支配它们如何运作。所以，我认为那个真的很酷。

So, one of the—I can go through a list of the ones we use in our paper. In Lenia by Bert Chan, and he basically created this continuous generalization of Game of Life which generalizes to like three channels, and you can dictate them to RGB. And they're basically just like a continuous differential equation update rule that dictates how they operate. So, I think that one's really cool.

这是 Lenia 的创造者 Bert Chan，讲述当他让规则连续时发生了什么。

Here is Bert Chan, the creator of Lenia, on what happens when he made the rules continuous.

所以，通过这种让一切连续的做法，会发生什么？嗯，当我把这个系统放进电脑时，一些诡异的事情发生了。一个模式在我的屏幕上蔓延。

So, by doing this kind of making everything continuous, what will happen? Well, when I put this system in my computer, something creepy happened. A pattern that creeping across my screen.

然后另一个很酷的是神经细胞自动机。

And then the cool—also another cool one is the neural cellular automata.

是的，由 Mordvintsev 提出。

Yeah, by Mordvintsev.

是的，你打算采访他吗？

Yeah, are you going to interview him?

呃，希望如此。他在 2020 年发表了这篇著名的论文，叫做生长神经细胞自动机。天哪，我们在 MLST 上谈过这个无数次了。所以那个壁虎的东西。

Uh hopefully. And he famously in 2020 released this paper called growing neural cellular automata. And my god, we talked about this on MLST millions of times. So the thing of the gecko.

是的。是的。

Yeah. Yeah.

它展示了这种涌现式优化和形态发生工程，你知道，本质上你训练一个 CNN，然后你只是——我的意思是，你可以把这有点看作一个卷积算子。所以你学习 CNN 的随机梯度下降参数。嗯，我猜它是被训练——它像是被涌现式地训练来重现一个图像。比如一只壁虎。

And it was showing this emergentist optimization and morphogenetic engineering, you know, essentially where you train a CNN and then you just—I mean because you can kind of think of this a bit like a convolution operator. So you learn the parameters of stochastic gradient descent of the CNN. Um I guess it was trained—it was like emergentistly trained to reproduce an image. So like a gecko.

是的。

Yeah.

是的，你训练局部规则，使它们重新组织以自组织，从而形成蜥蜴。是的。

Yeah, you're training the local rules such that they reorganize in order to self-organize in order to form the lizard. Yeah.

是的，然后你可以删除蜥蜴的部分。然后因为这东西在连续运行，对吧？所以只是卷积、卷积、卷积。它重建了蜥蜴。所以，再次，这就像我们在 AI 系统中想要的，你知道，我们不想要冒牌货。我们不想学习蜥蜴。我们想学习从构建块重建蜥蜴的能力。你知道，路径依赖的东西。所以它一次又一次地出现。就像那太重要了，你知道，而不是仅仅学习蜥蜴看起来像什么。

Yeah, and then you can delete parts of the lizard. And then because this thing's running continuously, right? So it's just convolution, convolution, convolution. And it reconstructs the lizard. So and again, this is like what we want in AI systems is, you know, we don't want imposters. We don't want to learn the lizard. We want to learn the ability to reconstruct the lizard from the building blocks. You know, the path dependency thing. So it's coming up again and again. Like that's just so important, you know, rather than just learning what the lizard looks like.

对，对，对。在此基础上，我想说的是，你其实并不想学习智能本身，而是想学习孕育智能的过程，对吧？

Yeah. Yeah. Yeah. And to build on top of that, I guess it's like you don't really want to learn the intelligence, you want to learn the process which grows the intelligence, sort of, right?

