Demis Hassabis:
Continual learning, long term reasoning, some aspects of memory, these are still unsolved. I think all of these are going to be required for AGI. Depending on what your AGI timeline is, you know, mine's like 2030 or something like this, then If you start off on a deep tech journey today, you have to just consider AGI appearing in the middle of that journey. It's not bad necessarily, but you have to take that into account. You have to have an active system that can actively solve problems for you to get to AGI. So agents are that path. And I think we're just getting going.
持续学习、长期推理，以及记忆的某些方面，这些问题仍然没有解决。我认为所有这些能力都是实现 AGI 所必需的。至于你认为 AGI 会在什么时候到来——你知道，我自己的判断大概是 2030 年左右，或者类似的时间点——那么，如果你今天开始一段深度科技之旅，你就必须考虑到：AGI 可能会在这段旅程的中途出现。这未必是坏事，但你必须把它纳入考虑。要实现 AGI，你必须拥有一个能够主动为你解决问题的主动系统。所以，智能体就是这条路径。我认为我们才刚刚开始。
Garry Tan:
Demis Hassabis has had one of the most unusual careers in tech. He was a chess prodigy as a kid, then designed his first hit video game, Theme Park, at 17. He then went back to school, got a PhD in cognitive neuroscience, published foundational work on how memory and imagination work in the brain, and then in 2010, co-founded DeepMind with one mission, solve intelligence. And I think they've done it since then. His lab has gone on to do things most people thought were decades away. AlphaGo beat a world champion at Go. AlphaFold cracked protein structure prediction, a 50-year grand challenge in biology, and they gave it away for free to every scientist on Earth. That work won him the Nobel Prize in Chemistry last year. Today, Demis leads Google DeepMind, where he's building Gemini and pushing toward the same goal he set when he was a teenager, artificial general intelligence. Please welcome Demis Hassabis.
Demis Hassabis 拥有科技行业中最不同寻常的职业经历之一。他小时候是国际象棋神童，17 岁时设计了自己的第一款成功电子游戏 Theme Park。之后他又回到学校，获得了认知神经科学博士学位，发表了关于大脑中记忆与想象如何运作的基础性研究。2010 年，他共同创立了 DeepMind，使命只有一个：解决智能问题。而我认为，从那以后，他们确实做到了。他的实验室后来完成了一些大多数人原本以为还要几十年才能实现的事情。AlphaGo 击败了围棋世界冠军。AlphaFold 攻克了蛋白质结构预测这个困扰生物学 50 年的重大难题，并且把它免费开放给地球上的每一位科学家。这项工作让他在去年获得了诺贝尔化学奖。今天，Demis 领导 Google DeepMind，正在打造 Gemini，并继续推进他十几岁时就设定的同一个目标：通用人工智能。请欢迎 Demis Hassabis。
So you've been thinking about AGI longer than almost anyone. When you look at the current paradigm, large-scale pre-training, RLHF, chain of thought, how much of the final architecture for AGI do you think we already have and what's fundamentally missing right now?
所以，你思考 AGI 的时间比几乎所有人都更长。当你看当前这套范式——大规模预训练、RLHF、思维链——你认为我们距离 AGI 的最终架构已经拥有了多少？而现在从根本上还缺少什么？
Demis Hassabis:
Well, first of all, thanks, Garry, for that great introduction. And it's great to be here. Thanks for welcoming me here. It's an amazing space, actually. I have to come back here often. Very inspiring that you will get to work in this space.
首先，Garry，谢谢你刚才精彩的介绍。很高兴来到这里，也感谢你们欢迎我。说实话，这个空间非常棒，我以后得经常来。你们能在这样的空间里工作，很有启发性。
So the question is, I think the components that you just mentioned I'm pretty sure will be part of the final architecture for AGI. So I think they've come such a long way now and we've proven out so many things about what they can do. I can't see a world in which we will sort of realise in a couple of years this was a dead end. That doesn't make sense to me. But there still might be one or two things missing on top of what we already know works.
关于你的问题，我认为你刚才提到的那些组成部分，我相当确定它们会成为 AGI 最终架构的一部分。它们现在已经走了很远，我们也已经验证了它们能够做到很多事情。我无法想象几年之后我们会突然发现这是一条死路。这在我看来不合理。但在我们已经知道有效的东西之上，可能仍然缺少一两个关键部分。
So continual learning, long term reasoning, some aspects of memory. These are still unsolved and how to get the systems to be more consistent across the board. I think all of these are going to be required for AGI. Now, it might be that the existing techniques can just scale up to that with some innovation and some incremental innovation. But it could be that there's still one or two big ideas left that need to be cracked. I don't think it's more than one or two, if there are out there. And I think, you know, my betting is about 50-50, if that's the case. So, of course, at DeepMind, at Google DeepMind, we work on both those things.
比如持续学习、长期推理，以及记忆的某些方面，这些问题仍然没有解决；还有如何让系统在整体上表现得更加一致。我认为这些都是 AGI 所必需的。现在，也许现有技术只要通过一些创新和渐进式创新，就能扩展到那个水平。但也可能还有一两个重大想法尚未被攻克。如果确实还缺什么，我认为也不会超过一两个。我的判断大概是五五开。所以当然，在 DeepMind，在 Google DeepMind，我们两条路径都会做。
Garry Tan:
I guess so. I mean, working with a bunch of identic systems, the wildest thing to me is to what degree it's the same weights over and over. So this idea of continual learning is so interesting because, like, right now, we're sort of cobbling it together with duct tape, these dream cycles at night and things like that.
我想是这样。我的意思是，在使用一堆相同系统时，最让我觉得不可思议的是，它在多大程度上其实是在一遍又一遍使用同一组权重。所以持续学习这个想法非常有意思，因为现在我们某种程度上像是用胶带把它临时拼起来，比如夜间的这些“梦境周期”之类的东西。
Demis Hassabis:
It's pretty cool, the dream cycles. We used to think about this with consolidation with episodic memory. Actually, that's what I studied for my PhD, is how the hippocampus works and integrates new knowledge gracefully into the existing knowledge base. So the brain does that amazingly well. It does it during sleep, especially things like REM sleep, replaying back episodes that are important so that you can learn from it.
这些梦境周期确实很有意思。我们过去会从情景记忆巩固的角度思考这个问题。事实上，这正是我博士期间研究的内容：海马体如何运作，以及如何把新知识优雅地整合进既有知识库。大脑在这方面做得极其出色。它会在睡眠中完成这件事，尤其是在快速眼动睡眠这类阶段，把重要经历重新回放一遍，让你能够从中学习。
In fact, our very first Atari program, DQN, one of the ways it was able to master Atari games was by doing experience replay. So we sort of borrowed that from neuroscience and replayed successful trajectories many times. Way back in 2013 now, in the dark ages of AI, it was a really important thing. And I agree with you, we're kind of using duct tape right now.
事实上，我们最早的 Atari 程序 DQN，之所以能够掌握 Atari 游戏，其中一种方法就是进行经验回放。所以我们某种程度上是从神经科学中借用了这个想法，多次回放成功的轨迹。那已经是 2013 年的事了，现在看像是 AI 的黑暗时代，但当时这是非常重要的东西。我同意你的说法，我们现在确实有点像是在用胶带拼接。
So like shove it all in the context window. This seems a bit unsatisfying, right? And actually, even though we're working on machines, not biological brains, and so potentially you could have, you know, millions or tens of millions size context window or memory, and it can be perfect. There's still a cost to looking it up. And finding the right thing. That's actually relevant for the specific decision you've got to make right now.
比如，把所有东西都塞进上下文窗口里。这看起来有点不令人满意，对吧？而且实际上，虽然我们处理的是机器，不是生物大脑，所以理论上你可以拥有数百万甚至数千万规模的上下文窗口或记忆，而且它可以是完美的；但查找它仍然有成本。找到正确的东西也有成本。更关键的是，你要找到的东西必须真正与眼下这个具体决策相关。
And that's non-trivial, that cost, even if you can potentially store it all. I think there's actually a lot of room for innovation in areas like memory.
即使你理论上可以把所有东西都存下来，这个成本也并不微不足道。我认为在记忆这类领域，其实还有很大的创新空间。
Garry Tan:
Yeah. I mean, the one thing is it feels like a million token context ones is actually bigger than, I mean, it's plenty big, honestly, you can do so.
是的。我的意思是，有一点是，100 万 token 的上下文窗口感觉其实已经大到——老实说，已经相当大了，很多事情都可以做了。
Demis Hassabis:
Well, it's plenty big for most things that it should be used for. I mean, if you think about the context windows sort of equivalent to working memory, you know, humans have, we have like a few digits, you know, it's like a dozen digits, maybe, you know, average of seven. We got a million or, you know, 10 million context windows. But the problem is, is that we're trying to store everything in that, you know, things that aren't and not important, things that are wrong.
