追寻治愈之道：同构实验室的人工智能药物设计

五年后，不使用人工智能进行药物设计，就像做任何科学不用数学一样。说实话，我认为整个科学领域都会如此。就像，如果你不使用人工智能，你还在做什么？

In five years' time, doing drug design without AI will be like doing any sort of science without maths. To be honest, I think the whole of science will be like this. It's like, if you're not using AI, what are you doing?

我认为，当两个领域的专家聚在一起，并对对方的领域充满好奇时，就会产生这种神奇的效果。

I think when you've got, like, experts in two fields and they come together and they're really curious about the other field, that's when you get this kind of magic.

欢迎回到《谷歌DeepMind播客》。我是汉娜·弗莱教授。你们已经知道，德米斯·哈萨比斯和约翰·詹珀因将人工智能应用于蛋白质折叠的研究，获得了2024年诺贝尔奖。人体的一切都归结为蛋白质，它们如何折叠，如何运作。但就在不久前，确定一个单一蛋白质的结构可能需要数月甚至数年时间。

Welcome back to Google DeepMind, the podcast. My name is professor Hannah Fry. Now you'll already know that Demis Esabes and John Jumper won the Nobel Prize in 2024 for their work applying artificial intelligence to protein folding. Now everything comes down to proteins in the human body, how they fold, how they function. But not very long ago, working out the structure for one single protein could take months or even years.

随着谷歌DeepMind开发的AlphaFold 2算法的发布，整个领域被彻底革新了。最近，AlphaFold 3能够以前所未有的准确性预测所有生命分子的结构，这对于药物设计来说至关重要。这些发展为从谷歌DeepMind分拆出来的一家新公司铺平了道路。它名为Isomorphic Labs，象征着生物学与人工智能的协同作用。今天与我一起的是该公司两位最引人注目的新成员。

And then with the release of AlphaFold two, the algorithms developed at Google DeepMind, the entire field has been completely revolutionized. More recently, AlphaFold three can predict the structure of all of life's molecules with unprecedented accuracy, which turns out to be absolutely pivotal for drug design. These developments have paved the way for a new company spun out of Google DeepMind. It's called Isomorphic Labs to represent the synergy between biology and AI. And joining me today are two of its most notable hires.

丽贝卡·保罗是药物设计负责人，在新药发现过程中拥有多年经验。马克斯·雅德伯格是首席人工智能官。一直关注本播客的听众会记得马克斯早年在DeepMind时，教智能体玩夺旗游戏和《星际争霸》。而今天，他认为智能体将在未来的药物发现中发挥关键作用。马克斯、丽贝卡，非常感谢你们加入我们。

Rebecca Paul is head of medicinal drug design with years of experience in the process of discovering new drugs. And Max Yaddeberg is its chief AI officer. Anyone who's been following this podcast will remember Max from his earlier days at DeepMind, teaching agents to play capture the flag and Starcraft. And today, he thinks that agents will be instrumental in the future of drug discovery. Max, Rebecca, thank you so much for joining me.

谢谢邀请我们。

Thank you for having us.

是的。很荣幸来到这里。谢谢。

Yes. A pleasure to be here. Thanks.

嗯，很高兴你能来，因为现在正发生一些大事。我知道很多人都在谈论AI将能解决所有疾病这一说法。马克斯，这现实吗？

Well, it's a delight to have you because there's some some big stuff happening. I know a lot has been made of this claim that AI is gonna be able to solve all diseases. Is that realistic, Max?

这不会一夜之间发生。我们要清楚这一点。但我认为令人兴奋的是，我们实际上可以看到或许有一条通往那个目标的可行路径。而且现在的情况与以往完全不同，因为我们拥有的这些AI机器学习模型以与以往截然不同的方式理解生物世界和生化世界。因此，这正在开辟大量我们以前认为难以攻克的疾病领域。

This isn't gonna happen overnight. Let's be clear. But I think the exciting thing is that we can actually see there's perhaps a practical path towards that point. And it's very different at this point in time than it has ever been because we've got these AI machine learning models that understand the biological world and the biochemical world in a completely different manner than what we've had before. And so that's opening up tons of disease space that we didn't think was tractable before.

而这仅仅是开始。所以随着我们进一步开发这些模型，这真的只是非常初期的阶段。

And that's just today. So as we start to develop these models further and further, it's really just the very beginning.

这是否意味着所有疾病都在考虑范围内，贝基？

Does that mean that every disease is on the table here, Becky?

所以我会说，目前没有什么是不可能的。对我来说，这段旅程意义重大。过去我在这个领域要保守得多，但来到Isomorphic Labs后，看到我们拥有的各种模型，那些我过去认为永远无法预测的事情，现在每天都能在五到十秒内完成，这完全改变了我的思维方式。所以现在我认为没有什么是不可能的。

So I would say nothing is off the table at this point. And the journey for me here has been a big one. Used to be much more conservative in this space, But having come to isomorphic labs, seeing the kind of models that we have, things that I thought in the past we would never be able to predict and now being able to do it every day within five, ten seconds, it's completely shifted my mindset. So now I would put nothing off the table.

这仅仅是在药物设计方面，还是AI也会影响临床试验？

And is it just in drug design or is AI gonna affect clinical trials as well?

我认为随着时间的推移，是的，它肯定会影响临床试验。目前我们非常专注于药物设计阶段。你可以想象这样一个世界：随着你在药物设计方面越来越擅长，实际上，越来越多的瓶颈会出现在临床开发方面。因此我们真的需要重新思考如何做到这一点。我认为我们已经很久很久没有真正改变临床开发的方式了。

I think over time, yes, it will definitely affect clinical trials. We're focusing really heavily on the drug design phase at the moment. You can imagine a world where as you start to get better and better at drug design, actually, you know, more and more of the bottleneck comes onto the clinical development side of things. And so we really need to rethink how we do that. I don't think we've really changed the way we do clinical development for a long, long time.

进展仍然非常缓慢。当然，出于充分理由存在大量监管。但随着我们对这些分子作用机制、疾病的真实机理，以及这些分子如何与身体其他部分相互作用并影响体内毒性等各方面的理解越来越深入，我们甚至可以重新思考如何进行临床试验设计，如何首次进入人体并开始测量这些分子的有效性。这对我们isomorphic来说还不是当前的重点，但绝对是未来的方向。

It's still very slow. There's of course a lot of regulation for good reason. But as we understand more and more about how these molecules work, the true mechanisms of disease, how these molecules also interact with the rest of the body and affect everything like toxicity inside of us, we can start to rethink even how we do those clinical trial designs, how we first go into people and start measuring the efficacy of these molecules. That's not right now for us at isomorphic, but it's very much in our future.

那么核心构想是什么？isomorphic的宏大愿景是什么？

What is the big idea then? What's the big ambition of isomorphic?

真正迈向可治愈疾病领域。这是在进入临床试验之前，甚至在开始人体测试之前，你需要有可测试的对象。对吧？你需要创造分子和药物。而药物是什么？

It really is stepping towards that solvable disease space. And so this is really before you go into clinical trial, before you even start testing out on people, need something to test. Right? You need to create a molecule and a drug. And a drug, what is that?

它是进入体内调节某些身体功能、细胞功能的东西。因此isomorphic实验室的关键第一阶段是：如何创建这个AI药物设计引擎，能够针对几乎任何疾病、任何与该疾病相关的蛋白质靶点，设计出能够进入体内调节这些蛋白质功能、细胞功能的分子，从而改变疾病状态，为患者带来积极疗效。

It's something that goes in and modulates some function of the body, some function in a cell. And so the key first phase of isomorphic labs is how can we create this AI drug design engine that can take pretty much any disease, any protein target that's implicated in that disease, and work out how to create a molecule that will go in and start modulating the function of these proteins, the function of cells, and then change the disease state for, you know, positive of patients.

问题是，过去我们已经治愈过疾病，研发出使某些疾病成为历史遗迹的药物方案。为什么有些疾病比其他疾病更难解决或治愈？

The thing is, I mean, diseases have been cured in the past, We have come up with like drug solutions that effectively make them a remnant of history. Why are some diseases so much more difficult than others to solve or cure?

某些癌症可以开发治愈性疗法，因为这些癌症可能非常稳定。比如可能只有一个驱动突变，用针对该特定突变的分子治疗就能治愈。但如果癌症不断进化，细胞演化出耐药性，你就需要持续更新治疗药物，这样的疾病就难解决得多。

So some types of cancers, you can develop a treatment to cure them because for example, those cancers might be quite stable. So there's maybe one mutation that drives that cancer and then you treat with that molecule that hits that particular mutation and that can be curative in that disease. But if you had a cancer which was continually evolving and the cells were evolving to overcome the drug that you're treating with, then you need to be continually evolving the drug that you're treating them with. So that would be then a much more difficult disease to solve.

我猜想是否还有其他情况，比如我们对体内发生的机制存在根本性的认知空白？

And I guess are there others where there's just like a real gap in our understanding of what's going on in the body?

是的，还有一个巨大的认知鸿沟在于，到底是什么驱动了这种疾病？是多种因素共同作用吗？我们对生物学还没有全部答案，生物学非常非常复杂。我们需要做的基本事情之一，就是在开发小分子或某种化学物质之前，先找出真正驱动疾病的生物学机制，这样才能调节正确的生物学过程，从而真正改善症状或对疾病进行修饰治疗。

Yeah, there's also that massive gap in understanding between what's actually driving this disease? Is it multiple things together? We don't have all the answers yet to biology, biology is so, so complex. One of the fundamental things we need to do is find out what biology is actually driving disease before we can then develop a small molecule or some kind of chemistry that then will modulate the right biology to actually make an improvement in the symptoms or a disease modification to that disease.

