#299 – 德米斯·哈萨比斯：DeepMind

以下是与DeepMind首席执行官兼联合创始人德米斯·哈萨比斯的对话，该公司发布并构建了计算机史上一些最不可思议的人工智能系统，包括完全自学并在围棋上超越全世界任何人类的AlphaZero，以及解决了蛋白质折叠问题的AlphaFold 2。这两项任务在很长一段时间内都被认为几乎不可能完成。德米斯被广泛认为是人工智能乃至整个科学和工程史上最杰出、最具影响力的人物之一。能最终与他坐下来进行这次对话，对我来说是莫大的荣幸和快乐，我相信我们未来还会多次交谈。现在，快速花几秒钟介绍一下每位赞助商。

The following is a conversation with Demes Hassabis, CEO and cofounder of DeepMind, a company that has published and built some of the most incredible artificial intelligence systems in the history of computing, including AlphaZero that learned all by itself to play the game of go better than any human in the world, and AlphaFold two that solved protein folding. Both tasks considered nearly impossible for a very long time. Demos is widely considered to be one of the most brilliant and impactful humans in the history of artificial intelligence and science and engineering in general. This was truly an honor and a pleasure for me to finally sit down with him for this conversation, and I'm sure we will talk many times again in the future. And now, a quick few second mention of each sponsor.

请在描述中查看它们。这是支持本播客的最佳方式。我们有用于邮件营销的Mailgun，用于健康长寿的InsightTracker，用于补充剂的Onnit，用于招聘的Indeed，以及用于早餐的Magic Spoon。朋友们，请明智选择。现在进入完整的广告播报。

Check them out in the description. It's the best way to support this podcast. We got Mailgun for email campaigns, InsightTracker for longevity, Onnit for supplements, Indeed for hiring, and Magic Spoon for breakfast. Choose wisely, my friends. And now on to the full ad reads.

一如既往，中间没有广告。我尽量让这些广告有趣，但如果你跳过了，也请去看看我们的赞助商。我喜欢他们的产品，也许你也会喜欢。本节目由Cinch旗下的Mailgun赞助，这是一项电子邮件投递服务，我已经使用了很多很多年，它提供了一个API，允许你通过编程方式发送邮件。

As always, no ads in the middle. I try to make these interesting, but if you skip them, please still check out our sponsors. I enjoy their stuff. Maybe you will too. This show is brought to you by Mailgun by Cinch, an email delivery service that I've used for many, many years to have an API that allows you to programmatically send emails.

如果你不知道API是什么，简单来说，它是一种让程序或代码与服务交互的方式。他们为交易邮件和营销邮件都提供了API。这些术语是比我懂行得多的人使用的，但我认为交易邮件是指针对个人的特定邮件，这大概是我用过的。它是一种通过邮件通知特定人士他们在网站上所做任何事情的状态的方式。然后是营销邮件，就是你向很多人发送相同邮件的时候。

If you don't know what an API is, the point is it's a way for programs for code to interact with the service. You have an API for both transactional and marketing emails. Those are terms used by people much smarter about the stuff than me, But I think transactional means specific to the person emails, which is what I've, I guess, used. It's a way to email certain people to notify them about the status of whatever the heck they're doing on the website. And then there's marketing emails, which is when you send an email to a lot of people, like the same email.

我想交易邮件是针对个人所采取行动的高度定制化邮件，而营销邮件就像是推送给很多人的群发邮件。这两类都是人们常用的邮件方式。Mailgun是一项服务，让你能非常轻松地做这类事情。你可以访问 lexfreedman.com/mailgun 了解更多信息。本节目也由InsideTracker赞助，这是一项我用来追踪生物数据的服务，这些数据来自我的身体。

I guess transactional is super customized to an action that a person took, and marketing is like a push email that you sent to a lot of people. And both of those two categories of how people often use email. And so Mailgun is a service that makes it super easy for you to do that kind of thing. You can go to lexfreedman.com/mailgun to learn more. This show is also brought to you by InsideTracker, a service I use to track biological data, data that comes from my body.

他们的很多计划都包含

A lot of their plans that you

你可以选择的项目包括血液检测。为什么是血液检测？因为很多非常有用的数据来自你的血液，然后他们使用机器学习算法来分析这些数据。这包括血液数据、DNA数据，甚至来自你健身追踪器的数据，为你提供体内情况的清晰图景。这就是未来。

can send out for include blood tests. Why blood tests? Because a lot of really useful data comes from your blood, and then they use machine learning algorithms to analyze that data. So that includes blood data, DNA data, and even data from your fitness tracker to provide you a clear picture of what's going on inside your body. This is the future.

你生活中所做的任何决定都应基于你实体、你自身的数据。这意味着你的生物身体。也许有一天这也意味着来自你的大脑。将会有一个像Neuralink这样的脑机接口设备，从你的大脑收集数据，并能够建议你服用哪种补充剂、做出哪种饮食改变。你可以访问insidetracker.com/lex，在限定时间内，作为本播客的听众，你将获得特别优惠。

Anything you decide to do in your life should be based on data from your entity, from your being. That means your biological body. Maybe one day that means from your brain as well. There'll be a brain computer interface device like Neuralink that collects data from your brain and is able to make suggestions of what kind of supplements to take, what kind of diet changes to make. You can go to insidetracker.com/lex, and for a limited time, you get special savings for being a listener of this very podcast.

本期节目也由Onnit赞助，这是一家营养补充和健身公司。他们有一款名为AlphaBrain的产品，是一种有助于提升记忆力、思维速度和专注力的益智药。你可能通过乔·罗根先生知道它，这大概是我第一次听说Onnit的广告。实际上，乔的播客是我最早虔诚收听的内容之一。我总体上非常喜欢有声书和播客，但乔所展现的真实感立刻与我产生了共鸣。

This episode is also brought to you by Onnit, a nutrition supplement and fitness company. They have a thing called AlphaBrain, which is a nootropic that helps you with the memory, mental speed, and focus. You might know it because of mister Joe Rogan, which is probably the first time I heard an on it read. It's one of the first podcast actually, Joe's, that I listened to religiously. I'm a huge fan of audiobooks and podcasts in general, but there's something about the authenticity that Joe projects that immediately connected with me.

随着时间的推移，他保持好奇心和同理心，善于倾听，但也能够改变想法，对一些疯狂的观点保持开放态度。总之，我提到这个是因为乔·罗根体验向我介绍了许多这类产品，比如Onnit，它们很快成为了我生活的一部分。我想播客广告是有效的。总之，如果你访问lexfreeman.com/onnit，可以获得AlphaBrain的特别折扣。本期节目也由招聘网站Indeed赞助。

And over time, his ability to be curious and empathetic, a good listener, but also be able to change his mind, keep an open mind to some crazy ideas. Anyway, I mentioned that because the Joe Rogan experience introduced me to a lot of these kinds of products like Onnit that quickly became part of my life. I guess podcast ad reads work. Anyway, you can get a special discount on Alpha Brain if you go to lexfreeman.com/onnit. This show is also brought to you by Indeed, a hiring website.

在我过去领导的团队招聘中，我曾多次使用他们的服务。他们有Indeed InstaMatch功能，能立即为你匹配到简历符合职位描述的高质量候选人。我以前说过，现在再说一次：生活中很少有比你所交往的人更重要的事情了。我是一个从工作中获得很多意义和快乐的人，即使是在西尔斯百货女鞋部卖鞋的时候。当你尝试掌握新技能时，那个小社区里的人对你的幸福、效率以及个人成长都至关重要。

I've used them as part of many hiring efforts I've done for the teams I've led. In the past, they have Indeed InstaMatch that gives you quality candidates whose resumes at Indeed fit your job description immediately. I've said this before, I'll say it again, there's very few things in life as important as the people you surround yourself with. I'm somebody for whom the work I've done brings a lot of meaning and joy to my life, even when I sold shoes at Sears Shoes in the women's section. That little community that you have as you try to figure out this new skill, those people are so instrumental to your happiness and to your effectiveness, to your growth as a human being.

所以招聘真的、真的、真的非常重要，这就是为什么你应该使用最好的工具来完成这项工作。Indeed就是这样一个工具。他们为本播客的听众提供特别优惠，仅限时开放。请访问indeed.com/lex查看。本期节目也由一个经典但好用的产品赞助，Magic Spoon，一种低碳水化合物、生酮友好的麦片。

So hiring is really, really, really important, and that's why you should use the best tools for the job. Indeed is one such tool. They have a special offer for listeners of this podcast, only available for a limited time. Check it out at indeed.com/lex. This episode is also brought to you by an oldie but a goodie, Magic Spoon, a low carb keto friendly cereal.

他们从一开始就在支持我们。我真的很喜欢Magic Spoon。它给我的心灵带来了很多快乐。我不在乎这是否是你的菜。请买来试试，让它成为你的最爱，因为它真的超级好吃。

They were there from the beginning. I really love magic spoon. It brings so much joy to my heart. I don't care if this is not your thing. Please get it and make it your thing because it's freaking delicious.

它拥有麦片的所有美味，却没有任何负面成分，比如所有的糖。它零克糖，13到14克蛋白质，仅4克净碳水化合物，每份140卡路里。就像我说的，它是生酮友好的。太神奇了。我不明白它是怎么做到的。

It has all the deliciousness of a cereal without any of the negative stuff, like all the sugar. It has zero grams of sugar, 13 to 14 grams of protein, only four net grams of carbs, a 140 calories in each serving. It's like I said keto friendly. It's magic. I don't understand how it works.

我不明白它怎么会这么美味，但他们有很多口味。对我来说，花生酱绝对能排进前三，可能是我第二喜欢的，但迄今为止，我最喜欢的是可可味。Magic Spoon提供100%满意保证，如果你不喜欢，他们会退款。访问magicspoon.com/lex并使用优惠码lex，即可享受订单折扣。

I don't understand how it could be so delicious, but they have a lot of flavors. I would say peanut butter is up there for me, maybe top three. Maybe it's my second favorite, but by far, my favorite is cocoa. Magic Spoon has a 100% happiness guarantee, so if you don't like it, they'll refund it. Get a discount on your order if you go to magicspoon.com/lex and use code lex.

这里是Lex Friedman播客。如需支持我们，请查看简介中的赞助商信息。现在，亲爱的朋友们，有请Demis Hassabis。

This is the Lex Friedman podcast. To support it, please check out our sponsors in the description. And now, dear friends, here's Demis Hassabis.

我们先从一个有点私人的问题开始。我是你编写的一个AI程序吗？专门用来采访别人，直到我足够优秀来采访你？

Let's start with a bit of a personal question. Am I an AI program you wrote to interview people until I get good enough to interview you?

嗯，如果你真是的话，我会很佩服。我会佩服我自己。我认为我们还没达到那个水平，但也许你是来自未来的Lex。

Well, I'd be impressed if if you were. I'd be impressed by myself if you were. I don't think we're quite up to that yet, but maybe you're from the future, Lex.

如果你真的编写了我，你会告诉我吗？告诉一个负责采访的语言模型它实际上是——这是好事吗？

If you did, would you tell me? Is that a is that a good thing to tell a language model that's tasked with interviewing that it is in fact

AI？也许我们正处在一种元图灵测试中。可能不告诉你是个好主意，这样就不会改变你的行为。

AI? Maybe we're in a kind of meta Turing test. Probably probably it would be a good idea not to tell you so it doesn't change your behavior.

对吧？这是一种

Right? This is a kind

就像海森堡不确定性原理的情况。如果我告诉了你，你的行为就会不同。是的。也许这正是发生在我们身上的事，当然。

of Feizenberg uncertainty principle situation. If I told you, you'd behave differently. Yeah. Maybe that's what's happening with us, of course.

这是一个来自未来的基准测试，他们将2022年重演为AI尚未足够强大的年份，现在我们想看看，它能否通过？没错。如果我是这样一个程序，你觉得你能分辨出来吗？所以回到图灵测试的问题，你谈过衡量智能的基准。什么会是最令人印象深刻的事情？

This is a benchmark from the future where they replay twenty twenty two as a year before AIs were good enough yet, and now we want to see, is it gonna pass? Exactly. If I was such a program, would you be able to tell, do you think? So to the Turing test question, you've you've talked about the benchmark for solving intelligence. What would be the impressive thing?

你提到过赢得诺贝尔奖，一个AI系统赢得诺贝尔奖。但我仍然认为图灵测试是一个引人入胜的测试。图灵测试的精神是一个 compelling 的测试。

You've talked about winning a Nobel Prize, an AI system winning a Nobel Prize. But I still return to the Turing test as a compelling test. The spirit of the Turing test is a compelling test.

是的。图灵测试当然有着难以置信的影响力，图灵是我永远的偶像之一。但我觉得如果你回顾1950年的原始论文，读读原文，你会发现我认为他并不是想把它作为一个严谨的正式测试。我认为它更像是一个思想实验，如果你看论文的风格，几乎有点像他在写的哲学。而且你可以看到他没有非常严格地规定细节。

Yeah. The Turing test, of course, it's been unbelievably influential, and Turing's one of my all time heroes. But I think if you look back at the nineteen fifty papers, original paper, read the original, you'll see I don't think he meant it to be a rigorous formal test. I think it was more like a thought experiment, almost a bit of philosophy he was writing if you look at the style of the paper. And you can see he didn't specify it very rigorously.

例如，他没有规定专家或法官所具备的知识。也没有规定他们需要花多少时间来调查这个。所以如果你想把它变成一个真正的正式测试，这些是重要的参数。而且，你知道，根据某些标准，有人声称图灵测试在十几年前就已经通过了。我记得有人声称用一个非常普通的逻辑模型就做到了，因为他们假装它是一个孩子。

So for example, he didn't specify the knowledge that the expert or judge would have. Not having to how much time would they have to investigate this. So these are important parameters if you were going to make it a true sort of formal test. And, you know, some by some measures, people claim the Turing test passed several you know, a decade ago. I remember someone claiming that with a with a kind of very bog standard normal logic model because they pretended it was a it was a kid.

所以法官们认为这台机器是个孩子。这与一个专业的AI人士审问一台机器并知道它是如何构建的等情况非常不同。所以我认为，我们可能应该放弃将其作为正式测试，转而更多地采用通用测试，我们在各种任务上测试AI的能力，看看它是否在可能数千甚至数百万个任务上达到或超过人类水平的表现，并覆盖整个认知空间。所以我认为在当时，它是一个惊人的思想实验。而且1950年代显然只是计算机时代的黎明。

So the the judges thought that the machine, you know, was was a was a child. So that would be very different from an expert AI person interrogating a machine and knowing how it was built and so on. So I think, we should probably move away from that as formal test and move more towards general tests where we test the AI capabilities on a range of tasks and see if it reaches human level or above performance on maybe thousands, perhaps even millions of tasks eventually, and cover the entire sort of cognitive space. So I think for its time, it was an amazing thought experiment. And also 1950s, obviously, it was barely the dawn of the computer age.

所以他当然只考虑了文本。而现在我们有更多不同的输入方式。

So of course, he only thought about text. And now we have a lot more different inputs.

所以，也许更好的测试方法是泛化能力，也就是跨多个任务的表现。但我认为，像Gato这样的系统最终可能会表明，这种能力会映射回语言本身。因此，你可能通过沟通自己跨任务泛化的能力来展示这种能力，这实际上就是我们通过对话跳跃话题时所做的事情。最终，对话中蕴含的不仅仅是你对知识的调动，而是你在不同理解模式间的转换，这些模式最终映射到你在所有领域成功运作的能力。

So, yeah, maybe the better thing to test is the generalizability, so across multiple tasks. But I think it's also possible as as systems like God will show that eventually that might map right back to language. So you might be able to demonstrate your ability to generalize across tasks by then communicating your ability to generalize across tasks, which is kind of what we do through conversation anyway when we jump around. Ultimately, what's in there in that conversation is not just you moving around knowledge. It's you moving around, like, these entirely different modalities of understanding that ultimately map to your ability to to operate successfully in all

这些领域，你可以将其视为任务？是的。我认为，我们人类确实将语言作为主要的泛化交流工具。所以我们最终用语言思考，并用语言表达解决方案。因此，这将是一个非常强大的模式，用来解释系统、说明它在做什么。

of these domains, which you can think of as tasks? Yeah. I think, certainly, we as humans use language as our main generalization communication tool. So I think we end up thinking in language and expressing our solutions in language. So it's going be very powerful mode in which to explain, you know, the system to explain what it's doing.

但我不认为这是唯一重要的模式。我认为会有很多不同的方式来展示能力，而不仅仅是语言。

But I don't think it's the only modality that matters. So I think there's gonna be a lot of, you know, there's there's a lot of different ways to express capabilities other than just language.

是的。视觉、机器人技术、身体语言。是的。行动是所有这一切的互动方面。

Yeah. Visual Yeah. Robotics, body language. Yeah. Action is the interactive aspect of all that.

这些都是其中的一部分。

That's all part of it.

但Gato有趣的地方在于，它似乎在将预测推向极致，比如将任意序列映射到其他序列，并预测接下来会发生什么。预测似乎是智能的基础。

But what's interesting with Gato is that it's it's it's it's sort of pushing prediction to the maximum in terms of, like, you know, mapping arbitrary sequences to other sequences and sort of just predicting what's going to happen next. Prediction seems to be fundamental to intelligence.

而你预测的具体内容并不那么重要。

And what you're predicting doesn't so much matter.

是的。看起来你可以很好地概括这一点。显然，语言模型预测下一个词，而Gato则可能预测任何动作或任何标记。这真的只是个开始。

Yeah. It seems like you can generalize that quite well. So obviously, language models predict the next word. Gato predicts potentially any action or any token. And it's just the beginning, really.

它是我们迄今为止最通用的智能体，可以这么说。但你知道，它本身还可以比我们现在所做的规模大得多。而且，显然，我们正在这个过程中。

It's our most general agent, one could call it so far. But, you know, that itself can be scaled up massively more than we've done so far. And, obviously, we're in the in the middle of doing that.