是的，类似。我的意思是，这也是深度学习最初兴起的原因之一，因为我们没有硬编码所有关于世界的知识，而是让它从数据本身学习，对吧？但也许还有其他问题。不过，回到神经细胞自动机，为了观众，基本上就像你有一个细胞自动机，但如果你把动力学规则改成从 3x3 的斑块到一个细胞的映射，对吧？这就是动力学规则告诉你的：如何从一个时间步到下一个时间步？基本上他们所做的就是用一个小神经网络来替代这个规则。这样你就有了一个可以搜索的可能物理规则的空间。嗯，所以我真的很喜欢神经细胞自动机。嗯，我可以简要介绍一下接下来的两个：boids 和粒子生命。是的，boids 是一个超级开创性的算法，灵感来自自然界中鱼群和鸟群的聚集。因为如果你看——我最近去了大阪水族馆，我在看这些视频，不是这些视频。我亲眼见过这种鱼。它们总是待在一起，后面有一条鲸鲨出现，它们就稍微向左转了一点。我就想，“它们怎么知道要这样做？”因为每条鱼——只有一条鱼知道鲸鲨的存在，它把其他鱼向左拉，其他鱼都意识到，“哦，我们应该都去那边。”这是自然界中集体智能的一个例子，对吧？所以 boids 基本上就是受此启发，你有一些像鸟一样的物体在 2D 或 3D 空间中流动，它们有非常简单的规则，它们只看邻居——比如我左边和右边有邻居，我基本上说我能看到四个邻居，好吧？我就根据它们的位置决定向左还是向右。就这样。如果你做出正确的决定，并且你们都做出相同的决定，你就能得到我们在自然界中看到的涌现的聚集行为。是的，我们所做的就是用一个小的神经网络来参数化这个，并搜索这个决策网络——如果 boids 做出不同的决策规则，集体的全局行为如何根据它们所做的局部决策规则而变化？粒子生命超级酷，因为在基于网格的模拟中，感觉像是这个空间的东西，但粒子生命基本上是说我们将有 3D 欧几里得空间或 2D 欧几里得空间，我们只是把粒子扔进去，就像普通粒子一样，但它们的物理规则不会像牛顿力学那样。它只是我们想要的任何东西，对吧？所以，在这个例子中，在粒子生命中，我受到了 Thomas More 的启发。他在 YouTube 上有一些关于这类东西的很棒的视频。基本上你有一组粒子，每个粒子可以采取六种类型之一，对吧？这些由它们的颜色决定。然后你可以有一个 6x6 的矩阵，告诉你不同类型的粒子如何相互作用。所以，可能是红色和红色排斥，但红色和橙色吸引。橙色和绿色是中性，对吧？你有一个 6x6 的矩阵，如果你改变这个物理矩阵的值——这又是世界的物理，对吧？或者你可以叫它元素周期表或随便什么。当你改变这个物理，模拟如何变化，对吧？有些非常平淡无聊，我不想看，但有些在演化、移动并做有趣的事情，有些实际上最终创造出看起来像细胞的小图案在移动。所以你可以 literally 看到它们。它们有像细胞屏障，像这层红色涂层，内部有紫色和黄色呈三角形，那是在内部，基本上这个物体——这个类细胞物体——正在利用物理在这个世界中创造一个持续存在的物体，对吧？而你刚刚采访了 Blaze。所以，生命只是持续存在的想法，对吧？就是生命基本上是关于持续存在的想法。这些生物只是试图弄清楚我们如何持续到下一个时间框架，它们利用了物理定律来做到这一点。

Yes. It's similar to that. I mean, that's one of the reasons why deep learning took off in the first place, because we're not hard coding all this knowledge about the world. We're letting it learn from the data itself, right? But there are other problems with that, maybe. But yeah, going back to neural cellular automata, just for the audience, it's basically just like you have the cellular automata, but what if you made it so that the dynamics rule was like this mapping from a 3 by 3 patch into one cell, right? That's what the dynamics rule tells you: how do you go from one time step to the next? And basically what they do there is they just replace this with a small neural network that dictates this. So that gives you a whole space of possible physics rules that you can search over. Um, so I really like neural cellular automata. Um, and I can briefly go over the next two, the boids and the particle life. Yeah, boids is like the super seminal algorithm inspired by how fish and how birds flock in nature. Because if you look at—I actually just went to the Osaka Aquarium and I was watching these videos, not these videos. I've seen this fish in person. And they were like always together and there's like a whale shark coming out behind them and they just like steered to their left a little bit. And I was like, "How do they know how to do that?" Because each fish—only one of them knew about the whale shark and it tugged the other ones to the left and the other ones all kind of just realized, "Oh, we should all go there." It's an example of collective intelligence in nature, right? So boids is basically an inspiration of that where you have these bird-like objects which flow through 2D or 3D space and they have super simple rules where they just look at the neighbors—like I have my neighbor to the left and to the right and I basically just say I have like four neighbors that I can see, all right? I'm just going to make a decision on whether to go left or right based on where their positions are. That's it. And if you make the right decision and you're all making the same decision, you can get the emergent flocking behavior that we see in nature. And yeah, so what we did is we just like parameterized this with a small neural network and just searched over this decision-making network—if the boids make different decision rules, how does the global behavior of the collective change according to the local decision rules that they're making? Particle life is super cool because in grid-based simulations it feels like this spatial thing, but particle life is basically saying that we're going to have this 3D Euclidean space or 2D Euclidean space, and we're just going to throw particles at it, just like vanilla particles, but their physics rule is not going to be like Newtonian mechanics or something like that. It's just going to be whatever we want it to be, right? So, in this case, in particle life, I was inspired by Thomas More. He has these great YouTube videos on this type of thing. It's basically you have this set of particles, and each particle can take on one of six types, right? And these are dictated by their colors. And then what you can have is that you have this 6 by 6 matrix, which tells you how different types of particles interact with each other. So, it could be that red and red repel, but red and orange attract. And orange and green, they are neutral, right? And you have this 6 by 6 matrix, and if you change the values of this physics matrix—this is the physics of the world again, right? Or you can call it a periodic table or whatever you want to call it. As you change this physics, how does the simulation change, right? Some of them are super bland and boring and I don't want to look at them, but some of them are like evolving and moving and doing interesting things, and some of them actually end up creating little patterns which move around that look like cells. So, you can literally see them. They have like a cell barrier, which is like this red coating, and inside they'll have like purple and yellow in a triangle, and that's like inside, and basically this object—this cell-like object—is exploiting the physics in a way to create a persisting object in this world, right? And you just interviewed Blaze. So, this idea that life is just the persistence, right? It's just that idea that life is basically about persistence. These organisms are just trying to figure out how do we persist into the next time frame, and they've exploited the laws of physics in order to do so.