是的，对于它本来应该被用来处理的大多数事情来说，它已经足够大了。我的意思是，如果你把上下文窗口看作某种相当于工作记忆的东西，你知道，人类的工作记忆容量其实只有几个数字，也就是十来个数字，也许平均是七个。而我们现在有 100 万，甚至 1000 万 token 的上下文窗口。但问题在于，我们试图把所有东西都存进去，包括那些不重要的东西，以及那些错误的东西。
It's pretty brute force currently. And that doesn't seem right. And then the problem is, if you're now trying to try and process live video, and you're just going to naively record all the tokens, then actually a million tokens isn't that much. It's only like 20 minutes. So actually you need more if you want something that's going to understand what's going on in your life over maybe a month or two.
目前这种方式相当蛮力，看起来并不对。另一个问题是，如果你现在想处理实时视频，而只是天真地把所有 token 都记录下来，那么 100 万 token 其实并没有多少。它大概只相当于 20 分钟。所以，如果你想要一个系统能够理解你一两个月生活中发生了什么，实际上你需要更多容量。
Garry Tan:
DeepMind has historically leaned into reinforcement learning and search, AlphaGo, AlphaZero and MuZero. How much of that philosophy is actually embedded in how you're building Gemini today? Is RL still underrated?
DeepMind 历史上一直很重视强化学习和搜索，比如 AlphaGo、AlphaZero 和 MuZero。今天你们构建 Gemini 的过程中，有多少这种理念真正嵌入其中？强化学习是否仍然被低估了？
Demis Hassabis:
Yeah, I think potentially it is. It sort of goes in ebbs and waves. We've worked on agents since the beginning of DeepMind. In fact, that was what we said we were working on.
是的，我认为有可能是这样。它某种程度上是起伏循环的。我们从 DeepMind 一开始就在研究智能体。事实上，那正是我们当初说自己在做的事情。
So all of the Atari work and AlphaGo, most specifically, they're agent systems. And what we meant by that is systems that are able to accomplish goals on their own. And make active decisions and make plans.
所以，所有 Atari 相关工作，尤其是 AlphaGo，本质上都是智能体系统。我们所谓的智能体，是指能够自主完成目标、主动做决策并制定计划的系统。
And so, of course, we were doing it in the domain of games to make it tractable. And then doing increasingly complex games, things like StarCraft, After AlphaGo, AlphaStar. So, we basically did all the games that are out there.
当然，我们当时是在游戏领域做这些事情，是为了让问题变得可处理。之后我们开始做越来越复杂的游戏，比如 StarCraft；AlphaGo 之后，又有 AlphaStar。所以基本上，外面能做的游戏我们都做了一遍。
And then, of course, the question is, can you generalize those models to be world models or models of language, not just models of simple games or even complex games? And that's what the last few years has been about.
然后，当然，问题就变成了：你能不能把这些模型泛化成世界模型，或者语言模型，而不只是简单游戏模型，甚至不只是复杂游戏模型？过去几年做的事情，核心就是这个。
But really, you can think of a lot of the things we're doing today, all the leading models with thinking modes and chain of thought reasoning as aspects of what was sort of pioneered with AlphaGo coming back now.
但实际上，你可以把我们今天做的很多事情——所有具备思考模式和思维链推理的领先模型——看作当年 AlphaGo 所开创的一些东西，如今以新的形式重新回来了。
And I actually think there's a lot of work we did back then that is relevant today and we're sort of re-looking at some of those old ideas at scale today in a more general way, including things like Monte Carlo tree search and other ways of augmenting the RL on top of the reinforcement learning we're ready to do today. And I think a lot of those ideas, both from AlphaGo and AlphaZero are really, really relevant to where we are with today's foundation models. And I think a lot of that is what we're going to see of the advances the next few years.
而且我确实认为，我们当年做的很多工作今天仍然非常相关。我们现在正在以更大规模、更通用的方式，重新审视其中一些旧想法，包括蒙特卡洛树搜索，以及在我们今天已经能够做的强化学习之上进一步增强强化学习的其他方法。我认为，来自 AlphaGo 和 AlphaZero 的许多思想，和今天基础模型所处的位置高度相关。我认为未来几年我们会看到的很多进展，都会来自这些方向。
Garry Tan:
Obviously, today you need bigger and bigger models to be smarter and smarter, but then we're also seeing distillation. And then smaller models can be like quite a bit faster. I think, you know, you guys have incredible flash models that are like, you're finding that they're 95% as good as the Frontier and at like one tenth the price. Is that right?
显然，今天要让模型越来越聪明，就需要越来越大的模型；但与此同时，我们也看到了蒸馏。于是较小的模型可以快得多。我想，你知道，你们有非常出色的 Flash 模型，好像你们发现它们能达到前沿模型 95% 的水平，但价格只有十分之一左右。是这样吗？
Demis Hassabis:
I think that's one of our core strengths is, I mean, you have to build the biggest models to have the Frontier capabilities. But I think one of our biggest strengths has been distilling and packing that power into smaller and smaller models very quickly. Obviously, we invented the kind of distillation process and people like Jeff and Aurel and others, and we're still world experts in that.
我认为这是我们的核心优势之一。我的意思是，要拥有前沿能力，你必须构建最大的模型。但我认为，我们最大的优势之一，就是能够非常快速地把这种能力蒸馏并压缩进越来越小的模型里。显然，我们发明了这类蒸馏过程，Jeff、Aurel 以及其他人都参与其中，而我们在这方面仍然是世界级专家。
And we also have a huge need to do it because we've got to serve the biggest probably AI surfaces there are. Obviously, there's search with AI overviews and AI mode, then there's Gemini app. And now increasingly, every single product at Google has, you know, Maps and YouTube and so on has some aspect of Gemini or Gemini related technology in it. And so that's billions of users, a dozen, more than a dozen billion user products.
而且我们也有巨大的现实需求去做这件事，因为我们必须服务的，可能是世界上最大的 AI 应用场景。显然，搜索里有 AI 摘要和 AI 模式，另外还有 Gemini 应用。现在，越来越多的 Google 产品——比如 Maps、YouTube 等等——都在某种程度上嵌入了 Gemini 或与 Gemini 相关的技术。所以这是数十亿用户的规模，是十几个、超过十几个十亿级用户产品。
And they have to be served extremely fast, extremely efficiently and cheaply and with low latency. So that gives us a really important incentive to make these flash and even smaller models, flashlight models, extremely efficient. And hopefully that ends up then being really useful for many of the workloads that all of you use for.
而这些服务必须极快、极高效、极便宜，并且具有很低的延迟。因此，这给了我们非常强的动力，把这些 Flash 模型，甚至更小的 Flashlight 模型，做到极其高效。希望最终它们能够对你们许多人使用的各种工作负载非常有用。
Garry Tan:
I'm curious about how much smarter these smaller models can actually be. Like, are there limits to the distillation process? Like, could a 50B or 400B model be as smart as like a Mythos for today?
我很好奇，这些小模型实际上还能变得多聪明。比如，蒸馏过程有没有极限？一个 500 亿或 4000 亿参数的模型，是否可能达到今天 Mythos 那样的智能水平？
Demis Hassabis:
Yeah, I don't, I don't see any, I don't think we've got to any kind of, or at least none of us know yet if we've got to any kind of information or limit. I mean, maybe at some point that will be the case where there's just an information density that we can't get beyond. But I think for now, the assumption we make is that a year later after one of our leading pro models or frontier models goes out, half a year later, a year later, you'll have them in the really tiny almost edge models.
是的，我没有看到任何这样的极限。我不认为我们已经碰到了某种——或者至少我们还没有人知道是否已经碰到了某种信息极限。我的意思是，也许在某个时点会出现这种情况，也就是存在某种我们无法突破的信息密度。但我认为，就目前而言，我们的假设是：在我们某个领先的 Pro 模型或前沿模型发布之后，半年到一年以后，你就会在那些非常小、几乎可以运行在边缘端的模型里看到类似能力。
And you also see some of that goodness in our Gemma models, which hopefully you're all enjoying our Gemma 4 models, which I think are really amazing power for their sizes. So again, that uses a lot of these distillation techniques and the idea of how to make things really efficient in these very small models. So I didn't really see any limit yet in terms of like some kind of theoretical limit. I think we're still pretty far off of that.
你们也可以在我们的 Gemma 模型里看到其中一些能力。希望你们都在使用并喜欢我们的 Gemma 4 模型，我认为以它们的规模而言，能力真的非常惊人。所以同样，这里面使用了大量蒸馏技术，也使用了许多关于如何让这些很小的模型变得极其高效的思路。因此，从某种理论极限的角度看，我还没有真正看到任何限制。我认为我们离那还很远。
Garry Tan:
That's amazing. I mean, that is really good.