但解决AGI（通用人工智能）的一大关键是创建基准测试，帮助我们越来越接近，也就是创建测试泛化能力的基准。有趣的是，艾伦·图灵是第一个，可能也是迄今为止唯一一个尝试——也许是从哲学角度——但确实在尝试制定一个可以遵循的基准的人。尽管它有些模糊，但仍然足够严谨，可以运行这个测试。我仍然认为，像图灵测试这样的东西最终将是真正让其他人类印象深刻的东西，以至于你可以有一个AI系统作为亲密朋友。为了让这个朋友成为一个好朋友，他们必须能玩《星际争霸》，还必须完成所有这些任务。

But the big part of solving AGI is creating benchmarks that help us get closer and closer, sort of creating benchmarks that test the generalizability. And it's just still interesting that this fella, Alan Turing, was one of the first and probably still one of the only people that was trying, maybe philosophically, but was trying to formulate a benchmark that could be followed. It is even though it's it's fuzzy, it's still sufficiently rigorous to where you can run that test. And I still think something like the Turing test will in at the end of the day, be the thing that truly impresses other humans so that you can have a close friend who's an AI system. For that friend to be a good friend, they're going to have to be able to play StarCraft, and they're gonna have to do all of these tasks.

给你拿啤酒，所以机器人任务，和你一起玩游戏，使用语言、幽默，所有这类事情，但最终都可以归结为语言。感觉上，不是在AI社区的层面，而是在通用智能对世界的实际影响层面，语言似乎将是

Get you a beer, so the robotics tasks, play games with you, use language, humor, all of those kinds of things, but that ultimately can boil down to language. It feels like in not in terms of the AI community, but in terms of the actual impact of general intelligence on the world, it feels like language will be the place where

它真正闪耀的地方。我认为是这样，因为它对我们来说是一种非常重要的输入输出方式。我觉得你是对的。我认为图灵测试及其背后的哲学思想是：机器能否模仿人类的行为？而且我认为这比仅仅语言和文本更广泛。

it truly shines. I think so because it's such an important kind of input output for us. I think you're right. I think the Turing test that what the the kind of the the philosophy behind it, is the idea of can can a machine mimic the behaviors of a human? And and I would say wider than just language and text.

然后，你知道，在行动和其他一切方面，创造力，所有这些事情，如果它能匹配或超越人类的认知能力，那么我认为我们就有了一个真正的智能。所以从这个角度来看，你是对的。我认为他确实制定了正确的那种框架。

Then, you know, in terms of actions and everything else, creativity, all these things, then if it can sort of match or exceed human cognitive capabilities, then I think we have a, you know, a true intelligence. So I thought from that perspective, you're right. I think he did formulate the right kind of setup.

我只是觉得，未来的AI系统回顾这次对话，思考图灵测试，并且到那时它们会知道是哪一年它们终于跨越了人类水平智能的门槛，然后想到我们人类仍然对整个问题感到困惑，这会是一种幽默。

I just I think there'll be a kind of humor in the AI systems of the future looking back to this conversation and thinking about the Turing test and also thinking about by that time, they would know which year they were finally able to sort of cross the threshold of human level intelligence and think how funny it is that we humans were still confused about this whole problem.

那已经

That has

那个问题早就解决了。好吧，向未来的AI智能体们问好。言归正传，回到你的经历上来，你最初是什么时候爱上编程的？

long ago been solved. Well, hello to the AI agents of the future. Anyway, so going back to your to your journey, when did you fall in love with programming first?

其实是在很小的年纪。你知道，我最初其实是游戏让我着迷。大约四岁开始学下国际象棋，后来大概八岁时用象棋比赛的奖金买了第一台象棋电脑。那是一台ZX Spectrum，当时在英国非常流行。这台机器很棒，因为它非常容易上手，我觉得它培养了英国整整一代程序员。

Well, it's pretty it's pretty young age actually. So, you know, I started off actually, games was my first love. So starting to play chess when I was around four years old and then it was actually with winnings from a chess competition that I managed to buy my first chess computer when I was about eight years old. It was a ZX Spectrum, which was hugely popular in The UK at the time. And it's amazing machine because I think it trained a whole generation of programmers in The UK because it was so accessible.

你知道，一开机就是BASIC提示符，马上就能开始编程。我父母对电脑一窍不通，但因为是用我比赛赢的钱，我可以坚持要买。然后我就去书店找编程的书，开始照着书上的代码敲进去。当然，一旦开始这么做，你就会开始调整代码，然后制作自己的游戏。

You know, you literally switched it on and there was the basic prompt and you could just get going. And my parents didn't really know anything about computers. So but because it was my money from a chess competition, I could I could say I I I wanted to buy it. And then, you know, I just went to bookstores, got books on programming, and started typing in, you know, the the programming code. And and then, of course, once you start doing that, you start adjusting it and then making your own games.

就是那时我爱上了电脑，意识到它们是非常神奇的设备。某种程度上，我当时可能无法解释清楚，但我觉得它们就像是思维的魔法延伸。我一直有这种感觉，也一直热爱电脑这一点：你可以让它们为你执行某个任务，然后去睡觉，第二天回来问题就解决了。这对我来说很神奇。

And that's when I fell in love with computers and realized that they were a very magical device. In a way, I kind of I don't would have been able to explain this at the time, but I felt that they were sort of almost a magical extension of your mind. I always had this feeling and I've always loved this about computers that you can set them off doing something, some task for you. You can go to sleep, come back the next day and it's solved. You know, that feels magical to me.

我的意思是，所有机器在某种程度上都是这样。它们都增强了我们的自然能力。显然，汽车让我们跑得比我们自己更快。但这是一台扩展思维的机器。当然，AI是机器可能做到或学习的终极体现。

So I mean, all machines do that to some extent. They all enhance our natural capabilities. Obviously, cars make us allow us to move faster than we can run. But this was a machine to extend the mind. And and then, of course, AI is the ultimate expression of what a machine may be able to do or learn.

所以很自然地，这个想法很快就延伸到了AI领域。

So very naturally for me, that thought extended into into AI quite quickly.

记得那个编程语言最初是在耶诞生的。

Remember the the programming language that was first started in Yeah.

专门针对机器的还是

Special to the machine or

是某种基础的东西。就是基础语言。我想就是 ZS Spectrum 上的基础语言。我不知道具体是哪种形式。后来我买了一台 Commodore Amiga，那是一台非常棒的机器。

was something that It basic. Was was just basic. I think it was just basic on the ZS Spectrum. I don't know what specific form it was. And then later on, I got a Commodore Amiga, which was a fantastic machine.

你现在就是在炫耀了。

Now you're just showing off.

是啊。嗯，我很多朋友有 Atari ST，但我设法搞到了 Amiga。它更强大一些，那真是太不可思议了。我过去用汇编语言编程，也用 Amos basic，这种特定形式的基础语言。实际上，那真的很棒。

So yeah. Well, lots of my friends had Atari STs, I I managed to get Amigas. It was bit more powerful, and and that was incredible. And I used to do programming in assembler and and also Amos basic, this this this specific form of basic. It was incredible, actually.

所以我所有的编程技能都是这样学来的。

So I learned all my coding skills.

那你是什么时候爱上人工智能的？你最初是什么时候开始意识到，你不仅能编写程序在你睡觉时为你做一些数学运算，还能创造出某种类似于赋予实体生命的东西，能够解决比简单数学运算更复杂的问题？

And when did you fall in love with AI? So when did you first start to gain an understanding that you can not just write programs that do some mathematical operations for you while you sleep, but something that's a keen to bringing an entity to life, sort of a thing that can figure out something more complicated than a than a simple mathematical operation?

是的。对我来说有几个阶段，都是在很年轻的时候。首先，在我努力提高国际象棋水平时，我担任过多个英格兰青少年国际象棋队的队长。大概在我10、11岁的时候，我打算成为一名职业棋手。那是我最初的梦想。

Yeah. So there was a few stages for me all all while I was very young. So first of all, as I was trying to improve at playing chess, I was captaining various England junior chess teams. And at the time when I was about, you know, maybe 10, 11 years old, I was gonna become a professional chess player. That was my first thought.

所以那个梦想就是，当然，当然。努力达到最高水平

And So that dream was there to Sure. Sure. Try to get to the highest level

的国际象棋水平。我在大约12岁时达到了大师级标准。当时我的世界排名仅次于朱迪特·波尔加，她后来成为了一位杰出的棋手并获得了世界女子冠军。在努力提高棋艺的过程中，你显然首先要努力改进自己的思维过程。这自然会引导你去思考“思考”本身。

of chess. So I was, you know, I got to when I was about 12 years old, I got to master standard. I was second highest rated player in the world to Judith Polgar, who obviously ended up being an amazing chess player and a world women's champion. And when I was trying to improve at chess, well, you know, what you do is you obviously, first of all, you're trying to improve your own thinking processes. So that leads you to thinking about thinking.

你的大脑是如何产生这些想法的？为什么会犯错？如何改进这个思维过程？但第二点是，这只是一个开始。那是在80年代早期，中期，国际象棋计算机刚开始出现。

How is your brain coming up with these ideas? Why is it making mistakes? How can you how can you improve that thought process? But the second thing is that you it was just the beginning. This is like in the in the '8 early eighties, mid eighties of chess computers.

如果你还记得，它们是实体棋盘，就像我们面前的这个，你需要按下方格。我记得卡斯帕罗夫有一个品牌版本，我也得到了一个。它们不像现在这么强大，但在当时已经相当厉害了，我们常用它们来练习开局和其他技巧。我记得我大概在11、12岁时得到了第一个，当时觉得这太神奇了。

If you remember, they were physical boards like the one we have in front of us and you press down the, you know, the the squares. And I think Kasparov had a branded version of it that I I I got. And you were, you know, used to they're not as strong as they are today, but they were they were pretty strong and used to practice against them to try and improve your openings and other things. And so I remember, I think I probably got my first one was around eleven or twelve. And I remember thinking, this is amazing.

怎么会有人编程让这个棋盘下棋呢？我买了一本很有启发性的书，叫《国际象棋计算机手册》，作者是大卫·利维，大概是1984年出版的。所以我应该是在11、12岁时得到的，它详细解释了这些象棋程序是如何制作的。我记得我的第一个人工智能程序是在我的Amiga电脑上编写的。

You know, how how has someone programmed this this chess board to play chess? And it was very formative book I bought, which was called the chess computer handbook by David Levy. This came came out in 1984 or something. So I must have got it when I was about 11, 12 and it explained fully how these chess programs are made. And I remember my first AI program being programming my Amiga.

它的性能不足以运行国际象棋程序，我写不出完整的象棋程序，但我为它编写了一个玩奥赛罗（黑白棋）的程序——在美国有时也叫翻转棋。这是一个比国际象棋稍简单的游戏，但我运用了所有象棋程序的原则：阿尔法-贝塔搜索等等。那就是我的第一个人工智能程序。

It couldn't it wasn't powerful enough to play chess. I couldn't write a whole chess program but I I wrote a program for it to play Othello Mhmm. Reversi, it's sometimes called I think in The US. And so a slightly simpler game than chess but I used all of the principles that chess programs had, alpha, beta, search, all of that. And that was my first AI program.

我记得很清楚。那时我大约12岁。那件事让我对AI产生了兴趣。第二部分是在我16、17岁的时候，当时我正在专业地编写游戏，设计游戏，写了一款叫《主题公园》的游戏，这款游戏将AI作为模拟游戏的核心玩法组成部分。它在全球售出了数百万份，人们很喜欢其中的AI，尽管以今天的AI标准来看相对简单，但它会根据你作为玩家的玩法做出反应。

I remember that very well. Was around 12 years old. So that that that brought me into AI. And then the second part was later on when I was around 16, 17 and I was writing games professionally, designing games, writing a game called Theme Park, which had AI as a core gameplay component as part of the simulation. And it sold millions of copies around the world and people loved the way that the AI, even though it was relatively simple by today's AI standards, was was reacting to the way you as the player played it.

所以它被称为沙盒游戏。它是与《模拟城市》一起最早出现的这类游戏之一，意味着你玩的每一局游戏都是独一无二的。

So it was called a sandbox game. So it's one of the first types of games like that along with SimCity, and it meant that every game you played was unique.

你能稍微岔开话题，从游戏设计和人类娱乐的角度谈谈真正令人印象深刻的AI吗？你在游戏中见过哪些真正令人印象深刻的AI？也许创建AI系统需要什么？这个问题有多难？所以有一堆问题。是的，就作为一个简短的小插曲。

Is there something you could say just on a small tangent about really impressive AI from a game design human enjoyment perspective? Really impressive AI that you've seen in games, and maybe what does it take to create AI system? And how hard of a problem is that? So a million questions Yeah. That would just as a brief tangent.

嗯，我认为游戏在我的生活中很重要，原因有三。首先，我小时候在玩它们，并通过游戏训练自己。然后我经历了一个设计游戏和为游戏编写AI的阶段。所以我专业编写的所有游戏都将AI作为核心组成部分。那主要是在九十年代。

Well, look, I think games games have been significant in my life for three reasons. So first of all, to to I was playing them and training myself on games when I was a kid. Then I went through a phase of designing games and writing AI for games. So all the games I I professionally wrote had AI as a core component. And that was mostly in the in the nineties.

我当时在游戏行业做这些是因为，我认为当时的游戏行业是技术的前沿。所以无论是像约翰·卡马克和《雷神之锤》那样的图形技术，还是AI，我认为所有的行动都发生在游戏中。我们现在仍在从中受益，甚至包括像GPU这样的东西，我觉得很讽刺的是，GPU显然是为计算机图形学发明的，但后来发现对AI非常有用。结果发现，整个世界似乎一切都是矩阵乘法。是的。

And the reason I was doing that in games industry was at the time, the games industry I think was the cutting edge of technology. So whether it was graphics with people like John Carmack and Quake and those kind of things or AI, I think actually all the action was going on in games. And we've seen we're still reaping the benefits of that even with things like GPUs, which, you know, I find ironic was obviously invented for graphics, computer graphics, but then turns out to be amazingly useful for AI. It just turns out everything's a matrix multiplication appear it appears, you know, in the way in the whole world. Yes.

所以我认为当时的游戏拥有最前沿的AI。很多我们参与编写的游戏，你知道的。有一款叫《黑与白》的游戏，我在早期阶段参与其中，我仍然认为它是计算机游戏中最令人印象深刻的强化学习范例。在那款游戏中，你训练一只小宠物动物，

So so I think games at the time had the most cutting edge AI. And a lot of the the the games we we you know, I was involved in writing. So there was a game called Black and White, which was one game I was involved with in the early stages of which I still think is the most impressive example of reinforcement learning in a computer game. So in that game, you know, you trained a little pet animal and

这是一款出色的游戏。

It's a brilliant game.

是的。而且它某种程度上是从你对待它的方式中学习的。所以如果你对它不好，它就会变得刻薄。

Yeah. And it sort of learned from how you were treating it. So if you treated it badly, then it became mean.

是的。

Yeah.

然后它就会对你的村民和你的人口，也就是你管理的那个小部落，变得刻薄。但如果你对它友善，它也会友善。人们对这种机制非常着迷，说实话，我也对它如何发展感到着迷。尤其是它映射善恶的方式。是的。

And then it would be mean to to your villagers and your and your population, the sort of little tribe that you were running. But if you were kind to it, then it would be kind. And people were fascinated by how that works and and so was I, to be honest with the way it kind of developed. And Especially the mapping to good and evil. Yeah.

我认为它让你意识到，让我意识到，你在某种程度上可以通过所做的选择来定义你的最终归宿，这意味着我们所有人都既有行善的能力，也有作恶的潜力。在通往那些目的地的轨迹上所做的每一个不同选择都至关重要。这很迷人。我的意思是，游戏能在哲学层面上对人产生这种影响，这很罕见。似乎确实很罕见。

I think made you made you realize, made me realize that you can sort of in the way in the choices you make can define the where you end up, and that means all of us are capable of the good, evil. It all matters in the different choices along the trajectory to those places that you make. It's fascinating. I mean, games can do that philosophically to you, and it's rare. It seems rare.

是的。嗯，我认为游戏是一种独特的媒介，因为你作为玩家，不仅仅是 passively 消费娱乐内容，对吧？你实际上是作为一个行动者积极参与其中。所以我认为这在某些方面让它比其他媒介，比如电影和书籍，更能让人有切身体验。

Yeah. Well, games are, I think, a unique medium because you as the player, you're not just passively consuming the the the entertainment. Right? You're you're actually actively involved as an as an agent. So I think that's what makes it in some ways can be more visceral than other other mediums like, know, films and books.

所以，第二点就是，你知道，在游戏中设计人工智能。然后我们使用的第三个AI应用是从一开始就在DeepMind进行的，即使用游戏作为测试场来验证AI算法和发展AI算法。这是我们DeepMind创立之初愿景的一个核心组成部分，就是我们会大量使用游戏作为我们的主要测试场，至少在初期是这样。因为使用游戏效率极高。而且，你知道，很容易设定指标来衡量你的系统改进得如何，你的想法正朝着什么方向发展，以及你是否在取得渐进式的改进。

So the second so that was, you know, designing AI in games. And then the third use I've we've used of AI is in DeepMind from the beginning, which is using games as a testing ground for proving out AI algorithms and developing AI algorithms. And that was a that was a sort of a core component of our vision at the start of DeepMind was that we would use games very heavily as our main testing ground, certainly to begin with. Because it's super efficient to use games. And also, you know, it's very easy to have metrics to see how well your systems are improving and what direction your ideas are going in and whether you're making incremental improvements.

而且因为这些游戏通常根植于人类长期以来所做的事情，它们已经有一套非常成熟的规则。就像，它本身就是一个非常好的基准测试。

And because those games are often rooted in something that humans did for a long time beforehand, there's already a strong set of rules. Like, it's already a damn good benchmark.