那么这就是你们正在尝试做的事情吗？比如像癌症，一旦深入到分子和蛋白质层面，就是那个层面上出了问题，然后逐步升级到人体规模。

Is that what you're trying to do then? So actually something like cancer, for instance, once you get down to like the level of molecules and proteins, it's that something's gone wrong at that level, which then escalates to the scale of the human body.

完全正确。你可以把蛋白质想象成细胞内的微型工厂或引擎。它们有功能，会执行某些任务。所以如果发生了突变或蛋白质发生了变化，比如它总是处于开启状态，而在正常细胞中它可能是时开时关，或者大部分时间关闭。

Absolutely. You can think of proteins as like mini factories or engines inside your cell. They have a function, they do something. So if you have a mutation or something that changes about that protein, that means, for example, it's always switched on. Where in a normal cell it might be going on and off, or maybe it's mostly off.

但在癌细胞中，发生了某些变化，它总是开启着。它会持续驱动一个信号，这个信号可能是生长、增殖，从而推动肿瘤的形成。

But now in a cancer cell, something's happened, it's always on. It's gonna just continually drive a signal, and that signal could be grow, proliferate, and that will then drive the formation of a tumour.

所以你的目标就是，既然知道那个蛋白质出了问题，你就用某种药物分子来靶向那个蛋白质。

And so your target then, you know that there's something going on with that protein, so you're targeting then that protein with a molecule of some medicine or drug.

没错。好的。所以如果你想阻止那个小工厂运转，你就要设计一个形状完美的扳手，扔进那套小齿轮里让它停止工作，这样它就不会再驱动那个信号了。

Exactly. Okay. So you want to, if you think of that little factory working away, you want to design like the perfect shaped wrench that you can throw into that little set of gears to stop it working, so that then it's gonna stop driving that signal.

这归根结底是形状问题吗？

Does it come down to shape?

形状非常重要。你可以把蛋白质想象成有口袋、凹槽或裂隙，而你确实需要设计一个分子，它能完美地嵌入那个凹槽中。

Shape's very important. You can think of proteins as having pockets or grooves or crevices, and you really need to design a molecule that perfectly fits inside that groove.

是为了阻断它吗？阻断它。哦，真的吗？

To block it? To block it. Oh, really?

确实就是这样。是的。

It's literally like that. Yes.

那所有已经存在的药物都是这样吗？比如扑热息痛，它也是这么起作用的吗？是的。所以扑热息痛是

And is that true of all drugs that that already exist? I mean, like paracetamol, for example, is is that what it's doing? Yes. So paracetamol is

一种小分子，它会结合到其蛋白质靶点上，阻断其功能。对于扑热息痛或止痛药来说，它是在阻止你感受到疼痛。所以，是的。所以在某种程度上，你有点像

a small molecule, binds to its protein targets, blocks the function, And in in the case of paracetamol or painkillers, it's preventing you from experiencing pain. So, yes. So in some ways then, you're sort of

在分子层面上玩乐高。

playing Lego at the at the molecular level.

完全正确。是的。我们非常喜欢这个过程。

Exactly. Yes. And we love it.

为什么这对AI来说是一个好情况？是什么让这成为一个非常适合AI解决的问题，Max？

Why would that be a good situation for AI? What what makes this a a a a well suited problem for AI, Max?

这实际上是AI和机器学习的一个完美应用。整个这种，你知道的，像玩乐高一样摆弄分子。过去五、六、七年里，我们看到像AlphaFold这样的模型崛起。我们有AlphaFold一代、AlphaFold二代来理解蛋白质结构。在AlphaFold二代之前，没有人能在不进入实验室、通过实验解析这些结构的情况下真正理解它们，而这可能需要数月时间。

It's actually such a perfect application of AI and machine learning. This whole, you know, playing Lego with molecules. We've seen over the last five, six, seven years the rise of models like AlphaFold. We had AlphaFold one, AlphaFold two understanding the structure of proteins. Before AlphaFold two, no one could really understand the structure without going into a lab and experimentally resolving these structures, and that can take months.

可能需要数年。有些蛋白质甚至根本不可能解析。显然，AlphaFold二代是化学领域获得诺贝尔奖的突破。现在，我们通过像AlphaFold三代这样的工具更进一步，现在我们可以理解蛋白质与小分子结合的结构。这些小分子就像是进来的小乐高积木，我们用作药物来抑制蛋白质的功能或改变蛋白质的功能。

It can take years. Sometimes it's not even possible for some proteins. Obviously, AlphaFold two, Nobel Prize winning breakthrough in chemistry. And now we've taken that even further with things like AlphaFold three, where now we can understand the structure of proteins with small molecules. And these small molecules are the little Lego blocks that come in and we use as drugs to inhibit the function of a protein or change the function of a protein.

这之所以是机器学习的一个绝佳领域，原因在于我们本质上试图预测的是这个生物分子系统的三维坐标。这非常、非常契合我们一些经典的监督学习建模领域。它也非常、非常契合我们的扩散建模框架，这些框架在图像生成或视频生成等领域已经取得了巨大成功。

And the the reason why this is such a good domain for machine learning is that what we're trying to do in essence is is predict the three d coordinates of this biomolecular system. And this fits really, really nicely into some of our classic supervised learning modeling domains. It fits really, really nicely into our diffusion modeling frameworks that have been so, so successful for things like image generation or video generation.

就适合之前已有的监督学习而言，是因为这里有一些成功的度量标准吗，比如某种方式上的'正确'（带引号）？

And in terms of being suited for the supervised learning stuff that has already gone before, is that because there is some metric of success here, like some way of being right in vertical commas?

这里有一个非常清晰的成败度量标准，这对于开发这些模型和研究非常有帮助。原因是，过去五十年里，人们一直在通过实验解析这些蛋白质结构，解析这些蛋白质与小分子、DNA、RNA结合的结构。他们一直在实验室里手动完成，然后将结果存入一个名为蛋白质数据库（Protein Data Bank, PDB）的大型数据库中。这提供了一个非常丰富的信息来源。你知道，它有几十万个三维结构。

There's a very clear metric of success here, which is super helpful for developing these models and for research. And the reason is, over the last fifty years, people have been experimentally resolving these protein structures, structure these proteins with small molecules, with DNA, with RNA. They've been doing it by hand in a lab, and then depositing the results into a big database called the Protein Data Bank, PDB. And this gives a really rich source of information. It's, you know, a couple of 100,003 d structures.

每个结构都有数千个原子坐标。所以信息密度非常高。这对于监督学习来说是完美的场景。不过，这不是网络规模的数据。所以它不是我们训练大型语言模型时可能习惯的那种数据规模。

And each structure has thousands of atom coordinates. So there's a really high information density. And that's a perfect scenario for supervised learning. Now this isn't web scale data. So it's not the scale of data that we might be used to for training large language models.

但令人难以置信的是，我们已经找到了设计这些神经网络架构和训练机制的方法，只需在几十万个结构上进行训练，就能激动人心地实现我们所谓的泛化能力。你看，我们发现这些模型能够泛化。它们可以应用于全新的蛋白质、全新的分子，这些是人类历史上从未见过的。当然，这对于药物设计至关重要。药物设计就是要创造出自然界中从未见过的新分子，以调节这些功能。

But incredibly, we've worked out ways to design these neural network architectures and these training regimes so that we can only train on a couple of 100,000 structures and excitingly get what we call generalization. You know, we see that these models can generalize. They can be applied to completely new proteins, completely new molecules that people have never seen before in history. And of course, that's essential if you're doing drug design. Drug design is about creating completely new molecules that we've never seen before in nature even to actually modulate these functions.

但我想，还有你可以设计的分子可能性。我的意思是，这个数字应该非常大吧。

But then also, I guess, the possibilities of molecules that you could design. I mean, it's sort of, well, very big number, I imagine.

是的。这个数字非常庞大。人们经常提到像10的60次方这样的数字，这是宇宙中可能存在的类药物分子的数量。所以这是一个巨大的组合问题。这也是人工智能一个非常令人兴奋的领域。

Yeah. That number is huge. People throwing out around numbers like 10 to the power of 60 is the the possible number of drug like molecules out there in the universe. So it's it's a huge combinatorial problem. And that's also a really exciting spot for AI.

我们可以创建出色的预测模型，预测这些分子如何结合，甚至它们结合的强度或它们的性质。但在10的60次方的设计空间中，我们达到了宇宙原子数量的级别。所以即使你有完美的预测模型知道它们如何结合，也无法穷尽搜索那个巨大的空间。那么该怎么办呢？好吧。

We can create great predictive models of how these molecules fit together, even how strongly they fit or the properties of them. But with a design space of 10 to the power of 60, we're reaching the level of, you know, atoms in the universe. So even if you had the perfect predictive models of how this fits together, you wouldn't be able to exhaustively search that massive space. So what do you do? Okay.

也许你可以对这个空间进行子采样，然后搜索一些大的数量。一千万，一百亿。

Maybe you subsample that space and you search through some large number. A 10,000,000, a 10,000,000,000.

这根本不算什么，对吧？

It's nothing, is it?

即使你...是的。你甚至还没有触及表面。这时我们就可以转向新型模型，比如生成模型、搜索方法、智能体，它们不是穷尽搜索整个分子空间，而是能够智能地探索整个空间，而无需穷尽搜索整个空间。

Even though you're not yeah. You're you're not scratching even scratching the surface. And that's where we can then fall to new types of models, things like generative models, search methods, agents, which instead of exhaustively searching the full molecular space, we can really smartly start to explore across that whole space, but without exhaustively searching the whole space.

