BuildSpeak每日 builder 文摘
今日归档生词本关于
🎙 播客Training Data· 2026 年 4 月 30 日· 4,769 词 · 约 24 分钟

Demis Hassabis on Building DeepMind, AlphaFold, and the Final Stretch to AGI

SPACE 播放 / 暂停·←→ 上一句 / 下一句
Speaker 100:02 - 00:06
Dennis, thank you so much. Exciting to be here. Thanks everyone for coming. It's great to be here.
Speaker 100:02 - 00:06
Dennis,非常感谢你。很高兴来到这里。也谢谢大家今天到场。能来这里真是太好了。
Speaker 200:06 - 00:09
We're so honored to have you at our chocolate factory.
Speaker 200:06 - 00:09
我们非常荣幸能邀请你来到我们的巧克力工厂。
Speaker 100:09 - 00:13
Yes, I just heard about that. Looking forward to the chocolate afterwards.
Speaker 100:09 - 00:13
是的,我刚刚才听说这件事。很期待之后的巧克力。
Speaker 200:13 - 00:36
Excellent. Well, Dennis, we're going to jump right in. We have one of the OGs in every way, original thinkers, founders, visionaries, and all things AI. A true believer, true scientist in Dennis. We're going to spend the beginning of the conversation about the early days, then early days of DeepMind, and then we'll get into the science and open up the room for some questions.
Speaker 200:13 - 00:36
太好了。那么,Dennis,我们就直接开始吧。今天我们请到的是一位在各方面都堪称 OG(元老级人物)的人物:原创思想家、founder(创始人)、visionary(远见者),以及 AI 领域的全能人物。Dennis 是一位真正的信徒,也是一位真正的科学家。我们会先聊聊早期经历,然后谈谈 DeepMind 的早期,再进入科学话题,最后也会开放现场让大家提一些问题。
Speaker 200:36 - 00:56
So let's jump right in. So Demis, you were a chess prodigy. You also were a founder of a gaming company. You're a neuroscientist, and you're the founder of DeepMind, and now leader of a really big consequential company. Those seem like pretty unrelated pieces, but you've said that there's a common thread.
Speaker 200:36 - 00:56
那我们就直接切入正题。Demis,你曾经是国际象棋神童;你也创办过一家游戏公司;你是一位 neuroscientist(神经科学家);你还是 DeepMind 的 founder(创始人),现在又在领导一家规模巨大且影响深远的公司。这些身份看起来彼此相当不相关,但你曾说过,它们之间其实有一条共同主线。
Speaker 200:56 - 00:57
Can you There walk us through
Speaker 200:56 - 00:57
你能带我们梳理一下吗——
Speaker 100:58 - 01:37
is a common thread and maybe I made it into a common thread, so it could be a post hoc sort of shaping. But I wanted to do AI for a long time, so I kind of decided it was the most important thing I could possibly and most interesting thing I could do in my teenage years. And then I picked things to study or do that I felt eventually would help me build a company like DeepMind. So I had that as a plan from about 15, 16 years old. I had a detour into games because actually that was in the 1990s, this is that's where all the most cutting edge technology was being done.
Speaker 100:58 - 01:37
确实有一条共同主线,不过也可能是我后来把它整理成了一条主线,所以某种程度上也许是一种 post hoc(事后)的塑形。但我很早就想做 AI,所以在我十几岁的时候,我基本上认定这是我可能去做的最重要、也最有趣的事情。之后我选择学习和从事的事情,都是那些我觉得最终能帮助我建立像 DeepMind 这样的公司的方向。所以大概从 15、16 岁开始,我就有了这样的计划。后来我绕去做游戏,其实是因为那是在 1990s,当时最前沿的技术基本都在那里诞生。
Speaker 101:37 - 02:21
Obviously, not just in AI, but in graphics especially, including hardware, of course, the GPUs we all use today, they were designed for graphics engines and we used the I mean, I was using the very first GPUs back in the day in the late 90s. So, there was a lot of really cutting edge technology and then all the games I made, including all the games I did for Bullfrog, but also my own company, Elixir Studios, all involved AI as the main gameplay component. So probably the most well known game I made was Theme Park when I was about 17 and that was a simulation of an amusement park and thousands of little people came into your theme park and played on your rides and decided what to buy from the shops. So there was a whole kind of economics AI model underneath. And it was one of the first games of its type along with SimCity.