是的。这有很多好处，因为你有清晰的标准来衡量人类在这些事情上能有多出色。没错。而且在某些情况下，比如围棋，我们已经玩了上千年了。是的。

Yes. It's really good for so many reasons because you've got you've got you've got clear measures of how good humans can be at these things. That's right. And in some cases, like Go, we've been playing it for thousands of years. Yeah.

而且通常它们有分数或至少是获胜条件。所以奖励学习系统很容易获得奖励，也很容易明确奖励是什么。最后，也很容易通过与世界顶级选手对弈来外部测试你的系统有多强大。因此，这有很多好处，而且还能在云端并行运行数百万次模拟，效率非常高。

And and often they have scores or at least win conditions. So it's very easy for reward learning systems to get a reward. It's very easy to specify what that reward is. And also at the end, it's easy to, you know, to test externally, you know, how strong is your system by, of course, playing against, you know, the world's strongest players at those games. So it's it's so good for so many reasons and it's also very efficient to run potentially millions of simulations in parallel on the cloud.

所以我认为我们在2010年初能如此成功是有很大原因的。为什么我们能进展这么快？因为我们利用了游戏。在DeepMind初期，我们还聘请了一些出色的游戏工程师，他们是我之前在游戏行业认识的，这帮助我们迅速起步。

So I think there's a huge reason why we were so successful back in starting out 2010. How come we were able to progress so quickly? Because we've utilized games. And at the beginning of DeepMind, we also hired some amazing game engineers who I knew from my previous lives in in the games industry and and that helped to bootstrap us very quickly.

而且，在哲学层面上，人与机器在棋盘上的对决非常引人入胜，尤其是考虑到整个AI的历史都是由人们说制造出能击败人类的机器是不可能的定义的。然后一旦实现了，在我接触AI时，人们确信围棋因为组合复杂性是无法解决的，无论摩尔定律如何发展，计算能力永远无法破解围棋。

And plus it's somehow super compelling almost at a philosophical level of man versus machine over over a chessboard or a go board, and especially given that the entire history of AI is defined by people saying it's gonna be impossible to make a machine that beats a human being in chess. And then once that happened, people were certain when I was coming up in AI that Go is not a game that could be solved because of the combinatorial complexity. It's just too it's it's it's, you know, no matter how much Moore's Law you have, compute is just never going to be able to crack the game of Go.

是的。

Yeah.

因此，从AI研究员和工程师的角度，面对这种不可能的任务很有吸引力，而作为人类观察这一切，看到你认为不可能的事情被打破，是令人谦卑的。意识到我们不如自己想象的聪明，意识到我们现在认为不可能的事情未来或许会实现，这很谦卑。游戏AI系统在游戏中与人类对决，将这一信息传递给数百万甚至数十亿人，尤其是在围棋案例中，这一点非常有力。

And so the then there's something compelling about facing sort of taking on the impossibility of that task from the AI researcher perspective, engineer perspective, and then as a human being, observing this whole thing, your beliefs about what you thought was impossible being broken apart. You it's it's humbling to realize we're not as smart as we thought. It's humbling to realize that the things we think are impossible now perhaps will be done in the in the future. There's something really powerful about a game, AI system being a human being in a game that drives that message home for like millions billions of people, especially in the case of Go.

当然。嗯，我认为这是一段迷人的旅程，尤其是我能从双方理解它，既是AI的创造者，又原本是游戏玩家。是的。所以，AlphaGo比赛对我来说是一个精彩但又有些苦乐参半的时刻，看到并深度参与其中。但如你所说，象棋一直是智能的果蝇，我认为卡斯帕罗夫这么称呼它是正确的。

Sure. Well, look, I think it's a I mean, it has been a fascinating journey and and especially as I I think about it from I can understand it from both sides, both as the AI, you you know, creators of the AI, but also as a games player originally. Yeah. So, you know, it was a it was a really intro you know, was I mean, it's it was a fantastic, but also somewhat bittersweet moment, the AlphaGo match for me seeing that and and and being obviously heavily heavily involved in that. But you know, as you say, chess has been the I mean, Kasparov, I think rightly called it the Drosophila of of intelligence.

对吧？所以我非常喜欢这个说法，我认为他是对的，因为从整个领域诞生之初，国际象棋就与人工智能形影不离。对吧？我认为每一位AI从业者，从图灵、克劳德·香农这些领域的奠基人开始，都尝试过编写国际象棋程序。我甚至收藏了克劳德·香农第一个象棋程序的原始手稿。

Right? So it's sort of I I love that phrase and and I think he's right because chess has been hand in hand with AI from the beginning of the the whole field. Right? So I think every AI practitioner starting with Turing and Claude Shannon and all those, the sort of forefathers of of of of the field, tried their hand at writing a chess program. I've got original audition of Claude Shannon's first chess program.

我记得大概是1949年的原始论文。他们都这么做了，图灵还著名地编写了一个象棋程序，但当时周围的计算机运行速度太慢，所以他不得不亲自充当计算机。对吧？他 literally 花了两三天时间，用铅笔和纸手动运行自己的程序，与朋友对弈他的象棋程序。

I think it was 1949, the the original sort of paper. And they all did that and and Turing famously wrote a chess program that that all the computers around them were obviously too slow to run it. So he had to run he had to be the computer. Right? So he literally, I think spent two or three days running his own program by hand with pencil and paper and playing playing a friend of his with his chess program.

当然，深蓝击败卡斯帕罗夫是一个重大时刻。但事实上，当那一刻发生时——我记得非常非常清楚，因为涉及象棋、计算机和AI，都是我热爱的事物，当时我还在上大学——但我印象更深刻的是卡斯帕罗夫的思维而非深蓝。因为卡斯帕罗夫凭借人类心智，不仅能与这个暴力计算机器大致在同一水平下棋，更重要的是他能做到人类能做的一切其他事情。

So of course, Deep Blue was a huge moment beating Kasparov. But actually, when that happened, I remember that very very vividly, of course, because it was, you know, chess and computers and AI, all the things I loved and I was at college at the time. But I remember coming away from that being more impressed by Kasparov's mind than I was by Deep Blue. Because here was Kasparov with his human mind, not only could he play chess more or less to the same level as as this brute of a calculation machine. But of course, Kasparov can do everything else humans can do.

骑自行车、说多种语言、从政，以及卡斯帕罗夫所做的所有其他惊人事情。而且都是用同一个大脑完成的。是的。而深蓝虽然在国际象棋上表现出色，但它是为象棋专门编写的程序，实际上浓缩了象棋大师的知识，但它做不了其他任何事。

Ride a bike, talk many languages, do politics, all the rest of the amazing things that Kasparov does. And so with the same brain. Yeah. And and yet deep blue of brilliant as it was at chess, it'd been hand coded for chess and actually had distilled the knowledge of chess grandmasters into a cool program. But it couldn't do anything else.

比如它甚至不能玩更简单的井字棋。所以在我看来，那个系统缺少了我们称之为智能的某种东西。我认为这就是通用性以及学习能力的概念。是的。因此我们尝试用AlphaGo来实现这一点。

Like it couldn't even play a strictly simpler game like tic tac toe. So something to me was missing from intelligence from that system that that we would regard as intelligence. And I think it was this idea of generality and and also learning. Yeah. So and as we try to do it out with AlphaGo.

是的。随着AlphaGo、AlphaZero、MuZero，以及Gato等系统的出现——我们稍后会讨论其中部分内容——这里有一条非常迷人的发展轨迹。但让我们先简要聚焦国际象棋的人类层面。

Yeah. With AlphaGo and AlphaZero, MuZero, and then Gato and all the things that we'll get into some parts of. There's just a fascinating trajectory here. But let's just stick on chess briefly. On the human side of chess.

你曾提出从游戏设计角度而言，国际象棋之所以引人入胜，是因为象与马之间存在一种创造性的张力。嗯。你能解释一下吗？首先思考什么使游戏具有持久吸引力确实很有趣，这让它能够跨越世纪经久不衰。

You've proposed that from a game design perspective, the thing that makes chess compelling as a game is that there's a creative tension between a bishop and the knight. Mhmm. Can you explain this? First of it's really interesting to think about what makes a game compelling. It makes it stick across centuries.

嗯。是的。我其实一直在思考这个问题，事实上很多顶尖棋手也不一定会从游戏设计师的角度来看待它。所以我正是戴着游戏设计的帽子在思考：为什么国际象棋如此引人入胜？

Mhmm. Yeah. I was sort of thinking about this, and actually a lot of even amazing chess players don't think about it necessarily from a games designer point of view. So it's this with my game design hat on that I was thinking about this. Why is chess so compelling?

我认为一个关键原因在于棋局动态的多样性——无论是封闭还是开放局面——都源于象和马的特性。当你思考象和马在移动方式上的能力差异时，会发现国际象棋的演变过程或多或少平衡了这两种能力。因此它们的价值都约为三兵。

And I think a critical reason is the the dynamicness of of of the different kind of chess positions you can have, whether they're closed or open and other things, comes from the bishop and the knight. So if you think about how different the the the the capabilities of the bishop and knight are in terms of the way they move and then somehow, chess has evolved to balance those two capabilities more or less equally. So they're both roughly worth three points each.

所以你认为这种动态性始终存在，而其他规则是为了稳定游戏？

So you think that dynamics is always there and then the rest of the rules are kinda trying to stabilize the game?

也许吧。我的意思是这有点像先有鸡还是先有蛋的问题，可能两者是共同形成的。但关键在于它们达到了美妙的平衡——象和马虽然能力迥异，但在所有可能棋局中的价值却近乎相等。人类用数百年时间实现了这种平衡。

Well, maybe. I mean, it's sort of I don't know it's chicken and egg situation probably both came together. But the fact that they it's got to this beautiful equilibrium where you can have the bishop and knight, they're so different in power but so equal in value across the set of the universe of all positions. Right? Somehow they've been balanced by humanity over hundreds of years.

我认为这种设计创造了创造性张力：你可以用象换马，价值相当却导向不同的局面。有马时你追求封闭局面，有象时你渴望开放局面。正是这种张力构成了象棋的创造性核心。

I think gives gives the game the creative tension that you can swap the bishop and knights for a bishop for a knight and you've you've they're more or less worth the same, but now you aim for a different type of position. If you have the knight, you want a closed position. If you have the bishop, you want an open position. So I think that creates a lot of the creative tension in chess.

一种受控的创造性张力。从AI角度来看，你认为人工智能系统最终能设计出对人类最具吸引力的游戏吗？这是个有趣的问题。有时候我被问到

So some kind of controlled creative tension. From an AI perspective, do you think AI systems could eventually design games that are optimally compelling to humans? Well, that's an interesting question. You know, sometimes I get asked

关于AI和创造力的问题，我的回答与此相关：我认为创造力存在不同层级。如果我们将创造力定义为产生具有实用价值的原创内容，那么最低层级的创造力就像是插值法——对所有示例进行平均。比如基础AI系统看过百万张猫图片后，可以生成一只「平均相貌的猫」。

about AI and creativity and and and this and the way I answer that is relevant to that question, which is that I think there are different levels of creativity, one could say. So I think if we define creativity as coming up with something original, right, that's that's useful for a purpose, Then, you know, I think the kind of lowest level of creativity is like an interpolation. So an averaging of all the examples you see. So maybe a very basic AI system could say you could have that. So you show it millions of pictures of cats and then you say, give me an average looking cat.

对吧？给我生成一只相貌平平的猫。我会称之为插值。然后是外推，就像AlphaGo展示的那样。AlphaGo与自己对弈了数百万盘围棋。

Right? Generate me an average looking cat. I would call that interpolation. Then there's extrapolation, which something like AlphaGo showed. So AlphaGo played, you know, millions of games of Go against itself.

然后它想出了像第二局第37手这样的绝妙新思路，引入了围棋中人类从未想到过的策略主题，尽管我们已经玩了数千年，职业化也有数百年历史。所以我称之为外推。但在此之上还有一个更高层次，可以称之为跳出框架的思考或真正的创新，即能否发明围棋？就像能否发明国际象棋，而不仅仅是走出一步妙棋或一手好着，而是真正创造出国际象棋或与之媲美的游戏？我认为有一天AI可以做到，但缺失的是我们现在该如何向程序描述这个任务？

And then it came up with brilliant new ideas like Move 37 in game two, bringing a motif strategies in Go that that no humans had ever thought of even though we've played it for thousands of years and professionally for hundreds of years. So that that I call that extrapolation. But then that's still there's still a level above that which is you know, you could call out of the box thinking or true innovation, which is could you invent Go? Like could you invent chess and not just come up with a brilliant chess move or brilliant Go move, but can you can you actually invent chess or something as good as chess or go? And I think one day AI could, but the what's missing is how would you even specify that task to a program right now?

如果我要告诉人类或游戏设计师这么做，我会这样说：去设计一个只需五分钟就能学会的游戏（就像围棋规则简单），但需要一生甚至几辈子才能精通（因为它如此深邃复杂），同时要具有美学吸引力，并且能在三四个小时内完成一局（适合人类日常安排）。你可以用这类高层次概念来定义要求。

And the way I would do it if if I was telling a human to do it or or games designer, a human games designer to do it is I would say something like go. I would say, come up with a game that only takes five minutes to learn, which go does because it's got simple rules, but many lifetimes to master, right, or impossible to master in one lifetime because it's so deep and so complex. And then it's aesthetically beautiful. And also, it can be completed in three or four hours of gameplay time, which is, you know, useful for our us, you know, in in a in a human day. And so you might specify these sort of high level concepts like that.

根据这些条件（可能还有其他一些），围棋恰好满足这些约束。但问题是我们尚无法向AI系统传达这种抽象的高层概念。我认为在真正理解高层次概念和可组合的抽象能力方面仍存在缺失。所以目前AI能做插值和外推，但还做不到真正的发明创造。

And then with that and there may be a few other things one could imagine that Go satisfies those constraints. But the problem is is that we're not able to specify abstract notions like that, high level abstract notions like that yet to our AI systems. And I think there's still something missing there in terms of high level concepts or abstractions that they truly understand and, you know, combinable and compositional. So for the moment, I think AI is capable of doing interpolation, extrapolation, but not true invention.

也就是说，设计规则集并围绕这些规则进行复杂目标优化，我们现在还做不到。但可以选取特定规则集，通过自我对弈实验观察AI系统从零开始学习的过程，看学习周期有多长。如果它能满足你提到的快速学习等条件，即使对AI系统也需要漫长掌握过程，或许就能说这是个有潜力的游戏。这有点像编程领域的AlphaCode思路。

So coming up with rule sets and optimizing with complicated objectives around those rule sets, we can't currently do. But you could take a specific rule set and then run a kind of self play experiment to see how long, Just observe how an AI system from scratch learns. How long is that journey of learning? And maybe if it satisfies some of those other things you mentioned in terms of quickness to learn and so on, and you could see a long journey to master for even an AI system, then you could say that this is a promising game. But it would be nice to do almost like alpha codes of programming Yeah.

规则。生成能自动化这部分创造过程的规则

Rules. So generating rules that kind of that that automate even that part of the generation

规则系统。我设想过一种令人惊叹的系统：游戏设计师可以让系统接管游戏，在一夜之间进行数千万次对局，自动优化规则平衡性。通过调整规则、公式和参数，使游戏单位更平衡——是的，部分规则可以被微调。这就像是提供基础设定后，让蒙特卡洛树搜索等算法进行探索优化。

of rules. So I have thought about systems actually that I think would be amazing in in in for a games designer if you could have a system that takes your game, plays it tens of millions of times, maybe overnight, and then self balances the rules better. So it tweaks the the rules and maybe the equations and the and the and the parameters so that the game is more balanced, the units in the game or Yes. Some of the rules could be tweaked. So it's a bit of like giving a base set and then allowing Monte Carlo tree search or something like that to sort of explore it.

对吧？而且我认为这实际上会成为一个非常强大的工具，用于游戏平衡，自动平衡游戏。通常这需要数千小时、数百场人类测试员的游戏才能平衡像《星际争霸》这样的游戏。你知道，暴雪在游戏平衡方面做得非常出色，但这花费了他们很多很多年的时间。所以可以想象，当这种技术足够高效时，你可能一夜之间就能完成这项工作。

Right? And I think that would be super powerful tool actually for balancing, auto balancing a game, which usually takes thousands of hours from hundreds of games, human games testers normally to balance some one, you know, game like StarCraft, which is, you know, Blizzard are amazing at balancing their games, but it takes them years and years and years. So one could imagine at some point when this this stuff becomes efficient enough to, you know, you might be able do that like overnight.

你认为由AI系统优化设计的游戏会非常像地球吗？

Do you think a game that is optimal designed by an AI system would look very much like a planet Earth?

也许吧。也许吧。这正是我喜欢制作的那种游戏，我也尝试过。在我的游戏设计生涯中，我的第一个大型游戏是设计一个主题公园，一个游乐园。然后像《共和国》这样的游戏，我试图设计整个城市并让你在其中游玩。当然，像威尔·赖特也创作过《模拟地球》这样的游戏，试图模拟整个地球。

Maybe. Maybe. It's only the sort of game I would love to make is is and I've tried, you know, in my in my in my games career, the games design career, you know, my first big game was designing a theme park, an amusement park. Then with games like Republic, I tried to, you know, have games where we designed whole cities and and allowed you to play in. So and of course, like Will Wright have written games like Sim Earth trying to simulate the whole of Earth.

相当棘手。我觉得《模拟地球》

Pretty tricky. I think Sim Earth,

我其实没玩过那个。那是什么游戏？是关于进化的还是

I haven't actually played that one. So what is it? Does it is it evolution or

是的。它有进化元素，它试图将其视为一个完整的生物圈，但从相当高的层面来看。我明白了。

Yeah. It has evolution and it sort of it tries to it it sort of treats it as an entire biosphere, but from quite high level. I see.