贝姬，那么请给我讲讲AlphaFold 3。在药物设计方面，它实际上能让你做什么？

Becky, tell me about AlphaFold three then. In terms of designing drugs, what does it actually allow you to do?

所以，作为药物化学家，我们一直希望能够可视化我们的分子如何与蛋白质结合。就像我们的乐高积木如何融入更大的乐高拼图中。

So, as a medicinal chemist, we always want to be able to visualise how our molecule binds to our protein. So how our Lego block fits into the bigger picture of Lego blocks.

谢谢你跟我一起用

Thank you for going with me on

这个类比。这个类比。我们需要这种可视化的原因是，当你优化小分子与蛋白质的结合时，你需要知道探索哪个方向。你需要在心里有个目标。我要探索口袋的那部分，或者我要研究蛋白质结构的那个区域，看起来是个不错的切入点。

this analogy. The analogy. The reason we need to have that visualisation is because when you're optimising a small molecule binding to a protein, you need to know which vector to explore. You need to have kind of some kind of target in mind. I'm gonna explore that part of the pocket or I'm going to, you know, that part of the protein structure, like that looks like a good place to go.

多年来，我们作为科学界一直在投资实现这一目标的方法。X射线晶体学是一种实验技术，你可以进入实验室，花很多时间结晶你的蛋白质，一旦它与你的小分子结合，你就用X射线照射它，然后你可以原子级别地可视化你的小分子如何与蛋白质结合。现在我们用AlphaFold3及其最新迭代模型，只需几秒钟就能做到。所以，这在我读博士或早期研究时可能需要几个月的事情，现在我在屏幕上随时都能看到。因此，你可以在计算机中反复迭代，直到找到看起来非常有前景的结果。

And so for years, we've invested as a kind of scientific community in ways to do that. So X-ray crystallography is an experimental technique where you can actually go into a lab and you can spend a lot of time crystallising your protein, you fire x rays at it once it's bound to your small molecule, and then you can actually visualize atom by atom how your small molecule is binding to your protein. Now we can do that with AlphaFold3 and sort of the latest iterations of that model, you know, in just seconds. And so something that might have taken me months when I was, you know, doing my PhD or in my early stage research, I'm now just seeing on my screen all the time. And so you can just iterate and iterate in Silico until you get to something that actually looks really quite promising.

然后你把它带到实验室去。

And then you take that into the lab.

那么它是如何运作的呢？你是说，好吧，我认为像这样、这样、这样的东西会有效。让我们试试看是否合适。嗯。还是反过来呢？

So which way around does it work then? Are you saying, okay, I think something like this, this, this and this would work. Let's try it out and see if it fits. Mhmm. Or is it the other way around?

这些模型是在告诉你这可能是一个合适的方案吗？

Are these models telling you this is something that might fit?

在我们ISO构建药物设计平台的方式中，你可以两者兼顾。所以我可以作为一名经验丰富的药物化学家介入，提出我的想法，并在几分钟内立即测试并获得反馈。但你也可以采取相反的方法。好吧，我其实不确定什么在这里会奏效。

So in the way that we've constructed our drug design platform at ISO, you can do both. So I can come in as an experienced medicinal chemist and I can say, I think this, and I can test it then and there couple of minutes and get that feedback. But you can also take the opposite approach. Okay. I don't actually know what's gonna work here.

所以我会应用我们现有的生成模型。我会进行一些我们称之为虚拟筛选的工作。我会选取一个商业上可获得的化学空间区域，将其针对我的蛋白质进行筛选，让模型告诉我那个子集中什么是最优的。

So I'm gonna apply the generative models we have. I'm gonna do some what we call virtual screenings. I'm gonna take an area of chemical space that's commercially available. I'm gonna screen that against my protein. I'm gonna get the models to tell me what's best from that subset.

我能看看你实际设计东西时的界面是什么样子吗？

Can I see what it looks like when you're actually designing something?

好的，让我看看。

Yeah. Let me see

让我描述一下这里的情况。所以这些卷曲状的东西是一种蛋白质。

let me describe what's going on here. So you've got the the sort of curly stuff is a protein.

是的，那是蛋白质。

Yes. That's the protein.

是的。而且它是折叠的，所以你得到了一种三维结构。确实如此。然后在这里，你有一个——我的意思是，这看起来像是你在GCSE化学中会做的那种东西。是的。

Yep. And it's folded, so you've got the kind of three-dimensional structure. Indeed. And then over here, you've got a I mean, this looks like the kind of thing you would do in GCSE chemistry. Yeah.

这种图表。

This kind of diagrams.

没错。这是一个小分子。

Exactly. This is a small molecule.

这算是插入了蛋白质中吗？

This is sort of plugged into the protein?

是的。所以这个小分子正好嵌入了蛋白质的一个小凹槽里，并且与蛋白质形成相互作用。你看到的虚线，就是那个小分子与蛋白质本身之间的相互作用。作为药物化学家，我们希望优化或增加相互作用的数量，因为这会增强小分子与蛋白质之间那种关系的强度。

Yeah. So the small molecule is fitting into this little groove in the protein and it's forming interactions with the protein. So the dotted lines you can see, those are interactions between that small molecule and the protein itself. And as medicinal chemists, we want to optimise or increase the number of interactions, because that's increasing the strength of that kind of relationship between the small molecule and the protein.

那它真的就像是三维拼图吗？我是说，真的吗？

And it really is like three d jigsaws then? Mean, Really?

是的，完全正确。

Yes, exactly.

这就是你们正在做的事情吗？

That's what you're doing?

是的，这就是我们正在做的事情。

It's what we're doing, yeah.

而那个裂缝可能就是让你感到疼痛的原因，你知道，可能是导致肿瘤生长或其他问题的根源。

And that crevice could be the thing that's making you feel pain the thing that's, you know, causing tumour growth or whatever it might be.

是的，导致你的疾病，没错。

Yeah, causing your disease, yep.

太神奇了。所以你们在尝试这个分子的不同版本，看看是否能最完美地嵌入那个小裂缝。

Amazing. So then you're trying different versions of this molecule to see if you can get the best possible fit in that little crevice.

没错，我可以很快给你展示一下。我们的平台上有许多不同的功能可以尝试。这是我最喜欢的一个，Max总是为此说我，因为我可以利用我作为药物化学家的专业知识，说，好吧，想对这个分子做特定的修改。我可以实时在三维视图中查看我正在做的事情。所以现在它会获取那个结构预测。

Exactly, and I can quickly show you. So we have lots of different functionality on the platform that you can try. This is my favorite, which Max always tells me off about, because I can use my expertise as a medicinal chemist and I can say, okay, want to make specific changes to this molecule. I can actually view what I'm doing in three d. So this is now going to fetch that structure prediction.

而且我实际上可以对这个分子进行修改，并实时看到预测的结构。

And I can actually make modifications to this molecule, and I can see the predicted structure in real time.

通常来说，这本来会花费很长时间，我是说，在人工智能出现之前。

And normally, this would have taken, I mean, before AI, a long time.

我的意思是，如果你要进实验室通过实验来确定这个，可能需要几周到几年的时间。

I mean, if you're gonna go into a lab and experimentally determine this, it could be anywhere from weeks to years.

你为什么因为这个责备她？

Why do you tell her off for this one?

我猜你指的是随着时间的推移，我们想要越来越多地从模型本身来做更多事情。啊。从我看来，真正令人兴奋的前沿是，会有很多场景中，贝基会想要手动进行这些更改，并测试她关于这个分子为何有效以及我们如何能改进它的非常具体的假设。然后我们还应该做的是，要求我们的生成模型和智能体说，嘿，我是这样思考这个问题的。

I guess you're referring to the fact that over time, we wanna do more and more from the model itself. Ah. The really exciting frontier from my perspective is there's gonna be lots of scenarios where Becky will wanna go in and make those changes by hand and test out very specific hypotheses that she has on why this molecule works and how we can make it better. And then what we also should be doing is asking our generative models and our agents to say, hey. This is how I'm thinking about the problem.

这些是我的设计约束。我想要一个能实现x y z功能、具有这些特性、看起来大致如此、或许能在那里相互作用并形成这种形状的分子。你能想出什么方案？也许设定好让它运行，然后离开，去喝杯咖啡，回家，第二天早上再过来看看智能体提出了什么方案。

These are my design constraints. I want a molecule that does x y zed and has these sort of properties and kind of looks like this and maybe interacts over there and makes this sort of shape. What can you come up with? And maybe set this running, go away, have a coffee, go home, come in come in the next morning and see what the agent has come up with.

我想，在DeepMind所有应用人工智能的项目中，大致都经历了这个过程，对吧？从最初依赖人类专业知识开始，然后逐渐在模型内部构建更多知识和专长。

I guess with all of the projects that you've applied AI to sort of generally in DeepMind, it has gone through that process, right, of starting off with human expertise and then kind of slowly building in more knowledge and expertise within the model itself.

是的。我认为这与我们在大语言模型领域经历的时刻有很多相似之处——我们拥有大语言模型已经很长时间了。十年前我就在研究它们，但当时它们还挺糟糕的。它们输出的东西看起来像语言，似乎有点道理，但又不太说得通。

Yeah. And I think there are a lot of analogies to that moment that we had with large language models where we had large language models for a long time. And I've been working on them ten years ago, but they were kind of rubbish. And they were spitting stuff out that looked like language. It kinda made sense, but it also didn't make sense.