这很惊人。我的意思是，这确实非常好。
Demis Hassabis:
Yeah.
是的。
Garry Tan:
Uh, you know, one of the weirder things that we're seeing right now is like engineers can do like 500 to a thousand times the amount of work that they were doing like six months ago, I guess. I mean, the people in this room, there are people who are doing about like a thousand X the work that like, I, Steve Yagi talks about this. It's like a thousand X the work that a Google engineer from the two thousands was doing.
你知道，现在我们看到的一个比较奇怪的现象是，工程师大概可以完成相当于六个月前 500 倍到 1000 倍的工作量。我的意思是，这个房间里就有人正在完成大约 1000 倍的工作量。Steve Yagi 也谈到过这一点。也就是说，相比 2000 年代的 Google 工程师，他们现在做的工作量大概是 1000 倍。
Demis Hassabis:
I think it's very exciting. I mean, I think the small models have many uses. One is obviously cost. But the speed can allow, you know, if you think about coding even or other things, you can iterate a lot faster also, especially if you're collaborating with the system. I think there's a lot of need for having fast systems that maybe are not quite frontier, like you said, like 95%, 90%, but that's plenty good enough and actually gain back more than the 10% on the iteration speed.
我认为这非常令人兴奋。我的意思是，小模型有很多用途。一个显然是成本。但速度也会带来很大价值。比如，即便是写代码，或者其他事情，你也可以更快地迭代，尤其是在你和系统协作的时候。我认为市场非常需要快速系统：它们也许还不是完全前沿，就像你说的，可能达到 95% 或 90%，但这已经足够好，而且通过迭代速度，你实际上能把那 10% 的差距赚回来，甚至赚得更多。
And then the other big thing I think is running these things on the edge. Again, for efficiency reasons, but also for privacy and security reasons too. If you think about different devices that you might run these systems on that process very personal information. You can also think about robotics as well, you know, robots in your house. I think you're going to want very efficient, very powerful local models, which may be orchestrated, you know, with some bigger models, frontier models that are in the cloud, but you only delegate to that in certain circumstances. And perhaps you know, you process all of the audiovisual feed, let's say, locally, and that stays local. I can imagine that would be a very good sort of end state.
然后我认为另一个重大方向，是让这些东西在边缘端运行。同样，这既是出于效率原因，也是出于隐私和安全原因。你可以想象，不同设备上运行这些系统时，它们会处理非常私人的信息。你也可以想到机器人，比如你家里的机器人。我认为你会需要非常高效、非常强大的本地模型，它们也许会和云端更大的模型、前沿模型协同调度，但只有在某些特定情况下才把任务委派给云端。也许，比如说，所有音视频输入都在本地处理，并且留在本地。我能想象，这会是一种非常好的最终状态。
Unknown Speaker:
YC Startup School is back. We're hand-selecting the most promising builders in the world and flying them out to San Francisco for July 25th and 26th to discuss the cutting edge of tech. Apply now for a spot. Okay, back to the video.
YC Startup School 回来了。我们正在亲自筛选世界上最有潜力的建设者，并把他们邀请到 San Francisco，参加 7 月 25 日和 26 日的活动，一起讨论科技前沿。现在就申请名额。好了，回到视频。
Garry Tan:
Going back to context and memory, models currently stateless, but, you know, continue, like, what would the developer experience even be like for someone who's using a continual learning model? Like, you know, any idea, like how you'd steer it?
回到上下文和记忆这个问题，目前模型是无状态的；但如果继续往前走，比如，对一个使用持续学习模型的人来说，开发者体验到底会是什么样？你知道，比如说，你会怎么引导它？
Demis Hassabis:
I think it's really interesting. I think that's one of the, not having continual learning currently is one of the things holding back agents from doing full tasks, you know, I think they're really useful for aspects of tasks right now and you can patch them together and do some really cool things, but they don't adapt well with the context that you're in.
我认为这非常有意思。我认为目前没有持续学习能力，是阻碍智能体完成完整任务的因素之一。你知道，我认为它们现在对任务的某些部分非常有用，你可以把它们拼接起来，做出一些很酷的东西，但它们不能很好地适应你所处的具体上下文。
And I think that's the missing piece for them being really kind of fire and forget and they'll figure it out themselves. You know, I think they need to be able to learn about the specific context that you're going to put them in. So I think we have to crack that to get full general intelligence.
我认为，这正是让它们真正变成那种“交给它就不用管，它会自己搞定”的系统所缺失的一块。你知道，我认为它们需要能够学习你将要把它们放进去的具体上下文。所以我认为，要实现完整的通用智能，我们必须攻克这个问题。
Garry Tan:
Where are we on reasoning? So models can do really impressive chain of thought now, but they still fail on things a smart undergrad wouldn't. What specifically needs to change and what progress do you expect in reasoning?
我们现在在推理方面处于什么位置？现在模型已经能够做出非常令人印象深刻的思维链，但它们仍然会在一些聪明本科生不会出错的问题上失败。具体来说，哪些地方需要改变？你预计推理能力会有哪些进展？
Demis Hassabis:
There's a lot of innovation left in the thinking paradigms, I would say. Again, I think we're doing fairly simplistic things, fairly brute force. One could imagine, I think there's a lot of scope, for example, in monitoring the chain of thought, maybe interjecting midway through a thought process.
我会说，在思考范式上仍然有很多创新空间。再说一次，我认为我们现在做的事情仍然相当简单，也相当蛮力。可以想象，比如在监控思维链方面还有很大空间，也许可以在思考过程的中途进行介入。
I often get the impression with our systems and our competitor systems that they're almost overthinking. They're almost getting into sort of loops of things. Like one thing I sometimes like to do is play chess against Gemini. And, you know, all the leading foundation models are pretty poor at games, which is quite interesting. It's very cool to kind of look at the thinking traces because obviously these can be well understood.
我经常从我们的系统以及竞争对手的系统中得到一种印象：它们几乎是在过度思考。它们几乎会进入某种循环。比如，我有时喜欢和 Gemini 下国际象棋。你知道，所有领先的基础模型在游戏上都相当差，这一点很有意思。观察它们的思考轨迹非常有意思，因为这些轨迹显然是可以被很好理解的。
You know, I can tell quite quickly if it's going off on a tangent and it's very sort of provable what the thinking is doing, whether it's useful or not. And so what we see is that, you know, sometimes it will consider a move, it will realize it's a blunder, but it can't find anything better. So it kind of goes back to that move and does it anyway. So, you know, you just shouldn't be seeing that happening in a very precise reasoning system.
你知道，我可以很快看出它是不是跑偏了，而且它的思考到底在做什么、有没有用，某种程度上是很容易验证的。所以我们看到的是，有时它会考虑某一步棋，意识到这是一个失误，但又找不到更好的走法。于是它会回到那一步，然后还是照样走下去。所以，你知道，在一个非常精确的推理系统里，你不应该看到这种事情发生。
So there's just sort of huge gaps, I think, still. But it may only be one or two tweaks that are required to fix those kind of gaps, just to be clear. But I think that's pretty obvious there are there. And that's why you get this kind of jagged intelligence.
所以我认为，这里面仍然存在很大的缺口。不过说清楚一点，要修补这类缺口，可能只需要一两个调整。但我认为这些缺口显然存在。这也就是为什么你会看到这种参差不齐的智能。
On the one hand, it can solve gold medal problems in IMO, which is super hard. But on the other hand, as we've all seen, it can still make basic elementary math errors if you pose the question in a certain way, or elementary reasoning errors. So there's just something to me about the almost an introspection about its own thought process that I feel like there's something maybe missing there.
一方面，它可以解决 IMO 金牌级别的问题，这非常难。但另一方面，正如我们都见过的，如果你以某种方式提出问题，它仍然可能犯基础的初等数学错误，或者犯初等推理错误。所以在我看来，这里面似乎缺少某种东西，几乎是一种对自身思考过程的内省能力。我感觉那里可能确实有东西还没有补上。
Garry Tan:
Agents are really big. Some would say they're hyped. I personally think they're just getting started. It's totally insane. What does DeepMind's internal research tell you about where agent capabilities actually are right now versus, you know, sort of the hype out there?
智能体现在非常火。有些人会说它们被炒得过热。我个人认为它们才刚刚开始，这件事完全疯狂。DeepMind 的内部研究告诉你，智能体能力现在实际上处在什么位置？和外面的那些炒作相比，真实情况是什么？
Demis Hassabis:
I think we are, I agree with you. I think we're just at the beginning. You have to have an active system that can actively solve problems for you to get to AGI. That was always clear to us.
我认为是这样，我同意你的看法。我认为我们才刚刚开始。要实现 AGI，你必须拥有一个能够主动为你解决问题的主动系统。这一点对我们来说一直很清楚。
So agents are that path and I think we're just getting going. I think all of us are getting used to how do we best work and you're leading the way in a lot of this in your own personal experiments and I saw many of you are doing that.