能够某种程度上

It'd be nice to be able to sort

放大缩小和

of zoom in zoom out and

放大。

zoom in.

没错。所以很明显，它做不到这一点。那是在晚上。我想他是在九十年代写的，所以它做不到，你知道，它当时没有那个能力。但那显然会是终极的沙盒游戏，当然。

Exactly. So obviously, it couldn't do that. That was in the night. I think he he wrote that in the nineties, so it couldn't it, you know, it wasn't wasn't able to do that. But that that would be obviously the ultimate sandbox game, of course.

说到这个话题，你认为我们生活在模拟世界中吗？

On that topic, do you think we're living in a simulation?

是的。嗯，好吧。所以

Yes. Well, so okay. So

我们将会从极其哲学的话题跳到——当然。

I We're gonna jump around from the absurdly philosophical to the Sure.

技术性的。当然。非常非常乐意。所以我认为我对这个问题的回答有点复杂，因为有模拟理论，显然是尼克·博斯特罗姆首次提出的，我想很有名。我并不完全相信那种意义上的理论。

Technical. Sure. Very very happy to. So I think my answer to that question is a little bit complex because there is simulation theory, which obviously Nick Bostrom, I think famously first proposed. And I don't quite believe it in in in that sense.

所以，从某种意义上说，我们是否处于某种电脑游戏中，或者我们的后代出于某种实验目的在二十一世纪重现了地球。我认为，但我确实认为，理解物理学和宇宙的最佳方式可能是从计算的角度出发。将其理解为一个信息宇宙，实际上信息是现实最基本的单位，而非物质或能量。物理学家会说，物质或能量，比如E=mc²，这些是宇宙的基本要素。

So in the sense that are we in some sort of computer game or have our descendants somehow recreated earth in the twenty first century and for some kind of experimental reason. I think that but I do think that we that that we might be that the best way to understand physics and the universe is from a computational perspective. So understanding it as an information universe and actually information being the most fundamental unit of reality rather than matter or energy. So physicists would say, you know, matter or energy, you know, e equals m c squared. These are the things that are are are the fundamentals of the universe.

我实际上会说信息，它当然可以指定能量或物质。物质实际上只是信息的一种排列方式，就像我们的身体和体内分子的排列一样。因此，我认为信息可能是描述宇宙的最基本方式。所以，你可以说我们因此处于某种模拟中。但我并不真正认同那种认为有无数模拟存在的观点。

I'd actually say information, which of course itself can specify energy or matter. Matter is actually just we're just out the way our bodies and the molecules in our body are arranged as information. So I think information may be the most fundamental way to describe the universe. And therefore, you could say we're in some sort of simulation because of that. But I don't I do I'm not I'm not really a subscriber to the idea that, know, these are sort of throw away billions of simulations around.

我认为这实际上非常关键且可能是独特的，这个模拟。这个特定的模拟？

I think this is actually very critical and possibly unique, this simulation. This particular one?

是的。所以，你的意思是，将宇宙视为一台处理和修改信息的计算机，是解决物理学、化学、生物学问题的一种好方法，或许还包括人类学等等。

Yes. So but and you you just mean treating the universe as a computer that's processing and modifying information is is a good way to solve the problems of physics, of chemistry Yeah. Of biology Yes. And perhaps of humanity and so on.

是的。我认为从信息论的角度理解物理学，可能是真正理解这里发生的一切的最佳方式。从

Yes. I think understanding physics in terms of information theory might be the best way to to really understand what's going on here. From

我们对通用图灵机的理解，对计算机的理解来看，你认为宇宙中是否存在计算机能力之外的东西？你在意识本质上与罗杰·彭罗斯有分歧。他认为意识不仅仅是计算。

our understanding of universal Turing machine, from our understanding of a computer, do you think there's something outside of the capabilities of a computer that is present in our universe? You have a disagreement with Roger Penrith about the nature of consciousness. He he thinks that consciousness is more than just a computation.

嗯。

Mhmm.

你认为这一切，整个复杂系统，能否被看作是一种计算？

Do you think all of it, the whole shebangs, can be can be a computation?

我曾与罗杰·彭罗斯爵士进行过许多引人入胜的辩论，显然，他以其著名观点闻名，我读过他的《皇帝新脑》等经典著作。这些书籍在九十年代相当有影响力。他认为需要某种更深层次的东西，比如量子层面的解释，才能阐明大脑中的意识。我思考我们在DeepMind的工作以及我的职业生涯，我们几乎像是图灵的捍卫者。因此，我们正在将图灵机或经典计算推向极限。

I've had many fascinating debates with Sir Roger Penrose, and obviously, he's he's famously and and I read, you know, Emperors of the New Mind and and and his books, his classical books. And they they were pretty influential in the, you know, the nineties. And he believes that there's something more, you know, something quantum that is needed to explain consciousness in the brain. I think about what we're doing actually at DeepMind and what my career is being, we're almost like Turing's champion. So we are pushing Turing machines or classical computation to the limits.

经典计算的极限是什么？同时，我也研究过神经科学，这正是我攻读博士学位的原因，旨在从神经科学或生物学角度探究大脑中是否存在量子现象。到目前为止，我认为大多数神经科学家和主流生物学家会表示，没有证据表明大脑中存在任何量子系统或效应。据我们所知，它大多可以用经典理论来解释。因此，从生物学角度出发，这成了一种探索。

What are the limits of what classical computing can do? Now, and at the same time, I've also studied neuroscience to see and that's why I did my PhD in was to see also to look at, you know, is there anything quantum in the brain from a neuroscience or biological perspective? And so far, I think most neuroscientists and most mainstream biologists and neuroscientists would say there's no evidence of any quantum systems or effects in the brain. As far as we can see, it's it can be mostly explained by classical classical theories. So and then so there's sort of the the the search from the biology side.

与此同时，经典图灵机的能力也在不断提升，包括我们新的人工智能系统。正如你早些时候提到的，我认为人工智能，尤其是在过去十多年里，一直是一个不断带来惊喜和成功的故事，推翻了一个又一个曾被认为不可能的理论，从围棋到蛋白质折叠等等。因此，我非常犹豫是否要赌经典图灵机和计算范式能走多远。我的赌注是，我们大脑中发生的一切很可能可以在经典机器上模拟或近似实现，而不需要某种形而上学或量子层面的东西。

And then at the same time, there's the raising of the water at the bar from what classical Turing machines can do and and and, you know, including our new AI systems. And as you alluded to earlier, I think AI, especially in the last decade plus, has been a continual story now of surprising events and surprising successes, knocking over one theory after another of what was thought to be impossible from Go to protein folding and so on. And so I think I would be very hesitant to bet against how far the universal Turing machine and classical computation paradigm can go. And my betting would be that all of certainly what's going on in our brain can probably be mimicked or approximated on classical machine. Not, you know, not requiring something metaphysical or or quantum.

我们将在AlphaFold的一些工作中实现这一点，我认为这开启了模拟生物学这个美丽而复杂世界的旅程。那么，你认为人类心智的所有魔力都来自于这几磅生物计算浆糊，类似于DeepMind一直在研究的神经网络（虽然不是直接相同，但精神上相似）？嗯，看，我

And we'll get there with some of the work with AlphaFold, which I think begins the journey of modeling this beautiful and complex world of biology. So you think all the magic of the human mind comes from this just a few pounds of mush of biological computational mush that's akin to some of the neural networks, not directly, but in spirit that DeepMind has been working with? Well, look, I

我认为，你说它是几磅的东西——当然，我认为宇宙中最大的奇迹就是我们颅骨中这几磅浆糊。是的。然而，我们的大脑也是我们所知的宇宙中最复杂的物体。因此，我们的大脑有着极其美丽和惊人的特质，我认为它是一个非常高效的机器，本质上是一种现象。建造人工智能的原因之一，我一直想建造人工智能，是因为我认为通过构建像AI这样的智能产物，并将其与人类心智进行比较，将帮助我们解锁自历史黎明以来一直让我们困惑的心智独特性和真正秘密，比如意识、梦境、创造力、情感。这些都是什么？

think it's you say it's a few you know, of course, it's this is the I think the biggest miracle of the universe is that it is just a few pounds of mush in our skulls Yeah. And yet it's also our brains are the most complex objects in the that we know of in the universe. So there's something profoundly beautiful and amazing about our brains and I think that it's an incredibly incredible efficient machine and it's a phenomenon basically. And I think that building AI, one of the reasons I want to build AI and I've always wanted to is I think by building an intelligent artifact like AI and then comparing it to the human mind, that will help us unlock the uniqueness and the true secrets of the mind that we've always wondered about since the dawn of history, like consciousness, dreaming, creativity, emotions. What are all these things?

对吧？自人类诞生以来，我们就一直在思考这些问题。我认为其中一个原因，也是我热爱哲学和心智哲学的原因，是我们发现很难真正科学地研究它，除了历史上非常聪明的哲学家通过内省之外，缺乏工具。但现在突然之间，我们有了大量工具。首先，有所有的神经科学工具，fMRI机器、单细胞记录等等。

Right? We've wondered about them since the dawn of humanity. And I think one of the reasons and you know, love philosophy and philosophy of mind is we found it difficult is there haven't been the tools for us to really other than introspection to from very clever people in history, very clever philosophers to really investigate this scientifically. But now suddenly we have a plethora of tools. Firstly, have all the neuroscience tools, fMRI machines, single cell recording, all of this stuff.

但我们也有能力，通过计算机和人工智能来构建智能系统。所以我认为，你知道，人脑的运作方式真的很神奇。我对此确实感到敬畏。而且我认为，凭借我们的人脑，我们能够创造出计算机这样的东西，甚至能够思考和研究这些问题，这也是对人脑能力的一种证明。

But we also have the ability, computers and AI to build intelligent systems. So I think that, you know, I think it is amazing what the human mind does. And and and I'm kind of in awe of it really. And and I think it's amazing that with our human minds, we're able to build things like computers and and actually even, you know, think and investigate about these questions. I think that's also a testament to the human mind.

是的。宇宙创造了人脑，而人脑现在正在建造计算机，帮助我们理解宇宙和我们自己的人脑。

Yeah. The universe built the human mind that now is building computers that help us understand both the universe and our own human mind.

没错。正是这样。我的意思是，我认为可以说，我们或许是宇宙试图理解自身的一种机制。

That's right. That's exactly it. I mean, I think that's one, you know, one could say we we are maybe we're the mechanism by which the universe is going to try and understand itself.

是的。这很美。那么让我们来看看生物学的基本构建模块，我认为这是另一个角度，你可以从中开始理解人脑和人体，这非常迷人，即从基本构建模块开始模拟和建模，如何通过这些模块构建越来越大、越来越复杂的系统，也许有一天能构建出整个人类生物学的全貌。所以这里有另一个曾被认为无法解决的问题，那就是蛋白质折叠。而AlphaFold，特别是AlphaFold 2，就做到了这一点。

Yeah. It's beautiful. So let's let's go to the basic building blocks of biology that I think is another angle at which you can start to understand the human mind, the human body, which is quite fascinating, which is from the basic building blocks, start to simulate, start to model how from those building blocks you can construct bigger and bigger, more complex systems, maybe one day the entirety of the human biology. So here's another problem that thought to be impossible to solve, which is protein folding. And AlphaFold or specific AlphaFold two did just that.

它解决了蛋白质折叠问题。我认为这是结构生物学史上，乃至整个科学领域最大的突破之一。也许从高层次来看，它是什么以及它是如何工作的？嗯。然后我们可以提出一些迷人的问题。

It solved protein folding. I think it's one of the biggest breakthroughs, certainly in the history of structural biology, but in general, in in science. Maybe from a high level, what is it and how does it work? Mhmm. And then we can ask some fascinating Sure.

之后再问。

Questions after.

当然。那么也许可以向不熟悉蛋白质折叠的人解释一下，首先解释蛋白质，你知道，蛋白质对所有生命都至关重要。你身体的每一项功能都依赖于蛋白质。它们有时被称为生物学的“主力军”。如果你深入研究它们，显然，作为AlphaFold的一部分，我在过去几年里一直在研究蛋白质和结构生物学。

Sure. So maybe to explain it to people not familiar with protein folding is, you know, first of all explain proteins, which is, you know, proteins are essential to all life. Every function in your body depends on proteins. Sometimes they're called the workhorses of biology. And if you look into them and I've, you know, obviously as part of AlphaFold, I've been researching proteins and structural biology for the last few years.

它们是令人惊叹的生物纳米机器蛋白质。如果你真的观看它们如何工作的小视频、它们如何工作的动画，它们是非常不可思议的。蛋白质由其称为氨基酸序列的基因序列所指定。所以你可以把它看作是它们的基因构成。然后在体内，在自然界中，当它们折叠成三维结构时。

They're amazing little bio nano machines proteins. They're incredible if you actually watch little videos of how they work, animations of how they work. And proteins are specified by their genetic sequence called the amino acid sequence. So you can think of it as their genetic makeup. And then in the body, in in nature, they when they when they fold up into a three d structure.

所以你可以把它想象成一串珠子，然后它们折叠成一个球。关键的一点是，你需要知道那个三维结构是什么，因为蛋白质的三维结构有助于决定它在你的身体里做什么，即它的功能。并且，如果你对药物或疾病感兴趣，你需要理解那个三维结构。因为如果你想用药物化合物靶向某个东西，以阻断蛋白质正在做的事情，你需要理解它将在蛋白质表面的哪个位置结合。所以显然，为了做到这一点，你需要理解三维结构。

So you can think of it as a string of beads and then they fold up into a ball. Now the key thing is you want to know what that three d structure is because the structure the three d structure of a protein is what helps to determine what does it do, the function it does in your body. And also, if you're interested in drug drugs or or disease, you need to understand that three d structure. Because if you wanna target something with a drug compound about to block something the protein is doing, you need to understand where it's gonna bind on the surface of the protein. So obviously, in order to do that, you need to understand the three d structure.

所以结构映射到了功能？

So the structure is mapped to the function?

结构映射到了功能。并且结构显然在某种程度上是由氨基酸序列指定的。这本质上就是蛋白质折叠问题：你能仅仅从氨基酸序列，从一维的字母串，立即通过计算预测出三维结构吗？

The structure is mapped to the function. And the structure is obviously somehow specified by the by the amino acid sequence. And that's the in essence, the protein folding problem is, can you just from the amino acid sequence, the one dimensional string of letters, can you immediately computationally predict the three d structure?

对。

Right.

这五十多年来一直是生物学领域的一个巨大挑战。我认为它最初是由1972年诺贝尔奖得主克里斯蒂安·安芬森在其诺贝尔奖获奖演讲中阐述的。他当时推测，从氨基酸序列推导出三维结构应该是可能的。他没有说具体怎么做。所以这就像……你知道，有人向我描述说这相当于费马大定理。

And this has been a grand challenge in biology for over fifty years. So I think it was first articulated by Christian Anfinsen, a Nobel Prize winner in 1972, as part of his Nobel Prize winning lecture. And he just speculated this should be possible to go from the amino acid sequence to the three d structure. He didn't say how. So it's like I I you know, I've been it's been described to me as as equivalent to Fermat's last theorem

是的。

Yeah.

但对于生物学来说。

But for biology.

没错。你应该像那些将来很可能获得诺贝尔奖的人一样，但除此之外，你应该多做这类事情。在空白处，随便写些东西。对。这会需要，大概，两百年

Right. You should as somebody that very well might win the Nobel Prize in the future, but outside of that, you you should do more of that kind of thing. In the margins, just put random things. Right. It'll take, like, two hundred years

才能解决。让人们忙活两百年。

to solve. Set people off for two hundred years.

这应该是可能的。正是。而且不要给出任何不体面的东西。没错。

It should be possible. Exactly. And just don't give any indecent. Exactly.

我觉得每个人都是这么想的。应该是这样。我得记住这一点以备将来。所以是的。他就这样用一个随口一提的话开启了，就像费马一样，你知道，他真的开启了这个整个五十年的计算生物学领域。

I think everyone's exactly. It should be. I'll I'll have to remember that for future. So yeah. So he set off, you know, with this one throwaway remark, just like Fermat, you know, he he set off this whole fifty year field really of of computational biology.

而且你知道，他们卡住了。在这方面并没有取得太大进展。直到现在，直到AlphaFold出现之前，这都是通过实验完成的，对吧，非常费劲。所以经验法则是，你必须让蛋白质结晶，这非常困难。有些蛋白质无法结晶，比如膜蛋白。

And and they had you know, they got stuck. They hadn't really got very far with doing this. And and until now, until AlphaFold came along, this is done experimentally, right, very painstakingly. So the rule of thumb is and you have to like crystallize the protein, which is really difficult. Some proteins can't be crystallized like membrane proteins.

然后你必须使用非常昂贵的电子显微镜或X射线晶体学仪器，进行极其繁琐的工作来获取三维结构并可视化。所以实验生物学的经验法则是，一个博士生需要整个博士生涯才能完成一个蛋白质的研究。而有了AlphaFold 2，我们能够在几秒钟内预测出三维结构。因此在圣诞节期间，我们完成了整个人类蛋白质组，也就是人体内的每一个蛋白质，全部20,000个蛋白质。所以人类蛋白质组就像是人类基因组在蛋白质空间的等价物。

And then you have to use very expensive electron microscopes or x-ray crystallography machines, really painstaking work to get the three d structure and visualize a three d structure. So the rule of thumb in experimental biology is that it takes one PhD student, their entire PhD, to do one protein. And with AlphaFold two, we were able to predict the three d structure in a matter of seconds. And so over Christmas, we did the whole human proteome or every protein in the human body, all 20,000 proteins. So the human proteome is like the equivalent of the human genome but on protein space.