你不得不进行修正，而且很明显那不是人类生成的。然后它们逐渐一点点变得更好，突然就突破了人类可感知的阈值，让你无法分辨这到底是人类生成的还是机器生成的。我们现在在分子设计模型上也达到了同样的阶段，之前有分子生成模型，它们会输出一些东西，但如果你把这些交给像贝基这样的化学家，她可能会急得抓狂。

You had to correct and it clearly wasn't human. And then they got steadily better little by little, and suddenly they just passed through this human perceptible threshold where you can't really tell whether this is generated by human or not. And we're getting to the same point with our molecule design models, where you had generative models of molecules and they spit stuff out, but you'd give them to a chemist like Becky and and she would probably like tear her hair out.

那么，你见过这类模型吗？

Well, did you see those kinds of models?

是的，我见过。没错。因为在加入Isomorphic Labs之前，我其实在一家AI公司工作过。我亲眼目睹了那种进步和发展的历程。

Yes, I did. Yeah. Because I actually worked at an AI company prior to joining Isomorphic Labs. I've seen like that progression and that journey.

告诉我它们都输出了什么样的东西？

And tell me what kind of stuff did they spit out?

你知道，很长一段时间里，它们输出的基本上都是些无意义的东西，因为你显然给模型设定了一个艰难的目标。比如你希望它能生成某种结合力非常强的分子。为了做到这一点，它可能只是把分子做得巨大。哦。但后来我们逐渐明白，这样的分子实际上无法通过肠道被吸收进入血液，因为你——

You know, for a long time it would just be kind of nonsense because you're obviously giving the model an uphill function. You want it to get to like something that's gonna bind really potently, for example. To do that, maybe it just makes the molecule massive. Oh. But then you know, we'll learn it's not gonna actually be absorbed through the intestine into the bloodstream because Because you've

需要考虑的不仅仅是分子本身。

got more to think about than just the molecule itself.

是的。所以还有很多——

Yeah. So there's a lot

在这里拼凑起来。哦，那很有趣。所以人工智能，如果我们回到乐高的比喻，就像是在蛋白质周围建造一堵巨大的乐高墙。是的。没错。

to piece together here. Oh, that's interesting. So the AI, if we go back to our Lego analogy, was just like building a massive Lego wall all around the protein. Yes. Yep.

我明白了。好的。基本上。现在这个，你有没有看到那个时刻，那个转折点，就像马克斯描述的语言模型那样？

I see. Okay. Essentially. And now this is, have you seen that moment, that tip over, like with the language models that Max describes?

是的，我对生成式人工智能产生的一些分子的质量感到非常惊讶。有时候出来的分子会让你想，哦，为什么

Yeah, I think I've been so surprised by the quality of some of the molecules that come out of the generative AI. And sometimes the molecules that come out, you think, oh, would

为什么我不会想到

have why wouldn't I have come

那个呢？就像，那真的非常惊人。当然，你不必去制造那个确切的分子，但你可以把它作为灵感来做些别的事情。所以

up with that? Like, that's really that's really amazing. And of course, you don't have to go and make that exact molecule, but you could then use that as inspiration to, like, do something else. So

所以它是在与你合作？

So it's working with you?

是的。你们可以一起工作。我的意思是，我猜在未来的某个时候，它会变得如此之好，以至于你会觉得，哦，我没什么需要改变的了。

Yeah. You can work together. I mean, I'm guessing at some point in the future, it will be so good that you'll be like, oh, there's nothing I would change.

我们有过一些非常有趣的时刻，比如，我们让一个模型盲提交分子，然后其他人查看这些分子并说，好吧，我们要送去测试什么？

We've had some really fun moments where, you know, for example, we've had a model submitting molecules blind and then other people looking at them and saying, okay. What are we gonna send off for testing?

哦，真的吗？就像是分子的图灵测试？

Oh, really? Like, Turing test for molecules?

人们看着这些分子，比如，非常有经验的药物化学家会说，哇，这个分子的设计背后有很多经验。实际上并不知道这是由一个通用模型设计的

And people looking at these molecules, like, very, very experienced medicinal chemists saying, wow. There's a lot of experience behind the design of this molecule. And actually not knowing that this was designed by a general

通用模型设计的。但好吧，让我理解一下，因为用大型语言模型的类比，某种程度上说得通，你有这些标记，你知道，你把单词分解成小片段，然后你可以从那里构建起来。你在化学上如何以有意义的方式做到这一点？我的意思是，你不是只取原子吧？

generalist instead. So but okay. Let let me understand this though, because using the the analogy of large language models, it sort of makes sense there that you you have these tokens, you know, you kind of break words down into to little little bites, and then you can kind of build up from there. How do you do it in a way that makes sense chemically? I mean, you're not just taking atoms, are you?

我们实际上就是只取原子。对于更大的东西，比如蛋白质，我们将其分块成氨基酸。所以每个氨基酸一个标记。所以，不是句子的字符、字母，而是蛋白质的氨基酸。对于小分子，我们将其分块成单个原子。

We are actually just taking atoms. For bigger things like proteins, we chunk up into amino acids. So one token per amino acid. So instead of characters of of a sentence, letters of a sentence, we have amino acids of a protein. And then for the small molecule, we chunk it up just into its individual atoms.

所以我们有一个氨基酸序列和一个原子序列，我们把它们放在一起，这就是一个大序列。然后我们通过一个结构模型，比如AlphaFold 3，来输入它。AlphaFold 3使用变换器，但不同于大型语言模型中变换器用于一维字符序列、字母序列，这里我们使用所谓的配对变换器，它操作于所有这些分子元素的二维交互网格上。所以我们可以考虑每一个氨基酸、蛋白质的每一部分、小分子的每一个原子以及它们之间所有可能的交互。然后这创建了神经网络特征，这些特征条件化一个扩散模型。

And so we have a sequence of amino acids and a sequence of atoms, and we put them together, and that's one big sequence. And then we feed it through a structure model like AlphaFold three. And AlphaFold three uses transformers, but unlike in large language models where transformers are used on one dimensional sequences of of characters, of letters, here we use what we call a pair former, which operates on a two dimensional interaction grid of all of these molecular elements. So we can consider every single possible interaction that could occur between every amino acid, every part of the protein, and every atom of the small molecule, and everything in between. And then this creates neural network features which condition a diffusion model.

扩散模型是生成模型。我们可能从这些惊人的图像生成模型、视频生成模型中了解它们。与生成图像的像素不同，我们的扩散模型生成的是整个生物分子系统的三维原子坐标。

Diffusion models are generative models. We probably know them from these amazing image generative models, video generative models. And instead of generating the pixels of an image, instead our diffusion models are generating the three d atom coordinates of this whole biomolecular system.

而它们恰好就是有效的。

And it just so happens they work.

而它恰好效果出奇地好。你会得到这些惊人的结构预测，当你去实验室通过实验解析这些结构时——我们偶尔会这样做——预测结果会非常惊人，比如一个全新的结合口袋或新的作用机制。我们走进实验室，想要验证这些模型是否与现实有任何基础关联？而反馈回来的结果是：确实如此。

And it just so happens this works phenomenally well. You get these amazing structure predictions that when you go to the lab and experimentally resolve these structures and, you know, we do this on occasion, something amazing is predicted like a completely new pocket or a new mechanism of action. We go into the lab. We wanna check, are these models grounded at all in reality? And what comes back is like, yeah.

这精确到一埃以内，你知道，这是最小的距离单位，精确度惊人。

This is like within one angstrom, like, you know, the tiniest unit of distance accurate, which is phenomenal.

是的。但通过那个训练过程，它是否设法提取了对化学工作原理的概念性理解？

Yeah. But then through that training process, does it sort of manage to extract a kind of conceptual understanding of how chemistry works?

在原子空间中思考概念非常困难，但我确实相信这些模型在分子和原子空间中进行了某种推理，因为我们从它们那里获得了大量的泛化能力。如果你考虑这些生成模型的训练目标，它们被训练来拟合你提供给它们的数据分布。在我们的案例中，我们提供了所有可能自然存在的分子，包括人们以前研究过和设计过的。当你的模型变得越来越好时，你会得到看起来像是之前可能被设计出来的东西，这已经超出了人类设计的可感知范围。

It's really hard to think about concepts in this atom space, but I do believe that there's some notion of reasoning in molecular and atomistic space that these models are doing because of the amount of generalization we're getting out of them. And if you think about what these generative models are trained to do, they're trained to fit to the data distribution that you give them. And so in our case, we give them all the molecules that might exist naturally that people have worked out before, that people have designed before. When your model gets better and better, you get things that look like they could have been designed before, which stops being perceptible from a human design.

你怎么看？

What do you think?

你确实可以从我们得到的分子中看到这一点。它们看起来像是我或其他人可能设计出来的分子。不过有时你确实会得到一些疯狂的东西。

You can certainly see that in the molecules that we get back. They look like molecules that myself or someone else might have designed. You do sometimes get something crazy though.

那么你呢？

So Do you?

是的，我们仍然会遇到这种情况。我们现在有办法过滤掉这些问题了。

Yeah. We still do. We kind of have ways of filtering that out now.

这某种程度上算是一种幻觉吗？

Is it sort of like a hallucination in a way?

是的，模型表现得非常自信，但它有点

Yeah, the model's like really confident, but it's kind

自信地错了。自信地错了。好吧。当它产生幻觉时是什么样子的？

of confidently wrong. Confidently wrong. Okay. What does it look like when it hallucinates?