所以，智能体就是这条路径，而我认为我们才刚刚起步。我觉得我们所有人都还在适应：到底应该怎样最好地使用它们。你在自己的个人实验中，在很多方面已经走在前面；我也看到你们很多人都在做类似的事情。
I think how do you incorporate it into your workflow in a way that isn't just sort of a nice to have. But actually starting to do fundamental things. My impression is at the moment we're all, you know, we're experimenting on lots of things, but we're only in maybe the last couple of months starting to find the really valuable places. And the technology is probably only getting good enough for that to be the case, right?
我认为关键问题是，怎样把它真正纳入你的工作流，而不是让它只是某种“有了挺好”的东西，而是开始做一些根本性的事情。我的印象是，目前我们所有人都还在很多事情上做实验，但也许只是最近几个月，我们才开始找到真正有价值的地方。而技术本身大概也是刚刚变得足够好，才让这种情况成为可能，对吧？
Whether it's not a kind of toy, nice demonstration, but actually really adding value to your time and efficiency. I often wonder, I see a lot of people working on like setting off, you know, dozens of agents for like 40 hours, but I'm not sure I've seen the output that yet of that quite justify that level of input going in, but I think it will come. So I still think we're in the experimentation phase.
也就是说，它不再只是一个玩具，或者一个漂亮的演示，而是真的在为你的时间和效率增加价值。我经常会想，我看到很多人在启动几十个智能体，让它们运行大约 40 个小时，但我还不确定自己已经看到相应的产出，足以证明这种投入是合理的；不过我认为这种结果会出现。所以我仍然认为，我们还处在实验阶段。
We haven't seen a AAA game that tops the App Store Charts that was sort of vibe coded yet. Right. I've seen and I've programmed and I'm sure many we've all done little nice demonstrations and it's like amazing.
我们还没有看到一款通过这种“氛围编程”做出来、并登上 App Store 排行榜榜首的 AAA 游戏，对吧。我见过，也自己编过，我相信我们很多人都做过一些漂亮的小演示，那确实很惊人。
I can do a prototype of theme park in half an hour now, which took me six months back when I was 17. It's kind of mind blowing. And I wish I got this feeling if I spent the whole summer working on it, you could make something really incredible, but it still needs craft. And, you know, human sort of soul into it and taste. I think that's something that you have to make sure you still bring that to whatever it is you're building.
现在我可以在半小时内做出 Theme Park 的原型，而我 17 岁时做这件事花了六个月。这有点令人难以置信。我有一种感觉：如果我花整个夏天来做，你可以做出真正惊人的东西。但它仍然需要手艺。你知道，它仍然需要人的灵魂和品味。我认为，无论你在构建什么东西，你都必须确保把这些带进去。
And I think it still shows like it's not quite there yet, because why haven't we seen a kid making a hit game that sells 10 million copies, right? That should be possible given the effort that's gone in. So something's still somehow missing.
我认为这也说明，它还没有完全到位。因为为什么我们还没有看到一个孩子做出一款卖出 1000 万份的爆款游戏，对吧？考虑到已经投入的努力，这本来应该是可能的。所以某种东西仍然缺失。
Maybe it's to do with the process, or maybe it's to do with the tools. I'm not quite sure. You all probably know better than me, because I'm sure you're all experimenting on that, but I haven't seen the result yet. Which I would expect once this is really delivering that full value, which I think will come in the next six to 12 months.
也许这和流程有关，也许和工具有关。我不太确定。你们可能比我更清楚，因为我相信你们都在这方面做实验，但我还没有看到那个结果。而一旦它真正释放出完整价值，我预期就会看到这种结果；我认为这会在未来 6 到 12 个月内发生。
Garry Tan:
Some of it is like how much of it will be autonomous versus, I mean, I don't think we'd see autonomous first. We would actually probably see people in this room operating a 1000X.
其中一部分问题是，到底有多少会是自主完成的。我的意思是，我不认为我们会先看到完全自主的版本。我们可能会先看到这个房间里的人以 1000 倍效率在运作。
Demis Hassabis:
And then that's what you should see first. And then many of you, you know, there'll be like, the games companies or, you know, other types of companies that have built some kind of best selling app, best selling game using these tools.
而这正是你应该先看到的。然后你们当中的许多人，你知道，可能会出现一些游戏公司，或者其他类型的公司，利用这些工具做出某种畅销应用、畅销游戏。
That's what you should see first. And then more of that will get automated.
这应该是你最先看到的东西。之后，其中越来越多的部分才会被自动化。
Garry Tan:
I mean, some of it is like there's a human in there and then the human doesn't want to say that the agents did it yet.
我的意思是，其中有些情况是，里面仍然有人参与，而这个人现在还不想说其实是智能体完成了这些事。
Demis Hassabis:
I think part of it might be, though, that we want to discuss like creativity. What I often say about that is like if we look at the things we've done like AlphaGo, so obviously very famously, you all know about the move 37 in game two.
不过我认为，其中一部分可能是，我们需要讨论创造力。关于这一点，我经常会说，如果我们看 AlphaGo 这类我们做过的东西，显然，非常有名的是第二局第 37 手，你们都知道那一步。
And for me, I was waiting for a moment like that to start the science projects like AlphaFold. So we started AlphaFold like the day we got back from Seoul, which is 10 years ago now.
对我来说，我当时就是在等待这样一个时刻，来启动像 AlphaFold 这样的科学项目。所以我们几乎是在从 Seoul 回来那天就启动了 AlphaFold，那已经是 10 年前的事了。
I'm going to Korea after this to celebrate the 10 year anniversary of AlphaGo. It's not enough to come up with Mu37. That's pretty cool. Very useful. But can it invent Go?
这次之后我要去韩国，庆祝 AlphaGo 十周年。能够下出第 37 手还不够。那当然很酷，也非常有用。但它能不能发明围棋？
That's what I want, a system that can invent Go if you give it a high-level description, you know, like a game you can learn the rules of in five minutes, but it takes many lifetimes to master.
这才是我想要的：一个系统，如果你给它一个高层次描述，比如说，一种五分钟就能学会规则、但需要很多辈子才能精通的游戏，它能够发明出围棋。
It's beautiful aesthetically, but you can play it in a few hours in an afternoon. So, you know, maybe you could imagine that would be the high-level description I would give, and then I'd want the return, the thing I get back is Go. Right.
它在美学上很漂亮，但你可以在一个下午用几个小时完成一局。所以，你知道，也许你可以想象，这就是我会给出的高层次描述，而我希望系统返回给我的东西，就是围棋。对吧。
And clearly today's systems, I think, can't do that. So the question is why? And I think there's something still missing there.
而很明显，我认为今天的系统还做不到这一点。所以问题是，为什么？我认为那里仍然缺少某种东西。
Garry Tan:
Well, someone in this room might make it.
嗯，这个房间里的某个人也许会做出来。
Demis Hassabis:
Then the answer would be there's nothing missing. It just was the way we were using the systems. And that might actually be the answer.
那答案就会是：其实没有什么缺失。问题只是我们使用这些系统的方式。而这也许真的就是答案。
It might be that our today's systems are capable of that with a brilliant enough creative person using it and providing that impetus, that's the soul of the project and being able to probably being Oh, fair enough with the tools to like almost be at one with the tools. I could imagine that would be happening if you experimented with the tools all day and all night like probably many of you are doing.
也许今天的系统已经具备这种能力，只要有足够出色、足够有创造力的人来使用它，并为它提供那种推动力——也就是项目的灵魂——而且这个人对工具足够熟练，几乎能和工具融为一体。我可以想象，如果你们像你们当中很多人可能正在做的那样，日夜不断地实验这些工具，这种事情是可能发生的。
And you combine that with proper deep creativity, something more incredible could be done.
如果再把它和真正深层的创造力结合起来，就有可能做出更加惊人的东西。
Garry Tan:
Switching gears to open source, I mean, or open weights. I mean, the recent release of Gemma, you're making highly capable, open and accessible ones that can actually run locally. What do you think that means for, will AI be something that is it in the hands of the users instead of primarily in the cloud? And does that change who gets to, you know, build with these models?
换个话题，谈谈开源，或者说开放权重。我的意思是，最近发布的 Gemma，你们正在打造能力很强、开放且可访问的模型，而且它们真的可以在本地运行。你认为这意味着什么？AI 会不会成为一种掌握在用户手里的东西，而不是主要存在于云端？这会不会改变哪些人能够基于这些模型进行构建？
Demis Hassabis:
We're huge proponents of, in general, of open source and open science. And you mentioned AlphaFold at the beginning, you know, we put that all out there for free and all of our science work, even still today, we publish in, you know, the big journals.