是的，Blaze 说这种持续存在有点像一种自然的——你知道，我在谈论这些自然收敛现象。这可能是一个例子，因为即使这些系统基本上是无根基的，它运行这个计算模拟。我们看到持续存在的涌现，事物只是想要随着时间的推移保持它们的形态，这难道不令人着迷吗？

Yeah, and Blaze was saying that this persistence is kind of like a natural—is what you know, I was talking about these natural convergent phenomena. That might be an example of one because even though these systems are basically ungrounded which it which is running this computational simulation. And isn't it fascinating that we see the emergence of like persistence that things just want to kind of keep their form over time.

是的，持续存在的涌现非常迷人。它并非在所有模拟中都会发生，只在某些中发生。但我认为这指向了关于生命是什么以及热力学第二定律告诉我们什么的更广泛的观点。有一个对持续存在的推广，就是——我喜欢这个，至少在我的心智模型中，我思考的方式是，热力学第二定律、生命和持续存在，它们基本上都在说——还有进化，我应该说是——它们都在说同一件事。它们在谈论现在存在什么以及未来可能存在什么。所以热力学第二定律说，你更可能走向更高熵的状态，对吧？就像如果你看任何模拟的宏观状态和微观状态，它就是这样运作的。但如果你从进化的角度看，未来更可能存在什么？是自我复制者，因为它们在未来自然会更频繁地存在，而非自我复制者。所以它们真的是——人们把它说成是哦生命就像对抗熵。它们实际上是同一枚硬币的两面，关于持续存在或未来可能存在什么。

Yeah, the emergence of persistence is very fascinating. It doesn't happen in all simulations. It only happens in some. But I think this points to the broader point about what life is and what the second law of thermodynamics tells us. There's a generalization to persistence which is this—I like this at least how I have in my mental model of how I think about it is the second law of thermodynamics and life and persistence they all basically are saying—and evolution I should say—they're all saying the same thing. They're talking about what exists right now and what is likely to exist in the future. So the second law of thermodynamics is saying that you're more likely to go towards higher entropy states, right? That's just like if you look at the macro states and micro states of any simulation, that's how it operates. But if you look at it from the standpoint of evolution, what is more likely to exist in the future? It's self-replicators because they'll naturally exist more often in the future versus non-self-replicators. So they're really—people make it out to be like oh life is like fighting entropy. They're really two sides of like the same idea of persistence or what is likely to exist in the future.

是的，迷人。所以在你的论文中，你基本上是在使用基础模型作为评论家。对，所以你是说我想有一种原则性的方式来搜索这个空间。

Yeah, fascinating. So in your paper essentially you are using foundation models as a critic. Right, so you're saying I want to have a principled way for searching this space.

是的，我们以三种不同的方式使用了它。

Yeah, we used it in three different ways.

所以从高层次来说，在那之前，我认为人工生命领域是一个非常酷的领域，但人们通常只模拟单个模拟，并试图从中得出结论。我觉得这很酷，但你确实需要搜索模拟空间，对吧？所以我认为我们的算法基本上是这样做的：不是硬编码单个模拟，而是参数化一个模拟空间，这要容易得多。这就像编写元素周期表——每个元素如何与每个元素相互作用——与只是说有 100 个元素，我不知道它们如何相互作用，你自己想办法。告诉算法自动搜索。所以只说有 100 个元素是一个容易得多的问题，对吧？所以一旦你描述了这个模拟空间，你就可以使用基础模型以自动方式搜索你想要的东西。

So at a high level, before that, I guess the field of artificial life is a very cool field, but often people only simulate single simulations and try to draw conclusions from them. I feel like it's cool, but you really need to be searching the space of simulations, right? So I think our algorithm basically does this: instead of hardcoding a single simulation, you parameterize a space of simulations, which is much, much easier. It's the difference between coding up the periodic table—how each element interacts with each element—versus just saying there are going to be 100 elements, and I don't know how they interact, you figure it out. Telling the algorithm to search for it automatically. So it's a much easier problem to just say there are 100 elements, right? So once you have the space of simulations described, then you can just search for what you want in an automatic way using a foundation model.

我喜欢 AISO 的原因是它捕捉到了计算不可约性的原理。你不需要知道这个物理规则会做什么，并试图预测哪一个是最好的。只需模拟它，看看会发生什么，将其插入基础模型，看看基础模型对发生的事情有什么看法，然后返回。关键的是，在这里，你让模拟展开，而不是试图预测会发生什么，对吧？所以你是在拥抱混沌和计算不可约性的理念。

And the reason I like AISO is because it captures this principle of computational irreducibility. You don't need to know what this physics rule is going to do and try to predict which one is the best one. Just simulate it, see what happens, plug it into a foundation model, and see what the foundation model thinks about what happened, and then go back. And the key thing is that here, you're letting the simulation unroll, and you're not trying to predict what's going to happen, right? So you're kind of embracing this idea of chaos and computational irreducibility.