并且在某种程度上彻底改变了结构生物学家的工作方式。因为他们现在不必担心那些费力的实验，你知道，是否值得投入那么多精力。他们几乎可以像谷歌搜索一样直接查找蛋白质的结构。

And sort of revolutionized really what a structural biologist can do. Because now they don't have to worry about these painstaking experimental, you know, should they put all of that effort in or not. They can almost just look up the structure of their proteins like a Google search.

所以它是在一个数据集上进行训练的，学习如何映射这个氨基酸序列。首先，令人难以置信的是，蛋白质这种微型化学计算机能够以某种分布式方式自行完成这种计算，并且速度非常快。

And so there's a dataset on which it's trained and how to map this amino acid sequence. First of all, it's incredible that a protein, this little chemical computer is able to do that computation itself in some kind of distributed way and do it very quickly.

嗯。

Mhmm.

这很神奇，它们之所以这样进化是因为，你知道，最初，我是说，这真是个伟大的发明，仅仅是蛋白质本身就

That's a weird thing, and they evolved that way because, you know, in the beginning, mean, I that's a great invention, just the protein itself of

我是说

I mean

然后我认为可能有一个历史过程，比如它们进化出许多这样的蛋白质，这些蛋白质自身学会了如何成为计算机，通过这种方式可以创建能够相互配合形成复合体以实现高级功能的结构。我是说，它们能摸索出这套系统真的很不可思议。

And then they there's, I think, probably a history of, like, they evolved to have many of these proteins, and those proteins figure out how to be computers themselves in such a way that you can create structures that can interact in complexes with each other in order to form high level functions. I mean, it's a weird system that they've figured it out.

确实如此。我是说，我们或许也该讨论生命起源，但蛋白质本身，正如我所说，是神奇而不可思议的微型生物纳米机器。实际上，另一位科学家莱文托——他是安芬森的同时代人——他提出了所谓的莱文托悖论，这正好印证了你的观点。他粗略计算出，一个约2000个氨基酸长度的普通蛋白质，可能拥有10的300次方种不同的构象。也就是说，这个蛋白质有10的300次方种不同的折叠方式。

Well, for sure. I mean, we you know, maybe we should talk about the origins of life too, but proteins themselves, I think, are magical and incredible, as I said, little little bio nano machines. And and and actually, Leventow, who is another scientist, a contemporary of Amfinson, he he he coined this Leventow, what became known as Leventow's paradox, which is exactly what you're saying. He calculated roughly a approach an average protein, which is maybe 2,000 amino acids bases long, is is is can fold in maybe 10 to the power 300 different confirmations. So there's 10 to the power 300 different ways that protein could fold up.

然而在自然界中，物理学却能在毫秒级时间内解决这个问题。蛋白质在你体内折叠，有时甚至只需几分之一秒。所以物理学在某种程度上解决了这个搜索问题。

And yet somehow in nature, physics solves this solves this in a matter of milliseconds. So proteins fold up in your body in you know, sometimes in in fractions of a second. So physics is somehow solving that search problem.

需要明确的是——是的。在许多情况下，如果我说错了请纠正，通常序列会以唯一的方式自我形成。是的。在如此庞大的可能性中，它找到了稳定存在的方式——当然有时会出现错误折叠等情况，这就导致了许多功能障碍之类的问题。

And just to be clear Yeah. In many of these cases, maybe correct me if I'm wrong, there's often a unique way for the sequence to form itself. Yes. So among that huge number of possibilities Yes. It figures out a way how to stably in some cases, there might be a misfunction and so on, which leads to a lot of the disorders and stuff like that.

但是，是的。

But Yes.

大多数时候，这是一种唯一的映射关系，而这种唯一映射并不显而易见。

Most of the time, it's a unique mapping, and that unique mapping is not obvious.

不，完全正确。

No. Exactly.

这正是问题所在。

Which is what the problem is.

没错。所以在健康状态下通常存在唯一映射。正如你所说，在疾病中——例如阿尔茨海默症，有一种推测认为是由错误折叠的蛋白质所致。β淀粉样蛋白以错误方式折叠后，会纠缠在神经元中。

Exactly. So there's a unique mapping usually in a healthy in if you if it's healthy. And as you say in disease, so for example, Alzheimer's, one one one conjecture is that it's because of misfolded protein. A protein that folds in the wrong way, amyloid beta protein. And then because it folds in the wrong way, it gets tangled up in your neurons.

因此，理解健康功能和疾病的关键在于了解这些蛋白质的作用方式及其结构机制。当然，下一步是认识到蛋白质在与某些物质相互作用时会发生构象变化，它们在生物学中并非总是静态不变的。

So it's super important to understand both healthy functioning and also disease is to understand what these things are doing and how they're structuring. Of course, the next step is sometimes proteins change shape when they interact with something. So they're not just static necessarily in biology.

或许您可以分享一些AlphaFold早期解决这个难题时的有趣或精彩之处。因为与游戏不同，这是真实的物理系统，不太适合自我博弈机制。数据集规模比理想情况要小，因此需要在某些方面非常巧妙地处理。您能否谈谈哪些问题特别难以解决，以及解决方案中有哪些精妙之处？

Maybe you can give some interesting sort of beautiful things to you about these early days of AlphaFold, of of solving this problem. Because unlike games, this is real physical systems that are less amenable to self play type of mechanisms. Sure. The the size of the dataset is smaller than you might otherwise like, so you have to be very clever about certain things. Is there something you could speak to what was very hard to solve and what are some beautiful aspects about the the solution?

是的。我认为AlphaFold是我们迄今构建的最复杂且最具意义的系统。过去两三年见证它的发展是一段非凡的历程——正如我们之前讨论的，我们最初从游戏领域起步，开发了AlphaGo和AlphaZero等系统。但真正的终极目标不仅是攻克游戏，更是要构建能引导通用学习系统的工具，并将其应用于现实世界的挑战。

Yeah. I I would say AlphaFold is the most complex and also probably most meaningful system we've built so far. So it it's been an amazing time actually in the last, two, three years to see that come through because as we talked about earlier, games is what we started on, building things like AlphaGo and AlphaZero. But really, the ultimate goal was to not just to crack games. It was just to build, use them to bootstrap general learning systems we could then apply to real world challenges.

具体而言，我的热情在于应对像蛋白质折叠这样的科学挑战。AlphaFold正是我们在这方面首个重大成果证明。就数据和创新投入而言，我们需要整合超过30种不同的组件算法来攻克蛋白质折叠难题。其中重要创新包括引入基于物理学和进化生物学的硬编码约束，例如限制蛋白质中的键角等参数，同时确保不影响学习系统的自主性。

Specifically, my passion is scientific challenges like protein folding. And then AlphaFold, of course, is our first big proof point of that. And so in terms of the data and the amount of innovations that had to go into it, it was like more than 30 different component algorithms needed to be put together to crack the protein folding. I think some of the big innovations were kind of building in some hard coded constraints around physics and evolutionary biology to constrain sort of things like the bond angles in the protein and things like that. But not to impact the learning system.

这样系统仍能从现有样本中自主学习物理规律。正如您所说，样本量只有约15万种蛋白质——即使经过40年的实验生物学研究，目前也仅解析出约15万种蛋白质的结构。这就是我们的训练集，规模远小于常规需求。但我们通过自蒸馏等技巧，将AlphaFold自身高置信度的预测结果反哺到训练集中扩大数据量。

So still allowing the system to be able to learn the physics itself from the examples that we had. And the examples, as you say, there are only about 150,000 proteins. Even after forty years of experimental biology, Only around 150,000 proteins have been the structures have been found out about. So that was our training set, which is much less than normally we would like to use. But using various tricks, things like self distillation, so actually using AlphaFold predictions.

这个策略对AlphaFold的成功至关重要。实际上需要大量此类创新才能最终攻克难题。AlphaFold一代产出的是距离矩阵（distogram），即蛋白质中所有分子间成对距离的矩阵。

Some of the best predictions that it thought was highly confident in, we put them back into the training set, right, to make the training set bigger. That was critical to AlphaFold working. So there was actually a huge number of different innovations like that that were required to ultimately crack the problem. AlphaFold one, what it produced was distogram. So a kind of a matrix of the pairwise distances between all of the molecules in the protein.

之后还需要独立的优化过程来生成三维结构。而AlphaFold二代实现了真正的端到端系统——直接从氨基酸碱基序列生成三维结构，无需中间步骤。我们在机器学习领域始终发现：系统越端到端，性能越优越。这可能因为系统最终比人类设计者更擅长自主领悟约束条件。

And then there had to be a separate optimization process to create the three d structure. And what we did for AlphaFold two is make it truly end to end. So we went straight from the amino acid sequence of bases to the three d structure directly without going through this intermediate step. And in machine learning, what we've always found is that the more end to end you can make it, the better the system. And it's probably because we in the end, the system is better at learning what the constraints are than than we are as the human designers of specifying it.

因此，任何时候你能让信息从头到尾流动，并直接生成你真正需要的东西——在这个例子中是三维结构——这都比设置一个中间步骤要好，因为后者还需要你手动设计后续步骤。所以，让梯度和学习从你期望的最终输出端一直流向输入端，贯穿整个系统，是更优的选择。

So anytime you can let it flow end to end and actually just generate what it is you're really looking for, in this case, the three d structure, you're better off than having this intermediate step, which you then have to handcraft the next step for. So so it's better to let the gradients and the learning flow all the way through the system from the endpoint, the end output you want to the inputs.

所以这是一个处理新问题的好方法：先手动设计一堆东西，加入大量人工约束，配上一小段端到端学习或小型学习模块，然后逐渐扩展这个学习模块，直到它覆盖整个流程。

So that's a good way to start on a new problem. Handcraft a bunch of stuff, add a bunch of manual constraints with a small end to end learning piece or a small learning piece and grow that learning piece until it consumes the whole thing.

没错。你可以看到，这是我们通过多个成功的Alpha项目——我们称之为Alpha X项目——逐渐形成的一种方法。最明显的例子就是AlphaGo到AlphaZero的演进。AlphaGo是一个学习系统，但它专门训练来下围棋。

That's right. And so you can also see, you know, this is a bit of a method we've developed over doing many sort of successful alpha, so we call them alpha x projects. Right? Is and the the easiest way to see that is the evolution of AlphaGo to AlphaZero. So AlphaGo was a learning system, but it was specifically trained to only play Go.

对吧？我们最初开发AlphaGo的目标是不惜一切代价达到世界冠军水平。然后，到了AlphaGo Zero，我们不再需要以人类棋谱作为起点。

Right? So and what we wanted to do in with first version of AlphaGo is just get to world champion performance no matter how we did it. Right? And then and then, of course, AlphaGo Zero, we we we remove the need to use human games as a starting point. Right?

它可以从随机初始状态开始自我对弈。这样就消除了对围棋人类知识的需求。最后，AlphaZero进一步泛化，移除了系统中的所有特定设置，包括围棋棋盘的对称性等，使得AlphaZero能够从零开始学习任何双人游戏。而MuZero作为这一系列的最新版本，更进一步，你甚至不需要提供游戏规则。

So it it could just play against itself from random starting point from the beginning. So that removed the the need for human knowledge about Go. And then finally, AlphaZero then generalized it so that any things we had in there, the system, including things like symmetry of the Go board, were removed. So that AlphaZero could play from scratch any two player game. And then MuZero, which is the final our latest version of that set of things, was then extending it so that you didn't even have to give it the rules of the game.

它会自行学习规则。因此，它既能处理棋盘游戏，也能应对电子游戏。

It would learn that for itself. So it could also deal with computer games as well as board games.

所以AlphaGo、AlphaGo Zero、AlphaZero、MuZero这一系列的发展轨迹，完整展示了从模仿学习到完全自监督学习的演进路径。

So that line of alpha go alpha go zero alpha zero mu zero, that's the full trajectory of what you can take from imitation learning to full self supervised learning.

是的，没错。并且从零开始学习你输入环境的整个结构，对吧？然后通过自我对弈来逐步提升它。

Yeah. Exactly. And learning learning the entire structure of the environment you put in from scratch. Right? And and and and boost trapping it through self play yourself.

但问题是，一开始就构建出AlphaZero或MuZero对我们来说是不可能的，或者非常困难。

But the thing is it would have been impossible, think, or very hard for us to build alpha zero or mu zero first out of the box.

甚至心理上也是如此，因为你必须长时间相信自己。你不断面对质疑，因为很多人说这是不可能的。

Even psychologically because you have to believe in yourself for a very long time. You're you're constantly dealing with doubt because a lot of people say that it's impossible to

光是下围棋就已经够难了。正如你所说，在2015年、2016年我们实现它时，所有人都认为那是不可能的，或者至少还要十年时间。所以，是的，心理上可能也非常困难，当然，我们通过先构建AlphaGo学到了很多，对吧？因此，我认为这就是为什么我称AI为一门工程科学。

it was hard enough just to do Go. As you were saying, everyone thought that was impossible or at least a decade away from when we when we did it in back in 2015, '24, you know, 2016. And and so, yes, it would have been psychologically probably very difficult as well as the fact that, of course, we learned a lot by building AlphaGo first. Right? So it's I think this is why I call AI an engineering science.

它是最迷人的科学学科之一。但它也是一门工程科学，因为与自然科学不同，你研究的现象在自然界中并不存在。你必须先构建它，所以你得先造出这个人工产物，然后才能研究它是如何运作的，以及如何拆解它。

It's one of the most fascinating science disciplines. But it's also an engineering science in the sense that unlike natural sciences, the phenomenon you're studying doesn't exist out in nature. You have to build it first. So you have to build the artifact first and then you can study how how and pull it apart and how it works.

这个问题很难问你，因为你可能会说每一样都重要，但我们试着思考一下，因为你处于一个非常有趣的位置：DeepMind既是AI史上一些最杰出思想的诞生地，也是一个卓越的工程实践场所。那么，在解决智能这一DeepMind的重大目标中，科学占多少？工程占多少？算法占多少？数据又占多少？

This is tough to ask you this question because you probably will say it's everything, but let's let's try let's try to think to this because you're in a very interesting position where DeepMind is a place of some of the most brilliant ideas in the history of AI, but it's also a place of brilliant engineering. So how much of solving intelligence, this big goal for DeepMind, how much of it is science? How much is engineering? So how much is the algorithms? How much is the data?

嗯。硬件计算基础设施占多少？软件计算基础设施又占多少？

Mhmm. How much is the hardware compute infrastructure? How much is it the software compute infrastructure?

是的。

Yep.

还有什么？人类基础设施占多少？还有，比如，人类以特定方式互动的因素？当然。在所有这些想法的范畴内，有多少可能是哲学层面的？关键是什么？

What else is there? How much is the human infrastructure? And, like, just the humans interacting certain kinds of ways Sure. In all in space of all those ideas, and how much is maybe, like, philosophy? How much what's the key?

如果，如果你回顾过去，比如，我们向前看两百年再回望，解决智能问题的关键是什么？是理论还是工程实践？

If if if you were to sort of look back, like, we go forward two hundred years and look back, what was the key thing that solved intelligence? Is that deals or the engineering?

是结合。首先，当然是所有这些因素的结合，但它们的比例随时间发生了变化。是的。即使在过去的十二年里，我们在2010年创立了DeepMind，现在很难想象，因为2010年只是短短的十二年前，但当时没人谈论AI。你知道，回想你在MIT的日子，没人讨论这个。

Combination. First of all, of course, it's a combination of all those things, but the the ratios of them changed over over time. So Yeah. So even in the last twelve years, we started DeepMind in 2010, which is hard to imagine now because 2010, it's only twelve short years ago, but nobody was talking about AI. You know, don't remember back to your MIT days, you know, no one was talking about it.

我当时在MIT做博士后，那时人们普遍认为，看，我们知道AI行不通。九十年代在MIT这样的地方我们努力尝试过，主要失败于使用逻辑系统和老式的所谓‘经典AI’，我们现在会这么叫。像Minsky、Patrick Winston这些人，你知道所有这些人物。对吧？

I I did a postdoc at MIT back around then and it was sort of thought of as a well, look, we know AI doesn't work. We tried this hard in the nineties at places like MIT. Mostly losing using logic systems and old fashioned sort of good old fashioned AI, we would call it now. People like Minsky and and and Patrick Winston, and you know all these characters. Right?

我曾经和他们中的一些人辩论，他们当时觉得我疯了，认为学习系统能带来新突破。我其实很高兴听到这个，因为至少你知道自己走的是独特的路。对吧？即使你所有的教授都说你疯了。当然。

And used to debate a few of them and they used to think I was mad thinking about that some new advance could be done with learning systems. I was actually pleased to hear that because at least you know you're on a unique track at that point. Right? Even if every all of yours, you know, professors are telling you you're mad. Sure.

当然在工业界，我们很难筹到资金，现在也很难想象，因为如今它是VC中最热门的话题，融资容易，诸如此类。所以在2010年，这非常困难。我们当时创立的原因，Shane和我经常讨论DeepMind的创立原则是什么。有多个因素，一个是算法进步。

And of course in industry, you couldn't we couldn't get know, it's difficult to get 2¢ together and which is hard to imagine now as well given that it's the biggest sort of buzzword in in VCs and fundraising is easy and all these kind of things today. So back in 2010, it was very difficult. And what we the reason we started then, and Shane and I used to discuss what were the sort of founding tenants of DeepMind. And it was various things. One was algorithmic advances.

所以深度学习，你知道，杰夫·辛顿和他的团队在学术界刚刚发明了这项技术，但工业界没人知道。我们热爱强化学习。我们认为它可以规模化。同时，对人脑的理解在过去十年里也取得了很大进展，这得益于fMRI机器和其他技术。

So deep learning, you know, Jeff Hinton and Co. Had just sort of invented that in academia, but no one in industry knew about it. We love reinforcement learning. We thought that could be scaled up. But also understanding about the human brain had advanced quite a lot in the decade prior with fMRI machines and other things.