你看，我们会得到一系列结构，模型认为这些结构是针对这个特定蛋白质口袋问题的良好解决方案。然后我们的工作是去筛选，好吧，我们应该选择哪些分子实际投入合成？所以这从来不是模型孤立运作的。这是一个模型与专家紧密合作的过程，专家会评估：在你提供的解决方案中，哪些是真正的精华？哪些分子能真正推动这个项目前进？

So, you know, we get back a number of structures that the models believe are, you know, good solutions for this problem, for this particular protein pocket. And then our job is to go, okay, well which of those molecules should we actually select to put into synthesis? And so this is never a model on its own in a silo. This is a model working really closely with an expert to say of the solutions you've given me, where's the gold in that? Where are the molecules that are actually going to push this project forward?

我们会将它们投入我们称之为化学合成的过程，制造并测试它们。但有时我们得到的结果显示化合物完全无法与蛋白质结合。这时模型基本上是完全产生了幻觉般的解决方案，甚至逼真到专家看了也会说‘是的，这看起来是个很好的方案’，置信度指标也会显示相同结果。所以这本质上就是一种幻觉。我们认为这是一个非常有趣的研究问题：如何从模型给我们的结果中筛选出真正有价值的东西？

And we'll put them into what we call chemical synthesis where we make them and we test them. But sometimes we get back results which actually the compound doesn't bind to the protein at all. So here the models essentially completely hallucinated a solution, so convincingly that actually an expert looks at it and goes, yeah, that looks like a really good solution, you know, confidence metrics would suggest the same. So it's essentially, it is a hallucination. And I think we find it a really fascinating research question to say, okay, how do we find the really good stuff that the model's giving us?

除了这个三维拼图或乐高（因为我们混合了隐喻），药物设计仅仅是关于结构吗？仅仅是找到能填补特定空缺的东西吗？还是说在设计药物时还有其他必须考虑的因素？

Beyond this three d jigsaw or Lego, because we're mixing our metaphors, is it just about structure? Is it just about finding something that will plug a particular hole? Or are there other considerations that you have to have as well when it comes to drug design?

还有很多其他考虑因素，这正是让这个问题变得极其复杂的原因。所以它甚至超越了形状，对吧？你可能有一个形状匹配的东西，但它必须与那个蛋白质紧密结合。而我们所说的结合亲和力预测实际上是不同的，你不能仅仅通过看图来评估这一点。

There are so many other considerations, that's what makes this problem just incredibly complex. So it's even beyond the shape, right? You can have something that maybe fits in the shape, but it's got to bind really strongly to that protein. And that prediction of binding affinity as we call it, is actually different. You can't really gauge that from just looking at a picture.

图像是一个有用的指南，但你需要能够单独预测那种结合亲和力。然后所有这些因素加在一起，就是你的分子如何与蛋白质结合的方式。你还必须考虑这个分子是否会与体内其他两万种蛋白质中的任何一种结合？因为如果答案是肯定的，那可能会导致你不想要的副作用，从而引发毒性。

The picture is as a helpful guide, but you need to be able to predict that binding affinity kind of separately. And then all of those things together, that's just how your molecule binds to your protein. You've also got to think about is that molecule gonna bind to any of the other 20,000 proteins in the body? Because if the answer is yes, that could drive a side effect that you don't want. That's gonna drive you some toxicity.

你知道，这个分子会稳定吗？它必须能在胃部极强的酸性环境中存活下来。它必须能经受住肝脏的代谢，就像穿越战区一样。肝脏会尽其所能地清除它，因为它觉得这是一个外来分子，需要将其排除。

You know, is this molecule gonna be stable? It's got to survive like the really acidic conditions of the stomach. It's gotta survive going through the liver, which is like going through a war zone. The liver wants to do everything it can. It's like, this is a foreign molecule, I need to like get rid of it.

所以你的分子必须非常坚固，能够挺过这段旅程。它还必须可溶。当你服药时，药片必须在胃中溶解，并且一路保持溶解状态直到肠道，否则它就无法被身体吸收。而且很多这些参数是相互矛盾的。为了让某物可溶并很好地溶解，它需要亲水；但为了与蛋白质结合并在那种口袋（乐高连接）中获得亲和力，它实际上需要疏水。

So your molecule's gotta be really robust, it's gotta survive that journey. It's gotta be soluble. When you take a pill, that pill's got to dissolve in your stomach, and it's got to stay dissolved all the way through your intestine, because otherwise it's not gonna absorb into your body. And a lot of these parameters are pulling against each other. So for something to be soluble and dissolve really well, it needs to be water loving, but for it to bind to the protein and to gain affinity in that kind of pocket, that Lego connection, it actually needs to be kind of water hating.

那么，如果你需要相反的特性，你如何可能解决这个问题呢？

So how do you possibly solve that, if you need opposing characteristics?

直到现在，在某种程度上，药物发现仍然是一个非常迭代的过程。人脑一次只能考虑这么多事情，所以你会想，好吧，我先解决一下结合亲和力的问题。然后我开始考虑，我的分子可溶吗？然后我再逐渐引入这些其他属性。说实话，这就像打地鼠游戏一样。

Up until this point and still now to some degree, drug discovery is a very iterative process. The human brain can only think about so many things at one time, so you're like, okay, I'm gonna solve this binding affinity problem a bit. And then I'm gonna start to think, okay, is my molecule soluble? And then I'm gonna start to bring in gradually these other properties. And it's honestly, it's like whack a mole.

这就像玩一场持续三年的打地鼠游戏，你解决了这个问题，欢呼一下。然后另一个问题又冒出来，你说好吧，我来解决那个，但之前的问题又变糟了。所以要在计算机模拟中预测所有这些情况，仍然是一个非常非常困难的问题。你们是否正在开发工具来

You play like this like three year game of whack a mole where you're like, okay, fix this problem, hooray. And then and then this other one pops up and you're like, okay, I'll fix that, but then the other one's gone bad again. So to be able to predict all these things in silico, it's still a really, really hard problem. Are you working on tools that

帮助解决这些方面的问题？

will help with those elements too?

是的，完全正确。在ISO，我们思考如何端到端地进行药物设计，这意味着要解决所有这些非常困难的问题：细胞渗透性、溶解性、毒性、肝脏清除率等等。这些问题都还没有解决，甚至连建模都非常困难。所以我们投入大量精力创建新模型来更好地理解这些。

Yeah, absolutely. So we think about at ISO, how do we do drug design end to end, which really means solving all of these very, very hard problems with cell permeability, solubility, toxicity, liver clearance, everything. And none of this is solved. These are really hard problems to even model. So we spend a lot of effort on creating new models to really understand this better.

然后正如贝基谈到的，我们如何开始在干草堆中寻找这些针尖般的分子，它们恰好平衡了所有属性，成为完美药物。这真的非常非常困难。实际上，这与我之前在DeepMind的工作有很多相似之处，比如在《夺旗》或《星际争霸》中，没有单一的策略能解决所有问题。你必须完全混合这些策略，为每个策略找出利用方法，基本上是在搜索这个巨大的组合策略空间。同样地，我们需要搜索这个巨大的分子组合空间。

And then as as Becky was talking about, how do we then start to find these needles in a haystack molecules that are somehow just balancing the properties just right to be a perfect drug. And it's really, really hard. And actually, there there are a lot of analogies to maybe what I used to do at DeepMind in, for example, Capture the Flag or StarCraft is there's not just one agent or strategy that solves StarCraft or a particular game like Go. You have to completely start mixing up these strategies and working out exploits for each individual strategy and basically searching this huge combinatorial strategy space. In the same way, we need to be searching this huge combinatorial molecule space.

就像在围棋游戏中可能有树搜索，每一步你都推演出其他可能的走法，然后开始深入搜索这个策略树。就像你可以为围棋走法这样做，你可以想象为设计分子做类似的事情。所以你从分子的一个部分开始，假设可以添加或移除什么，然后得到一整棵可能未来的树，你可以评分并计算出与之相关的价值，从而为这个特定适应症创造出完美分子。

So just like you might have tree search in a game of Go, where at every move you elucidate some other possible moves and you start searching through that tree of possible strategies going deeper and deeper. And just like you can do that for moves in a game of Go, you can imagine doing a similar thing for designing a molecule. So you start with a part of a molecule and you start to hypothesize what are the different things I could add or take away from this molecule? And you get to a whole tree of possible futures that you can then score and work out a value associated with that, to create that perfect molecule for this very specific indication.

但你怎么知道完美分子存在呢？也许蛋白质的某些裂缝就是无法适配。

But how do you even know that the perfect molecule exists? Like maybe there's just some crevices in the proteins that just are unfittable.

我们确实有'不可成药蛋白质'的概念，那些裂缝非常平坦，无法让任何东西抓住。这些蛋白质可能需要不同的解决方案。实际上我们有一个新兴领域叫做分子胶，就是让两个蛋白质结合在一起，它们结合时形成的口袋实际上更合适。所以现在你需要设计一个分子坐在它们中间，把它们粘在一起。

We do have this concept of like undruggable proteins, where the crevice is like really flat, you can't really get anything to grip in there. And those proteins might need different solutions. So actually we have a whole emerging field which we call molecular glues. And this is where, you know, you have two proteins that come together and the pocket that's formed when they come together is actually a much more suitable pocket. So now you need to design a molecule that sits in the middle of them and glues them together.

所以现在各种不同的模式都爆发式涌现，这让这个领域变得极其令人兴奋。

So there's like this whole explosion of all these different modalities now, which makes this an incredibly exciting field to work in.

