从规模到创新：通往 AGI 之路

实际上，你可以认为我们 50%的努力放在 Scaling（规模扩张）上，50%放在创新上。我打赌，要达成 AGI（通用人工智能），两者缺一不可。我一直觉得，如果我们造出 AGI，用它作为心智的模拟，再与真实心智比较，就能看出差异，以及人类心智还有什么独特和剩余之处。也许是创造力，也许是情感，也许是梦境。关于意识有很多假设，关于什么可计算、什么不可计算也有很多假说。这又回到了图灵机的问题：图灵机的极限是什么？

We effectively you can think of as 50% of our effort is on scaling, 50% of it is on innovation. My betting is you're going to need both to get to AGI. I've always felt this that if we build AGI and then use that as a simulation of the mind and then compare that to the real mind, we will then see what the differences are and potentially what's special and remaining about the human mind, right? Maybe that's creativity, maybe it's emotions, maybe it's dreaming. There's a lot of consciousness. There's a lot of hypotheses out there about what may or may not be computable. And this comes back to the Turing machine question of like what is the limit of a Turing machine.

所以，没有什么是在计算范畴内做不到的。

So there's nothing that cannot be done within the sort of computational.

嗯，没人这么说过。到目前为止，宇宙中还没有发现任何不可计算的东西。

Well, no one's put it this way. Nobody's found anything in the universe that's non-computable so far.

到目前为止。

So far.

欢迎收听 Google DeepMind 播客，我是主持人 Hannah Fry 教授。今年对 AI 来说是非凡的一年。我们看到重心从大型语言模型转向了智能体式 AI。AI 加速了药物发现，多模态模型被集成到机器人和无人驾驶汽车中。这些话题我们都在播客中详细探讨过。但在今年的最后一期节目中，我们想放眼更广阔的视角，超越头条和产品发布，思考一个更大的问题：这一切究竟走向何方？哪些科学和技术问题将定义下一阶段？而花大量时间思考这些问题的，正是 Google DeepMind 的 CEO 兼联合创始人 Demis。欢迎回到播客，Demis。

Welcome to Google DeepMind the podcast with me, Professor Hannah Fry. It has been an extraordinary year for AI. We have seen the center of gravity shift from large language models to agentic AI. We've seen AI accelerate drug discovery and multimodal models integrated into robotics and driverless cars. Now, these are all topics that we've explored in detail on this podcast. But for the final episode of this year, we wanted to take a broader view, something beyond the headlines and product launches to consider a much bigger question. Where is all this heading really? What are the scientific and technological questions that will define the next phase? And someone who spends quite a lot of their time thinking about that is Demis, CEO and co-founder of Google DeepMind. Welcome back to the podcast, Demis.

很高兴回来。

Great to be back.

过去一年发生了很多事。

I mean, quite a lot's happened in the last year.

是的。

Yes.

你认为最大的转变是什么？

What's the biggest shift do you think?

哇，就像你说的，发生了太多事，感觉一年浓缩了十年。我认为变化很多。对我们来说，模型的进步——我们刚发布了 Gemini 3，非常满意。多模态能力等各方面都进展得很好。然后，大概在夏天，最让我兴奋的是世界模型的进展。我们肯定会聊到这个。

Oh wow. I mean it's just so much has happened as you said, it feels like we've packed in 10 years in one year. I think a lot's happened. I mean certainly for us, the progress of the models, we've just released Gemini 3 which we're really happy with. The multimodal capabilities, all of those things have just advanced really well. And then probably the thing I guess over the summer that I'm very excited about is world models being advanced. I'm sure we're going to talk about that.

是的，绝对会。我们待会儿再详细聊这些。我记得第一次采访你时，你谈到根节点问题，即用 AI 解锁下游收益的想法。不得不说，你兑现了承诺。你能给我们更新一下进展吗？哪些事情即将实现，哪些已经解决或接近解决？

Yeah, absolutely. We will get on to all of that stuff in a bit more detail in a moment. I remember the very first time that I interviewed you for this podcast and you were talking about the root node problems, about this idea that you can use AI to unlock these downstream benefits. And you've made pretty good on your promise, I have to say. And do you want to give us an update on where we are with those, what are the things that are just around the corner and the things that we've sort of solved or near solved?

是的。当然，最大的证明是 AlphaFold，想想 AlphaFold（至少是 AlphaFold 2）问世快五周年了，真是不可思议。那证明了解决这类根节点问题是可能的。我们现在正在探索其他问题。材料科学方面，我很想做出室温超导体和更好的电池，这些都有可能。各种更好的材料。我们还在研究核聚变，因为刚宣布了一个新合作。

Yeah. Well, of course, obviously the big proof point was AlphaFold and it's crazy to think we're coming up to like 5-year anniversary of AlphaFold being announced to the world, AlphaFold 2 at least. So that was the proof I guess that it was possible to do these root node type of problems. And we're exploring all the other ones now. I think material science, I'd love to do a room temperature superconductor and better batteries these kinds of things. I think that's on the cards. Better materials of all sorts. We're also working on fusion because there's a new partnership that's been announced.

核聚变。

Fusion.

是的，我们刚宣布了与一家更深入的合作。我们之前就在合作，但现在与 Commonwealth Fusion 的合作更深了，我认为他们可能是最优秀的初创公司，至少在做传统托卡马克反应堆。他们最接近做出可行方案，我们想帮助加速，帮助他们用磁体约束等离子体，甚至可能涉及一些材料设计。这很令人兴奋。我们还与量子团队合作，他们在 Google 的量子 AI 团队做着出色的工作，我们帮助他们用机器学习做纠错码，也许有一天他们也会帮助我们。

Yeah, we've just announced a partnership with a deeper one. We already were collaborating with them, but it's a much deeper one now with Commonwealth Fusion who, I think, are probably the best startup working on at least traditional tokamak reactors. So they're probably closest to having something viable and we want to help accelerate that, helping them contain the plasma in the magnets and maybe even some material design there as well. So that's exciting. And then we're collaborating also with our quantum colleagues who are doing amazing work at the quantum AI team at Google and we're helping them with error correction codes where we're using our machine learning to help them and then maybe one day they'll help us.

完美。确实。核聚变尤其如此，它能为世界带来的改变是巨大的。

That's perfect. Exactly. The fusion one is particularly, I mean the difference that that would make to the world that would be unlocked by that is gigantic.

是的。核聚变一直是圣杯。当然，太阳能也很有前景，有效利用天上的聚变反应堆。但如果我们能有模块化聚变反应堆，几乎无限的可再生清洁能源的承诺显然会改变一切。那就是圣杯。当然，这也是我们帮助应对气候问题的方式之一。如果能做到，很多现有问题都会消失。确实，它开启了很多可能性。这就是为什么我们认为它是根节点。它直接有助于能源、污染和气候危机。而且，如果能源真的可再生、清洁、超便宜或几乎免费，那么很多其他事情也会变得可行。比如水资源获取，因为我们可以到处建海水淡化厂。甚至制造火箭燃料。海水中含有氢和氧，那基本上就是火箭燃料，但需要大量能量才能分解出氢和氧。但如果能源便宜、可再生、清洁，为什么不做呢？你可以 24/7 生产。

Yeah. I mean fusion's always been the holy grail. Of course I think solar is very promising too, effectively using the fusion reactor in the sky. But I think if we could have modular fusion reactors, this promise of almost unlimited renewable clean energy would obviously transform everything. And that's the holy grail. Of course, that's one of the ways we could help with climate. It does make a lot of our existing problems sort of disappear if we can do that. Definitely, it opens up many things. This is why we think of it as a root node. Of course it helps directly with energy and pollution and the climate crisis. But also, if energy really was renewable and clean and super cheap or almost free, then many other things would become viable. Like water access because we could have desalination plants pretty much everywhere. Even making rocket fuel. There's lots of seawater that contains hydrogen and oxygen, that's basically rocket fuel, but it just takes a lot of energy to split it out into hydrogen and oxygen. But if energy is cheap and renewable and clean, then why not do that? You could have that producing 24/7.

你也看到 AI 在数学领域的很多变化，对吧？在国际数学奥林匹克竞赛中获奖，但同时这些模型会在高中数学上犯很基本的错误。为什么会有这种矛盾？

You're also seeing a lot of change in the AI that is applying itself to mathematics, right? Winning medals in the International Math Olympiad and yet at the same time, these models can make quite basic mistakes in high school math. Why is there that paradox?

是的，我觉得这其实很迷人。这是最迷人的事情之一，可能也是需要解决的关键问题之一，因为我们还没达到 AGI。如你所说，其他团队在国际数学奥林匹克竞赛中取得了金牌等很多成功。那些题目非常难，只有世界顶尖学生才能做。另一方面，如果以某种方式提问，我们都在日常生活中用聊天机器人实验过，它会在逻辑问题上犯一些相当琐碎的错误。它们还不能好好下国际象棋，这很令人惊讶。

Yeah, I think it's fascinating actually. One of the most fascinating things and probably that needs to be fixed as one of the key things while we're not at AGI yet. As you said, we've had a lot of success in other groups on getting like gold medals at the International Math Olympiad. You look at those questions and they're super hard questions that only the top students in the world can do. And on the other hand, if you pose a question in a certain way, we've all seen that with experimenting with chatbots ourselves in our daily lives that it can make some fairly trivial mistakes on logic problems. They can't really play decent games of chess yet, which is surprising.