因此我们可以获得一些关于大脑使用的架构、算法和某种表示的线索。这是在系统层面，而不是实现层面。另外两个重要因素是计算能力和GPU。我们可以看到计算能力将非常有用，并且通过游戏行业，它已经变得商品化。我们可以利用这一点。

So we could get some good hints about architectures and algorithms and sort of representations maybe that the brain uses. So at a systems level, not at an implementation level. And then the other big things were compute and GPUs. So we could see a compute was going to be really useful and it got to a place where it become commoditized mostly through the games industry. And and that could be taken advantage of.

最后一点是智能的数学和理论定义。比如AIXI、AIXE，这是肖恩和他的导师马库斯·赫特研究的，它实际上是一种通用智能的理论证明，本质上是一个强化学习系统。嗯。在极限情况下，它假设无限的计算能力和无限的内存，就像图灵机证明的那样。但我也在等待类似的东西出现，你知道，像图灵和香农等人提出的图灵机和计算理论是现代计算机科学的基础。

And then the final thing was also mathematical and theoretical definitions of intelligence. So things like AIXI, AIXE, which Shane worked on with his supervisor Marcus Hutter, which is this sort of theoretical proof really of universal intelligence, which is actually a reinforcement learning system. Mhmm. In the limit, I mean, it assumes infinite compute and infinite memory in the way, know, like a Turing machine proves. But I was also waiting to see something like that too To you know, like Turing machines and and computation theory that people like Turing and Shannon came up with underpins modern computer science.

你知道，我一直在等待这样一个理论来支撑AGI研究。所以当我遇到肖恩，看到他在研究类似的东西时，对我来说就像是拼图的最后一块。所以在早期，我认为想法是最重要的。对我们来说，就是深度强化学习，规模化深度学习。当然，我们也看到了Transformer的出现。

You know, I was waiting for a theory like that to sort of underpin AGI research. So when I met Shane and saw he was working on something like that, that to me was a sort of final piece of the jigsaw. So in the early days, I would say that ideas were the most important. And for us, was deep reinforcement learning, scaling up deep learning. Of course, we've seen transformers.

所以我会说，从2010年到现在，有了巨大的飞跃，大概有三四次。比如AlphaGo这样的巨大进化。可能还需要一些更多的突破。但随着我们越来越接近AI、AGI，我认为工程和数据变得越来越重要。因为规模，当然还有GPT-3和所有大型语言模型以及包括我们自己的大模型的最新成果表明，规模和大模型显然是AGI解决方案的必要部分，但可能还不够充分。

So huge leaps I would say, three or four if you think from 2010 till now. Huge evolutions, things like AlphaGo. And maybe there's a few more still needed. But as we get closer to AI, AGI, I think engineering becomes more and more important and data. Because scale and of course the recent results of GPT-three and all the big language models and large models including our ones has shown that scale is and large models are clearly going to be a necessary but perhaps not sufficient part of an AGI solution.

正如你所说，在整个过程中，我想向你表示衷心的感谢。你是这方面的先驱之一，坚持像强化学习这样的想法，认为它实际上可以成功，尽管阿克沙伊过去取得的成功有限。而且，虽然我们仍然不知道，但自豪地拥有世界上最好的研究人员，并谈论解决智能问题。所以谈论无论你称之为AGI还是类似的东西，这在MIT是你会提到的，而不是你不会提起的事情。

And throughout that, like you said, and I'd like to give you a big thank you. You're one of the pioneers in this is sticking by ideas like reinforcement learning, that this can actually work, given Akshay's limited success in the past. And also, which we still don't know, but proudly having the best researchers in the world and talking about solving intelligence. So talking about whatever you call it, AGI or something like this, that's speaking of MIT, that's that's just something not you wouldn't bring up.

不。

No.

不，也许你在四五十年前做过。

Not not maybe you did in, like, forty, fifty years ago.

嗯。

Mhmm.

但那时候人工智能是一个让你捣鼓的地方，规模很小，项目也不怎么宏大。是的。也许最宏大的项目是在机器人领域，比如DARPA挑战赛。当然。但解决智能的任务并相信你能做到，这真的非常非常强大。

But that was AI was a place where you do tinkering, very small scale, not very ambitious projects. Yeah. And maybe the biggest ambitious projects were in the space of robotics and doing, like, the DARPA challenge. Sure. But the task of solving intelligence and believing you can, that's really, really powerful.

所以为了让工程发挥作用，让优秀的工程师构建伟大的系统，你必须拥有那种信念，贯穿始终的信念，即你实际上可以解决一些这些不可能的挑战。

So in order for engineering to do its work, to to have great engineers build great systems, you have to have that belief, that threads throughout the whole thing that you can actually solve some of these impossible challenges.

是的，没错。而且早在2010年，你知道，我们的使命宣言，今天仍然是，它曾经是第一步，解决智能。是的。第二步，用它来解决其他一切问题。

Yeah. That's right. And and and back in 2010, you know, our mission statement and still is today, you know, is it was used to be solving step one, solve intelligence. Yeah. Step two, use it to solve everything else.

是的。所以你可以想象在2010年向风投推销这个时，你知道我们得到的那种眼神。我们设法找到了一些古怪的人支持我们，但这很棘手。而且我到了我们不会向任何教授提及的地步，因为他们只会翻白眼，认为我们职业自杀了。所以我们有很多事情必须做。

Yes. So if you can imagine pitching that to a VC in 2010, you know, the kind of looks we we got. We managed to, you know, find a few kooky people to back us, but it was it was tricky. And and I and I got to the point where we we wouldn't mention it to any of our professors because they would just eye roll and think we committed career suicide. So there's a lot of things that we had to do.

但我们一直相信它。顺便说一句，我一直相信强化学习的一个原因是，如果你看神经科学，那就是灵长类大脑学习的方式。主要机制之一是多巴胺系统实现了某种形式的TD学习。这是90年代末一个非常著名的结果，他们在猴子身上看到了这一点，并传播预测误差。所以再次在极限情况下，这就是我认为你可以利用神经科学来做任何数学的地方。

But we always believed it. By the way, one reason I've always believed in reinforcement learning is that if you look at neuroscience, that is the way that the primate brain learns. One of the main mechanisms is the dopamine system implements some form of TD learning. It's a very famous result in the late 90s where they saw this in monkeys and propagating prediction error. So again in the limit, this is what I think you can use neuroscience for is at any mathematics.

当你从事像解决智能这样雄心勃勃的蓝海研究时，没有人知道该怎么做，你需要利用任何证据或信息来源来引导你走向正确的方向，或给你信心确认方向正确。这就是我们在这方面如此努力的一个原因。回到你之前关于组织的问题，我认为我们在DeepMind为鼓励发明和创新所做的另一项重大创新是我们建立并至今保持的多学科组织。DeepMind最初是神经科学最前沿知识与机器学习、工程学、数学和游戏领域的交汇。自那以后，我们进一步扩展了它。

When you're doing something as ambitious as trying to solve intelligence and it's blue sky research, no one knows how to do it, you need to use any evidence or any source of information you can to help guide you in the right direction or give you confidence you're going in the right direction. So that was one reason we pushed so hard on that. And just going back to your earlier question about organization, the other big thing that I think we innovated with at DeepMind to encourage invention and innovation was the multidisciplinary multidisciplinary organization we built and we still have today. So DeepMind originally was a confluence of the most cutting edge knowledge in neuroscience with machine learning, engineering, and mathematics, and gaming. And then since then, we've built that out even further.

所以我们这里有哲学家和伦理学家，还有其他类型的科学家、物理学家等等。这正是我试图构建的一种新型贝尔实验室，但处于其黄金时代。对吧？并且以一种新的表现形式来尝试培育这种不可思议的创新机器。所以谈到机器中的人类，DeepMind本身就是一个学习机器，里面有许多出色的人类智慧汇聚在一起，试图构建这些学习系统。

So we have philosophers here and ethicists, but also other types of scientists, physicists, and so on. And that's what brings together I tried to build a sort of new type of Bell Labs, but in its golden era. Right? And and a new expression of that to try and foster this incredible sort of innovation machine. So talking about the humans in the machine, DeepMind itself is a learning machine with a lots of amazing human minds in it coming together to try and build these learning systems.

如果我们回到AlphaFold那个宏大的雄心梦想，它可能只是生物学漫长旅程的早期步骤。你认为同样的方法能否用于预测更复杂生物系统的结构和功能？比如多蛋白质相互作用，然后，我的意思是，你可以从这里扩展出去。是的，就是模拟越来越大的系统，最终模拟像人脑或人体这样的东西。

If we return to the big ambitious dream of AlphaFold that may be the early steps on a very long journey in in biology. Do you think the same kind of approach can use to predict the structure and function of more complex biological systems? So multi protein interaction, and then, I mean, you can go out from there. Yeah. Just simulating bigger and bigger systems that eventually simulate something like the human brain or the human body.

这只是一大团乱麻，生物学那美丽而坚韧的混乱。你……你认为这是一个长期的愿景吗？

It's just the big mush, the mess of the beautiful resilient mess of biology. Do you do you see that as a long term vision?

我认为是的。而且我想，你知道，如果你考虑一旦我们拥有足够强大的系统，我最想将AI应用于哪些事情，生物学、治愈疾病和理解生物学就在我的清单上名列前茅。这是我个人推动AlphaFold及其相关工作的原因之一。但我认为AlphaFold虽然惊人，却仅仅是个开始。我希望它能证明计算方法所能达到的成就。

I do. And I think, you know, if you think about what are the things top things I wanted to apply AI to once we had powerful enough systems, Biology and curing diseases and understanding biology was right up there, top of my list. That's one of the reasons I personally pushed that myself and with AlphaFold. But I think AlphaFold, amazing as it is, is just the beginning. And I hope it's evidence of what could be done with computational methods.

所以AlphaFold解决了蛋白质结构这个巨大问题。但生物学是动态的。因此，我从这里设想的是，并且我们现在正在研究所有这些事情，就是蛋白质-蛋白质相互作用，蛋白质-配体结合，也就是与分子反应。然后你想要构建到通路层面。最终实现一个虚拟细胞。

So AlphaFold solved this huge problem of the structure of proteins. But biology is dynamic. So really, what I imagine from here and we're working on all these things now is protein protein interaction, protein ligand binding, so reacting with molecules. Then you want to build up to pathways. And then eventually a virtual cell.

这是我的梦想，也许在未来十年内。实际上我一直在与我的许多生物学家朋友讨论，比如保罗·纳斯，他运营着克里克研究所，是一位杰出的生物学家，诺贝尔奖得主。我们二十年来一直在讨论虚拟细胞。你能构建一个细胞的虚拟模拟吗？如果能，那对生物学和疾病发现将是不可思议的，因为你可以在虚拟细胞上进行大量实验，然后只在最后阶段在湿实验室中进行验证。

That's my dream, maybe in the next ten years. And I've been talking actually to a lot of biologist friends of mine, Paul Nurse, who runs the Crick Institute, amazing biologist, Nobel Prize winning biologist. We've been discussing for twenty years now virtual cells. Could you build a virtual simulation of a cell? And if you could, that would be incredible for biology and disease discovery because you could do loads of experiments on the virtual cell and then only at the last stage validate it in the wet lab.

因此，在发现新药的搜索空间方面，从确定靶点到获得候选药物大约需要十年时间。如果大部分工作能在计算机模拟中完成，这个时间或许能缩短一个数量级。为了实现虚拟细胞，我们必须逐步理解生物学的不同部分及其相互作用。每隔几年，我们都会讨论这个问题，我和保罗也讨论过。最终，在AlphaFold问世后，我去年说，现在终于可以着手进行了。

So in terms of the search space of discovering new drugs, takes ten years roughly to go from identifying a target to having a drug candidate. Maybe that could be shortened to by an order of magnitude if you could do most of that that work in silico. So in order to get to a virtual cell, we have to build up understanding of different parts of biology and the interactions. And and so we, know, we every every few years, we talk about this with I talked about this with Paul. And then finally, last year after AlphaFold, I said, now's the time we can finally go for it.

AlphaFold是证明这可能实现的第一个证据点。他非常兴奋，我们与他的实验室有一些合作。实际上，他们就在我们马路对面。能在国王十字区这里，与马路对面的克里克研究所为邻，真是太棒了。我认为下一步，基于AlphaFold等技术，生物学将取得一些惊人的进展。

And and AlphaFold is the first proof point that this might be possible. And he's very excited, and we have some collaborations with his with his lab. They're just across the road actually from us. So it's just, you know, wonderful being here in Kings Cross with the Crick Institute across the road. And I think the next steps, you know, I think there's going to be some amazing advances in biology built on top of things like AlphaFold.

在我们开源并发布后，我们已经看到社区在这方面取得的成果。我常说，如果数学是物理学的完美描述语言，那么人工智能最终可能是生物学的完美描述语言。因为生物学如此混乱、涌现性强、动态且复杂，我很难相信我们会找到像牛顿运动定律那样优雅的法则来描述一个细胞，对吧？

We're already seeing that with the community doing that after we've open sourced it and released it. And I often say that I think if you think of mathematics is the perfect description language for physics, I think AI might end up being the perfect description language for biology. Because biology is so messy, it's so emergent, so dynamic and complex. I think I find it very hard to believe we'll ever get to something as elegant as Newton's laws of motions to describe a cell. Right?

这实在太复杂了。所以我认为人工智能是解决这个问题的合适工具。

It's just too complicated. So I think AI is the right tool for for this.

你必须从基本构建块开始，利用人工智能来模拟所有这些构建块。因此，需要一种非常强大的方法来预测，给定这些构建块，会产生什么样的生物学功能以及该生物系统的演化。这几乎就像细胞自动机。你必须运行它，无法从高层次进行分析。

You have to you have to start at the basic building blocks and use AI to run the simulation for all those building blocks. So have a very strong way to do prediction of what given these building blocks, what kind of biology how the the function and the evolution of that biological system. It's almost like a cellular automata. You have to run it. You can't analyze it from a high level.

你必须从基本成分入手，找出规则，然后让它运行。但在这种情况下，规则非常难以确定。

You have to take the basic ingredients, figure out the rules Yeah. And let it run. But in this case, the rules are very difficult to figure out.

是的。

Yes.

必须学习它们。

Have to learn them.

正是如此。所以生物学太复杂了，难以找出规律。它太涌现、太动态了，比如说，与像行星运动这样的物理系统相比。是的，对吧？

That's exactly it. So it's the biology is too complicated to figure out the rules. It's it's too emergent, too dynamic, say, compared to a physics system like the motion of a planet. Yeah. Right?

因此你必须学习这些规则，而这正是我们正在构建的系统类型。

And and so you have to learn the rules, and that's exactly the type of systems that we're building.

所以你提到你们开源了AlphaFold甚至相关数据。对我个人而言，也非常高兴并衷心感谢你们开源了Mujoco，那个常用于机器人研究等的物理模拟引擎。我认为这是一个相当霸气的举动。我的意思是，很少有公司或个人会做这种事。这背后的哲学是什么？

So you you mentioned you've open sourced AlphaFold and even the data involved. To me personally, also really happy and a big thank you for open sourcing Mujoco, the physics simulation engine that's that's often used for robotics research and so on. So I think that's a pretty gangster move. So what what what's the what's I mean, this very few companies or people do that kind of thing. What's the philosophy behind that?

你知道，这是具体情况具体分析的。在这两种情况下，我们都认为这样做对人类和科学界能带来最大利益。一个是机器人物理界，通过Mujoco。所以我们收购了它。我们收购它是为了，是的。

You know, it's a case by case basis. And in both those cases, we felt that was the maximum benefit to humanity to do that and and the scientific community. In one case, the robotics physics community with Majoco. So We purchased it. We purchased it for to yes.

我们收购它的明确原则就是为了将其开源。所以，你知道，我希望人们能欣赏这一点。听到你确实欣赏，这很棒。第二件事是，我们主要这么做是因为构建它的人无法再承担支持它的工作了，因为它对他来说变得太大了。他是一位最初构建它的杰出教授。

We purchased it for the express principle to open source it. So so, you know, I hope people appreciate that. It's great to hear that that you do. And then the second thing was and mostly we did it because the person building it is would not able was not able to cope with supporting it anymore because it was it got too big for for him. He's an amazing professor who who built it in the first place.

所以我们在这方面帮助了他。至于AlphaFold，我认为它甚至更大。在那种情况下，我们认定AlphaFold有太多下游应用，我们甚至无法想象它们全部是什么。因此，加速药物发现和基础研究的最佳方式就是免费提供所有数据以及系统本身。你知道，看到人们在短短一年内（这在科学上是很短的时间）所取得的成果，真是令人无比欣慰。

So we helped him out with that. And then with AlphaFold, it's even bigger, I would say. And I think in that case, we decided that there were so many downstream applications of AlphaFold that we couldn't possibly even imagine what they all were. So the best way to accelerate drug discovery and also fundamental research would be to to give all that data away and and and the and the and the system itself. You know, it's been so gratifying to see what people have done that within just one year, which is a short amount of time in science.

它已被超过50万研究人员使用。我们认为这几乎是世界上所有的生物学家。我认为全球大约有50万名专业生物学家用它来研究他们感兴趣的蛋白质。我们已经看到了令人惊叹的基础研究成果。几周前，《科学史》特刊的封面就是核孔复合体，它是人体中最大的蛋白质之一。

And it's been used by over 500,000 researchers have used it. We think that's almost every biologist in the world. I think there's roughly 500,000 biologists in the world, professional biologists, have used it to to look at their proteins of interest. We've seen amazing fundamental research done. So a couple of weeks ago, front cover there was a whole special history of science, including the front cover, which had the nuclear pore complex on it, which is one of the biggest proteins in the body.