总体上，我们是开源和开放科学的坚定支持者。你一开始提到了 AlphaFold，你知道，我们把它全部免费开放出来。直到今天，我们所有的科学工作也仍然会发表在那些重要期刊上。
We wanted to create world-leading models for their sizes. And so that's what hopefully we've done with Gemma and we're very committed to that path. And hopefully you all experiment and build and enjoy using Gemma. I think it's been like 40 million downloads now and just in two and a half weeks. So we're really excited about that.
我们希望创造出在各自规模下世界领先的模型。希望这正是我们用 Gemma 做到的事情，而我们非常坚定地走在这条路上。也希望你们都去实验、构建，并享受使用 Gemma 的过程。我记得它现在大概已经有 4000 万次下载了，而且只用了两个半星期。所以我们对此非常兴奋。
And I also think it's important for there to be Western stacks on Open. And we're here to talk about open source. You know, obviously, a lot of the Chinese models are excellent. And they're currently what we're leading in open source. And we think Gemma is very competitive for its sizes in all those respects.
我也认为，开放体系中存在西方技术栈是很重要的。我们今天在这里谈开源。你知道，显然很多中国模型非常出色，而且它们目前在开源领域处于领先位置。我们认为，就相应规模而言，Gemma 在所有这些方面都非常有竞争力。
And for us, I mean, there is a question of resources, talent and compute, like nobody has enough spare compute to just make two, Frontier models at maximum size, right, with different attributes. So that's pretty difficult.
对我们来说，这里也涉及资源、人才和算力的问题。没有谁有足够多的闲置算力，可以随便做两个最大规模、但属性不同的前沿模型，对吧？所以这非常困难。
But also, for now, what we've decided is that our edge models, the things we want to use for Android and glasses and robotics, it's best that they're open models because they're vulnerable anyway once you put them out on the surfaces. So they might as well be actually fully open.
但同时，就目前而言，我们已经决定，我们的边缘端模型——也就是我们想用于 Android、眼镜和机器人等场景的模型——最好是开放模型，因为一旦你把它们放到这些应用表面上，它们无论如何都会比较容易被接触和分析。所以不如让它们真正完全开放。
Right, so we've sort of made a decision to kind of unify that at the kind of, we call it nano size level. And so that actually works for us strategically as well. And, you know, we hope as many people as possible build on it. And of course, we'll be building on that too.
所以，我们某种程度上已经决定，在我们称为 nano 尺寸级别的模型上，把这件事统一起来。这在战略上对我们也有好处。你知道，我们希望尽可能多的人基于它进行构建。当然，我们自己也会基于它继续构建。
Garry Tan:
Earlier, before we came on, I got to show you a demo of my version of Samantha from Her, which is harrowing for me to try to demo something to you. And it worked, which is amazing.
早些时候，在我们上台之前，我给你演示了我自己版本的《Her》里的 Samantha。对我来说，向你演示一个东西实在有点惊心动魄。而且它成功运行了，这很神奇。
Gemini was built multimodal and I spent a lot of time with a bunch of the models. And I mean, the depth of the context and the tool use with Speech directly to model. Yeah, there's nothing like bar none, like the best one. Yeah.
Gemini 从一开始就是多模态构建的，而我花了很多时间使用一堆不同模型。我的意思是，它的上下文深度，以及直接从语音到模型的工具使用能力。是的，完全没有别的东西能比，它就是最好的。是的。
Demis Hassabis:
Yeah, I think that's a sort of still a slightly underappreciated aspect of the Gemini series is. We started it being multimodal from the start.
是的，我认为 Gemini 系列有一个仍然稍微被低估的方面，就是我们从一开始就把它设计成多模态。
That made it a little bit more difficult, actually, to begin with, because then just focusing on text, for example. But we believe we're going to gain from that in the long run.
这在一开始实际上让事情变得稍微更难一些，因为如果只专注于文本，比如说，事情会更简单。但我们相信，从长期看，我们会因此受益。
And I think we're seeing that now for things like world model building. So stuff like Genie that we build on top of Gemini. I think it's going to be really important for things like robotics.
我认为我们现在已经在世界模型构建这类事情上看到这种收益。比如我们在 Gemini 之上构建的 Genie。我认为这对机器人这类领域会非常重要。
So this is why Gemini Robotics, which many of you probably played around with, I think it's going to be built on multimodal foundation models, the robotics models. And we think we have a sort of competitive advantage with Gemini being so strong at multimodal.
这就是为什么 Gemini Robotics——你们很多人大概都试过——我认为机器人模型将建立在多模态基础模型之上。而我们认为，Gemini 在多模态方面如此强大，会给我们带来某种竞争优势。
We're using it increasingly in things like Waymo. But also, if you imagine devices and assistants, digital assistants that come with you into the real world, you know, maybe on your phone or glasses or some other device, it needs to understand the physical world around you. And intuitive physics and the physical context you're in.
我们也越来越多地把它用于 Waymo 这类场景。同时，如果你想象一下那些会陪你进入现实世界的设备和助手，数字助手，也许是在你的手机、眼镜或其他某种设备上，它就需要理解你周围的物理世界，理解直觉物理，以及你所处的物理上下文。
And that's what our systems are extremely good at. And I think you found that's why you've enjoyed using it in your setup. We're planning to continue on that. And I think we're far and away the strongest models on those types of problems.
而这正是我们的系统极其擅长的事情。我想你也发现了，这就是为什么你在自己的设置中使用它会觉得很顺手。我们计划继续沿着这个方向推进。而且我认为，在这类问题上，我们的模型遥遥领先，是最强的。
Garry Tan:
So the cost of inference is dropping fast. What becomes possible when inference is essentially free? And how does that change what your team is actually optimizing for?
所以，推理成本正在快速下降。当推理基本上免费时，会有什么事情变得可能？这又会如何改变你们团队实际优化的目标？
Demis Hassabis:
Yeah, I'm not sure inference will ever be essentially free. I mean, there's sort of Jevron's paradox and other things about like, I think we'll just end up using, all of us will end up using whatever we can get our hands on.
是的，我不确定推理是否真的会变得基本免费。我的意思是，这里有某种 Jevons 悖论，以及类似的东西。我认为最终我们会使用一切能够拿到的算力；我们所有人都会把能用的东西用满。
And you can imagine millions of agents, swarms of agents working together on things. That's one way to use the inference. Or you could imagine single agents or groups, smaller groups of agents thinking for in multiple directions and then ensembling that. So we're experimenting with all these things. Probably many of you are. All of that will use up any inference I think that's available.
你可以想象数百万个智能体，像蜂群一样的智能体共同处理事情。这是一种使用推理算力的方式。或者你也可以想象，单个智能体，或者较小的一组智能体，沿着多个方向思考，然后把这些结果集成起来。所以我们正在实验所有这些东西。你们当中很多人大概也在做。我认为，所有这些都会消耗掉任何可用的推理能力。
I mean, one day maybe it can be almost cost zero, certainly the energy if we solve fusion or, you know, superconductors or, you know, optimal batteries or some set of those things, which I think we will do with material science.
我的意思是，也许有一天它可以几乎接近零成本。至少在能源方面，如果我们解决了核聚变，或者超导体，或者最优电池，或者这些问题中的某些组合——我认为我们会通过材料科学做到这些——那么能源成本可能会接近零。
Energy costs will be essentially zero, but there'll still be the physical creation of the chips and other things. There'll be some bottleneck. At least for the next few decades, I think.
能源成本会基本接近零，但芯片和其他东西的实体制造仍然存在。总会有某种瓶颈。至少在未来几十年，我认为都会如此。
And so if that's the case, there'll still be rationing on the inference side. You'll still have to use it, I think, efficiently.
所以，如果情况是这样，那么推理端仍然会存在配给约束。我认为，你仍然必须高效地使用它。
Garry Tan:
Well, luckily the smaller models are getting smarter and smarter, which is fantastic. We got a lot of bio and biotech founders in the audience. I can see a few. AlphaFold3 took us beyond proteins to a broad spectrum of biomolecules. How close are we to modeling full cellular systems? Or is that still a fundamentally harder problem in a class of its own?
好在小模型正在变得越来越聪明，这非常好。现场有很多生物和生物科技领域的创始人，我能看到几位。AlphaFold3 已经让我们从蛋白质扩展到更广泛的生物分子。我们距离建模完整的细胞系统还有多近？还是说，这仍然是一个根本上更难、独立成类的问题？
Demis Hassabis:
Well, Isomorphic Labs, which we spun out from DeepMind after we did AlphaFold2, which is going amazingly well, it's trying to build out not just AlphaFold, it's just one piece of the drug discovery process, as many of you know, but we're trying to do the adjacent biochemistry and chemistry to design the right compounds with the right properties and so on.