所以这些系统在一致性方面还缺少一些东西。我认为这正是通用智能，也就是 AGI 系统所应具备的——全面的一致性。有时人们称之为锯齿状智能。它们在某些方面非常出色，甚至达到博士水平，但在其他方面可能连高中水平都不到。所以这些系统的表现仍然很不均衡。在某些维度上它们非常令人印象深刻，但在其他方面仍然相当基础，我们必须缩小这些差距。这背后有理论和原因，取决于具体情况。甚至可能跟图像被感知和分词化的方式有关。有时它甚至无法识别所有字母，所以当你在单词中数字母时，它有时会出错，但它可能并没有看到每个单独的字母。所以这些问题的原因各不相同，每个都可以被修复，然后你就能看到还剩下什么。但我认为一致性是一方面。另一个方面是推理和思考。我们现在有了思考系统，在推理时它们会花更多时间思考，并且能更好地输出答案。但在是否以有用的方式利用思考时间来双重检查输出结果方面，它还不是非常一致。我认为我们正在路上，但可能只走了 50%。

So there's something missing from these systems in terms of their consistency. And I think that's one of the things you would expect from a general intelligence, an AGI system, is that it would be consistent across the board. Sometimes people call it jagged intelligences. They're really good at certain things, maybe even PhD level, but other things they're not even high school level. So it's very uneven still, the performance of these systems. They're very impressive in certain dimensions, but still pretty basic in others, and we've got to close those gaps. There are theories and reasons why, depending on the situation. It could even be the way an image is perceived and tokenized. Sometimes it doesn't even get all the letters, so when you count letters in words, it sometimes gets that wrong, but it may not be seeing each individual letter. So there are different reasons for some of these things, and each one can be fixed, and then you can see what's left. But I think consistency is one thing. Another thing is reasoning and thinking. We have thinking systems now that at inference time spend more time thinking and are better at outputting answers. But it's not super consistent yet in terms of whether it's using that thinking time in a useful way to double-check and use tools to double-check what it's outputting. I think we're on the way, but maybe we're only 50% of the way there.

我也在想 AlphaGo 和 AlphaZero 的故事，你拿走了所有人类经验，结果发现模型反而提升了。在你正在创建的模型中，是否有类似的科学或数学版本？

I also wonder about that story of AlphaGo and then AlphaZero, where you took away all of the human experience and found that the model actually improved. Is there a scientific or mathematical version of that in the models you're creating?

我认为我们今天试图构建的东西更像 AlphaGo。这些大型语言模型，这些基础模型，从所有人类知识开始——我们放在互联网上的东西，如今几乎包罗万象——并将其压缩成某种有用的产物，它们可以查询并从中进行泛化。但我确实认为，在拥有像 AlphaGo 那样的搜索或思考能力方面，我们仍处于早期阶段。AlphaGo 必须利用模型来引导有用的推理轨迹、有用的规划思路，然后针对当时的问题得出最佳解决方案。所以我不觉得我们目前受到互联网这种人类知识极限的约束。我认为主要问题是我们还不知道如何像使用 AlphaGo 那样完全可靠地使用这些系统。当然，那要容易得多，因为它只是一个游戏。我认为一旦你有了一个类似 AlphaGo 的系统，你就可以回过头来做一个 AlphaZero，让它开始自己发现知识。那将是下一步，但那显然更难，所以我认为最好先创建第一步，即某种类似 AlphaGo 的系统，然后再考虑类似 AlphaZero 的系统。但这也是当今系统所缺少的东西之一：在线学习和持续学习的能力。我们训练这些系统，进行预训练、后训练，然后它们被部署到世界中，但它们不会像我们那样在世界中继续学习。我认为这是 AGI 之前另一个关键缺失的部分。

I think what we're trying to build today is more like AlphaGo. These large language models, these foundation models, start with all of human knowledge—what we put on the internet, which is pretty much everything these days—and compress that into some useful artifact that they can look up and generalize from. But I do think we're still in the early days of having search or thinking on top, like AlphaGo had to use that model to direct useful reasoning traces, useful planning ideas, and then come up with the best solution to whatever the problem is at that point. So I don't feel we're constrained at the moment by the limit of human knowledge like the internet. I think the main issue is we don't know how to use those systems in a reliable way fully yet, the way we did with AlphaGo. But of course that was a lot easier because it was just a game. I think once you have an AlphaGo-like system, you could go back and do an AlphaZero where it starts discovering knowledge for itself. That would be the next step, but that's obviously harder, so I think it's good to create the first step first with some kind of AlphaGo-like system, and then we can think about an AlphaZero-like system. But that is also one of the things missing from today's systems: the ability to online learn and continually learn. We train these systems, we pre-train them, we post-train them, and then they're out in the world, but they don't continue to learn out in the world like we would. And I think that's another critical missing piece that will be needed before AGI.

就所有这些缺失的部分而言，我知道目前有一场发布商业产品的大竞赛，但我也知道 Google DeepMind 的根源确实在于科学研究这个理念。我找到你最近说过的一句话：‘如果按我的意思来，我们会把 AI 留在实验室更久，做更多像 AlphaFold 这样的事情，也许能治愈癌症之类的。’你认为我们没有走那条更慢的路，是否失去了什么？

In terms of all those missing pieces, I know there's a big race at the moment to release commercial products, but I also know that Google DeepMind's roots really lie in that idea of scientific research. I found a quote from you where you recently said, 'If I had had my way, we would have left AI in the lab for longer and done more things like AlphaFold, maybe cured cancer or something like that.' Do you think we lost something by not taking that slower route?

我认为我们既有失去也有获得。那会是更纯粹的科学研究方法。至少那是我 15 或 20 年前的原始计划，当时几乎没有人研究 AI。我们正要创立 DeepMind。人们认为研究 AI 是疯狂的事。但我们相信它，想法是如果我们取得进展，我们会继续逐步构建 AGI，非常小心每一步及其安全性，分析系统在做什么等等。但与此同时，你不必等到 AGI 到来才让它有用。你可以将这项技术分支出来，用于对社会真正有益的方式，即推进科学和医学。就像我们用 AlphaFold 所做的那样，它本身不是基础模型，但使用了相同的技术——Transformer 和其他东西——并将其与更特定于该领域的东西融合。所以我设想完成一大堆这样的事情，它们将非常有益，像 AlphaFold 一样发布给世界，并确实做像治愈癌症之类的事情，同时我们在实验室里研究 AGI。现在的结果是，聊天机器人可以大规模实现，人们觉得它们有用，它们已经演变成这些基础模型，可以做比聊天和文本更多的事情，包括 Gemini。它们可以处理图像、视频和各种事情。这在商业和产品方面也非常成功，我也很喜欢。我一直梦想拥有终极助手，能在日常生活中帮助你，提高你的生产力，甚至保护你的大脑空间免受注意力消耗，让你能够专注并进入心流状态，因为如今社交媒体只是噪音。我认为为你工作的 AI 可以帮助解决这个问题。所以我认为这是好事，但它也造成了一种相当疯狂的竞赛状态，许多商业组织甚至国家都在争先恐后地改进和超越对方，这使得同时进行严谨的科学变得困难。我们试图两者兼顾，我认为我们正在取得平衡。另一方面，这种情况也有很多好处：更多的资源涌入这个领域，这无疑加速了进展。

I think we lost and gained something. That would have been the more pure scientific approach. At least that was my original plan 15 or 20 years ago, when almost no one was working on AI. We were just about to start DeepMind. People thought it was a crazy thing to work on. But we believed in it, and the idea was if we made progress, we would continue to incrementally build towards AGI, be very careful about each step and the safety aspects, analyze what the system was doing, and so on. But in the meantime, you wouldn't have to wait until AGI arrived before it was useful. You could branch off that technology and use it in really beneficial ways to society, namely advancing science and medicine. Exactly what we did with AlphaFold, which is not a foundation model itself, but uses the same techniques—transformers and other things—and blends them with more specific things to that domain. So I imagined a whole bunch of those things getting done, which would be hugely beneficial, released to the world just like we did with AlphaFold, and indeed do things like cure cancer, while we worked on the AGI track in the lab. Now it's turned out that chatbots were possible at scale and people find them useful, and they've morphed into these foundation models that can do more than chat and text, including Gemini. They can do images, video, and all sorts of things. That's also been very successful commercially and as a product, and I love that too. I've always dreamed of having the ultimate assistant that would help you in everyday life, make you more productive, maybe even protect your brain space from attention drain so you can focus and be in flow, because today with social media it's just noise. I think AI that works for you could help with that. So I think that's good, but it has created a pretty crazy race condition where many commercial organizations and even nation states are all rushing to improve and overtake each other, and that makes it hard to do rigorous science at the same time. We try to do both, and I think we're getting that balance right. On the other hand, there are lots of pros to the way it's happened: a lot more resources are coming into the area, so that's definitely accelerated progress.

嗯，而且我认为，有趣的是，普通公众实际上只比最前沿落后几个月就能使用到这些技术。所以每个人都有机会亲身感受 AI 会是什么样子。我认为这是件好事，然后政府也能更好地理解这一点。

Um and also um I think the general public are actually interestingly only a couple of months behind the absolute frontier in terms of what they can use. So everyone gets the chance to sort of feel for themselves what AI is going to be like. And I think I think that's a good thing and then governments sort of understanding this better.

奇怪的是，去年这个时候，有很多关于 Scaling（规模扩张）最终会遇到瓶颈、数据会用完的讨论，但现在我们录制时，Gemini 3 刚刚发布，它在各种不同的基准测试中都领先。这怎么可能？不是应该存在 Scaling（规模扩张）遇到瓶颈的问题吗？

The thing that's strange is that I mean this time last year I think there was a lot of talk about um you know scaling eventually hitting a wall about us running out of data and yet you know we're recording now Gemini 3 has just been released and it's leading on this whole range of different benchmarks. Um how how has that been possible? Like wasn't there supposed to be a problem with scaling hitting a wall?