核孔复合体是一种控制所有营养物质进出细胞核的蛋白质。它们就像小小的门户，开合以让物质进出细胞核。因此它们非常重要。但它们非常巨大，因为它们是巨大的环形结构。几十年来，人们一直在试图解析其结构。

The nuclear pore complex is a protein that governs all the nutrients going in and out of your cell nucleus. So they're like little gateways that open and close to let things go in and out of your cell nucleus. So they're really important. But they're huge because they're massive donut ring shaped things. And they've been looking to try and figure out that structure for decades.

他们拥有大量实验数据，但分辨率太低，有些部分缺失。他们能够像拼巨型乐高拼图一样，利用AlphaFold的预测加上实验数据，将这两个独立的信息源结合起来。实际上，全球有四个不同的研究小组几乎同时使用AlphaFold预测将其组装起来。这真是太棒了。

And they have lots of experimental data, but it's too low resolution. There's bits missing. And they were able to, like a giant Lego jigsaw puzzle, use AlphaFold predictions plus experimental data and combined those two independent sources of information. Actually, four different groups around the world were able to put it together more or less simultaneously using AlphaFold predictions. So that's been amazing to see.

几乎每家制药公司、每位与我交谈过的药企高管都表示，他们的团队正在使用AlphaFold来加速他们试图发现的任何药物。因此，我认为AlphaFold所产生的影响，其连锁效应是巨大的。而且它

And pretty much every pharma company, every drug company executive I've spoken to has said that their teams are using AlphaFold to accelerate whatever drugs they're trying to discover. So I think the knock on effect has been enormous in terms of the impact that AlphaFold has made. And it's

可能正在吸引并培养生物学家。它正吸引更多人进入这一领域，无论是出于对其兴奋点还是所涉及的技术技能。它几乎像是进入生物学领域的‘入门药’。

probably bringing in it's creating biologists. It's bringing more people into the field, both on the excitement and both on the technical skills involved. And it's almost like a gateway drug to biology.

是的，确实如此。它也能让更多计算领域的人参与进来。并且我认为对我们来说，正如我所说，下一阶段，未来我们还必须考虑其他因素。我们正在AlphaFold的基础上，以及我与你讨论过的关于蛋白质相互作用、基因组学等其他想法上进行构建。并非所有东西都会开源。

Yes. It is. You can get more computational people involved too, And and I think for us, you know, the next stage as I said, you know, in future, we have to have other considerations too. We're building on top of AlphaFold and these other ideas I discussed with you about protein protein interactions and and genomics and other things. And not everything will be open source.

其中一些会进行商业化，因为这将是为其争取最多资源和产生最大影响的最佳方式。在其他方面，一些其他项目将以非营利方式进行。此外，对于未来的事物，我们还必须考虑安全与伦理问题，比如合成生物学。你知道，存在双重用途，我们也必须考虑到这一点。对于AlphaFold，我们咨询了30位不同的生物伦理学家以及该领域的其他专家，以确保其在发布前是安全的。

Some of it will will do commercially because that will be the best way to actually get the most resources and impact behind it. In other ways, some other projects will do nonprofit style. And also we have to consider for future things as well, safety and ethics as well, like, you know, synthetic biology. There are, you know, there is dual use and we have to think about that as well. With AlphaFold, we, you know, we consulted with 30 different bioethicists and and other people expert in this field to make sure it was safe before we released it.

所以未来还会有其他考量。但就目前而言，我认为AlphaFold是我们献给科学界的一份礼物。

So there'll be other considerations in the future. But for right now, you know, I think AlphaFold is a a kind of a a gift from us to to to the scientific community.

所以我非常确信像AlphaFold这样的成果未来会成为诺贝尔奖的一部分。但我们人类当然不擅长归功，所以最终肯定会把奖项颁给人类。你认为会有那么一天，AI系统无法被否认它赢得了诺贝尔奖吗？你觉得我们会在21世纪看到这一幕吗？

So I'm pretty sure that something like AlphaFold will be part of Nobel Prizes in the future. But us humans, of course, are horrible with credit assignment, so we'll, of course, give it to the humans. Do you think there will be a day when AI system can't be denied that it earned that Nobel Prize? Do you think we will see that in twenty first century?

这取决于我们最终构建出什么样的AI。对吧？取决于它们是否是能够设定目标、提出假设、决定要攻克哪些问题的目标导向型智能体。

It depends what type of AIs we end up building. Right? Whether they're, you know, goal seeking agents who specifies the goals, who comes up with the hypotheses, who, you know, who determines which problems to tackle. Right?

所以推特上我思考过这个问题。宣布

So Tweets I think about it. Announcement of

研究成果。宣布成果本身就是其中的一部分。所以我认为目前当然是由惊人的人类智慧在背后驱动这些系统。而在我看来，系统只是一个工具。就像说伽利略和望远镜，智慧和功劳应该归于望远镜一样荒谬。

the results. It's announcing the results exactly as part of it. So I think right now, of course, it's it's it's it's amazing human ingenuity that's behind these systems. And then the system, in my opinion, is just a tool. You know, it'd be a bit like saying with Galileo and telescope, the ingenuity, the credit should go to the telescope.

我的意思是，显然是伽利略制造了这个工具然后使用它。所以即使这些工具能自主学习，我至今仍以同样视角看待它们。我认为像AlphaFold和我们正在构建的系统，是科学和获取新知识的终极工具，帮助科学家获得新认知。我相信总有一天，AI系统可能会自主解决或提出像广义相对论这样的突破性理论，而不是仅仅通过整合互联网或PubMed上的所有信息——虽然看到那种结果也会很有趣。

I mean, it's clearly Galileo building the tool which he then uses. So I still see that in the same way today even though these tools learn for themselves. I think of things like AlphaFold and the things we're building as the ultimate tools for science and for acquiring new knowledge to help us as scientists acquire new knowledge. I think one day there will come a point where an AI system may solve or come up with something like general relativity of its own bat, not just by averaging everything on the internet or averaging everything on PubMed. Although that would be interesting to see what that would come up with.

这在我看来有点像我们之前关于创造力的辩论，是发明围棋而不仅仅是下出一步好棋。所以我认为，如果要授予它诺贝尔奖级别的荣誉，它需要像凭空发明围棋那样，独立提出全新猜想，而不是由人类科学家或创造者指定方向。因此目前它绝对只是一个工具。

So that to me is a bit like our earlier debate about creativity, inventing Go rather than just coming up with a good Go move. And so I think solving I think to you know, if we wanted to give it the credit of like a Nobel type of thing, then it would need to invent Go and sort of invent that new conjecture out of the blue rather than being specified by the the the human scientists or the human creators. So I think right now that's it's definitely just a tool.

尽管如你所说，通过平均互联网上的一切能走多远确实很有趣，因为你知道，很多人确实认为科学总是站在巨人的肩膀上。问题是，你究竟在多大程度上真正超越了巨人的肩膀？也许只是吸收了过去的各种成果，并最终以这种新视角给了你这个突破性的想法。但这个想法可能并不新颖，以至于它无法在互联网上被发现。也许未来一百年的诺贝尔奖都已经在互联网上等着被发现了。

Although it is interesting how far you get by averaging everything on the Internet, like you said, because, you know, a lot of people do see science as you're always standing on the shoulders of giants. And the question is, how much are you really reaching up above the shoulders of giants? Maybe it's just assimilating different kinds of results of the past with ultimately this new perspective that gives you this breakthrough idea. But that idea may not be novel in the way that it can't be already discovered in the Internet. Maybe the Nobel Prizes of the next hundred years are already all there on the Internet to be discovered.

它们可能是。它们可能是。我的意思是，我认为这是一个巨大的谜团，首先，我相信未来几十年甚至过去十年即将到来的许多重大新突破将出现在不同学科领域的交叉点，在那里会发现看似无关领域之间的新联系。正如我之前所说，人们甚至可以将DeepMind视为神经科学思想和人工智能工程思想之间最初的一种跨学科结合。所以我认为有这一点。

They they could be. They could be. I mean, I think this is one of the big mysteries I think is that I I first of all, I believe a lot of the big new breakthroughs that are gonna come in the next few decades and even in the last decade are gonna come at the intersection between different subject areas where there'll be some new connection that's found between what seemingly were disparate areas. And one can even think of DeepMind, as I said earlier, as a sort of interdisciplinary between neuroscience ideas and AI engineering ideas originally. And so I think there's that.

然后我们今天无法想象的一件事，也是我认为人们对大型模型效果如此之好感到惊讶的原因之一是，实际上，我们人类有限的心智很难理解阅读整个互联网会是什么样子。对吧？我认为我们可以做一个思想实验。我以前常这样做，比如，如果我读完了整个维基百科，我会知道什么？

And then one of the things we can't imagine today is and one of the reasons I think people we were so surprised by how well large models worked is that actually it's very hard for our human minds, our limited human minds, to understand what it would be like to read the whole internet. Right? I think we can do a thought experiment. And I used to do this of like, well, what if I read the whole of Wikipedia? What would I know?

我认为我们的心智也许勉强能理解那会是什么样子。但整个互联网是超出理解范围的。所以我认为我们只是不明白，能够同时在脑海中容纳所有那些信息并可能同时激活，然后那里可能有哪些可用的联系。所以我认为毫无疑问，有巨大的东西就像那样等着被发现。

And I think our minds can just about comprehend maybe what that would be like. But the whole Internet is beyond comprehension. So I think we just don't understand what it would be like to be able to hold all of that in mind potentially, right? And then active at once and then maybe what are the connections that are available there. So I think no doubt there are huge things to be discovered just like that.

但我确实认为存在另一种创造力，即真正的新知识火花，前所未有的新想法，无法从已知事物中平均得出。当然，一切都，你知道，没有人能在真空中创造。所以线索一定在某个地方。但只是以一种独特的方式将这些事物组合起来，我认为历史上一些最伟大的科学家已经展示了这一点。尽管回到他们的时代，很难确切知道当他们提出那些东西时究竟已知些什么。

But I do think there is this other type creativity of true spark of new knowledge, new idea never thought before about can't be averaged from things that are known. That really, of course, everything come, you know, nobody creates in a vacuum. So there must be clues somewhere. But just a unique way of putting those things together, I think some of the greatest scientists in history have displayed that I would say. Although it's very hard to know going back to their time what was exactly known when they came up with those things.

虽然，你，你真的让我思考，因为仅仅是深入了解100个维基百科页面的思想实验。

Although, I the you're you're making me really think because just the thought experiment of deeply knowing a 100 Wikipedia pages.

我

I

我不认为我能...我对维基百科在技术主题上的表现印象非常深刻。是的。所以如果你了解一百页或一千页的内容，我不认为你能直观地理解那是一种什么样的智能。是的。如果你知道如何正确使用和整合这些信息，它是一种相当强大的智能。

don't think I can I've been really impressed by Wikipedia for for technical topics. Yeah. So if you know a 100 pages or a thousand pages, I don't think you can visually comprehend what's what kind of intelligence that is. Yeah. It's a pretty powerful intel if you know how to use that and integrate that information correctly

是的。

Yes.

我认为你可以走得很远。是的。你或许可以基于此构建思想实验，比如模拟不同的想法。嗯。所以如果这是真的，让我运行这个思想实验，那么也许这个也是真的。

I think you can go really far. Yeah. You can probably construct thought experiments based on that, like simulate different ideas. Mhmm. So if this is true, let me run this thought experiment, then maybe this is true.

这其实不是发明。更像是直接获取知识并用它来构建一个非常基础的世界模拟。我的意思是，有些人认为这某种程度上是浪漫的，但爱因斯坦也会用思想实验做类似的事情。

It's not really invention. It's like just taking literally the knowledge and using it to construct a very basic simulation of the world. I mean, some argue it's romantic in part, but Einstein would do the same kind of things with the thought experiments.

对吧？可以想象系统性地对数百万维基百科页面加上PubMed等所有内容这样做。我认为通过这种方式可以发现许多非常有用的东西。你知道，可以想象，我希望我们在材料科学方面做一些这样的事情，比如室温超导体是我某天的目标之一。我希望有一个AI系统来帮助构建更优化的电池。

Right? One could imagine doing that systematically across millions of Wikipedia pages plus PubMed, all these things. I think there are many many things to be discovered like that that are hugely useful. You know, you could imagine and I want us to do some of these things in material science like room temperature superconductors is something on my list one day. I'd like to have an AI system to help build better optimized batteries.

所有这些机械类的东西。我认为由模型引导的系统性搜索可能会极其强大。

All of these sort of mechanical things. I think a systematic sort of search could be guided by a model, could be could be extremely powerful.

说到这个，你有一篇关于核聚变的论文，通过深度强化学习对托卡马克等离子体进行磁控制。所以你试图用深度强化学习解决核聚变问题，也就是控制高温等离子体。你能解释一下这项工作吗？AI最终能解决核聚变吗？

So speaking of which, you have a paper on nuclear fusion, magnetic control of tokamak plasmas through deep reinforcement learning. So you you're seeking to solve nuclear fusion with DeepRL, so it's doing control of high temperature plasmas. Can you explain this work? And can AI eventually solve nuclear fusion?

过去一两年非常有趣且富有成效，因为我们启动了许多我梦想中的项目，可以说是我多年来收集的科学领域想法，我认为如果我们帮助加速这些领域，它们可能会带来变革，并且这些问题本身也是极具吸引力的科学挑战。这就是能源领域。是的，没错。所以是能源和气候。

It's been very fun last year or two and very productive because we've been taking off a lot of my dream projects, if you like, of things that I've collected over the years of areas of science that I would like to I think could be very transformative if we helped accelerate and really interesting problems, scientific challenges in of themselves. So this is energy. So energy. Yes, exactly. So energy and climate.

我们之前谈到疾病和生物学是我认为人工智能能发挥最大作用的领域之一。我认为能源和气候是另一个重要领域。所以这两者可能是我最关注的两个方向。聚变是我认为人工智能能提供帮助的一个领域。聚变面临许多挑战，主要是物理学、材料科学和工程学方面的挑战，需要建造这些巨大的聚变反应堆并约束等离子体。

So we talked about disease and biology as being one of the biggest places I think AI can help with. I think energy and climate is another one. So maybe they would be my top two. And fusion is one area I think AI can help with. Now fusion has many challenges, mostly physics, material science and engineering challenges as well to build these massive fusion reactors and contain the plasma.

每当我们进入一个新领域应用我们的系统时，我们会与领域专家交流。我们努力寻找世界上最好的人合作。在聚变领域，我们与瑞士的EPFL（瑞士联邦理工学院）合作，他们非常出色。他们有一个测试反应堆，愿意让我们使用，我向团队确认过，我们会小心安全地使用它。

And what we try to do whenever we go into a new field to apply our systems is we talk to domain experts. We try and find the best people in the world to collaborate with. In this case, in Fusion, we collaborated with EPFL in Switzerland, the Swiss Technical Institute, are amazing. They have a test reactor that they were willing to let us use, which, you know, I double checked with the team. We were gonna use carefully and safely.

我很佩服他们能说服对方让我们使用它。他们那里的测试反应堆非常棒，他们会在上面尝试各种相当疯狂的实验。我们通常会审视的是，当我们进入像聚变这样的新领域时，所有的瓶颈问题是什么？就像从第一性原理思考，今天仍然阻碍聚变工作的所有瓶颈问题是什么？然后我们让聚变专家告诉我们。

I was impressed they managed to persuade them to let us use it. And and it's a it's an amazing test reactor they have there, And they try all sorts of pretty crazy experiments on it. What we tend to look at is if we go into a new domain like fusion, what are all the bottleneck problems? Like thinking from first principles, what are all the bottleneck problems that are still stopping fusion working today? And then we get a fusion expert to tell us.

然后我们审视这些瓶颈，看看哪些是目前我们的AI方法可以处理的。并且从研究角度、从我们的视角、从AI的角度来看会很有趣。并且这能解决他们的一个瓶颈。在这个案例中，等离子体控制是完美的选择。所以等离子体，它的温度大约有一百万摄氏度。

And then we look at those bottlenecks and we look at the ones which ones are amenable to our AI methods today. And would be interesting from a research perspective, from our point of view, from an AI point of view. And that would address one of their bottlenecks. And in this case, plasma control was perfect. So the plasma, it's a million degrees Celsius, something like that.

它比太阳还热。显然没有任何材料能容纳它。所以他们必须用非常强大的超导磁场来约束它。但问题是，等离子体相当不稳定，正如你所想象的。你就像是在反应堆里 holding 着一个迷你太阳、迷你恒星。

It's hotter than the sun. And there's obviously no material that can contain it. So they have to be containing these magnetic, very powerful superconducting magnetic fields. But the problem is, plasma is pretty unstable as you imagine. You're kind of holding a mini sun, mini star in a reactor.

所以你需要在等离子体行动之前预测它的行为，以便在几毫秒内移动磁场来基本上约束它接下来的动作。所以如果你把它看作一个强化学习预测问题，这似乎是一个完美的问题。所以你有一个控制器。你来移动磁场。在我们介入之前，他们用的是传统的运筹学类型的控制器，这些控制器是手工制作的。

So you kind of want to predict ahead of time what the plasma is going to do so you can move the magnetic field within a few milliseconds to basically contain what it's going to do next. So it seems like a perfect problem if you think of it for like a reinforcement learning prediction problem. So you got controller. You're to move the magnetic field. And until we came along, they were doing it with traditional operational research type of controllers, which are kind of handcrafted.

问题当然在于，它们无法实时响应等离子体的变化。必须是硬编码的。而且，考虑到这通常是我们首选的解决方案，我们更希望能学会自适应调整。他们还有一个等离子体模拟器，所以有很多标准与我们想要使用的条件相匹配。

And the problem is, of course, they can't react in the moment to something the plasma is doing. Have to be hard coded. And again, knowing that that's normally our go to solution is we would like to learn that instead. And they also had a simulator of these plasma. So there were lots of criteria that matched what we we like to to to to use.

那么人工智能最终能解决核聚变问题吗？

So can AI eventually solve nuclear fusion?