Isomorphic Labs 是我们在完成 AlphaFold2 之后从 DeepMind 拆分出来的公司，目前进展非常好。它试图构建的不只是 AlphaFold。正如你们许多人知道的，AlphaFold 只是药物发现过程中的一个环节；而我们正在做相邻的生物化学和化学工作，用来设计具有正确性质的正确化合物等等。
We'll have some big announcements very soon to talk about on that front. I think that's going really well.
我们很快会在这个方向上发布一些重大消息。我认为进展非常顺利。
Eventually, you want a whole virtual cell.
最终，你想要的是一个完整的虚拟细胞。
So I've talked about this in many of my science talks about a full working simulation of a cell that you can perturb and then the outputs of that would be close enough to experimental that it's useful.
我在很多科学演讲中都谈过这个目标：一个完整可运行的细胞模拟系统，你可以对它进行扰动，而它输出的结果要足够接近实验结果，从而具备实际用途。
You could skip out a lot of the search steps and generate lots of synthetic data to train other models that then would predict things about real cells.
这样你就可以省掉大量搜索步骤，并生成大量合成数据，用来训练其他模型，然后让这些模型预测真实细胞中的情况。
I think we're about 10 years away probably from something like a virtual cell, like a full virtual cell. We're working on the DeepMind side, science side on a virtual nucleus. Cell nucleus first, because relatively self-contained.
我认为，我们距离类似虚拟细胞这样的东西，距离完整虚拟细胞，大概还有 10 年左右。在 DeepMind 的科学团队这边，我们正在研究虚拟细胞核。先从细胞核开始，因为它相对自洽。
The trick with all of these things is, can you pick a slice of the complexity? You know, eventually you want to model a human body, but can you model it down to the right level of detail?
所有这些事情的技巧在于：你能不能从复杂性中切出一个合适的片段？你知道，最终你想要建模整个人体，但你能不能把它建模到正确的细节层级？
And what slice can you take out of it that will be self-contained enough? You can kind of model and approximate the inputs and outputs into that self-contained system and then just focus on the self-contained system.
你能从中切出哪一块，使它足够自成系统？你可以对进入这个自成系统的输入和输出进行建模和近似，然后只专注于这个自成系统本身。
So a nucleus is quite interesting from that perspective. Then the other issue is just there's not enough data yet. So you need data.
所以从这个角度看，细胞核非常有意思。另一个问题是，目前数据仍然不够。所以你需要数据。
And I talked to various, you know, top scientists about who work on electron microscopes and other imaging things. If we could image a live cell without killing the cell, that would be game changing, obviously, because then you could convert it into a vision problem.
我和许多研究电子显微镜以及其他成像技术的顶尖科学家谈过。如果我们能够在不杀死细胞的情况下对活细胞成像，那显然会改变游戏规则，因为那样你就可以把它转化成一个视觉问题。
Which we would know how to solve, right? But at the moment, there are at least, I'm not aware of any techniques that can give you a kind of, you know, nanometer resolution, but without destroying, but in, you know, in a live dynamic cell. So you can see all the interactions, right?
而视觉问题是我们知道如何解决的，对吧？但目前，至少据我所知，还没有任何技术能够在不破坏细胞的情况下，在一个动态活细胞中提供纳米级分辨率，让你看到所有相互作用，对吧？
You can take static images at that resolution, obviously, really detailed now, and that's quite exciting. But it's not enough to turn it just into a complex vision problem. So that's one way it could be solved.
显然，现在你可以用那种分辨率拍摄静态图像，而且已经非常细致，这很令人兴奋。但这还不足以把问题直接转化成一个复杂的视觉问题。所以，这是一种可能的解决路径。
So it could be a hardware-driven, data-driven solution, or it could be that we build better learned simulators of these dynamical systems. So that's the more modeling way of solving it.
因此，它可能是一种由硬件和数据驱动的解决方案；也可能是我们构建出更好的、通过学习得到的动态系统模拟器。后者就是更偏建模的解决方式。
Garry Tan:
You've been looking at all kinds of science, not just bio. There's material science, drug discovery, climate modeling, mathematics. If you had to rank which scientific domain will transform the most dramatically in the next five years, what's in your list?
你一直在关注各种科学领域，不只是生物学。还有材料科学、药物发现、气候建模、数学。如果你必须给出一个排序：未来五年，哪些科学领域会发生最剧烈的转变？你的名单里会有哪些？
Demis Hassabis:
Well, they're all so exciting. And that's why, I mean, that for me has been my main passion and always the reason why I've worked on AI for my whole career for 30 plus years now. Is to use AI as the ultimate tool.
它们都非常令人兴奋。也正因为如此，对我来说，这一直是我最主要的热情所在，也是我整个职业生涯——到现在已经 30 多年——一直从事 AI 的原因：把 AI 用作终极工具。
I always thought AI would be the ultimate tool for science and to invite such advanced scientific understanding, scientific discovery and things like medicine and just our understanding of the universe around us.
我一直认为，AI 会成为科学的终极工具，推动极其先进的科学理解、科学发现，推动医学等领域的发展，也推动我们对周围宇宙的理解。
So actually, when you mentioned our original way we used to articulate our mission statement, which is still. The way we think about it is there was two steps to it. Step one was solve intelligence, i.e. build AGI.
所以实际上，当你提到我们最初表达使命的方式时，那到今天仍然是我们思考这个问题的方式。它分成两步。第一步是解决智能问题，也就是构建 AGI。
And then step two was use it to solve everything else. We had to change that a bit over time because people were like, do you really mean solve everything else? And we did mean that.
然后第二步，是用它解决其他一切问题。后来我们不得不稍微调整这个说法，因为人们会问：你们真的指“解决其他一切问题”吗？而我们的确是这个意思。
And I think people are sort of understanding what that means today. But specifically, I was meaning solve other what I call root node problems in science. So areas of science that would unlock whole new branches or avenues of discovery.
我认为今天人们开始有点理解这句话的含义了。但更具体地说，我指的是解决科学中那些我称为“根节点问题”的问题。也就是那些一旦被解决，就会打开全新研究分支或发现路径的科学领域。
And AlphaFold is the prototypical example of what we want to do. So over 3 million researchers around the world, pretty much every biology researcher in the world uses AlphaFold now.
AlphaFold 就是我们想做的事情的原型例子。现在全世界有超过 300 万研究人员使用 AlphaFold，几乎每一位生物学研究者现在都在使用 AlphaFold。
And I was told by some of my, you know, pharma executive friends that, you know, almost every drug discovered from now on will have used AlphaFold at some point in the drug discovery process.
我的一些制药行业高管朋友告诉我，从现在开始，几乎每一种被发现的药物，都会在药物发现流程的某个环节使用 AlphaFold。
So that's something we're very proud of and it's the sort of impact that we hope to have with AI. But I do think it's just the beginning. I don't really see any area of science or engineering that this won't be able to help be helpful with.
这是我们非常自豪的事情，也是我们希望 AI 能够产生的那类影响。但我确实认为，这只是开始。我几乎看不到任何科学或工程领域，是 AI 不能提供帮助的。
And the ones you mentioned, I think we're almost like an alpha fold one moment. So we've got very promising results, but it's not quite solved the grand challenge yet in that domain.
至于你刚才提到的那些领域，我认为我们差不多处在类似 AlphaFold 1 的时刻。也就是说，我们已经有非常有前景的结果，但还没有真正解决那个领域里的重大挑战。
But I think we're going to have a lot to talk about in the next couple of years on all those areas you mentioned, materials, which I think is very exciting, all the way to mathematics.
但我认为，在未来几年里，围绕你刚才提到的所有这些领域，我们都会有很多东西可以讨论。从材料科学——我认为这非常令人兴奋——一直到数学，都会如此。
Garry Tan:
In science, I mean, it feels Promethean. It's like, here is this capability.
在科学领域，我的意思是，这感觉像是普罗米修斯式的。就像是：这里突然出现了这种能力。
Demis Hassabis:
I think so. I mean, of course, along with that, including the parable of Prometheus, we have to also be careful. With how we use that and what we use it for and also the misuse that can happen with those same tools.
我认为是这样。我的意思是，当然，与此同时，包括普罗米修斯这个寓言本身也提醒我们，我们也必须小心。我们要小心如何使用这种能力、把它用于什么目的，以及同样的工具可能被如何滥用。
Garry Tan:
A lot of people in this room are trying to build companies, applying AI to science. For them, what's the difference between a startup that actually advances the frontier in your view versus one that's just wrapping an API around a foundation model and calling it AI for science?
这个房间里有很多人正在尝试创办公司，把 AI 应用于科学。对他们来说，在你看来，一家真正推进前沿的初创公司，和一家只是给基础模型套上一层 API、然后称之为“科学 AI”的公司，区别在哪里？
Demis Hassabis:
Well, look, I think that's one of the things I would recommend. I'm trying to think about, and I think you mentioned this to me before, what would I do today myself if I was sitting in your place in Y Combinator? You know, looking at things.