我认为很多人这么想，尤其是其他公司可以说进展较慢，但我认为我们从未真正看到任何瓶颈。我想说的是，也许存在收益递减，当我说这个时，人们只想到“哦，那就没有收益了”，像是零或一，要么是指数增长，要么是渐近线。但实际上，这两种情况之间有很大的空间，我认为我们正处于中间。所以并不是每次发布新版本时，所有基准测试的性能都会翻倍。也许在早期，三四年前是这样。但你会看到显著的改进，就像我们在 Gemini 3 上看到的，这些改进非常值得投资，而且投资回报也在持续。我们没有看到任何放缓。确实存在一些问题，比如可用数据是否会用完，但有办法解决，比如合成数据。这些系统足够好，可以开始生成自己的数据，尤其是在编码和数学等某些领域，你可以在某种程度上验证答案，从而产生无限的数据。所有这些都是研究问题，我认为这是我们一直以来的优势：我们始终以研究为先。我认为我们拥有最广泛、最深入的研究团队，一直如此。回顾过去十年的进步，无论是 Transformer 还是 AlphaZero，还是我们讨论过的任何东西，都来自 Google 或 DeepMind。所以我一直说，如果需要更多的创新，尤其是科学创新，我会押注于我们，就像过去 15 年我们取得许多重大突破一样。我认为这正是正在发生的事情。我实际上很喜欢当难度增加时，因为那时你不仅需要世界级的工程能力（这已经够难了），还必须结合世界级的研究和科学，这正是我们擅长的。此外，我们还有世界级的基础设施优势，比如 TPU 和其他我们长期大量投资的东西。这种组合使我们能够处于创新和 Scaling（规模扩张）的前沿。你可以认为我们 50%的努力在 Scaling（规模扩张）上，50%在创新上。我打赌，要到达 AGI（通用人工智能），两者都需要。

I think a lot of people thought that especially as other companies have sort of had slower progress should we say but I think we've never really seen any wall as such like what I would say is um maybe there's like diminishing returns and people when I say that people think only think like oh so there's no returns like it's zero or one it's either exponential or or it's asmtopic no actually there's a lot of room between those two regimes and I think we're in in between those so it's not like you're going double the performance on all the benchmarks every time you release a new iteration. Maybe that's what was happening in the early very early days, you know, three four years ago. But you are getting significant improvements like we've seen with Gemini 3 that are well worth the investment and the return on that investment and doing. So I that we haven't seen any slowdown on. There are issues like are we running out of just available data but there are ways to get around that you know synthetic data uh generating your you know these systems are good enough they can start generating their own data especially in certain domains like coding and math where you can verify the answer in some sense you could produce unlimited data so all of these things though are research questions and I think that's the advantage that we've always had is that um we've we've always been sort of research first and We I think we have the broadest and deepest research bench always have done. Um and if you look back at the last decade of advances whether that's transformers or alpha zero any of the things we just discussed that they all came out of Google or deep mind. So I've always said like if if more innovations are needed uh scientific ones then I would back us to be the place to do it just like we were you know in the previous sort of 15 years for a lot of the big breakthroughs. So I think that's just what's transpiring and I actually really like it when the terrain gets harder because then it's not just worldass engineering you need which is already hard enough um but you have to ally that with worldclass research and science which is what we specialize in uh and on top of that we also have the advantage of world-class infrastructure with our TPUs and and other things that we've invested in a lot for a long time um and so that combination I think allows us to uh uh sort be at the frontier of the innovations as well as the scaling part and we effectively you can think of as 50 50% of effort is on scaling 50% of it is on innovation and I think my betting is you're going to need both to get to AGI

我的意思是，即使在 Gemini 3 这样的优秀模型中，我们仍然看到幻觉问题。我记得有一个指标显示，它仍然会在本应拒绝回答时给出答案。你能构建一个系统，让 Gemini 像 AlphaFold 那样给出置信度分数吗？

I mean one thing that we are still seeing even in Gemini 3 which is an exceptional model is uh this idea of hallucinations so I think um there was one metric that said uh it can still give an answer when actually it should decline um I mean could you build the system where Gemini gives a confidence score in the same way that Alpha Fold does.

是的，我认为可以。而且我认为我们确实需要这个。这算是缺失的一块。我们正在接近。模型越好，它们就越了解自己知道什么，如果这说得通的话。因此，我们可以更可靠地依赖它们进行某种内省或更多思考，并自己意识到它们不确定或对这个答案存在不确定性。然后我们必须找出如何训练它，使其能够输出一个合理的答案。我们在这方面越来越好，但有时它仍然会强迫自己回答本不该回答的问题，这可能导致幻觉。所以我认为目前很多幻觉都是这种类型。这是一个必须解决的缺失环节。你说得对，我们在 AlphaFold 中解决了这个问题，但显然是在一个更有限的范围内。

Yeah, I think so. And I think we need that actually. And I think that's sort of one of the missing things. I think we're getting close. I think the better the models get, the more they know about what they know, if that makes sense. And so, and I think the more reliable we could sort of rely on them to actually introspect in some way or do more thinking and actually realize for themselves that they're uncertain or there's there's there's uncertainty over this answer. Uh, and then we've got to sort of work out how to train it in a way that where it can it can output that as a as a reasonable answer. Um, we're getting better at it, but it still sometimes, you know, it sort of forces itself to answer when it probably shouldn't. Um, and then that can lead to a hallucination. So, I think, you know, a lot of the hallucinations are of that type currently. So, there's a missing piece there that that sort of has to be solved. And you're right, as we did solve it with Alpha Fold, but in in obviously a much more limited way

因为大概在幕后，存在某种对下一个词可能性的概率度量。

cuz presumably behind the scenes there is some sort of measure of probability of whatever the next token might be.

是的，对于下一个词是有概率的。这就是它的工作原理。但这并不能告诉你整体的情况，比如你对整个事实或整个陈述有多自信。这就是为什么我们需要这个。我认为我们需要利用思考步骤和规划步骤来回顾刚刚输出的内容。目前，这些系统有点像和一个状态不好的人交谈，他们只是把脑子里第一个想到的东西说出来。大多数时候这没问题，但有时遇到非常困难的问题时，你会希望他们停下来，思考一下，回顾并调整要说的内容。也许现在世界上这种情况越来越少，但这仍然是更好的交流方式。所以，你可以这样理解。这些模型需要在这方面做得更好。

Yes, there is of the next token. That's how it all works. But that doesn't tell you the overall arching piece is this is you know how confident are you about this entire fact or this entire um statement. And I think that's why you'll need this. I I think we'll need to use the thinking steps and the planning steps to go back over what you just output. At the moment, it's a little bit like the systems are just it's like talking to some a person and they just, you know, when when they're in on a bad day, they're just literally telling you the first thing that comes to their mind. Most most of the time that would be okay. But then sometimes when it's very difficult thing, uh you'd want to like stop pause for a moment and maybe go over what you were about to say and adjust what you were about to say. But perhaps that's happening less and less in the world these days, but um that's still the better way of having a discourse. So, you know, I think you can think of it like that. These models need to do that better.

我还非常想和你谈谈模拟世界以及在其中放置智能体，因为我们今天早些时候和你的 Genie 团队聊过。

I also really want to talk to you about um the the simulated worlds and putting agents in them because we got to talk to your genie team earlier today.

告诉我你为什么关心模拟。世界模型能做什么语言模型做不到的事情？

Tell me why you care about simulation. What What can a world model do that that a language model can't?

嗯，实际上，这可能是我最持久的热情所在，就是世界模型和模拟。

Well, look, I it's it's actually been it's probably my longest standing passion is world models and simulations.

除了 AI 之外，当然我们最近的工作如 Genie 也融合了这些。我认为语言模型能够理解很多关于世界的信息，实际上比我们预期的更多，比我预期的更多，因为语言可能比我们想象的更丰富。它包含的世界信息可能比语言学家想象的还要多。这些新系统已经证明了这一点。但关于世界的空间动态，空间意识以及我们所在的物理环境如何机械地运作，还有很多难以用语言描述，通常也不在语料库中描述。很多这类知识来自经验，在线经验。有很多东西你无法真正描述，你必须亲身体验。传感器等很难用语言表达，无论是电机角度、气味还是这类传感器。用任何语言描述都非常困难。所以我认为围绕这一点有一整套东西。如果我们想让机器人工作，或者让一个通用助手在日常生活中陪伴你，也许在眼镜上或手机上，帮助你的日常生活，而不仅仅是在电脑上，你就需要这种世界理解。世界模型是核心。

In addition to AI, and of course it's all coming together in our most recent work like Genie. I think language models are able to understand a lot about the world, actually more than we expected, more than I expected, because language is actually probably richer than we thought. It contains more about the world than maybe even linguists imagined. And that's proven now with these new systems. But there's still a lot about the spatial dynamics of the world, how spatial awareness and the physical context we're in, and how that works mechanically, that is hard to describe in words and isn't generally described in corpuses of words. A lot of this is allied to learning from experience, online experience. There are many things you can't really describe; you have to just experience them. The sensors and so on are very hard to put into words, whether that's motor angles, smell, these kinds of sensors. It's very difficult to describe that in any kind of language. So I think there's a whole set of things around that. And I think if we want robotics to work, or a universal assistant that maybe comes along with you in your daily life, maybe on glasses or on your phone, and helps you in your everyday life, not just on your computer, you're going to need this kind of world understanding. World models are at the core of that.