从我的角度来看，考虑到问题的难度和复杂性，我们实际上已经找到了一些药物，这本身就令人充满希望。我们基本上一直依靠一点人类直觉和大量随机筛选实验测试，尽管设计空间巨大，我们还是设法找到了分子。这实际上给了我很多希望，因为这意味着化学空间可能存在大量冗余。可能有很多不同的解决方案都可行，我们只需要找到它们。

From my perspective, the fact that we've actually found any drugs at all already, given how hard and complex the problem is, and we've basically been doing a bit of human intuition and a lot of random screening experimental testing, and we've managed to find molecules even though the design space is huge. Actually, that gives me a lot of hope because that means that there's probably a lot of redundancy in chemical space. I there's, like, probably lots of different solutions that could work, but we've just gotta find them.

你们实际上尝试制造过这些分子吗？还是目前它们只存在于屏幕上？

Have you actually tried to make any of these molecules? Or do at the moment, do they just exist on the screen?

哦，不，我们制造了很多分子。我们有一个庞大的实验体系。是的。

Oh, no, we make a lot of molecules. We've got a huge experimental footprint. Yeah.

它们的结果符合你们的预期吗？我是说，结果怎么样？

And do they turn out how you expect? I mean, what are the results like?

我们在一些项目中取得了令人难以置信的成功，我会在马克斯桌前找到他，然后说，你看到这个东西了吗？我们俩都会对此感到非常震惊。

So we've had like some incredible success in some of our projects where I'll find Max at his desk and I'll be like, have you seen this thing? And it would just both be like really kinda mind blown about it.

当你得到那个分子时，它确实有效。

That when you get the molecule, it actually works.

是的，没错。我们有自己的药物设计项目，这些都是我们自己从零开始研发的。我们还与制药公司合作伙伴合作，比如礼来和诺华。在这些合作中，你会得到具体的目标来开展工作。

Yeah. Exactly. And we have our own drug design programs, so things that we've started from scratch ourselves. We also work with pharma company partners, people like Eli Lilly and Novartis. In these collaborations, you'll get specific targets to work on.

这些是这些公司高度确信并可能有大量证据支持的目标。我们参与的一些合作中，被分配了非常、非常困难的目标。要知道，这些是有些人有时已经研究了十多年却未能取得重大进展，以至于无法将产品推向市场的东西。

These are ones that these companies have high conviction behind and probably a bunch of evidence behind. Some of the collaborations we're in, we've been given very, very hard targets. Know, these are these are things that people have worked on sometimes for over a decade and not made significant progress to the point where you've got something on the market.

比如癌症之类的东西。

Things like cancer and that sort of stuff.

涵盖了一系列治疗领域和疾病领域。然后贝基和团队坐下来，开始用这些模型进行设计，能够开始为全新的机制找到全新的化学物质，这是以前没有人真正发现过的，对我这个计算机科学家来说真是令人惊叹。

A whole host of therapeutic areas and disease areas. And then Becky and team sit down, start designing with these models, and can start finding completely novel chemical matter for completely novel mechanisms that that no one's really discovered before, which is mind blowing for me as a computer scientist.

是的，这对我来说也令人惊叹。有一些可以说是职业生涯定义的时刻，你知道，人工智能会给你一个假设。对吧？它会建议一些东西，你会想，我不确定我会那样做，但你知道，模型告诉我这个，而且它对这个东西相当确信。

Yeah. That's mind blowing for me. There's been some, like, career defining moments where, you know, the AI will give you a hypothesis. Right? It will it will suggest something and you think, I'm not convinced I would do that, but you know, the model's telling me this thing and it's really quite convinced about this thing.

所以也许我应该测试这个假设。然后实际上发现模型是对的，你测试它是完全正确的，它真的推动了你的项目甚至整个领域的发展。所以我认为对我来说，不是关于我们如何信任模型，而是关于我们如何对它们提出的假设持开放态度进行测试，而不是说，哦，这不符合我的世界观，所以我就不测试它。

So I should maybe just test this hypothesis. And then actually it turns out that the model was right and you were absolutely right to test it and it's really pushed forward your project or even like that kind of field. So I think for me, it's not about how we trust the models, it's about how we are open to testing the hypotheses that they put in front of us and not sort of going, oh, that doesn't fit with my world for you, so I'm not gonna test it.

但你们也是人类，对吧？所以我确实想知道，如果你看到模型有很多成功的发现，如果模型连续提出很多好东西，你是否会开始可能比更信任它而不是自己？

But then you are also human, right? So I do wonder whether if you see lots of hits with the model as it were, if the model is coming up with lots of good stuff in a row, do you start sort of maybe trusting it more than yourself?

实际上我们对模型给予了很大的信任。比如我们有些模型采用了相当严格的截断值。

We actually put a lot of trust in the models. For example, some of the models we have, we use the most quite strict cutoffs.

什么样的截断值？

What kind of cutoff?

比如说，我们有个称为结合概率的模型，其数值范围从0到1。1表示模型确信你的分子一定会与蛋白质结合，0则表示模型判定完全不会结合。随着时间推移建立起一定信心后，我们发现低于0.7的成功概率非常低。因此我们将其设定为截断值，决定不在实验室测试任何概率低于此值的分子——因为模型很可能是对的，这些分子大概率效果不佳。

Like for example, we have a model which we call binding probability and it goes from zero to one. So one is the model is convinced your molecule is definitely going to bind to your protein. Zero, the model is telling you this is definitely not gonna bind. And you know, when you can build a little bit of confidence over time that the model really does understand, anything below 0.7 is really got very low probability of success. So we just define that as a cutoff and be like, we're not gonna put anything in the lab that's got a probability of less than this because actually the model's quite likely to be right, it's probably not gonna be any good.

这对化学家来说挺难的，当你设计出一个自以为绝妙的分子时，却看到模型不认可它。

And that's quite hard because as a chemist, you design something and you think that was a really clever idea that I just came up with. And like, why doesn't it like it?

但如果模型生成难以理解或不合理的结果呢？它需要自我解释吗？

But then what if it makes something that you don't understand or that doesn't make sense? I mean, do you does it need to explain itself?

目前可解释性确实很重要，因为整个过程仍由人类主导，尚未实现端到端自动化。我们需要介入并决定后续步骤，若缺乏可解释性，就无法确定下一步方向。

I think at the moment, that explainability is quite important for now because the process is kind of quite driven still by the human. It's not end to end yet. We have to go in there and we have to sort of say what what comes next? So if there's no explainability there, you don't know what would be next. Right?

这样的工作模式会非常困难。可以想象未来实现更端到端的流程时，比如模型能一步生成药物分子。

And so that would be very difficult to work with. I can imagine like in a future state where actually the process is a bit more end to end, like in one step, the Here's model a drug.

这是一种药物。

Here's a drug.

那么实际上，也许你并不需要那种可解释性。但当你需要以人类身份介入、进行迭代并施加更多方向性指导时，可解释性就变得重要了。这就是我认为AlphaFold模型的价值所在——当模型预测某个分子有效时，我可以通过实际观察来验证其合理性，并清楚下一步该怎么做。

Then actually that maybe you don't need that explainability. But when you've gotta go in there as a human and you've gotta iterate and you've gotta do a bit more of that directionality, then that explainability is important. And that's where I think for me, the alpha fold models really come in because, okay, models predicting this molecule is gonna be good, I can rationalize that with what I'm actually seeing, I know what I'd do next.

你对可解释性的看法略有不同，是吗？

You have a slightly different view on explainability, don't you?

我确实对可解释性有不同看法，但本质上我认为只有当模型不够好时才需要可解释性。目前我们还没有完美的模型，因此可解释性仍有很大存在空间。但每当我听到对可解释性的呼吁时，总觉得我们应该优先改进模型本身。有趣的是，可解释性确实能帮助理解模型的缺陷和偏见，根据已知科学知识找出错误所在，从而不断修补优化，最终实现完全基于计算机的端到端药物设计，或许只需在最后进行一轮实验室验证。

I do have a slightly different view on explainability, but I think you need explainability when your model sucks, basically. We don't have perfect models yet, so I think, you know, there's there's a good amount of room for explainability. But I always, like, hear the call for explainability and think, look, we need we need to make this model better. And actually, the interesting thing about explainability is it can help you understand the pathologies that this model has, the biases that it has, where is that wrong given the science that we know about? And so we can start patching that and make it better and better and get to this point where, yeah, actually, can just do end to end design purely in silico, and maybe just do a final round of verification in the lab at the end.

所以你的意思是直接说：为我设计治疗X疾病的药物，系统就能给出所需分子结构？你觉得这可能实现吗？

So I mean, you literally say, make me a drug for x disease, off it goes, says here's the molecule you need. Yeah. Yeah. Do you think that's possible?

我认为是可能的。所有趋势都指向这个方向。我们正在不断进步，已经减少了实验周期和实验室时间的需求，而这仅仅是个开始。

I think it's possible. I think everything is pointing in that direction. We're getting better and better. We're already reducing the amount of experimental cycles you need and reducing the amount of lab time you need. And, yeah, this is just the beginning.

确实非凡。不过我想这取决于是否清楚要靶向的蛋白质对吧？

Absolutely extraordinary. I mean, suppose it does it does depend on knowing what protein you're targeting to. Right?

是的。

Yes.

所以那些我们尚未完全理解的疾病仍然会很难攻克。

So the diseases that we don't have a full understanding of is still gonna be difficult.