你看，我认为这正是我会建议大家思考的问题之一。我也在想——我记得你之前也跟我提过——如果今天是我坐在你们的位置上，在 Y Combinator 里看这些机会，我自己会做什么。
One thing you have to do is obviously intercept where the AI tech is going. So that's one hard part of it. But I do think there's huge scope for combining where AI is going with some other deep technology area.
你必须做的一件事，显然是判断并截住 AI 技术正在前进的方向。这本身就是很难的一部分。但我确实认为，把 AI 的发展方向与某个其他深度科技领域结合起来，空间非常大。
I just think that that sweet spot is whether it's materials or medicine or other really hard areas of science. I think that those kinds of interdisciplinary teams, especially if it involves the world of atoms as well, there's not going to be a shortcut to that, at least in the foreseeable future.
我认为那个甜蜜点可能在材料、医学，或者其他真正困难的科学领域。我认为这类跨学科团队，尤其是如果它还涉及原子世界，那么至少在可预见的未来，这里面不会有什么捷径。
Those are areas that are pretty safe from just getting swarmed by whatever the next update is to the foundation models. So I think if you're looking for things like that, that's one of the more defensible areas, I would say. And I've always loved deep tech, so I'm kind of biased towards deep tech things.
这些领域相对不太容易被基础模型的下一次更新直接淹没。所以我会说，如果你在寻找这类机会，那是更有防御性的领域之一。而且我一直喜欢深度科技，所以我对深度科技类事情本身有些偏好。
I think nothing that's really long lasting and worthwhile is easy. And so I'm always being drawn to deep technologies. Obviously, AI was like that back in 2010 when we started out, right?
我认为，真正长期、真正值得做的事情，没有什么是容易的。所以我一直被深度技术吸引。显然，2010 年我们刚开始时，AI 就是这样的，对吧？
It was thought to just, we know it doesn't work kind of thing is what I was told by investors and even in academia, it was considered to be a very niche subject that we sort of tried in the 90s and we know doesn't work.
当时投资人告诉我的基本意思是：“我们知道这东西行不通。”甚至在学术界，它也被认为是一个非常小众的课题，大家觉得 90 年代已经试过了，而且我们知道它不行。
But if you, you know, if you have belief and conviction in your idea why it's different this time or what special combination from your background that you had, ideally you're expert in both those areas, both the machine learning and the other area you're applying it to, or you can create a founding team with that expertise.
但如果你对自己的想法有信念和确信，知道为什么这一次会不一样，或者你的背景里有什么特殊组合；理想情况下，你既懂机器学习，也懂你要应用它的那个领域，或者你能组建一个具备这种专业能力的创始团队。
I think there's huge impact to be made there and huge value to be built there.
我认为，在那里可以产生巨大的影响，也可以创造巨大的价值。
Garry Tan:
That's a really important message. I mean, it's hard. It's easy to forget. Like, basically, once you've done it, you've done it. But before you've done it, people are arrayed against you.
这是一个非常重要的信息。我的意思是，这很难，也很容易被忘掉。基本上，一旦你做成了，它就成了；但在你做成之前，很多人都会站在你的对面。
Demis Hassabis:
Oh, sure. I mean, no one believes in it, which is why I think you've also got to work in things that you're genuinely passionate about. Like, for me, I would have worked on AI no matter what happened, I just decided from a very young age, it was the thing that could be the most consequential thing I could think of.
当然。我的意思是，没人相信它。也正因为如此，我认为你必须从事那些你真正有热情的事情。比如对我来说，无论发生什么，我都会研究 AI。我从很小的时候就决定，AI 是我能想到的最可能产生重大影响的事情。
It's turned out that way, but it might not. Maybe we would have been 50 years too early. And it was also the most interesting thing I could think of working on.
后来事实证明确实如此，但它本来也可能不是这样。也许我们会早了 50 年。而且，它也是我能想到的最有意思、最值得投入的事情。
And so I would still be working on AI today, even if we were still, you know, in a little garage somewhere and it still wasn't quite working. I would have still been trying to find, maybe I'd have been back in academia or something, but I would have found some way of continuing to work on it.
所以，即使今天我们仍然在某个小车库里，AI 仍然还没有真正跑通，我也依然会继续研究 AI。我仍然会想办法继续做这件事，也许我会回到学术界之类的地方，但我一定会找到某种方式继续研究它。
Garry Tan:
So, I mean, AlphaFold was like an example of a spike that you pursued and it worked. You know, what makes a scientific domain ripe for an AlphaFold style breakthrough? And is there a pattern, a certain objective function?
所以，我的意思是，AlphaFold 就像是你们追求的一个尖峰方向，而且它成功了。你知道，什么样的科学领域已经成熟到可以出现 AlphaFold 式的突破？这里有没有某种模式，某种特定的目标函数？
Demis Hassabis:
The way I, I should write this up at some point when I have five minutes spare. But the lesson I've learned from all the alpha projects we've done, specifically AlphaGo and AlphaFold, is I think the techniques we have and the problems I like to look for are great if the situation can be described as massive combinatorial search space. The more massive, the better in some ways.
我的想法是——等我哪天有五分钟空闲时，我应该把这个写下来。但从我们做过的所有 Alpha 项目中，尤其是 AlphaGo 和 AlphaFold，我学到的经验是：如果一个问题可以被描述为巨大的组合搜索空间，那么我们现有的技术，以及我喜欢寻找的这类问题，就非常适合处理它。从某种意义上说，这个空间越巨大越好。
So no brute force or special case algorithm will solve it. And that's true of Go moves and of different configurations of proteins. Far more than the atoms in the universe, both of those. And then you have a clear objective function.
也就是说，蛮力方法或特殊情形算法都无法解决它。围棋走法是这样，蛋白质的不同构型也是这样。两者的可能性数量都远远超过宇宙中的原子数量。然后，你还需要一个清晰的目标函数。
So, you know, you could think of as minimizing the free energy in the proteins or, you know, winning the game of Go. So you need to specify your objective function clearly so you can hill climb.
比如，你可以把蛋白质问题理解为最小化自由能，或者把围棋问题理解为赢得比赛。所以，你需要清楚地指定目标函数，这样系统才能进行爬山式优化。
And then enough data and or simulator that can generate you lots of in-distribution synthetic data. If those things are true, then I think with today's methods, you can go a long way into tackling and finding the kind of needle in the haystack that you need for the solution that you're trying to look for.
然后，你还需要足够的数据，或者一个模拟器，能够为你生成大量同分布的合成数据。如果这些条件成立，那么我认为，凭借今天的方法，你就可以在很大程度上去处理这类问题，并在稻草堆中找到你为了解决问题所需要的那根针。
And I think of just drug discovery, by the way, in the same way, right? There is a compound out there that would solve this disease if one could find it, if one could only find it, right? And that wouldn't have any side effects and so on. And as long as the laws of physics allows it, then the only question is how do you find it in an efficient way, in a tractable way?
顺便说一句，我也用同样的方式看待药物发现，对吧？某个地方存在一种化合物，如果你能找到它，它就能治疗这种疾病；只要你能找到它，对吧？而且它还不会有副作用等等。只要物理定律允许它存在，那么唯一的问题就是：你如何以高效、可处理的方式找到它？
I think we showed for the first time actually with AlphaGo. And we're excited that these systems could find those kinds of needles in a haystack. In that case, you know, the perfect go move.
我认为，我们实际上是第一次通过 AlphaGo 展示了这一点。我们很兴奋地看到，这些系统能够找到那种稻草堆里的针。在那个例子里，你知道，就是那一步完美的围棋走法。
Garry Tan:
I guess to get a little meta, I mean, we were talking about humans using these methods to create alpha fold, but then there's a meta level, which is humans using AI to explore the space of possible hypotheses. How close are we to AI systems that can do genuine scientific reasoning, not just pattern matching on data?
我想稍微上升到更元的层面。我的意思是，我们刚才谈的是人类使用这些方法创造 AlphaFold，但还有一个更高层次的问题：人类使用 AI 去探索可能假设的空间。我们距离能够进行真正科学推理的 AI 系统还有多近？也就是说，不只是对数据进行模式匹配。
Demis Hassabis:
I think we're close. We're working on these general systems like we have this system called CoScientist and we have other algorithms like AlphaVolv that can go a little bit beyond what the basic Gemini will do.
我认为我们已经接近了。我们正在研究这些通用系统，比如我们有一个叫 CoScientist 的系统，还有其他算法，比如 AlphaVolv，它们能够稍微超出基础版 Gemini 所能做到的范围。
And obviously all the frontier labs are experimenting in this way. I've yet to see anything so far. And we all tinker with same things, you know, some math problems that are a little bit harder than IMO and so on.