我们所说的世界模型是指这种理解世界机械因果关系的模型，即直观的物理，但事物如何运动，如何表现。现在我们实际上在视频模型中看到了很多这一点。测试你是否拥有这种理解的一种方法是：你能生成逼真的世界吗？因为如果你能生成它，那么在某种意义上你一定理解了它；系统必须封装了很多世界的机械原理。这就是为什么 Genie、VO 以及这些模型，我们的视频模型和交互式世界模型，令人印象深刻，也是展示我们拥有通用化模型的重要步骤。希望将来某个时候我们可以将其应用于机器人和通用助手。当然，我最喜欢的事情之一，我将来一定要做的，就是将其重新应用于游戏和游戏模拟，创造终极游戏，这当然可能一直是我潜意识的计划。

What we mean by a world model is this sort of model that understands the cause and effect of the mechanics of the world, right, intuitive physics, but how things move, how things behave. Now we're seeing a lot of that in our video models actually. And one way to test if you have that kind of understanding is: can you generate realistic worlds? Because if you can generate it, then in a sense you must have understood it; the system must have encapsulated a lot of the mechanics of the world. So that's why Genie and VO and these models, our video models and our sort of interactive world models, are really impressive but also important steps towards showing we have generalized models. And then hopefully at some point we can apply it to robotics and universal assistance. And then of course, one of my favorite things I'm definitely going to have to do at some point is reapplying it back to games and game simulations, and create the ultimate games, which of course was maybe always my subconscious plan.

那科学呢？你能在那个领域使用它吗？

What about science though? Could you use it in that domain?

是的，可以。在科学中，构建科学复杂领域的模型，无论是原子层面的材料、生物学，还是一些物理事物如天气。理解这些系统的一种方法是从原始数据构建这些系统的模拟。所以你有大量原始数据，比如关于天气的，显然我们有一些很棒的天气项目在进行。然后你有一个模型学习这些动态，并能比暴力方法更高效地重现这些动态。所以我认为模拟和世界模型有巨大潜力，也许是为科学和数学的某些方面专门定制的。

Yes, you could. In science, building models of scientifically complex domains, whether that's materials on an atomic level, in biology, but also some physical things like weather. One way to understand those systems is to build simulations of those systems from raw data. So you have a bunch of raw data, let's say about the weather, and obviously we have some amazing weather projects going on. Then you have a model that learns those dynamics and can recreate those dynamics more efficiently than doing it by brute force. So I think there's huge potential for simulations and world models, maybe specialized ones for aspects of science and mathematics.

但然后，你也可以将一个智能体放入那个模拟世界，对吧？

But then also, you can drop an agent into that simulated world too, right?

是的。你的 Genie 3 团队有一句很棒的引言：'几乎没有任何重大发明是在预见到该发明的情况下做出的。'他们谈到将智能体放入这些模拟环境，让它们以好奇心为主要动力进行探索，对吧？所以这是这些世界模型的另一个令人兴奋的用途。我们有另一个项目叫 Simma。我们刚刚发布了 Simma 2。模拟智能体，你有一个化身或智能体，把它放入一个虚拟世界。它可以是一个普通的商业游戏，或者像《无人深空》这样非常复杂的开放世界太空游戏。然后你可以指导它，因为它底层有 Gemini；你可以直接对智能体说话，给它任务。但后来我们想，如果把 Genie 接入 Simma，把一个 Simma 智能体放入另一个实时创建世界的 AI 中，那岂不是很有趣？所以现在两个 AI 在彼此的思维中互动。Simma 智能体试图导航这个世界，而对 Genie 来说，那只是一个玩家和化身；它不在乎有另一个 AI。它只是围绕 Simma 正在尝试做的事情生成世界。看到它们两者互动真是令人惊叹。我认为这可以成为一个有趣的训练循环的开始，你几乎有无限的训练样本，因为无论 Simma 智能体试图学习什么，Genie 基本上都可以实时创建。所以你可以想象一个设定和解决任务的世界，自动完成数百万个任务，而且它们变得越来越难。所以我们可能会尝试建立这样一个循环。当然，这些 Simma 智能体也可以成为很好的游戏伙伴。此外，它们学到的一些东西可能对机器人技术有用。

Yes. Your Genie 3 team had this really lovely quote: 'Almost no prerequisite to any major invention was made with that invention in mind.' And they were talking about dropping agents into these simulated environments and allowing them to explore with curiosity being their main motivator, right? So that's another really exciting use of these world models. We have another project called Simma. We just released Simma 2. Simulated agents where you have an avatar or an agent and you put it down into a virtual world. It can be a normal commercial game or something very complex like No Man's Sky, an open-world space game. Then you can instruct it because it's got Gemini under the hood; you can just talk to the agent and give it tasks. But then we thought, wouldn't it be fun if we plug Genie into Simma and drop a Simma agent into another AI that was creating the world on the fly? So now the two AIs are kind of interacting in the minds of each other. The Simma agent tries to navigate this world, and as far as Genie is concerned, that's just a player and an avatar; it doesn't care there's another AI. So it's just generating the world around whatever Simma is trying to do. It's kind of amazing to see them both interacting together. And I think this could be the beginning of an interesting training loop where you almost have infinite training examples because whatever the Simma agent is trying to learn, Genie can basically create on the fly. So I think you could imagine a whole world of setting and solving tasks, just millions of tasks automatically, and they're just getting increasingly more difficult. So we might try to set up a kind of loop like that. As well as obviously those Simma agents could be great as game companions. Also, some of the things they learn could be useful for robotics.

是的。基本上是无聊 NPC 的终结。

Yeah. The end of boring NPCs basically.

没错。这对这些游戏来说将是惊人的。

Exactly. It's going to be amazing for these games.

但你创建的那些世界，如何确保它们真的逼真？我的意思是，如何确保你不会得到看起来合理但实际上错误的物理？

Those worlds that you're creating though, how do you make sure that they really are realistic? I mean, how do you ensure that you don't end up with physics that looks plausible but is actually wrong?

是的，这是个好问题，可能是个问题。这基本上又是幻觉。所以有些幻觉是好的，因为这也意味着你可能会创造出有趣和新颖的东西。所以实际上，有时如果你试图做创造性的事情，或者试图让你的系统创造新事物、新颖事物，一点幻觉可能是好的，但你希望它是故意的，对吧？所以你现在可以打开幻觉，或者创造性探索。但是，当你试图训练一个 Simma 智能体时，你不希望 Genie 幻觉出错误的物理。

Yeah, that's a great question and can be an issue. It's basically hallucinations again. So some hallucinations are good because it also means you might create something interesting and new. So in fact, sometimes if you're trying to do creative things or trying to get your system to create new things, novel things, a bit of hallucination might be good, but you want it to be intentional, right? So you kind of switch on the hallucinations now, or the creative exploration. But yes, when you're trying to train a Simma agent, you don't want Genie hallucinating kind of physics that are wrong.

所以实际上我们现在正在做的，几乎是在创建一个物理基准测试，我们可以使用物理非常精确的游戏引擎来创建大量相当简单的场景，就像你在物理 A-level 实验课中会做的那样，对吧？比如让小球沿着不同轨道滚动，看它们速度有多快，从而在非常基础的层面上检验牛顿三大运动定律。这些模型——无论是 VO 还是 Genie——是否已经完美地封装了这些物理规律？目前还没有，它们只是近似。当你随意看的时候，它们看起来很逼真，但还不够精确，无法用于机器人等领域。所以我们有了这些非常有趣的模型。对于物理，我认为这可能需要生成大量真实数据，比如简单的单摆视频，两个单摆相互绕转的情况，但很快你就会遇到三体问题，这本身就无法解析求解。所以我觉得这会很有趣。但已经令人惊叹的是，当你观察像 VO 这样的视频模型处理反射和液体的方式时，至少肉眼看来已经非常精确了。所以下一步是超越人类业余观察者的感知能力，看看它是否真的能经得起严格的物理实验检验。

So actually what we're doing now is we're almost creating a physics benchmark where we can use game engines which are very accurate with physics to create lots of fairly simple things like what you would do in your physics A-level lab lessons, right? Like rolling little balls down different tracks and seeing how fast they go, so really teasing apart on a very basic level Newton's three laws of motion. Has it encapsulated? Whether it's VO or Genie, have these models encapsulated the physics of that 100% accurately? And right now they're not, they're kind of approximations. They look realistic when you just casually look at them, but they're not accurate enough yet to rely on for say robotics. So now we've got these really interesting models. And with physics, I think that's going to probably involve generating loads and loads of ground truth. Simple videos of pendulums, you know, what happens when two pendulums go around each other, but then very quickly you get to like three-body problems which are not solvable anyway. So I think it's going to be interesting. But what's amazing already is when you look at the video models like VO and just the way it treats reflections and liquids, it's pretty unbelievably accurate already, at least to the naked eye. So the next step is actually going beyond what a human amateur can perceive and would it really hold up to a proper physics-grade experiment?

我知道你思考这些模拟世界已经很长时间了。我回顾了我们第一次采访的文字记录，你在其中提到你很喜欢一个理论，即意识是进化的结果——在我们进化史上的某个时刻，理解他人内部状态成为一种优势，然后我们将其内化。这是否让你对在模拟中运行一个智能体的进化过程感到好奇？

I know you've been thinking about these simulated worlds for a really long time and I went back to the transcript of our first interview and in it you said that you really like the theory that consciousness was a consequence of evolution, that at some point in our evolutionary past there was an advantage to understanding the internal state of another and then we sort of turned it in on ourselves. Does that make you curious about running an agent in evolution inside of a simulation?