是的，从某种意义上说，这是超越了 Moravec 的神经细胞自动机想法的一步，因为他做的是这种涌现测试优化，你说：‘好吧，当宏观事物出现时，我希望它看起来像这个特定的事物。’而你现在插入一个基础模型，你说：‘我想问基础模型它是否类似于我感兴趣的东西。’

Yeah, and in a sense, it's a step past Moravec's idea with the neural cellular automata, because he was doing this emergent test optimization, where you're saying, 'Okay, when the macroscopic thing emerges, I want it to look like this specific thing.' And you're now plugging in a foundation model, and you're saying, 'I want to ask the foundation model whether it resembles something I'm interested in.'

是的，是的，没错。有两个原因使它更好。一是更容易指定你想要什么。你可以直接说：‘我想要一只猫。’我不需要有一张猫的图片并以像素级的方式监督它。我可以只是笼统地说我想要一只猫，对吧？这是一种方式。但重要的是，它也使得优化和搜索容易得多，因为找到一个自组织规则来创建这个特定猫的特定图像真的非常难。如果你在康威的生命游戏中这样做，那不会发生。但如果你只是说：‘哦，我不想要那个特定的猫。我只想要一只猫。给我任何猫。’那是一个更容易满足问题的空间。所以搜索起来容易得多。我们亲身经历了这一点，在开放式模拟中，我们不是在搜索我们想要的特定状态配置。我们只是说我们想要一些开放式模拟。它可以以任何方式开放，只要基础模型认为它是开放的，它就会工作，它就会做。酷的事情是——我想说我们使用基础模型的原因不仅仅是因为它是最新最伟大的东西，而是因为我们真正关心人类对这些模拟的看法。因为很难描述什么是复杂性，什么是涌现，拥有一个非常有趣的系统意味着什么。你无法轻易地将其数学化。如果你这样做了，并且你为那些模拟进行优化，你会遇到古德哈特定律。你会找到你要求的东西，但不是你想要的东西，对吧？所以我们真正想要的是人类以这种主观模糊的方式对此的看法。我们不能把人类放在循环中，因为那太昂贵了。所以次优的选择是尝试找到一个与人类相匹配的表示空间。有很多证据——至少在统计上——表明这些基础模型在统计上匹配人类的表示，对吧？

Yeah, yeah, exactly. And there are like two reasons it's better. One is just easier to specify what you want. You can just be like, 'I want a cat.' I don't need to have a picture of a cat and supervise it in a pixel-wise manner. I can just say I want a cat in a general way, right? That's one way. But importantly, it also makes the optimization and the search much easier, because it's really dang hard to find a self-organizing rule that creates this particular image of this particular cat. If you do that in Conway's Game of Life, that's not going to happen. But if you just say, 'Oh, I don't want that specific cat. I just want a cat. Give me any cat.' That's a much easier space of things that satisfy that problem. So it's much easier to search for. And we experienced this first-hand that in the open-ended simulation, we're not searching for particular configurations of states that we want. We're just saying we want some open-ended simulation. It can be open-ended in however it wants to be, as long as the foundation model thinks it's open-ended, it'll work, it'll do. And the cool thing about—I want to say the reason we use foundation models is not just because it's the newest and greatest thing that you can use, it's because we really care about what a human thinks about these simulations. Because it's really hard to describe what complexity is, what emergence is, what it means to have a very interesting system. You can't really mathematize that really easily. And if you did, and you optimize for those simulations, you'll find Goodhart's law. You'll find something that you asked for, but not something that you wanted, right? So what we really want is what a human thinks about this in this subjective vague way. And we can't put a human in the loop because that's too expensive. So the next best thing is you try to find a representation space which matches that of a human. And there's a lot of evidence—at least statistically—that these foundation models are matching human representations in a statistical manner, right?

是的。所以我现在要在屏幕上展示一张图。这是你做的第一个实验，当时你有一个监督目标。我们有一个多细胞生物、一个神经元网络、一个病毒和一个真菌群落。在每一行中，我们可以看到线性、boids 和粒子生命的涌现现象是什么。但这也引出了一个问题：一定存在某种程度的模糊性，因为在计算空间中有太多可能的地方，基础模型会说这是一个匹配，并且它在不同的模拟器中表现不同。我们之前讨论过这个概念，就像这个冒名顶替者。所以有很多不同的路径来产生这种现象。你会把这些东西称为冒名顶替者，还是你认为它们是不同的？

Yes. So I'm going to show a graphic on the screen now. So this is the first experiment you did when you had a supervised target. And we've got a multi-celled organism, a network of neurons, a virus, and a fungal colony. And in every single row, we can see what the emergent phenomena was for linear and for boids and for particle life. But it also leads to the question: there must be some degree of ambiguity because there are so many possible places in computational space where the foundation model would say that this was a match, and it's represented differently in the different simulators. And we had this concept we were talking about, like this imposter. So there are many different paths to produce the phenomena. Would you call these things imposters or do you think that they're different?

所以这些是——这是个好问题。所以你基本上是在问，让我们谈谈 boids 那个。你基本上是在说：每个 boid 的决策规则是一个冒名顶替者，还是它是给你真实东西的真实东西，对吧？

So these are—that's a good question. So you're basically asking, let's talk about the boids one. You're basically saying: is the decision-making rule of each boid an imposter, or is it the real thing that gives you the real thing, right?