显然，所有前沿实验室都在沿着这个方向实验。到目前为止，我还没有看到真正达到那个水平的东西。我们都在摆弄类似的问题，比如一些比 IMO 稍微更难一点的数学问题等等。
I haven't seen anything yet that is a true, genuine, you know, massive discovery. That's my personal opinion. I think it's coming. I think it may be related to this earlier, this thing we discussed about creativity and actually going on beyond the bounds of what's known.
我还没有看到任何真正意义上的、重大的发现。这是我的个人看法。我认为它会到来。我认为它可能和我们前面讨论过的创造力有关，也就是能够真正超越已知边界。
So clearly that's just not pattern matching at that point because there is no pattern to match to. And it's a bit more than extrapolation. It's some kind of analogical reasoning. And I don't think these systems have that, or at least we're not using them in the right way to do that.
显然，到那一步就不只是模式匹配了，因为根本没有现成的模式可以匹配。它也不只是外推，而是某种类比推理。我不认为这些系统已经具备这种能力，或者至少我们还没有以正确方式使用它们来做到这一点。
So the way I often say that in science is, can it come up with a hypothesis that's really interesting, not just solve one? When I say just, we're now talking about just like solving the Riemann hypothesis or something.
所以我在科学领域经常这样说：它能不能提出一个真正有意思的假设，而不只是解决一个已有假设？当我说“不只是”时，我们现在谈的“只是”，已经是解决 Riemann 假设之类的问题了。
This would be obviously amazing, or one of the Millennium Prize problems. And maybe we're a couple of years out from doing that. But I'd like to solve P equals NP. That's my favourite one.
这当然会非常惊人，或者解决某个千禧年大奖难题也会非常惊人。也许我们距离做到这一点只有几年时间。但我最想解决的是 P 是否等于 NP，这是我最喜欢的一个问题。
But even harder than that would be to come up with a new set of Millennium Prize problems that were regarded by top mathematicians to be as, you know, deep and meaningful and worthy of lifetime of study and effort to solve.
但比这更难的是，提出一组新的千禧年大奖难题，并且让顶尖数学家认为它们同样深刻、同样有意义，值得一个人用一生去研究和努力解决。
I think that's another level harder. And we don't have, you know, I still don't think we know how to do that. I don't think it's magical, though. I do think these systems will eventually be able to do that.
我认为这又难了一个层级。而我们现在还没有——你知道，我仍然不认为我们知道该如何做到这一点。不过我不认为这是什么魔法。我确实认为，这些系统最终会能够做到。
Maybe we're missing one or two things. And then the way we would test that is, you know, I sometimes call it my Einstein test, which is, you know, can you train a system with the knowledge of cutoff of 1901?
也许我们还缺少一两个东西。然后，我们可以用一种方式测试它。你知道，我有时把它称为我的 Einstein 测试：也就是说，你能不能训练一个系统，把它的知识截止在 1901 年？
And then will it come up with, you know, what Einstein did in 1905, including special relativity, you know, his Annus Mirabilis? Can it do that, right? And then I think we could run that test.
然后，它能不能提出 Einstein 在 1905 年提出的那些东西，包括狭义相对论，也就是他的“奇迹年”成果？它能做到吗，对吧？我认为我们可以运行这个测试。
Maybe we should just run that test and keep seeing if that's possible. And once that is, then I think we're on the verge of these systems being able to invent something new, truly novel.
也许我们就应该运行这个测试，并持续观察它是否可能做到。一旦它能够做到，我认为我们就站在这样一个边缘上：这些系统将能够发明真正新的东西，真正原创的东西。
Garry Tan:
So last, last question for the people who are deeply technical in this room, who want to work on something, you know, even close to the scale that what you've created with, you know, it's one of the largest AI efforts in the world and you've been a pioneer for all these years. So for that, I think everyone in this room thanks you and the folks at DeepMind very, very deeply from the bottom of our hearts.
所以，最后一个问题，问给这个房间里那些技术很深、也想做一些事情的人——你知道，哪怕只是接近你所创造的规模。你参与打造的是世界上最大的 AI 努力之一，而且这么多年来你一直是这个领域的先驱。因此，我想这个房间里的每个人，都从心底里非常、非常深地感谢你，也感谢 DeepMind 的所有人。
Thank you. What's the thing that you know now about building at the frontier that you wish you'd known at 25?
谢谢你。关于在前沿领域进行构建这件事，你现在知道、但希望自己 25 岁时就知道的东西是什么？
Demis Hassabis:
I think we covered some of it in terms of actually you work out that going off to hard problems and deep problems is no more difficult in some ways than going off to a shallower, simpler, more superficial problem.
我想我们前面已经谈到其中一部分。实际上你会发现，从某些方面看，去做困难而深刻的问题，并不比去做一个更浅、更简单、更表层的问题更难。
They're just differently difficult. There's different things that are hard about each of those things. But I think given life's very short and you only have so much time and energy.
它们只是难在不同的地方。每一种问题都有各自困难的部分。但我认为，人生很短，你拥有的时间和精力都有限。
You might as well put your life force into something that will really make a difference if you hadn't done it, if you hadn't been there to push it. So I would just think of it through that lens.
所以你不妨把自己的生命力投入到真正能产生差异的事情上：如果不是你去做，如果不是你在那里推动，它就不会发生。我会用这个视角来思考。
And then the other thing is if you're, if you are, and we talked about deep tech and I love interdisciplinary work and I think that's going to be even more prevalent in the next few years in combinations of fields and finding the connections between those fields.
然后还有一件事是，如果你——我们刚才谈到了深度科技，而我非常喜欢跨学科工作。我认为在未来几年里，不同领域的组合，以及寻找这些领域之间的联系，会变得更加普遍。
And it's going to be even easier to do that with AI. And then the only other thing I would say is if, you know, if you have your, depending on what your AGI timeline is, you know, mine's like 2030 or something like this.
而有了 AI，这件事会变得更容易。除此之外，我还想说的是，这取决于你自己的 AGI 时间表。你知道，我自己的判断大概是 2030 年左右，或者类似的时间点。
Then if you start off on a deep tech journey today, usually that you're talking about a 10 year journey for true deep tech, in my opinion. So then now you have to just consider AGI appearing in the middle of that journey.
那么，如果你今天开始一段深度科技旅程，通常来说，真正的深度科技在我看来是一段 10 年旅程。所以现在你就必须考虑：AGI 可能会在这段旅程的中途出现。
So what does that mean? It doesn't, it's not bad necessarily, but you have to take that into account. Right to will it be able to leverage it? What will the AGI system do with it?
这意味着什么？这未必是坏事，但你必须把它纳入考虑。对吧？它能不能利用 AGI？AGI 系统会怎样处理它？
And it goes a little bit back to what you said earlier about AlphaFold and general AI systems. So one thing I can think see happening is Gemini, Claude or one of these general systems making use of AlphaFold like specialized systems as tools.
这又有点回到你前面关于 AlphaFold 和通用 AI 系统的说法。我能想象的一种情况是，Gemini、Claude 或其他某个通用系统，会把 AlphaFold 这类专用系统当作工具来使用。
I don't think we're going to have it just in one giant brain because it will have too much regression. And if I put all the proteins into, you know, Gemini, that wouldn't make sense. We don't need Gemini to do protein folding.
我不认为我们会把一切都放进一个巨大的大脑里，因为那会带来太多退化。如果我把所有蛋白质相关知识都塞进 Gemini，这并不合理。我们不需要 Gemini 自己去做蛋白质折叠。
Going back to your information efficiency, it would definitely affect its language skills or something like that, right, in a bad way.
回到你说的信息效率问题，那肯定会以不好的方式影响它的语言能力，或者类似能力，对吧？
So much better, I think, is to have really good general purpose tool usage models that will then, maybe they could even train those specific tools, but they would be in a separate system.
所以我认为，更好的方式是拥有非常优秀的通用工具使用模型。它们之后也许甚至可以训练那些专用工具，但这些专用工具会存在于一个独立系统里。
So I think that's kind of interesting to think through the implications of that and then what you might build today. Also physical things too, like what kinds of factories would you build, what sorts of finance systems and so on.
所以我认为，认真推演这一点的含义非常有意思，也就是据此思考你今天应该构建什么。还包括实体层面的东西，比如你会建造什么样的工厂，什么样的金融系统，等等。
So I just think you need to really take that seriously on the one hand and imagine what that world would look like and then build something that would be useful if that comes in halfway through.
所以我认为，一方面你真的需要严肃看待这件事，想象那个世界会是什么样；然后构建一种东西，即使 AGI 在中途到来，它仍然会有用。
Garry Tan:
Demis Hassabis, everyone.
各位，Demis Hassabis。