当然。我是说，我很想在某个时候做这个实验。重新运行进化，也重新运行社会动态。就像圣塔菲研究所过去在小型网格世界上做的那些很酷的实验。我以前很喜欢其中一些，但他们大多是经济学家，试图运行小型人工社会，他们发现如果让智能体在正确的激励结构下运行足够长时间，就会发明出各种有趣的东西，比如市场、银行等等疯狂的事物。所以我认为这非常酷，而且也能帮助理解生命起源和意识起源。这也是我一开始就对 AI 充满热情的原因之一：我认为你需要这类工具才能真正理解我们从哪里来，这些现象是什么。而模拟是最强大的工具之一，因为你可以用统计方法来做——你可以多次运行模拟，控制初始条件略有不同，然后可能运行数百万次，从而以非常受控的实验方式理解细微差异，这在现实世界中对于任何我们想回答的真正有趣的问题都是非常困难的。所以我认为精确的模拟将对科学产生难以置信的推动作用。

Sure. I mean, I'd love to run that experiment at some point. Kind of rerun evolution, rerun almost social dynamics as well. Like the Santa Fe Institute used to run lots of cool experiments on little grid worlds. I used to love some of these, but they're mostly economists and they were trying to run like little artificial societies and they found that all sorts of interesting things got invented, like if you let agents run around for long enough with the right incentive structures, markets and banks and all sorts of crazy things. So I think it would be really cool and also just to understand the origin of life and the origin of consciousness. And I think that is one of the big passions I had for working on AI from the beginning: I think you're going to need these kinds of tools to really understand where we came from and what these phenomena are. And I think simulations are one of the most powerful tools to do that because you can then do it statistically, because you can run the simulation many times with controlled slightly different initial starting conditions and then maybe run it millions of times and then understand what the slight differences are in a very controlled experiment sort of way, which of course is very difficult to do in the real world for any of the really interesting questions we want to answer. So I think accurate simulations will be an unbelievable boon to science.

鉴于我们已经发现这些模型具有涌现特性，拥有我们未曾预料到的概念理解，你在运行这类模拟时是否也需要非常小心？

Given what we've discovered about emergent properties of these models, having conceptual understanding that we weren't expecting, do you also have to be quite careful about running this sort of simulation?

我认为必须小心，是的。但这也是模拟的另一个好处：你可以在相当安全的沙盒中运行它们，也许最终你还需要进行物理隔离，而且你可以全天候监控模拟中发生的一切，并访问所有数据。所以我们可能需要 AI 工具来帮助我们监控模拟，因为它们会非常复杂，里面会发生很多事情。想象一下，大量 AI 在模拟中运行，任何人类科学家都难以跟上，但我们可能可以用其他 AI 系统来自动分析并标记模拟中任何有趣或令人担忧的事情。我的意思是，我想我们谈论的这些东西仍然属于中长期范畴。

I think you would have to be, yes. But that's the other nice thing about simulations: you can run them in pretty safe sandboxes, maybe eventually you want to air gap them, and you can of course monitor what's happening in the simulation 24/7 and you have access to all the data. So we may need AI tools to help us monitor the simulations because they'll be so complex and there'll be so much going on in them. If you imagine loads of AIs running around in a simulation, it will be hard for any human scientist to keep up with it, but we could probably use other AI systems to help us analyze and flag anything interesting or worrying in those simulations automatically. I mean, this I guess we're still talking sort of medium to long term in terms of this stuff.

那么回到我们目前所处的轨迹。我还想和你谈谈 AI 和 AGI 将对更广泛社会产生的影响。上次我们谈话时，你说你认为 AI 在短期内被过度炒作，但长期来看被低估了。我知道今年有很多关于 AI 泡沫的讨论。如果出现泡沫并破裂，会发生什么？

So just going back to the trajectory that we're on at the moment. I also want to talk to you about the impact that AI and AGI are going to have on wider society. And last time we spoke you said that you thought AI was overhyped in the short term but underhyped in the long term. And I know that this year there's been a lot of chatter about an AI bubble. What happens if there is a bubble and it bursts?

嗯，我仍然认为它在短期内被过度炒作，而在中长期其变革性仍被低估。当然现在有很多关于 AI 泡沫的讨论。在我看来，这不是一个二元问题：我们是不是在泡沫中？我认为 AI 生态系统的某些部分可能处于泡沫中。举个例子，一些初创公司的种子轮，它们甚至还没起步，就获得了数百亿美元的估值。这很有趣，这怎么能持续呢？我猜可能不行，至少总体上不行。所以那是其中一个领域。然后人们担心大型科技公司的估值和其他事情。我认为这背后有很多真实的业务。所以还有待观察。我的意思是，对于任何新的、难以置信的变革性和深刻的技术——AI 可能是其中最深刻的——你都会在某种程度上看到这种过度修正。当我们创办 DeepMind 时，没人相信它。没人认为这是可能的。人们还在想 AI 到底有什么用。而现在，10 到 15 年后，显然它似乎是商业中人们唯一谈论的话题。所以你会看到对之前低估的过度反应。我认为这很自然。我们在互联网上看到过，在移动领域也看到过，我认为我们正在或将在 AI 上再次看到。

Well, look, I still subscribe to it's overhyped in the short term still and still underappreciated in the medium to long term how transformative it's going to be. There is a lot of talk of course right now about AI bubbles. In my view, it's not one binary thing: are we or aren't we? I think there are parts of the AI ecosystem that are probably in bubbles. What one example would be, you know, just seed rounds for startups that basically haven't even got going yet and they're raising at tens of billions of dollars valuations just out of the gate. It's sort of interesting to see how can that be sustainable? My guess is probably not, at least not in general. So there's that area. Then people are worrying about the big tech valuations and other things. I think there's a lot of real business underlying that. So it remains to be seen. I mean, I think maybe for any new unbelievably transformative and profound technology, of which AI is probably the most profound, you're going to get this overcorrection in a way. So when we started DeepMind no one believed in it. No one thought it was possible. People were wondering what's AI for anyway. And then now fast forward 10, 15 years and now obviously it seems to be the only thing people talk about in business. So you're sort of going to get an overreaction to the underreaction. I think that's natural. I think we saw that with the internet. I think we saw with mobile and I think we're seeing or going to see it again with AI.

我不太担心我们是否处于泡沫中，因为从我的角度来看，正如你所知，领导 Google DeepMind，以及 Google 和 Alphabet 整体，我们的工作（也是我的工作）是确保无论哪种情况，我们都能强势胜出。我认为无论哪种情况，我们都处于极好的位置。所以如果像现在这样继续下去，那太好了。我们会继续我们正在做的所有这些伟大的事情、实验以及迈向 AGI 的进展。如果出现收缩，那也没问题。那样的话，我认为我们也处于有利位置，因为我们有自己的 TPU 堆栈。我们还有所有这些令人难以置信的 Google 产品以及它们产生的利润，可以把我们的 AI 接入其中。我们正在搜索中这样做，搜索已经被 AI 概览、AI 模式彻底革新，底层是 Gemini。我们还在看 workspace、电子邮件、YouTube。所以 Chrome 中有所有这些令人惊叹的东西。AI 已经可以看到很多这些低 hanging fruit 来应用 Gemini 2，当然还有 Gemini 应用，它现在也做得很好，以及通用助手的理念。所以有新产品的，我认为随着时间的推移它们会变得非常有价值，但我们不必依赖这个。我们可以直接增强我们现有的生态系统，这差不多就是过去一年发生的事情。我们现在已经非常高效了。

I don't worry too much about whether we're in a bubble or not because from my perspective, as you know, leading Google DeepMind and also with Google and Alphabet as a whole, our job and my job is to make sure either way we come out of it very strong. I think we're tremendously well positioned either way. So if it continues going like it is now, fantastic. We'll carry on all these great things we're doing, experiments and progress towards AGI. If there's a retrenchment, fine. Then also, I think we're in a great position because we have our own stack with TPUs. We also have all these incredible Google products and the profits that all makes to plug our AI into. And we're doing that with search, which is totally revolutionized by AI overviews, AI mode, with Gemini under the hood. We're looking at workspace, email, YouTube. So there are all these amazing things in Chrome. There are a lot of these amazing things that AI can already see as low-hanging fruit to apply Gemini 2 as well, of course, as the Gemini app, which is doing really well now, and the idea of a universal assistant. So there are new products, and I think they will in the fullness of time be super valuable, but we don't have to rely on that. We can just power up our existing ecosystem, which is sort of what's happened over the last year. We've got that really efficient now.

就目前人们能接触到的 AI 而言，我知道你最近说过，不要为了最大化用户参与度而构建 AI，以免重蹈社交媒体的覆辙，这很重要。但我也在想，我们是否已经在某种程度上看到了这种情况。我的意思是，人们花太多时间与他们的聊天机器人交谈，以至于最终陷入自我激进的螺旋。你如何阻止这种情况？你如何构建 AI，让用户成为自己宇宙的中心——这在很多方面正是其意义所在——但又不会制造出单人的回音室？

In terms of the AI that people have access to at the moment, I know you said recently how important it is not to build AI to maximize user engagement just so we don't repeat the mistakes of social media. But I also wonder whether we are already seeing this in a way. I mean people spending so much time talking to their chatbots that they end up kind of spiraling into self-radicalizing. How do you stop that? How do you build AI that puts users at the center of their own universe, which is sort of the point of this in a lot of ways, but without creating echo chambers of one?