是的，在这个模拟器中，冒名顶替者意味着什么？

Yeah, and what would an imposter mean in this simulator?

是的，这是个好问题。所以我认为冒名顶替者的概念——我喜欢将这个想法建立在它在未来任务上的表现以及它如何适应我们希望它适应的事情上。这就是我喜欢定义它的方式。我不太喜欢将其建立在模块化或类似的具体事物上。我喜欢将其建立在一组未来任务上。

Yeah, that's a great question. So I think the idea of imposter—I like to ground this idea in its performance on future tasks and how it adapts to things we want it to adapt to. That's how I like to coin it. I don't really like to ground it in modularity or something specific like that. I like to ground it in a set of future tasks.

当你继续运行模拟时，它会退相干吗？

When you continue to run the simulation, does it decohere?

是的，在这种情况下，它们对那种事情相当稳健。但在这个情况下，我的意思是，可能是小变化——我想我会称之为——你可以想象，如果这些 boids 真的是 UFR 或很好地分解，那么对它们网络的小改变会以良好的方式改变集体的全局行为，比如从螺旋变成三角形或其他东西。也许不是三角形，而是其他酷的形状。但在这种情况下，它们不会那样做。它们实际上会退化。如果你扫描参数，它们基本上会变得随机。所以从那个意义上说，它有点像是这个螺旋的冒名顶替版本。没错，没错。问题是，在这种情况下，它们被评估并期望保持对模拟未来的持久性。所以从那个意义上说，它们结构非常好，因为它们能够无限期地持续存在。你可能会运行这个模拟数百万个时间步，它们仍然会做它们正在做的事情。是的，但如果你想让它们做其他事情，比如通过扫描一些权重来制造一个三角形，那么它们就会以那种方式失败。

Yeah, in this case they're pretty robust to that sort of thing. But in this case, I mean, it could be that small changes—I guess what I'd call—you can imagine that if these boids were really UFR or nicely factored, then small changes to their network change the global behavior of the collective in nice ways where it goes from a spiral to triangles or something. Maybe not triangles but some other cool shape. But in this case they don't do that. They actually degrade. If you sweep the parameters, then they become basically kind of random. So in that way, it kind of is an imposter version of this spiral. Exactly, exactly. And the thing is, in this case they're evaluated on and they're expected to maintain persistence to the future of this simulation. So in that way they're very, very well structured because they are able to persist indefinitely. You could run this simulation probably for millions of time steps and they'll still be doing what they're doing. Yeah, but if you want them to do something else, like make a triangle by sweeping some weight, then they would fail in that way.

但公平地说，它们一开始就没有被训练去做那件事。它们的演化谱系没有理由让它们想要那样做，对吧？所以我想指出，我们在论文中有一个图，展示了通过基础模型搜索模拟这一可能方向的未来会是什么样子。我们有一个有趣的例子——我把它调出来了。最下面那个特别吸引人。我们基本上是在搜索元胞自动机，比如神经元胞自动机，它从自我复制的分子到漂浮在原始汤中的细胞，再到捕食者-猎物动态，再到四处漫游的外星动物。这只是一个我们使用的有趣玩具例子。在这个例子中，元胞自动机只是记住了那个帧序列。但你可以想象，如果你用像千万亿粒子的大粒子模拟来做这件事，并且你有正确的基础模型，你搜索这个事件序列，那么会发生什么？也许你最终会得到这个大粒子系统的某种物理规则，它真的能给你整个序列——最后你得到动物，就像我们宇宙中发生的那样，对吧？

But then to be fair, they're not trained to do that in the first place. Their evolutionary lineage doesn't—there's no reason for them to want to do that, right? So I want to point out that we have this figure in our paper which showed what the future of this possible direction of searching for simulations via foundation models could look like. And we have this fun example—I have it pulled up right here. The bottom one is particularly fascinating. We're basically trying to search for cellular automata, like neurocellular automata, which go from self-replicating molecules to cells floating in primordial soup to predator-prey dynamics to alien animals roaming around. And this is just a fun toy example that we used. In this case, the cellular automata just memorized that sequence of frames. But you can imagine that if you did this with a big particle simulation with quadrillions of particles and you have the right foundation model and you searched for this sequence of events, then what would happen? Maybe you end up with some physics rule for this big particle system which really does give you this entire sequence—you get animals at the end, just like what happened in our universe, right?

是的，但这不是很迷人吗？你拿了你之前认为是破碎纠缠表示的东西。所以，基础模型有破碎纠缠的表示。但它们产生模仿者，但你只需要模仿者，因为你所做的只是对模仿者进行批评，然后你利用它来引导创建一个因子化表示。

Yeah, but isn't it fascinating that you've taken something which you previously argued was a fractured entangled representation. So, the foundation model has fractured entangled representations. But they produce imposters, but you only need the imposter because all you're doing is you're acting as a critic on the imposter and then you're using that to bootstrap the creation of something which is a factored representation.