是的，我们称之为靶点识别领域，你需要真正确定导致疾病的蛋白质。这实际上是药物发现中非常重要的一部分，因为如果你从一开始就没有找准正确的生物学靶点，即使设计出世界上最好的分子，当它进入人体时也无法达到预期效果。所以在靶点识别领域还有很多工作要做，我认为人工智能在这方面也能发挥重要作用。

Yeah, and that kind of, we call it target ID space, where you actually need to identify the protein that's causing your disease. It's actually like a really important part of drug discovery, because if you're not hitting the right biological target from the start, you can design the best molecule in the world. It's not gonna do what you want it to do when you put it into a human. So there's a lot to be done in that target ID space that I think AI has got a big role to play there as well.

人工智能在生物学领域的一大前沿就是真正理解疾病的驱动机制是什么。我们能否开始理解DNA中的突变如何转化为RNA表达的变化，这又如何改变蛋白质的类型和表达水平，这些蛋白质如何相互作用并构建成信号通路，以及这些信号通路的变化又如何改变疾病状态。当然，如果我们能开始理解这些环节，就能找出需要调控这个生物系统的位置。但所有这些都非常非常困难，该领域正在取得一些惊人的突破——更好地理解DNA，更好地理解这种转化过程，甚至通过理解蛋白质的相互作用来更好地构建这些相互作用网络。这正是我们在ISO进行的一些非常激动人心的前沿研究。

It's one of the big frontiers of AI for biology is really understanding what are those driving mechanisms of disease? Can we start to understand how mutations in our DNA translate into changes of expression of RNA and how that changes the type of proteins and expression levels of proteins, how those proteins interact with each other and build up into these signaling pathways and how changes in those signaling pathways change the disease states as well. And of course, if we can start to understand these bits, we can start to work out where do we need to modulate this biological system. But all of this is really, really hard and there's there's some amazing breakthroughs happening in the field, understanding DNA better, understanding this translation better, even through understanding how proteins interact, can we build up these interaction networks better? This is some of the really exciting frontier research that we're also doing at ISO.

所以你们也有团队在这个领域工作吗？

So you have a team working in that space as well?

是的，没错。我们有一整个计算生物学团队，一整个专注于这个领域的机器学习建模团队。是的。

Yeah, that's right. We have a whole computational biology team, whole machine learning modeling team focused in this space. Yeah.

那么个性化医疗呢？我的意思是，因为我觉得每个人在某些方面都是不同的。

But then what about personalized medicine? Mean, because I guess each person is different in some ways.

嗯。我的意思是，你知道，这是一个真正激动人心的潜在未来，我们可以更深入地了解，例如癌症、个体肿瘤中的突变，并通过生成式人工智能和设计代理，能够设计出专门针对这类突变的分子。现在存在一个完整的问题，即我们如何实际运作这一点，将这些药物送到患者手中并批准这个框架，但我们正在朝着这项技术可能实现的方向迈进。

Mhmm. I mean, you know, this is the really exciting potential future where we can understand much more about, for example, cancer, individuals' mutations in their tumour, and through generative AI and design agents, be able to come up with molecules that work specifically for these sort of mutations. Now there's a whole question of how do we actually operationalize that and get these drugs to patients and approve this framework, but we're moving towards a place where that technology could be potentially there.

我的意思是，我在这里想到化疗药物，它们伴随着非常严重的副作用。你认为这类事情真的有希望吗？

Mean, I'm thinking here about chemotherapy drugs, come with really devastating side effects. You think there's real hope on the horizon for that kind of thing?

是的，当我们考虑化疗药物时，它们基本上是非特异性的药物。它们进入体内，试图阻止快速的细胞增殖。但我们现在有能力思考实际的特定目标，即我们想要抑制的蛋白质靶点，我们想要阻止其功能。这可能产生相同的效果，但你不是普遍使用对快速分裂细胞非常有毒的东西。是的，它会阻止你的肿瘤细胞分裂，但它也会阻止你胃和肠道的细胞，让你感到恶心和不适，它会阻止你的毛囊，你会掉头发。

Yeah, think when we think about chemotherapy drugs, they're basically drugs that are like non specific. So they're going into the body and they're kind of trying to halt that rapid cell proliferation. But what we have now is an ability to think about actually what's the specific target, the protein target that we want to inhibit, we want to stop its function. And that might have the same effect but you're not just generally using something like very toxic to rapidly dividing cells. Yes, it's gonna stop your tumour cells dividing but it's also gonna stop the cells that line your stomach and your intestine, gonna make you feel nauseous and sick, It's gonna stop your hair follicles, you're gonna lose your hair.

然而，我们现在知道我们可以进入，针对一个非常特定的蛋白质，即真正导致疾病的那个。如果你抑制那个特定的蛋白质，希望如果你做对了，它不会引起所有这些其他副作用。

Whereas we now know we can go in, we can target a very specific protein, the one that's actually causing the disease. And if you inhibit that particular protein, that's not gonna cause, hopefully, if you get it right, that's not gonna cause all these other side effects.

你还可以做一些非常酷的事情，针对特定细胞。所以如果你知道某种特定细胞类型在其表面表达某种东西，你可以开始编程像抗体这样的东西进入并找到那些特定的受体。这样你就可以将你的有效载荷直接递送到那种特定细胞类型，而不是更广泛地到全身。

You can do some also really cool stuff of targeting particular cells. So if you know that like a particular cell type is expressing something on its surface, you can start programming things like antibodies to come in and and find those particular receptors. And so you're delivering your payloads directly to that particular cell type and not more broadly to the body.

我想回到你之前提到的观点，贝基，关于一旦你有了药物设计，然后将其用于人体，就会出现所有其他潜在问题。因为，我的意思是，以前有过这样的例子，药物被制造出来，看起来非常好，但一旦实际用于人体，就会引起一些重大问题。我记得有一种药物，人们对其对疼痛的影响非常兴奋，但结果发现那种蛋白质对于确保心脏持续跳动也相当关键。你如何减轻这种风险，或者在这个阶段还不能？

I just wanna go back to the point that you made earlier, Becky, about once you've got the drug design, then once put it into the human, there's all of these other potential problems. Because, I mean, there have been examples of this before, where drugs have been made and looked like they were very good, and then once you actually put it into a human, it causes some massive problem. I think there was one which people were very excited about the impact it was gonna have on pain, but it turned out that protein also was quite crucial to making sure your heart kept beating. How do you mitigate against that, or can you not at this stage?

嗯，我们面临的问题之一是，我们经常使用动物模型将事物转化到临床，而动物模型实际上并不能很好地复制人类生理学。当我们在发现和临床前空间工作时，也就是进入人体之前的所有空间。我们使用不同的动物模型，这些模型可能模拟我们感兴趣的疾病。我们在寻找在这些动物模型中有效的分子。我们必须证明它们在那些动物模型中没有毒性。

Well, of the problems we have is that we often use animal models to then translate things into the clinic and animal models, they don't replicate human physiology very well at all actually. When we're working in the kind of discovery and preclinical space, which is all of that space before you go into human. We're working with different animal models, which might model the disease we're interested in. And we're looking for molecules which have an effect in those animal models. And we have to show that they're not toxic in those animal models.

然后我们利用那套证据资料去找药品监管机构，说：好了，我们准备好进入人体试验阶段了。但从那一步到上市，失败率高达百分之九十。哇，百分之九十。所以之前的所有投入，那可是巨大的。

And then we use that bank of evidence to go to the drug regulatory bodies and say, right, we're ready to go into a human. But from that point until the market, there's a ninety percent failure rate. Wow, ninety percent. So all that investment up to that point, which is huge.

是什么导致它们这么容易失败？

What makes them so likely to fail?

分子在临床试验中会因为毒性问题失败，也会因为疗效不足失败。我认为很大程度上归根于我们使用的动物模型——它们并不擅长模拟人类生理机制。

So molecules fail in the clinic for toxicity, they fail in the clinic for lack of efficacy. And I think a lot of it comes back to, the animal models we use, just are not very good at replicating human physiology.

因为老鼠和人类不一样。

Because a mouse is different to a human.

老鼠确实不同，所以我们能治好老鼠的疾病，可能在这方面还挺在行。

A mouse is different, so we can cure mouse disease, probably be quite good at that.

我们在工作中已经有很多导航了。是的。

We've got loads of navigation at work. Yeah.

是的。但没错，这种转化正是科学中我们需要解决的重要部分。

Yeah. But yeah, that translation is a big part of science that we need to fix.

AI在这里也能提供帮助吗？我的意思是，如果动物模型是这种成功率极低的绊脚石，你们能对此做些什么呢？

Can AI help here as well? I mean, if animal models are this sort of stumbling block with such a low level of success, what can you do about it?

嗯，这正是我们可以实际运用我们一直在开发的一些技术和模型的地方，然后思考，好吧，我们如何能更好地理解毒性，更好地理解对人类细胞的影响，并观察这如何转化到器官层面。如果你考虑一些脱靶效应，可能有很多药物进入临床试验后，它们击中了你的目标靶点并治愈了疼痛，但同时却击中了心脏中另一个蛋白质的靶点，从而停止了心脏功能，这就是脱靶效应。

Well, this is this is where we can actually use some of the technology and models we've been developing and and think about, okay, how can we understand toxicity better, understand the effect on human cells better and see how that translates to organs. If you think about some of these off target effects, probably there are many drugs that you go into the clinic and you're hitting your target of interest and it's curing your pain, but then it's hitting another target that's in another protein in your heart and stopping the function of your heart, that's an off target effect.

我的意思是，副作用在广义上就是一种脱靶效应。

I mean, side effects in general are sort of off target effects.