是的，这是一个非常微妙的平衡，我认为这是我们行业必须做对的最重要的事情之一。所以我认为我们已经看到了一些过于谄媚的系统会发生什么，然后你会得到这些回音室强化，对个人非常不利。所以我认为部分原因在于，实际上我们想用 Gemini 构建的——我对 Gemini 3 的角色非常满意，我们有一个很棒的团队在开发，我个人也参与了——就是这种近乎科学的个性：温暖、有帮助、轻松，但简洁切题，并且会以友好的方式反驳不合理的事情。而不是试图强化你，比如你说地球是平的，然后它说“好主意”，我认为如果发生这种情况，对社会总体来说是不好的。但你必须平衡人们的需求，因为人们希望这些系统能提供支持，对他们的想法和头脑风暴有帮助。所以你必须把握好这个平衡。我认为我们正在发展一种关于个性和角色的科学，比如如何衡量它在做什么，以及我们希望它在真实性、幽默感等方面处于什么位置。然后你可以想象它有一个基础个性。然后每个人都有自己的偏好。你希望它更幽默还是更不幽默，更简洁还是更冗长？人们喜欢不同的东西。所以你在上面添加了额外的个性化层。但每个人仍然有一个核心基础个性，对吧？它试图遵循科学方法，这正是这些系统的意义所在。我们希望人们将这些用于科学、医学、健康问题等等。所以我认为这是让这些大语言模型正确的科学的一部分。我对我们目前的方向非常满意。

Yeah, it's a very careful balance that I think is one of the most important things that we as an industry have got to get right. So I think we've seen what happens with some systems that were overly sycophantic, and then you get these echo chamber reinforcements that are really bad for the person. So I think part of it is, and actually what we want to build with Gemini, and I'm really pleased with the Gemini 3 persona that we had a great team working on and I helped with too personally, is just this sort of almost like a scientific personality that's warm, helpful, light, but it's succinct to the point and it will push back on things in a friendly way that don't make sense. Rather than trying to reinforce you, the idea that the earth's flat and you said it and it's like wonderful idea, I don't think that's good in general for society if that were to happen. But you've got to balance it with what people want because people want these systems to be supportive, to be helpful with their ideas and their brainstorming. So you've got to get that balance right. And I think we are sort of developing a science of personality and persona of how to measure what it's doing and where we want it to be on authenticity, humor, these sorts of things. And then you can imagine there's a kind of base personality that it ships with. And then everyone has their own preferences. Do you want it to be more humorous or less humorous, or more succinct or more verbose? People like different things. So you add that additional personalization layer on it as well. But there's still the core base personality that everyone gets, right? Which is trying to adhere to the scientific method, which is the whole point of these. And we want people to use these for science, medicine, health issues, and so on. So I think it's part of the science of getting these large language models right. And I'm quite happy with the direction we're going in currently.

几周前我们和 Shane Legg 聊过，特别是关于 AGI，涉及目前 AI 领域发生的一切，语言模型、世界模型等等。什么最接近你对 AGI 的愿景？

We got to talk to Shane Legg a couple weeks ago about AGI in particular, across everything that's happening in AI at the moment, the language models, the world models, and so on. What's closest to your vision of AGI?

我认为实际上，显然有 Gemini 3，我认为它非常强大，但我们上周还推出了 Nano Banana Pro 系统，这是我们图像创建工具的高级版本。它真正令人惊叹的是，它底层也用了 Gemini。所以它不仅能理解图像，还能理解这些图像中语义上发生了什么。人们才玩了一个星期，但我在社交媒体上看到了很多关于人们用它做什么的酷东西。例如，你可以给它一张复杂飞机的图片，它可以标记所有不同部件的图表，甚至以所有部件暴露的形式可视化它。所以它对力学以及物体的组成部分、材料有某种深刻的理解。而且它现在可以非常准确地渲染文字。所以我认为这正在走向一种用于成像的 AGI。我认为它是一种通用系统，可以处理图像方面的任何事情。所以我认为这非常令人兴奋。然后是世界模型的进展，Genie 和 Simma 以及我们正在做的。最终我们必须将所有这些不同的项目汇聚起来——目前它们是不同的项目，但相互交织——我们需要将它们整合到一个大模型中，然后那可能开始成为原始 AGI 的候选。

I think actually the combination of obviously there's Gemini 3, which I think is very capable, but the Nano Banana Pro system we also launched last week, which is an advanced version of our image creation tool. What's really amazing about that is it has also Gemini under the hood. So it can understand not just images, it sort of understands what's going on semantically in those images. And people have been only playing with it for a week now, but I've seen so much cool stuff on social media about what people are using it for. So for example, you can give it a picture of a complex plane or something like that and it can label all the diagrams of all the different parts of the plane and even visualize it in a form with all the different parts sort of exposed. So it has some kind of deep understanding of mechanics and what makes up parts of objects, what's materials. So it's a sort of, and it can render text really accurately now. So I think that's sort of getting towards a kind of AGI for imaging. I think it's a kind of general purpose system that can do anything across images. So I think that's very exciting. And then the advances in world models, Genie and Simma and what we're doing there. And then eventually we've got to kind of converge all those different projects—they're kind of different projects at the moment and they're intertwined, but we need to converge them all into one big model, and then that might start becoming a candidate for proto-AGI.

我知道你最近读了很多关于工业革命的书。我们能从那里发生的事情中学到什么，来试图减轻当 AGI 到来时我们可以预期的一些颠覆？

I know you've been reading quite a lot about the industrial revolution recently. Are there things that we can learn from what happened there to try and mitigate against some of the disruption that we can expect when AGI comes?

我认为我们可以学到很多东西。这是你在学校会学到的东西，至少在英国是这样，但只是非常表面的层面。对我来说，深入了解这一切是如何发生的，它从什么开始，背后的经济原因——比如纺织业——然后第一台计算机实际上是缝纫机，对吧？然后它们变成了早期 Fortran 计算机、大型机的打孔卡。

I think there's a lot we can learn. It's something you sort of study in school, at least in Britain, but on a very superficial level. It was really interesting for me to look into how it all happened, what it started with, the reasons behind the economic reasons behind that, which is like the textile industry, and then the first computers were really the sewing machines, right? And then they became punch cards for the early Fortran computers, mainframes.

有一段时间它非常成功。英国成为了纺织业的中心，因为他们可以通过自动化系统以极低的成本制造出质量极高的产品。然后蒸汽机等一切随之而来。工业革命带来了许多令人难以置信的进步：儿童死亡率下降、现代医学、卫生条件以及工作与生活的分离都在那个时期得以解决。但它也伴随着挑战。这花了大约一个世纪的时间，劳动力市场的不同部分在特定时期受到冲击。必须建立工会等新组织来重新平衡。看到整个社会逐渐适应并最终形成现代世界，这很迷人。工业革命有利有弊，但考虑到食物、现代医学、交通等方面的富足，没有人愿意回到前工业时代。我们可以从那些冲击中吸取教训，并更早或更有效地缓解它们。而且我们可能不得不这样做，因为这一次的规模可能是工业革命的 10 倍，速度也快 10 倍——更像是一个十年而不是一个世纪。

And for a while it was very successful. Britain became the center of the textile world because they could make amazingly high-quality things very cheaply due to automated systems. Then the steam engines and all of those things came in. There were many incredible advances from the Industrial Revolution: child mortality went down, modern medicine, sanitary conditions, and the work-life split were all worked out during that period. But it also came with challenges. It took about a century, and different parts of the labor force were dislocated at certain times. New organizations like unions had to be created to rebalance things. It was fascinating to see society adapt over time, leading to the modern world. There were pros and cons, but no one would want to go back to pre-industrial times given the abundance of food, modern medicine, transport, etc. We can learn from those dislocations and mitigate them earlier or more effectively this time. And we'll probably have to, because this time it's likely 10 times bigger and 10 times faster—more like a decade than a century.

Shane 告诉我们，当前这种用劳动换取资源的经济体系在 AGI 之后的社会中将无法以同样的方式运作。你对社会应该如何重新配置有什么设想吗？

Shane told us that the current economic system, where you exchange your labor for resources, won't function the same way in a post-AGI society. Do you have a vision of how society should be reconfigured?

是的，我现在花更多时间思考这个问题。Shane 正在领导一项工作，思考 AGI 之后的世界可能是什么样子，以及我们需要准备什么。但整个社会都需要花更多时间思考这个问题——经济学家、社会科学家和政府。就像工业革命一样，整个工作世界和工作周都从前工业时代的农业发生了变化。我认为至少那种程度的变化会再次发生。如果需要新的经济体系和模式来帮助转型并确保利益广泛分配，我不会感到惊讶。像全民基本收入这样的东西可能是解决方案的一部分，但我不认为那是完整的——那只是我们现在能模拟出来的。可能会有更好的系统，比如直接民主，你用积分投票决定你想要什么。这在地方社区层面已经发生：这里有一笔预算，你想要一个游乐场还是一个网球场？然后社区投票。你甚至可以衡量结果，并给那些持续投票支持受欢迎项目的人更多影响力。我听到一些经济学家朋友在集思广益，如果能在这方面做更多工作就好了。还有哲学层面：工作会改变，但也许聚变问题会得到解决，带来丰富的免费能源，所以我们进入后稀缺时代。钱会变成什么样？每个人可能都过得更好，但目标呢？很多人从工作和养家糊口中获得目标感。这些问题从经济问题几乎变成了哲学问题。

Yeah, I'm spending more time thinking about this now. Shane is leading an effort here to think about what a post-AGI world might look like and what we need to prepare for. But society in general needs to spend more time thinking about that—economists, social scientists, and governments. As with the Industrial Revolution, the whole working world and work week changed from pre-industrial agriculture. I think at least that level of change will happen again. I wouldn't be surprised if we need new economic systems and models to help with that transformation and ensure benefits are widely distributed. Things like universal basic income might be part of the solution, but I don't think that's complete—it's just what we can model now. There might be better systems, like direct democracy where you vote with credits for what you want. This happens at the local community level: here's a budget, do you want a playground or a tennis court? Then the community votes. You could even measure outcomes and give more influence to those who consistently vote for well-received projects. I hear economist friends brainstorming this, and it would be great to have more work on that. Then there's the philosophical side: jobs will change, but maybe fusion will be solved, giving abundant free energy, so we're post-scarcity. What happens to money? Everyone might be better off, but what about purpose? Many people get purpose from their jobs and providing for their families. These questions blend from economic into philosophical.

你是否担心人们没有足够关注或行动不够快？需要什么才能促成这方面的国际合作？

Do you worry that people aren't paying attention or moving as quickly as you'd like? What would it take for international collaboration on this?