是的，是的，这是一个好观点，因为这些基础模型在这类任务上的表现将是无瑕的，对吧？因为这正是你只关心输出行为的那类事情。你不在乎它的内部表示。你不在乎创造力，因为人类是注入那些提示的人。而你正在使用这种统计智能来引导一个真正的自然生成——像有机粒子模拟，它应该具有我们现实世界的所有属性，对吧？

Yeah, yeah, that's a good point because the performance of these foundation models for this type of tasks will be flawless, right? Because this is the exact type of thing where you only care about the output behavior. You don't care about its internal representation. You don't care about creativity in the sense because the humans are the ones injecting those prompts. And you're using this statistical intelligence to bootstrap like a real natural gen—like organic particle simulation which should have all the properties of our real world, right?

是的，正是如此。然后你做的另一个实验是创建多样性搜索。你说计算空间的某些部分是开放式的。所以你之前提到的这个金发姑娘区。也许我们正处于计算空间的这样一个区域，所有这类生命般的东西都发生了，这种开放性也发生了。但你是否在空间中看到了类似的事情，我们谈论的这些属性存在于这个计算空间的一小条中？

Yeah, exactly. And then so the other experiment you did was creating a diversity search. You said some parts of the computational space are open-ended. So there's this Goldilocks zone you were speaking to before. Maybe we're in such a region of the computational space where all of this life-like stuff happened and this open-endedness happened. But did you see similar things in the space where these properties that we're talking about like existed in a sliver of this computational space?

是的，绝对如此。如果你指的是计算空间，我想我们可以谈论物理定律。那是你的空间，对吧？那样的话绝对如此。我的意思是我把它调出来了。找到了基本上——它们看起来比康威的生命游戏更开放式的模拟。是的，那个。就是那个，是的。基本上我们找到了比生命游戏更开放式的模拟，对吧？但我们想更进一步，问，可能规则的空间看起来像什么？哪些规则是开放式的，哪些不是，对吧？所以如果你滚动到论文，我们这里有这个图，我们基本上试图绘制所有可能的规则。有 260,000 条规则，我们绘制了所有。我们给它们着色——我们所做的是运行规则，我们将这个图像放入 CLIP 并在 2D 中绘制这个 CLIP 空间，但我们根据它们的开放式程度给每个点着色。所以我们发现确实有一个大的非开放式解决方案岛，还有一个小岛，所有最酷的模拟都在那里。所以这样这个图大致勾勒出我们正在谈论的——可能模拟的空间——我们想找到这个空间的哪部分是这样的，是的。你可以看到有不同的岛。有一个大岛，一个较小的岛，还有一个微小的岛，上面有一个模拟在行为属性上与其他所有模拟非常不同。

Yeah, definitely. If you mean in the computational space, I guess we can talk about the laws of physics. That's your space, right? In that way definitely. I mean I have it pulled up right here. Found simulations which are basically—they look more open-ended than Conway's Game of Life. Yeah, that one. That exact one, yeah. Basically we found simulations which are more open-ended than Game of Life, right? But we wanted to go further than that and ask, what does the space of possible rules look like? What rules are open-ended, what rules are not, right? So if you scroll down to the paper, we had this graph right here where we basically tried to plot all possible rules. There's 260,000 rules and we plotted all of them. We colored them—what we did is we ran the rules, we put this image into CLIP and plotted this CLIP space in 2D, but we colored each of the points based on how open-ended they are. So we found that there's literally a big island of solutions which are not open-ended and there's a small island of solutions where all of the coolest simulations lie. So in that way this graph kind of outlines exactly what we're talking about—the space of possible simulations—and we want to find what part of this space is like that, yeah. And you can see that there's different islands. There's one big island and one smaller island and one tiny island with one simulation that's very different from all the rest of them in its behavioral properties.

那么，我们有一个问题：在构建这些人工生命模拟和构建具有实用性的通用智能之间似乎存在脱节。那么，我们如何利用这类技术来构建 AGI（通用人工智能）？

So then we have the problem of there seems to be a disconnect between building these artificial life simulations, and we want to build a general-purpose intelligence which has utility. So, how can we leverage this type of technology to build AGI?

是的，是的。所以，那是大问题，对吧？这就是我喜欢处理它的方式。所以，人工生命领域，我真的认为是对智能的一个非常长期的赌注。但是，有两个注意事项。第一，它们仍然值得研究，因为它们给了我们很多关于我们宇宙和涌现以及所有其他科学事物的见解，即使我们不创造智能，我们仍然关心这些。但第二，我认为 Ken 经常谈到这一点，我们并不真的需要在较低层次模拟每一个可能的粒子并让一切引导。我们真正想要做的是创建自然进化的抽象和这些模拟的抽象，并在 10 年后找到一个抽象，第二，它以正确的方式给我们一个抽象，让你得到你想要的，即智能，通用智能。这是一个开放问题，也是我的研究议程之一，即找到我们从自然进化和人工生命中提取的原则，以创建更好的 AI 系统。

Yeah, yeah. So, that's the grand question, right? That's how I like to go about it. So, the field of artificial life, I really think is a very long-term bet on intelligence. But, there's two caveats. One, they're still interesting to study because they give you lots of insights about our universe and emergence and all these other scientific things that we still care about even if we don't create intelligence. But, second of all, I think Ken talks about this a lot is that we don't really need to simulate every possible particle at a lower level and get everything to bootstrap. What we really want to do is create an abstraction of natural evolution and an abstraction of these simulations, and find an abstraction 10 years from now, and two, it gives us an abstraction in the right way where you get what you want, which is intelligence, general intelligence. And that is an open problem, and that's one of my research agendas is finding the principles that we extract from natural evolution and artificial life in order to create better AI systems.