是的，没错。但你可以想象，如果我们一直在构建能够很好地理解这个分子如何与你的目标靶点相互作用的模型，你也可以问这个问题：这个分子如何与人体中所有其他靶点相互作用？你知道，所有两万个蛋白质。然后你可以开始构建这个分子，也就是你的药物分子，在全身产生的相互作用指纹。这样就能给你提供线索，甚至可能是关于这个分子毒性或副作用的具体信号。

Yeah, exactly. But you can imagine that if we've been building models that understand really well how this molecule interacts with your target of interest, you could also ask the question, well, how does this molecule interact with every other target in the human body? You know, all 20,000 proteins. And you can start building up this fingerprint of interactions that this molecule, your drug molecule, is having across the body. And so that can give you clues, maybe even concrete signal into the toxicity or side effects of this molecule.

而好处是，我们可以在你进入人体试验之前，实际上是在设计过程的最最早期就获得这个信号。所以当你完成了所有分子设计，到达你想要进入人体试验的阶段时，你已经以一种非常理性的方式思考这些副作用很长时间了。因此，希望实际上遇到其中一些副作用的几率会大大降低。

And the nice thing is we can get that signal not when you're going into humans, but actually at the very, very beginning of the design process. So by the time you've gone through all of your molecule design and you get to the point where you're like, yeah, I want to go into humans. You've been thinking about these side effects in a very rational way for a long time. And so hopefully, chances of of actually hitting some of those radically reduces.

你是在拿你的乐高积木结构或拼图，然后尝试它与人体内可能遇到的每一个其他可能的组合。

You're taking your structure of LEGO bricks or your jigsaw, and you're just trying it with every other possible combination that it might encounter in a human body.

是的。我们将尝试所有可能的乐高组合。

Yeah. We're gonna make every possible LEGO combination.

那么，如果那是设计阶段的话，贝基，我的意思是，还得把这个推进临床试验。如果可以的话，请给我们讲讲临床试验的过程。

Well, if that's the design stage then, Becky, I mean, also have to put this into clinical trials. Just talk us through the process of clinical trials, if you could.

所以，你的分子首次进入人体时，那就是一期临床试验。通常会涉及少量患者，其中一些可能实际上是健康志愿者，他们不一定患有相关疾病。你要观察的是，你的药物是否真的达到了在患者体内产生效果所需的暴露水平？以及药物是否耐受良好？

So the first time that your molecule ever goes into a human, that's a phase one clinical trial. So it'll be a small number of patients. Some of those might actually be healthy volunteers, they don't necessarily have the disease interested in. And what you're looking to see is, does your drug actually reach the level of exposure in the patient that would be needed to generate an effect? And is the drug well tolerated?

或者你是否突然开始看到一些未曾预料到的副作用？如果一切顺利，你会进入二期临床试验，这时你将针对实际患有该疾病的人群，并且样本量更大。你真正要回答的问题是，你的分子是否真的对你关注的疾病有效？而这里正是我们看到高失败率的地方。所以，进入二期的分子中有百分之七十实际上无法进入三期。

Or do you suddenly start to see some side effects that you weren't anticipating? If all is good, you'll proceed to a phase two clinical trial, which is now you're going into people who actually have the disease and you're going into larger numbers. You're really looking to answer the question, does your molecule actually have efficacy against the disease that you're interested in? And And this is where we do see that big failure rate. So seventy percent of molecules going into phase two, don't actually pass through into phase three.

对于那些确实进入三期的药物，你将面对更庞大的患者群体，观察药物在这个更大群体中是否有效。它不仅需要安全，还必须优于标准治疗方案。医生要真正开处方给患者，他们必须确信这种药物比目前使用的更好或更安全。

For those that do pass into phase three, that's where you're going into much bigger patient populations, seeing if your drug is effective across that bigger population. It's not just gotta be safe, but it's gotta be better than the standard of care. For doctors to actually prescribe this to their patients, they've got to say, this drug is better or this drug is safer than what I currently use.

正如你所说，这整个过程的失败率高达百分之九十。是的。我的意思是，这是否意味着在这个领域工作的人有些从未成功过？

And this whole thing has like a ninety percent failure rate, as you said. Yeah. I mean, does that mean that there are people who work in this space who never succeed?

没错，我是一名药物化学家，我们经常听到这个数字：实际上每20名药物化学家中只有一人能将一种药物推向市场。所以，每20人中有19人在整个职业生涯中从未成功让药物上市。是的，我们是一个习惯于面对重大失败的职业。

Yep, so I'm a medicinal chemist and we often have this number where actually only one in twenty medicinal chemists will ever get a drug to market. So 19 of us out of every 20 will never get a drug onto the market through our careers. So yeah, we are a profession where we're used to seeing significant failure.

你们对失败如此坦然，我们从中学到... 非凡，真是难以想象。你认为第一款由AI设计的药物上市还需要多久？因为所有这些额外的阶段确实需要时间，对吧？

You're comfortable with failure as We a learn from Extraordinary, extraordinary to imagine. How long do you think it will be until the first AI designed drug is on the market? Because all of these additional levels really take some time, don't they?

所以现在已经有AI设计的药物进入临床阶段进行试验了，这些现有药物中AI的参与程度各不相同。我想在未来五年左右，我们会看到其中一种药物获得批准。但对我来说，最重要的是AI何时能真正充实这条管线，快速将药物推入临床，并真正开始为患者提供分子药物？那才是AI产生真正重大影响的时刻。

So there's AI designed drugs that are in the clinic now in clinical trials, the different levels of AI input into those current drugs. I would imagine that in the next five years or so, we're going to see an approval of one of those medicines. But for me, the big thing is going to be when can AI start to really fill out this pipeline and start to get drugs into the clinic really quickly and really start to deliver molecules for patients? That for me will be when AI is having a really big impact.

就是当你能说'这是靶点'，然后它就能在最后'蹦'出一种药物的时候。

When you can start to say, here's the target, and then it sort of pops out a drug at the end.

没错，而且你可以直接将其投入临床。

Yeah, and you can put that straight into the clinic.

并且确信它不会对人造成任何伤害。

And be confident that it's not gonna cause any damage to a person.

即使比我们现在所处的信心水平只有略微提高，也会产生相当大的影响。

And even a slightly improved level of confidence in where we are now would be quite impactful.

是的。正如贝基所说，临床上已经存在被AI触及、以某种方式由AI赋能的分子的药物。我们只会看到越来越多这样的情况。五年后，不使用AI进行药物设计就会像做任何不使用数学的科学一样，你知道吗？而且我认为，对我们来说，整个科学领域都会是这样。

Yeah. Because as Becky said, there's already molecules in the clinic that have been touched by AI, that have been enabled by AI in some way. We're just gonna see more and more of that. In five years time, doing drug design without AI will be like doing any sort of science without maths, you know? And and to us, I think the whole of science will be like this.

就像，如果你不使用AI，你还在做什么呢？对吧？那里有太多信息可以获取。正如贝基所说，更重要的是，我们如何真正看到在我们能够攻克的疾病领域或最终能够解锁的靶点方面实现快速进展，也就是我们能够帮助的患者。这非常令人兴奋。

It's like, if you're not using AI, what what are you doing? Right? There's just so much information to be gained there. As Becky said, it's more like, how do we actually see that rapid increase in disease areas that we're able to tackle or targets that we're able to unlock ultimately, like patients that we're able to help. And that's super exciting.

太棒了。真的非常吸引人。Becky、Max，非常感谢你们加入我。

Amazing. It was really fascinating. Becky, Max, thank you so much for joining me.

谢谢你的邀请。我们玩得很开心。

Thank you for having us. It was so much fun.

是的。来到这里真的很棒。

Yeah. It's been great to be here.

我想我现在意识到，药物化学是世界上最艰难的工作之一。设计一种药物需要数年时间。即使进入临床试验，其中百分之九十都会失败。而且你的同事中只有二十分之一的人能亲眼看到他们的药物改善患者的生活。但奇怪的是，我认为这正是这个领域如此令人兴奋的原因。

I think I now realize that medicinal chemistry is one of the hardest jobs in the world. It takes years to design a drug. Even if you get it to clinical trials, ninety percent of them fail. And only one in twenty of your colleagues ever manages to see their medicine improving the lives of patients. But strangely, that is precisely what I think is so exciting about this space.

因为如果我们迄今为止所做的一切都像是在黑暗中工作，缓慢而费力地在广阔可能性领域中探索极其微小的区域，那么现在就像是有人打开了一盏探照灯。当然，我们离一个能解决所有疾病的大AI按钮还非常非常遥远。但这里有很大的改进空间，有很大的推动进展的余地，同时也有巨大的机会直接影响我们所有人的生活。您一直在收听由我——Hannah Fry教授主持的Google DeepMind播客。如果您喜欢这一期节目，请订阅我们的YouTube频道或在您喜欢的播客平台上留下评论。

Because if everything we've done up until now has effectively been like working in the dark, slowly, laboriously navigating the most infinitesimally small areas of the vast landscape of possibilities, it's like someone has just turned on a floodlight. And, okay, of course, we are still very, very far away from a big AI button that's just gonna solve all diseases. But there is so much headroom here for improvement, so much scope to move the dial, and simultaneously so much opportunity to directly impact the lives of all of us. You have been listening to Google DeepMind the podcast with me, professor Hannah Fry. If you enjoyed this episode, then do subscribe to our YouTube channel or leave a review on your favorite podcast platform.

当然，我们还有更多涵盖各种主题的节目即将推出，请务必关注。下次见。

And, of course, we have plenty more episodes on a whole range of topics to come, so do check those out. See you next time.