我对此感到担忧。在理想情况下，早就应该有更多的合作，特别是国际合作，以及更多关于这些话题的研究和讨论。我很惊讶没有更多的讨论，因为即使我们的时间线——5 到 10 年——对于建立应对此事的机构来说也很短。一个担忧是现有机构支离破碎，影响力不足。目前可能没有合适的机构来处理这个问题。而且随着地缘政治紧张局势，合作比以往任何时候都更难。看看气候变化，达成任何协议都多么困难。我们拭目以待。随着风险越来越高，系统越来越强大——也许它们被产品化的一个好处是，普通人会感受到能力和力量的提升，这会影响政府。也许随着我们接近 AGI，他们会明白过来。

I am worried about that. In an ideal world, there would have been a lot more collaboration already, especially internationally, and more research and discussion on these topics. I'm surprised there isn't more discussion given that even our timelines—which are 5 to 10 years—are short for building institutions to handle this. One worry is that existing institutions are fragmented and not influential enough. There may not be the right institutions to deal with this currently. And with geopolitical tensions, collaboration is harder than ever. Look at climate change and how hard it is to get any agreement. We'll see. As stakes get higher and systems become more powerful—maybe one benefit of them being in products is that everyday people will feel the increase in power and capability, which will reach governments. Maybe they'll see sense as we get closer to AGI.

你认为需要发生一个事件才能让所有人关注吗？

Do you think it will take an incident for everyone to pay attention?

我不知道。我希望不会。大多数主要实验室都相当负责任。我们尽量做到最负责任。正如你多年来所了解的，这始终是我们所做一切的核心。

I don't know. I hope not. Most of the main labs are pretty responsible. We try to be as responsible as possible. That's always been at the heart of what we do, as you know if you've followed us over the years.

这并不意味着我们能把每件事都做对，但我们会尽可能以深思熟虑和科学的方式来处理。我认为大多数主要实验室都在努力负责任。实际上，商业压力也促使我们负责任。想想智能体，如果你把智能体租给另一家公司去做某事，那家公司会想知道这些智能体的限制、边界和护栏是什么，比如它们能做什么、不能做什么，以及不会搞乱数据等等。所以我认为这是好事，因为那些比较鲁莽的运营者不会得到业务，企业不会选择他们。所以我认为资本主义体系在这里实际上会有助于强化负责任的行为，这很好。但也会有恶意行为者，可能是恶意国家、恶意组织，或者是在开源基础上构建的人。我不知道，显然很难阻止他们。然后可能会出问题，希望只是中等规模的问题，那将是对人类的一个警告。那时可能就是倡导国际标准或国际合作的时候了，至少在一些我们想要并同意的高层基本标准上。我希望这能成为可能。

Doesn't mean we'll get everything right, but we try to be as thoughtful and as scientific in our approach as possible. I think most of the major labs are trying to be responsible. Also, there's good commercial pressure actually to be responsible. If you think about agents, and you're renting an agent to another company, let's say, to do something, that other company is going to want to know what the limits are and the boundaries are and the guardrails are on those agents, in terms of what they might do and not just mess up the data and all of this stuff. So, I think that's good because the more cowboy operations, they won't get the business because the enterprises won't choose them. So I think the kind of capitalist system will actually be useful here to reinforce responsible behavior which is good. But then there will be rogue actors, maybe rogue nations, maybe rogue organizations, maybe people building on top of open source. I don't know, obviously it's very difficult to stop that. Then something may go wrong and hopefully it's just sort of medium-sized and then that will be a kind of warning shot to humanity across the bow. And then that might be the moment to advocate for international standards or international cooperation or collaboration, at least on some high-level basic standards we would want and agree to. I'm hopeful that that will be possible.

从长远来看，超越 AGI 走向 ASI（人工超级智能），你认为有没有一些事情是人类能做而机器永远无法做到的？

In the long term, so beyond AGI and towards ASI, artificial super intelligence, do you think that there are some things that humans can do that machines will ever be able to manage?

嗯，我认为这是个重大问题，而且我觉得这与我钟爱的主题之一——图灵机——有关。我一直觉得，如果我们构建了 AGI，然后用它来模拟心智，再与真实心智比较，我们就会看到差异，以及人类心智独特和剩余的部分，对吧？也许是创造力，也许是情感，也许是梦境。有很多关于意识、关于什么可计算什么不可计算的假说。这又回到了图灵机的问题：图灵机的极限是什么？我认为这是我一生中的核心问题，自从我了解到图灵和图灵机以来。这就是我爱上的东西，是我核心的热情所在。我认为我们一直在做的所有事情，都是在推动图灵机能力的极限，包括蛋白质折叠，对吧？结果是我并不确定极限在哪里，也许根本没有极限。当然，我那些量子计算的朋友会说存在极限，你需要量子计算机来处理量子系统，但我真的不太确定。实际上我和一些量子领域的人讨论过，也许我们需要从这些量子系统中获取数据才能进行经典模拟。然后这又回到了心智：它完全是经典计算，还是另有其事？你知道，罗杰·彭罗斯认为大脑中存在量子效应。如果存在，并且意识与之相关，那么机器将永远无法拥有意识，至少经典机器不行。我们得等待量子计算机。但如果没有，那么可能就没有任何极限。也许宇宙中一切都是计算可处理的，因此如果你以正确的方式看待，图灵机或许能模拟宇宙中的一切。如果非要我猜，我会猜是这样，并且我基于这个假设在工作，直到物理学证明我错了。

Well, I think that's the big question and I feel like this is related to, as you know, one of my favorite topics is Turing machines. I've always felt this that if we build AGI and then use that as a simulation of the mind and then compare that to the real mind, we will then see what the differences are and potentially what's special and remaining about the human mind, right? Maybe that's creativity, maybe it's emotions, maybe it's dreaming. There's a lot of consciousness. There's a lot of hypotheses out there about what may or may not be computable. And this comes back to the Turing machine question of like what is the limit of a Turing machine? And I think that's the central question of my life really ever since I found out about Turing and Turing machines. And I think that's what I fell in love with. That's my core passion. And I think everything we've been doing has been sort of pushing the notion of what a Turing machine can do to the limit, including folding proteins, right? And so it turns out I'm not sure what the limit is, maybe there isn't one. And of course my quantum computing friends would say there are limits and you need quantum computers to do quantum systems, but I'm really not so sure. And I've actually discussed that with some of the quantum folks and it may be that we need data from these quantum systems in order to create a classical simulation. And then that comes back to the mind: is it all classical computation or is there something else going on? You know, Roger Penrose believes there are quantum effects in the brain. If there are, and that's what consciousness is to do with, then machines will never have that, at least the classical machines. We'll have to wait for quantum computers. But if there isn't, then there may not be any limit. Maybe in the universe everything is computationally tractable and therefore if you look at it in the right way, Turing machines might be able to model everything in the universe. I'm currently, if you were to make me guess, I would guess that, and I'm working on that basis until physics shows me otherwise.

所以没有什么是在这种计算范畴内无法做到的……

So there's nothing that cannot be done within these sort of computational...

嗯，没人这么说过。到目前为止，还没有人在宇宙中发现任何不可计算的东西。

Well, no one's put it this way. Nobody's found anything in the universe that's non-computable so far.

到目前为止。

So far.

对吧？而且我认为我们已经证明，你可以远远超越通常复杂性理论家关于经典计算机今天能做什么的 P vs NP 观点。比如蛋白质折叠和围棋等等。所以我认为没人知道那个极限是什么。这实际上就是我们在 DeepMind 和 Google 所做的，也是我试图做的——找到那个极限。但在这个想法的极限处，对吧，我们坐在这里，脸上感受到灯光的温暖，听到背景中机器的嗡嗡声，手下的桌面触感。

Right? And I think we've already shown you can go way beyond the usual complexity theorist P vs NP view of what a classical computer could do today. Things like protein folding and Go and so on. So I don't think anyone knows what that limit is. And that's really, if you boil down to what we're doing at DeepMind and Google and what I'm trying to do, is find that limit. But then in the limit of that idea, right, is that we're sitting here, there's the warmth of the lights on our face. We kind of hear the whir of the machine in the background. There's the feel of the desk under our hands.

所有这些都可以被经典计算机复制。

All of that could be replicable by a classical computer.

是的。嗯，我认为最终，我对此的看法也是我热爱 AI 的原因：我两个最喜欢的哲学都是心智的构造。我认为这是真的。所以，是的，你提到的所有那些东西，它们进入我们的感官，感觉不同，对吧？灯光、灯光的温暖、桌面的触感，但最终，它们都是信息。我们是信息处理系统。我认为生物学就是这样。这就是我们在 Isomorphic 试图做的：我认为我们最终会通过将生物学视为信息处理系统来治愈所有疾病。而且我认为最终，这将是……我在业余时间，那两分钟的业余时间，研究物理理论，比如信息是宇宙中最基本的单位，不是能量，不是物质，而是信息。所以最终它们可能都是可互换的，对吧？但我们只是感知它，以不同的方式感受它。但据我们所知，我们拥有的这些惊人传感器，它们仍然可以被图灵机计算。但这就是为什么你的模拟世界如此重要，对吧？

Yes. Well, I think in the end, my view on this is why I love AI as well: all of my two favorite philosophies are a construct of the mind. I think that's true. And so, yes, all of those things you mentioned, they're coming into our sensory apparatus and they feel different, right? The light, the warmth of the light, the feel, the touch of the table, but in the end, they're all information. And we're information processing systems. And I think that's what biology is. This is what we're trying to do with Isomorphic: that's how I think we'll end up curing all diseases, by thinking about biology as an information processing system. And I think in the end, that's going to be... and I'm working on my spare time, my 2 minutes of spare time, physics theories about things like information being the most fundamental unit, should we say, of the universe, not energy, not matter, but information. And so it may be that these are all interchangeable in the end, right? But we just sense it. We feel it in a different way. But as far as we know, these amazing sensors that we have, they're still computable by a Turing machine. But this is why your simulated world is so important, right?