我们想看看如果你使用 LLM 来进化基本上相互竞争的小型汇编程序会发生什么。所以，我可以给你一些背景。有一个编程游戏发明于 1980 年代，人们过去常玩。它叫做 Core War。基本想法是有一个类似汇编的语言叫做 Redcode，过程是这样的：我创建一个汇编程序，你创建一个汇编程序，每个人都创建一个汇编程序。我们把它们放入一个虚拟机，我们把它们放在这个虚拟机的内存地址中的随机位置，然后我们让机器运行。它只是运行程序，它们可以做任何它们想做的事。它就像一个图灵完备的语言，目标是我的程序将是最后一个运行的。所以，我的程序需要做的是，它基本上需要在你的程序执行线程前面注入一个无效操作。当它这样做时，它会通过崩溃来停止你的程序——你的程序会崩溃。然后我的程序将是最后一个运行的，我将赢得这个游戏。

We wanted to see what would happen if you used LLMs to evolve basically small little assembly programs that compete against each other. So, I can give you some background. So, there's this programming game invented in like the 1980s that people used to play. It's called Core War. And basically the idea is there's this assembly-like language called Redcode, and how it goes down is I create an assembly program, you create an assembly program, and everyone creates an assembly program. We put them into a virtual machine, and we put them in the memory address of this virtual machine in a random location, and we just let the machine go. It's just going to run the programs and they can do whatever they want. It's like a Turing complete language and the goal is that my program will be the last one to run. So, what my program needs to do in order to do that is it needs to basically inject an invalid operation in front of your program's execution thread. And when it does that, it's going to halt your program by crashing—your program will crash. And then my program will be the last one running and I will win this game.

所以，为了做到这一点，你的程序基本上需要防御他人的攻击并攻击他人。所以，这就像是一个你可以直接运行的非常酷的游戏。我们所做的是，我们看到人们长期以来一直在制作这些人类战士。我们问，如果我们启动 LLM 来进化这些程序并让它们相互竞争，会发生什么。我们能否由此创建出这种开放式的程序进化军备竞赛？

And so, in order to do this, you need to basically your program needs to defend itself against other people attacking it and attack other people. So, it's like this really cool game that you can just run. And what we did was we saw that people have been making these war human warriors for such a long time. We asked what would happen is if we launched LLMs to evolve these programs and have them compete against each other. And can we create like this open-ended evolutionary arms race of programs from this?

太棒了。所以，这就像是《权力的游戏》图灵版。

That's amazing. So, it's like Game of Thrones Turing edition.

是的。是的，没错。这个说法很好。是的。

Yeah. Yeah, exactly. That's a good way to put it. Yeah.

是的。好吧，什么，告诉我更多。那么，嗯，多少代，种群规模是多少？我是说，你是怎么设置这个的？

Yeah. Well, what what tell me more. So, um how how many generations how what was the population size? I mean, how did you set this thing up?

是的，所以实际的算法非常简单。我们所做的是从一个程序开始。这可以是人类编写的程序，也可以是 LLM 生成的程序。然后你基本上只是用 LLM 作为变异算子，在进化循环中优化一个程序来击败这个程序。所以，你只是尝试这样做，我们最终使用了 map-elites，因为你优化的空间非常具有欺骗性。它非常非局部。如果你只是做普通的梯度下降——不是梯度下降，是贪婪的遗传算法——它不会起作用。你需要使用 map-elites。但然后我们最终得到一个战士——这些程序在 Core War 中被称为战士。你最终得到一个击败你初始战士的战士。然后你可以优化第三个战士来击败这两个。然后你可以优化第四个战士来击败这三个，你一直这样做，最终得到一长串战士，对吧？然后问题是，这种对抗性的进化轮次会把你带向何方？我得到的一个很酷的结果是——我想我们有两个主要结果——第一个是在第一轮训练中，我们已经可以击败人类战士数据集中的 96%。有一个大约 300 个人类战士的数据集，我们仅仅通过使用 LLM 来进化它们，就能击败其中 96%。重要的是，LLM 并不擅长 red code。如果你只是尝试零样本这个语言，它很糟糕。如果你做 best of n 采样，它很糟糕。但如果你做这种进化并且有验证器，它就会变得非常好。这是来自 Alpha evolve 的相同教训，对吧？这些 LLM 不擅长零样本某些类型的证明等，但如果你让它们在循环中与验证器一起进化，它们可以非常非常强大。这说明了进化的力量和模型的先验。但我们有的第二个结果，我最兴奋的，结果是如果你做这些轮次的进化，对吧？那么你最终得到的这些战士，在更远的未来，它们变得越来越通用，也就是说它们更有可能击败它们以前没见过的人类战士。所以，我认为这是一个相当酷的结果。但更重要的是，更酷的部分是，如果我们有这个描述这些战士行为的向量，就像它们的行为特征，我们可以详细讨论我们如何测量行为，但我们有这个向量，这些战士行为的方差随着轮数的增加而下降。所以，这有点意味着它们变得更通用，并且趋势是减少方差。所以，它们在行为上变得像一个单一的通用战士。至少趋势是朝向那个方向。