嗯，针对这个问题，我们在去年的《自然》论文中发表了成果，通过将等离子体控制在特定形状中实现了聚变。实际上，这几乎就像是在将等离子体雕刻成不同形状

Well, so we with this problem, and we published it in the Nature paper last year, we held the fusion that we held the plasma in specific shapes. So actually, it's almost like carving the plasma into different shapes

好的。

Okay.

并控制并将其维持在那里创纪录的时间。所以这算是解决了聚变的一个难题。

And control and hold it there for a record amount of time. So so that's one of the problems of of fusion sort of solved.

所以是有一个控制器能够，无论什么形状

So have a controller that's able to, no matter the shape

约束住它。

Contain it.

包含

Contain

它。是的。将其约束并保持在结构中，存在更适合能量产生的不同形状，称为液滴等等。所以这非常重大。现在我们正在与许多聚变初创公司交流，看看我们在聚变领域能解决的下一个问题是什么。

it. Yeah. Contain it and hold it in structure, there's different shapes that are better for for the energy productions called droplets and and and so on. So so that was huge. And now we're looking we're talking to lots of fusion startups to see what's the next problem we can tackle in the fusion area.

那么，在题为《通过解决分数电子问题推动密度泛函前沿》的论文中，另一个引人入胜的地方是。你们正在承担电子量子力学行为的建模与模拟工作。是的。能否解释这项研究，以及人工智能未来能否模拟任意量子力学系统？

So another fascinating place in a paper titled Pushing the Frontiers of Density Functionals by Solving the Fractional Electron Problem. So you're taking on modeling and simulating the quantum mechanical behavior of electrons. Yes. Can you explain this work and can AI model and simulate arbitrary quantum mechanical systems in the future?

是的。这是我关注了十年或更久的另一个问题，即模拟电子的特性。如果你能做到这一点，基本上就能描述元素、材料和物质的工作原理。所以，如果你想推动材料科学的发展，这就像是基础工作。我们有薛定谔方程，然后有它的近似方法——密度泛函理论。

Yeah. So this is another problem I've had my eye on for a decade or more, which is sort of simulating the properties of electrons. If you can do that, you can basically describe how elements and materials and substances work. So it's kind of like fundamental if you want to advance material science. And we have Schrodinger's equation, and then we have approximations to that density functional theory.

这些理论很有名。人们尝试编写这些泛函的近似形式，并提出电子云的描述，预测它们将如何运动，当两种元素结合时它们将如何相互作用。我们试图做的是学习一种模拟，学习一种能描述更多化学类型、更多化学反应规律的泛函。到目前为止，你可以运行昂贵的模拟，但只能模拟非常小的分子，非常简单的分子。

These things are famous. And people try and write approximations to these functionals and kind of come up with descriptions of the electron clouds, where they're going to go, how they're to interact when you put two elements together. And what we try to do is learn a simulation, learn a functional that will describe more chemistry, types of chemistry. So until now, you can run expensive simulations. But then you can only simulate very small molecules, very simple molecules.

我们希望模拟大型材料。而目前还没有办法做到这一点。我们正在努力构建近似薛定谔方程的泛函，从而让你能够描述电子的行为。所有材料科学和材料性质都受电子及其相互作用方式的支配。

We would like to simulate large materials. And so today there's no way of doing that. And we're building up towards building functionals that approximate Schrodinger's equation and then allow you to describe what the electrons are doing. And all material sort of science and material properties are governed by the electrons and how they interact.

所以，通过泛函得到一个很好的模拟总结，但这个总结仍然接近实际模拟会得出的结果。那么，这项任务有多困难？涉及哪些内容？是运行那些复杂的模拟吗？是的。并且学习从初始条件和模拟参数映射到泛函应该是怎样的任务？

So have a good summarization of the simulation through the functional, but one that is still close to what the actual simulation would come out with. So what how difficult is that task? What's involved in that task? Is it running those those complicated simulations Yeah. And learning the task of mapping from the initial conditions and the parameters of the simulation, learning what the functional would be?

是的，所以这相当棘手。但好在我们可以在计算集群上运行大量的分子动力学模拟。这样就会产生大量数据。在这种情况下，数据是生成出来的。

Yeah. So it's pretty tricky. And we've done it with you know, the nice thing is we there are we can run a lot of the simulations that the the molecular dynamic simulations on our compute clusters. And so that generates a lot of data. So in this case, the data is generated.

所以我们喜欢这类系统，这也是我们使用游戏的原因。这是模拟器生成的数据。我们基本上可以随心所欲地生成任意数量。只要云端有闲置的计算资源，我们就运行这些计算，对吧？

So we like those sort of systems, and that's why we use games. It's simulator generated data. And we can kind of create as much of it as we want, really. And just let's leave some you know, if any computers are free in the cloud, we just run we run some of these calculations. Right?

计算集群运算。

Compute cluster calculation.

我很喜欢闲置计算时间被用来研究量子力学的做法。

I like how the the the free compute time is used up on quantum mechanics.

没错。量子力学。确切地说是模拟，包括蛋白质模拟等。当人们不在YouTube上看猫咪视频时，我们正把这些计算机有效地用于量子化学研究。

Yeah. Quantum mechanics. Exactly. Simulations and protein simulations and other things. And so and so, you know, when you're not searching on YouTube for video, cat videos, we're using those computers usefully in quantum chemistry.

就是这个理念。让它们物尽其用。然后所有这些生成的计算数据，我们可以尝试从中学习函数关系——一旦掌握了函数关系，效率将远高于直接运行那些模拟。

It's the idea. And and putting them to good use. And then, yeah, and then all of that computational data that's generated, we can then try and learn the functionals from that, which of course are way more efficient once we learn the functional than running those simulations would be.

你认为有一天人工智能能否让我们实现类似破解物理学奥秘的突破？比如实现超光速旅行？

Do you think one day AI may allow us to do something like basically crack open physics? So do something like travel faster than the speed of light.

我毕生致力于AI的终极目标，也是我个人一生研究AI的原因，是为了打造一个工具来帮助我们理解宇宙。这意味着真正要研究物理学，以及现实的本质。所以我认为我们目前还没有能够做到这一点的系统。但当我们迈向通用人工智能（AGI）时，我认为这应该是我们首先应用AGI的领域之一。

My ultimate aim has always been with AI is the reason I am personally working on AI for my whole life. It was to build a tool to help us understand the universe. So I wanted to and that means physics really, and the nature of reality. So I don't think we have systems that are capable of doing that yet. But when we get towards AGI, I think that's one of the first things I think we should apply AGI to.

我想测试物理学的极限以及我们对物理学的认知。有太多我们不知道的事情。科学有一点让我着迷。作为一个科学方法的坚定拥护者，我认为它是人类有史以来最伟大的思想之一，让我们能够不断推进知识。但作为一个真正的科学家，我发现你了解得越多，就越意识到我们所知甚少。

I would like to test the limits of physics and our knowledge of physics. There's so many things we don't know. There's one thing I find fascinating about science. And as a huge proponent of the scientific method as being one of the greatest ideas humanity's ever had and allowed us to progress with our knowledge. But I think as a true scientist, I think what you find is the more you find out, the more you realize we don't know.

我一直觉得奇怪的是，为什么没有更多人为此感到困扰。你知道吗，每晚我都会思考所有这些我们时刻与之互动、却完全不知道其工作原理的事物。时间、意识、引力、生命。我的意思是，这些都是自然界最根本的东西。

And I always think that it's surprising that more people aren't troubled. You know, every night I think about all these things we interact with all the time that we have no idea how they work. Time, consciousness, gravity, life. We can't I mean, these are all the fundamental things of nature.

我认为我们

I think the way we

我们其实并不知道它们是什么。

We don't really know what they are.

为了生活，我们给它们贴上某些假设的标签，并把这些假设当作事实来对待。

To live life, we pin certain assumptions on them and kind of treat our assumptions as if they're fact.

是的。这让我们能够以某种方式将它们框定起来。

Yeah. That allows us to sort of Box them off somehow.

是的。把它们框起来

Yeah. Box them

以某种方式隔开。

off somehow.

嗯，是的，现实是当你思考时间时，你应该提醒自己，你应该把它从架子上拿下来，然后意识到，不，我们有一堆假设。现在仍然有很多争议。关于时间究竟是什么，还有很多不确定性。是否存在一个时间纪元？有很多基本问题，你不能仅仅做出假设。

Well, yeah, the reality is when you think of time, you should remind yourself, you should put it off the take it off the shelf and realize, like, no, we have a bunch of assumptions. There's still a lot of there's even now a lot of debate. There's a lot of uncertainty about exactly what is time. Is there an era of time? There's a lot of fundamental questions that you can't just make assumptions about.

也许人工智能让你不用把任何东西放在架子上。不做任何硬性假设，真正地开放它，看看

And maybe AI allows you to not put anything on the shelf. Not make any hard assumptions and really open it up and see what

正是如此。我认为我们应该对此真正保持开放的心态，而不是对某个特定理论教条化。它也将使我们能够构建更好的工具，最终是实验工具，然后可以测试某些今天可能无法测试的理论，比如我们一开始谈到的关于宇宙的计算本质。如果那是真的，人们可能会如何去测试它。对吧？

Exactly. I think we should be truly open minded about that and exactly that, not be dogmatic to a particular theory. It'll also allow us to build better tools, experimental tools eventually that can then test certain theories that may not be testable today about things about like what we spoke about at the beginning about the computational nature of the universe. How one might if that was true, how one might go about testing that. Right?

还有多少，你知道，有些人已经推测，像斯科特·阿伦森和其他人，你知道，一个特定的普朗克时空单位能包含多少信息。对吧？所以人们或许能够思考如何测试这些想法。如果有人工智能帮助你构建一些新的精密的实验工具。这就是我想象的，你知道，几十年后我们将能够做到。

And and how much, you know, there are people who've conjectured people like Scott Aronson and others about, you know, how much information can a specific planck unit of space and time contain. Right? So one might be able to think about testing those ideas. If you had AI helping you build some new exquisite experimental tools. This is what I imagine, you know, many decades from now we'll be able to do.

以及运行它们的模拟可以回答什么样的问题。所以你可以想象有很多物理模拟可以运行。嗯。以某种高效的方式，很像你在量子模拟工作中所做的那样。甚至可能是生命的起源。所以弄清楚在AlphaFold的工作开始之前，这一切是如何从一块石头中涌现出来的。是的。

And what kind of questions could be answered to running a simulation of of them. So there's bunch of physics simulations you can imagine that could be run Mhmm. In an so some some kind of efficient way, much like you're doing in the quantum simulation work. And perhaps even the origin of life. So figuring out how going even back before the work of AlphaFold begins, of how this whole whole thing emerges from a rock Yes.

从一个静态的事物说起。你认为人工智能将让我们能够——这是你关注的方向吗？它试图理解生命的起源。首先，你自己认为，生命究竟是如何在地球上起源的？

From a static thing. What do you what do you do you think AI will allow us to is that something you have your eye on? It's trying to understand the origin of life. First of all, yourself, what what do you think how the heck did life originate on Earth?

是的。嗯，也许我稍后会谈到这一点。但我认为人工智能的终极用途是最大限度地加速科学发展。所以我把它想象成一棵知识之树——假设这是宇宙中所有可获取的知识总和。

Yeah. Well, maybe we I'll come to that in a second. But I think the ultimate use of AI is to kind of use it to accelerate science to the maximum. So I think of it a little bit like the tree of all knowledge. If you imagine that's all the knowledge there is in the universe to attain.

尽管人类自启蒙时代以来取得了不错的进展，但迄今为止我们仅仅触及了这棵树的表层。而人工智能将像AlphaFold那样全面加速这一进程。我希望尽可能探索这棵知识之树的更多领域。这需要人工智能帮助我们理解或发现模式，也可能涉及设计和构建新的实验工具，还包括运行模拟和学习模拟。

And we sort of barely scratched the surface of that so far, even though we've done pretty well since the Enlightenment as humanity. And I think AI will turbocharge all of that like we've seen with AlphaFold. And I want to explore as much of that tree of knowledge as is possible to do. And I think that involves AI helping us with understanding or finding patterns, but also potentially designing and building new tools, experimental tools. I think that's all and also running simulations and learning simulations.

所有这些我们现在都还处于蹒跚学步的阶段。但我可以想象，在未来几十年里，这种思维方式的全面绽放将会真正令人惊叹。

All of that we're sort of doing at baby steps level here. But I can imagine that in in the decades to come as, you know, what's the full flourishing of of that line of thinking. It's gonna be truly incredible, I would say.

当我可视化这棵知识之树时，有种直觉告诉我：人类的知识之树在所有可能的知识树集合中其实非常渺小——受限于我们的认知能力，即使借助工具，我们仍无法理解许多事物。而这或许正是非人类系统能够更深入探索的领域，它们不仅是工具，更能自主理解并反馈某些知识。

If I visualize this tree of knowledge, something tells me that that knowledge tree of knowledge for humans is much smaller. In the set of all possible trees of knowledge, it's actually quite small given our cognitive limitations, limited cognitive capabilities that even with the with the tools we build, we still won't be able to understand a lot of things. And that's perhaps what nonhuman systems might be able to reach farther, not just as tools, but in themselves understanding something that they can bring back.

确实有可能。你刚才的表述包含了很多层次。我认为首先需要区分两个不同维度：一是我们当前的理解程度？

Yeah. It could well be. So I mean there's so many things that are sort of encapsulated in what you just said there. I think first of all, there's two different things there. It's like what do we understand today?

二是人类心智能够理解什么？三是可被理解的总体范畴是什么？这三个 concentric 圈层可以看作不断扩大的知识树或更多分支的探索。借助人工智能，我们将大规模探索这些领域。现在的问题是：如果考虑可被理解的总体范畴，可能存在某些基础物理学限制——比如模拟之外或宇宙之外的事物可能是无法理解的。

What could the human mind understand? And what is the totality of what is there to be understood? And so there's three concentrators, know, you can think of them as three larger and larger trees or exploring more branches of that tree. And I think with AI, we're going to explore that whole lot. Now the question is if you think about what is the totality of what could be understood, there may be some fundamental physics reasons why certain things can't be understood like what's outside a simulation or outside the universe.

也许从宇宙内部是无法理解的。所以可能存在一些这样的硬性约束，你知道的。

Maybe it's not understandable from within the universe. So that's there may be some hard constraints like that, you know.

可能是更小的约束，比如我们认为时空是基本的。我们人类大脑非常习惯于这种三维世界加时间的概念。

It could be smaller constraints like we think of space time as fundamental. Our human brains are really used to this idea of a three-dimensional world with time.

对。但也许我们的工具可以超越这个限制。是的，它们不一定有那种局限。它们可以在11维、12维或任何需要的维度中思考。

Right. Maybe But our tools could go beyond that. Yeah. They wouldn't have that limitation necessarily. They could think in 11 dimensions, 12 dimensions, whatever is needed.

但我们仍然可能通过几种不同的方式理解这一点。我经常举的例子是，当我与加里·卡斯帕罗夫下快棋时，或者我们讨论过国际象棋这类事情。你知道，如果你国际象棋下得还不错，你无法像加里那样想出他走的那一步棋，但他可以向你解释。

But we could still maybe understand that in several different ways. The example I always give is when I, you know, play Gary Kasparov for speed chess or we've talked about chess and these kind of things. You know, he if you if you if you're reasonably good at chess, you can you can't come up with the move Gary comes up with in his move, but he can explain it to you.

而你能理解。

And you can understand.

并且你能事后理解其中的推理。

And you can understand post hoc the reasoning.

是的。

Yeah.

所以我认为还有更深一层的含义，就像，也许你无法发明那个东西。但回到使用语言的角度，或许你可以理解和欣赏它。就像你能欣赏维瓦尔第或莫扎特的作品一样，即使你自己无法创作出那样的音乐，也能欣赏其美妙之处，对吧？

So so I think there's a there's an even further level of like, well, maybe you couldn't have invented that thing. But but using that going back to using language again, perhaps you can understand and appreciate that. Same way that you can appreciate, you know, Vivaldi or Mozart or something without you can appreciate the beauty of that without being able to to construct it yourself. Right? Invent the music yourself.

所以我认为我们在所有生命形式中都能看到这一点。所以这将是，你知道，乘以一百万倍。但你可以想象，智力的一个标志是能够清晰简洁地解释事物。对吧？你知道，像理查德·费曼这样的人，他是我永远的偶像之一，就经常这么说，对吧？

So I think we see this in all forms of life. So it'll be that times, you know, a million. But it would you can imagine also one sign of intelligence is the ability to explain things clearly and simply. Right? You know, people like Richard Fiehm, another one of my all time heroes used to say that, right?

如果你不能，你知道，如果你不能简单地解释某个复杂的话题，那说明你还没有真正理解它——简单解释复杂话题的能力是理解它的最佳标志之一。

If you can't, you know, if you can explain it something simply, then you that's that's the best sign, a complex topic simply, then that's one of the best signs of you understanding it.

是的。所以我能想象自己以那种方式对AI系统说废话。是的。它会因为我在解释事情时显得多么愚蠢而感到沮丧。我就想，嗯，那意味着你不够智能，因为如果你足够智能，你就能简单地解释清楚。

Yeah. So I can see myself talking trash in the AI system in that way. Yes. It's it's it's it gets frustrated how dumb I am in in trying to explain something to me. I was like, well, that means you're not intelligent because if you were intelligent, you'd be able to explain it simply.

是的。当然，你知道，还有另一种可能性。当然，我们可以通过我们的设备来增强自己。我们已经与我们的计算设备，比如手机和其他东西，形成了一种共生关系，对吧。而且，你知道，像Neuralink这样的技术可能会进一步推动这一点。

Yeah. Of course, you know, there's there's also the other option. Of course, we could enhance ourselves and and with our devices. We we are already sort of symbiotic with our compute devices, right, with our phones and other things. And, you know, there's stuff like Neuralink and etcetera that could be could could advance that further.