是的，完全正确。因为那将是达到它的途径之一。我们能模拟的极限是什么？因为如果你能模拟它，那么在某种意义上，你就理解了它。

Yes. Exactly. Because that would be one of the ways to get to it. What's the limits of what we can simulate? Because if you can simulate it, then in some sense, you've understood it.

我想以一些个人反思来结束，关于身处这一前沿的感受。这种情感上的重量是否曾让你感到沉重？是否曾感到相当孤立？

I wanted to finish with some personal reflections of what it's like to be at the forefront of this. I mean, does the emotional weight of this ever sort of weigh you down? Does it ever feel quite isolating?

是的。你看，我睡得不多，部分原因是工作太多，但也因为我难以入睡。要处理非常复杂的情感，因为这令人难以置信地兴奋。我基本上在做我梦想的一切。

Yes. Look, I don't sleep very much, partly because it's too much work, but also I have trouble sleeping. It's very complex emotions to deal with because it's unbelievably exciting. I'm basically doing everything I ever dreamed of.

我们在科学的绝对前沿，无论是应用科学还是机器学习。这令人振奋，所有科学家都知道那种处于前沿、首次发现某样东西的感觉。而我们几乎每个月都在经历这种时刻，这太棒了。但当然，我们这些——Shane 和我以及其他长期从事这项工作的人——比任何人都更清楚即将到来的事情有多么重大，而这一点仍然被低估了。实际上，未来十年内会发生的事情，包括哲学层面的问题，比如成为人类意味着什么，什么才是重要的。所有这些问题都会浮现出来。所以这是一个巨大的责任。但我们有一个出色的团队在思考这些。不过，对我自己来说，这也是我一生都在为之训练的事情。从我早年下棋，到后来研究计算机、游戏、模拟和神经科学，这一切都是为了这一刻。而且这大致符合我的想象。所以我能应对的一部分原因就是训练有素。

And we're at the absolute frontier of science in so many ways, applied science as well as machine learning. And that's exhilarating, as all scientists know that feeling of being at the frontier and discovering something for the first time. And that's happening almost on a monthly basis for us. So, which is amazing. But then of course, we, as Shane and I and others who've been doing this for a long time, we understand it better than anybody, the enormity of what's coming, and this is still underappreciated. In fact, what's going to happen in more of a 10-year time scale, including things like the philosophical, you know, what it means to be human, what's important about that. All of these questions are going to come up. And so it's a big responsibility. But we have an amazing team thinking about these things. But also it's something I guess at least myself I've trained for my whole life. So, ever since my early days playing chess and then working on computers and games and simulations and neuroscience, it's all been for this kind of moment. And it's roughly what I imagined it was going to be. So that's partly how I cope with it is just training.

有没有一些方面比你预想的更让你难以承受？

Are there parts of it that have hit you harder than you expected though?

是的，当然。一路走来，即使是 AlphaGo 那场比赛，对吧？看到我们如何攻克围棋，但围棋曾经是那么美丽的谜题，而它改变了它。所以那很有趣，也带点苦乐参半。我觉得即使是最近的语言和图像生成，以及它对创造力的意义，我对此有极大的尊重和热情，我自己也做过游戏设计，我和电影导演聊过，这对他们来说也是一个有趣的矛盾时刻。一方面，他们有了这些惊人的工具，能将创意原型速度提升十倍，但另一方面，它是否在取代某些创造性技能？所以我认为这种权衡无处不在，对于像 AI 这样强大且具有变革性的技术来说，这是不可避免的，就像过去的电力和互联网一样。我们已经看到，这就是人类的故事：我们是制造工具的动物，我们热爱这样做。而且出于某种原因，我们的大脑还能理解科学、从事科学，这很神奇，同时也充满永不满足的好奇心。我认为这是人类本质的核心。而我从一开始就有这种冲动。我试图回答这个问题的表达方式就是构建 AI。

Yes, for sure. On the way, I mean, even the AlphaGo match, right? Just seeing how we managed to crack Go, but Go was this beautiful mystery and it changed it. So that was interesting and kind of bittersweet. I think even the more recent things of like language and then imaging and what does it mean for creativity. I have huge respect and passion for the creative arts and having done game design myself, and I talk to film directors, and it's an interesting dual moment for them too. On one hand they've got these amazing tools that speed up prototyping ideas by 10x, but on the other hand, is it replacing certain creative skills? So I think there are these trade-offs going on all over the place, which I think is inevitable with something as powerful and transformative as AI, just as in the past electricity was and the internet. And we've seen that that is the story of humanity: we are tool-making animals, and that's what we love to do. And for some reason we also have a brain that can understand science and do science, which is amazing, but also insatiably curious. I think that's the heart of what it means to be human. And I think I've just had that bug from the beginning. And my expression of trying to answer that is to build AI.

当你和其他 AI 领导者同处一室时，你们之间是否有一种团结感，认为这是一群都知道利害关系、真正理解这些事情的人，还是竞争让你们彼此疏远？

When you and the other AI leaders are in a room together, is there a sense of solidarity between you that this is a group of people who all know the stakes, who all really understand these things, or does the competition kind of keep you apart from one another?

嗯，我们都认识彼此。我和几乎所有的人都相处得来。有些其他人之间则不太合得来。这很难，因为我们也处于可能是有史以来最激烈的资本主义竞争中。我的一些投资人和风投朋友，他们在互联网泡沫时期就在，说这比那时激烈十倍。在很多方面，我喜欢这样。我的意思是，我为竞争而生。从下棋时代起我就一直热爱竞争。但退一步说，我希望每个人都明白，有比公司成功之类更重大的事情处于危险之中。

Well, we all know each other. I get on with pretty much all of them. Some of the others don't get on with each other. And it's hard because we're also in the most ferocious capitalist competition there's ever been, probably. Investor friends of mine and VC friends who were around in the dotcom era say this is like 10x more ferocious and intense than that was. In many ways, I love that. I mean, I live for competition. I've always loved that since my chess days. But stepping back, I hope everyone understands that there's a much bigger thing at stake than just company successes and that type of thing.

展望未来十年，有没有一些即将到来的重大时刻是你个人最担心的？

When it comes to the next decade, when you think about it, are there big moments coming up that you're personally most apprehensive about?

我认为目前的系统，我称之为被动系统。用户投入能量，提出问题或任务，然后它们提供一些总结或答案。所以很大程度上是人类主导，人类投入能量和想法。下一阶段是基于智能体的系统，我认为我们即将开始看到。我们现在已经看到一些，但它们还很初级。在未来几年内，我认为我们会开始看到一些真正令人印象深刻且可靠的系统。如果你把它们当作助手之类的，它们会非常有用和强大，但它们也会更加自主。所以我认为这类系统的风险也会增加。所以我非常担心这些系统在两三年后可能能够做什么。因此我们正在研究网络防御，为这样一个可能有数百万智能体在互联网上游荡的世界做准备。

I think right now the systems are, I call them passive systems. You put the energy in as the user, the question or the task, and then they provide you with some summary or answer. So very much it's human-directed and human energy going in, and human ideas going in. The next stage is agent-based systems, which I think we're going to start seeing. We're seeing now, but they're pretty primitive. In the next couple of years, I think we'll start seeing some really impressive reliable ones. And I think those will be incredibly useful and capable if you think about them as an assistant or something like that, but also they'll be more autonomous. So I think the risks go up as well with those types of systems. So I'm quite worried about what those sorts of systems will be able to do maybe in two, three years time. So we're working on cyber defense in preparation for a world like that where maybe there's millions of agents roaming around on the internet.

那你最期待的是什么？是否有一天你能退休，知道你的工作已经完成，还是说还有超过一生的工作要做？

And what about what you're most looking forward to? Is there a day when you'll be able to retire sort of knowing that your work is done, or is there more than a lifetime's worth of work left to do?

我绝对可以休个长假，而且我会用它来做些事情。是的，一周假，甚至一天假也好。但你看，我认为我的使命一直是帮助世界为全人类安全地实现 AGI。所以我认为当我们达到那个点时，当然还有超级智能、后 AGI 以及我们讨论过的所有经济和社会问题，也许我能在某些方面提供帮助。但我认为那将是我使命的核心部分，我的人生使命就算完成了——我的意思是这只是一个小工作，就是把它实现，或者帮助世界实现它。我认为这需要像我们之前谈到的合作。我是一个相当善于合作的人，所以我希望我能从我所在的位置为此提供帮助。

I could definitely do with a sabbatical, and I would spend it doing stuff. Yeah, a week off, even a day would be good. But look, I think my mission has always been to help the world steward AGI safely over the line for all of humanity. So I think when we get to that point, of course there's then superintelligence and post-AGI and all the economic stuff we were discussing and societal stuff, and maybe I can help in some way there. But I think that will be the core part of my mission, my life mission will be done if it's a—I mean it's only a small job, just get that over the line or help the world get that over the line. I think it's going to require collaboration like we talked earlier. And I'm quite a collaborative person, so I hope I can help with that from the position that I have.

然后你就可以休假了。

And then you get to have a holiday.

然后我就能休一个应得的长假了。

And then I'll have a well-earned sabbatical.

是的，绝对。Demis，非常感谢。一如既往地有帮助。

Yeah, absolutely. Demis, thank you so much. Helpful as always.

好了，本季的 Google DeepMind 播客到此结束，我是 Hannah Fry 教授。但请务必订阅，这样你就能第一时间听到我们在 2026 年回归的消息。与此同时，为什么不重温我们庞大的剧集库呢？因为今年我们涵盖了太多内容，从无人驾驶汽车到机器人技术，从世界模型到药物发现。足够让你忙个不停。再见。

Well, that is it for this season of the Google DeepMind podcast with me, Professor Hannah Fry. But be sure to subscribe so you will be among the first to hear about our return in 2026. And in the meantime, why not revisit our vast episode library because we have covered so much this year. From driverless cars to robotics, world models to drug discovery. Plenty to keep you occupied. See you soon.