Yeah, so the actual algorithm is really really simple. So, what we do is we start off with a program. This can be a human program or you can be like just a LLM generated program. And then you basically just optimize a program with an LLM as the mutation operator in evolutionary loop to beat this program. So, you just try to do that and we actually ended up using map-elites because the space that you're optimizing is very deceptive. It's very non-local. If you just do a vanilla gradient descent and not gradient descent a greedy um genetic algorithm, it won't work. You need to use map-elites. But then we end up with a warrior these are programs are called warriors in Core War. You end up with a warrior that beats your initial warrior. Then what you can do is optimize a third warrior to beat both of these two. and then you can optimize a fourth warrior to beat these three and you just keep on doing this and you end up with a long sequence of warriors, right? And then the question is like where does this like adversarial like rounds of evolution, where does it take you? And one of the cool results I got was I guess we have two main results is the first one is in this first round of training, we can already beat like 96% of a data set of human warriors. There's like a data set of like 300 human warriors, we can beat 96% of them just by using a LLM to evolve them. And importantly, the LLM is not good at red code. If you just try to zero shot this language, it's terrible. If you do best of best of n sampling, it's terrible. But if you do this evolution and you have this verifier, it turns out to be really good. This is the same lesson from Alpha evolve, right? These LLMs are not good at zero shotting certain kinds of proofs and stuff, but if you let them evolve with a verifier in the loop, they can be really, really powerful. That speaks to the power evolution and the priors from the model. But the second result that we have that I'm the most excited about, it turns out if you do this rounds of evolution, right? Then these warriors that you end up with uh further down the line, they become more and more general as in they're more likely to beat human warriors that they haven't seen before. So, I think that's a pretty cool result. But even more than that, the cooler part is is if we if we have this vector describing the behavior of these warriors, just like their behavioral characteristics, and we can get into details of how we measure the behavior, but we have this vector, the variance in the behavior of these warriors goes down as you increase the number of rounds. So, that kind of means that they're getting more general and they're trending towards like kind of like reducing variance. So, they're kind of behaviorally becoming like a single generalist warrior. At least that's what the trend is towards.

没错，我注意到 AI 领域的人通常认为进化是一种非常愚蠢的朴素算法，当你无法做梯度下降时才会使用，对吧？他们基本上称之为随机搜索。我想澄清的是，进化绝不是随机的。它绝不是随机的，进化可能是我们所知的解决离散组合问题的最强大算法，你需要基本上先解决拼图的各个部分，然后才能解决整个问题。进化中的选择机制被严重低估了，因为如果你解决了拼图的一部分并坚持它，然后搜索其他部分，你可以将指数搜索问题转化为线性搜索问题。所以，进化是一个真正强大的算法，但它只需要你的变异有 1% 的时间是好的，对吧？如果你只是天真地在程序空间这样做并进化——我想有一篇很酷的论文，大约在 LLM 时代之前的 AI，我想是 auto ML zero，他们做程序搜索但用随机变异，那真的很难正常工作，因为任何变异给你有用代码的机会几乎为 0%。但 LLM，当然它们不能给你 100% 的成功，但它们可以给你远超过 1%。所以，基本上，每当你成功时，你可以坚持那个成就并在其基础上构建，这就是为什么进化加 LLM 的组合比单独使用 LLM 强大得多。甚至不是接近的。

Exactly, and I think people in AI I've noticed tend to think of evolution in general as this very dumb naive algorithm that you use when you can't do gradient descent, right? And they basically call it just random search. And I would just want to clarify that evolution is anything but random. It's anything but random and evolution is probably the most powerful algorithm that we know to solve like discrete combinatorial problems where you need to basically get pieces of the puzzle right before you get the whole thing right. And the selection mechanism in evolution is really really underrated because if you solve like even one part of a puzzle and you hang on to that and you search for the other ones, you can turn your exponential search problem into a linear search problem. So in that way evolution is a really really powerful algorithm, but all it needs is your mutations to be good like 1% of the time, right? If you just do this with like naively in like program space and just evolve I think there was a cool paper from like about before LLM era of AI like I think auto ML zero where they did program search but with random mutations and that's like really hard to get working properly because there's like literally around a 0% chance that any mutation will be give you some useful code. But LLMs, sure they can't give you 100% success, but they can give you way more than 1%. So then you can basically it whenever you are successful, you can hang on to that achievement and build on top of it and that's what gives this combination of evolution plus LLMs a lot like a significant more amount of power than if you just do LLMs by themselves. It's not even close.

Akash Kumar，非常荣幸能邀请你上节目。

Akash Kumar, it's been an absolute honor having you on the show.

太棒了。感谢邀请我。

Awesome. Thanks for having me.

是的，太棒了。这是一次很棒的对话。

Yeah, it's been amazing. This was a great conversation.

是的。

Yeah.