Speaker 101:37 - 02:21
显然,不只是 AI,尤其是在 graphics(图形)方面也是如此,当然还包括 hardware(硬件)。我们今天都在使用的 GPUs,当年最初就是为 graphics engines(图形引擎)设计的;我的意思是,我在 90 年代后期就已经在使用最早的一批 GPUs 了。所以当时有大量非常前沿的技术。而且我做过的所有游戏——包括我在 Bullfrog 做的那些,以及我自己公司 Elixir Studios 的作品——都把 AI 作为核心 gameplay component(游戏机制组件)。所以我做过的、可能最知名的一款游戏是 Theme Park,那时我大概 17 岁。那是一款游乐园模拟游戏,成千上万的小人会进入你的主题公园,去玩你的游乐设施,并决定要从商店里买什么。因此其底层其实有一整套类似 economics AI model(经济学 AI 模型)的系统。它和 SimCity 一样,都是这一类型中最早的一批游戏。
Speaker 102:22 - 02:51
And when I saw it sold 10,000,000 plus copies and when I saw how delighted people were to interact with the AI, that was one of the things that made me think about spending my whole career on it. And then of course neuroscience is to get inspiration from how the brain works and different algorithmic ideas from that and then just bringing all those different things together for the start of DeepMind when the timing we felt was right. And then of course we use games as the early proving ground for our AI ideas.
Speaker 102:22 - 02:51
当我看到它卖出了超过 10,000,000 份,又看到人们与 AI 互动时是多么欣喜,这让我开始认真考虑把自己的整个职业生涯都投入到这件事上。当然,neuroscience(神经科学)这部分,是为了从大脑的工作方式中获得启发,从中提炼出不同的算法思路;然后就是在我们觉得时机成熟的时候,把所有这些不同的东西整合起来,创办 DeepMind。当然,我们也把 games(游戏)作为早期验证 AI 思路的试验场。
Speaker 202:51 - 03:12
So we've got a room full of founders here, and you can relate because you're a founder not just once, but twice. Take us back to the first time, Elixir Studios. What was that like? It's not the startup that you're most known for, but it was one that you had incredible success with. How did you lead that and what did it teach you about building?
Speaker 202:51 - 03:12
这里坐着满满一屋子的 founders(创业者),而你很能让大家共鸣,因为你不是创办过一次公司,而是两次。带我们回到第一次吧,Elixir Studios。那段经历是什么样的?它不是你最广为人知的 startup(初创公司),但你在那里取得了惊人的成功。你当时是如何领导它的?它又让你学到了哪些关于 building(创业与打造产品/公司)的东西?
Speaker 103:12 - 03:58
Well look, I started Elixir Studios straight out of college and I was lucky enough to work at Bullfrog Productions, which for those of you know games, it was a kind of legendary game studio in the early days of the game industry, probably the best one in The UK in Europe. And I wanted to do something that combined pushed AI, so effectively I was funding AI back in those days through the backdoor through games development and then push the forefront of that and combine it with cutting edge creativity. And I think that's still relevant today with the way we do our blue sky research. But maybe the biggest lesson I learned was you want to be five years ahead of your time, not fifty years ahead. So we tried to do a game called Republic at Elixir Studios, which simulated a whole country.
Speaker 103:12 - 03:58
你看,我大学一毕业就创办了 Elixir Studios,而且我很幸运曾在 Bullfrog Productions 工作。对了解游戏的人来说,那是在游戏产业早期一个近乎传奇的 game studio(游戏工作室),大概也是 The UK 乃至 Europe 最好的之一。我想做的是,把推动 AI 发展这件事结合进去;某种意义上说,那些年我是借由游戏开发这条“后门”来为 AI 提供资金支持,同时推进这一领域的前沿,并把它和最尖端的创造力结合起来。我觉得这一点在今天仍然适用,也体现在我们做 blue sky research(蓝天式研究)的方法上。不过,也许我学到的最大教训是:你要比时代领先五年,而不是五十年。所以我们当时在 Elixir Studios 试图做一款叫 Republic 的游戏,它会模拟整个国家。
Speaker 103:58 - 04:33
And then the idea of the game is you could sort of overthrow, I think there was a dictator in charge of the country in any number of different ways. And we basically simulated living breathing cities. And this is bearing in mind, this is like the late 90s on a Pentium. So we had to get all the graphics and all the AI for a million people working on home PC at the time. So it was a little bit ambitious and maybe it was too ambitious and it caused some issues and I took that lesson with me of like you want to be ahead of your time, you don't want to be obviously when it's obvious to everyone, it's too late.
Speaker 103:58 - 04:33
这款游戏的设定大致是:你可以用许多不同方式推翻那个国家的统治者,我记得应该是个 dictator(独裁者)。而我们实际上模拟的是一个个有生命力、会呼吸的城市。要知道,这还是在 90 年代末,而且运行平台只是 Pentium。所以我们得让一百万人的所有 graphics(图形)和 AI,都在当时的家用 PC 上跑起来。所以这目标有点太雄心勃勃了,也许确实过于雄心勃勃,并因此带来了一些问题。我后来一直记着这个教训:你要领先于时代,但不能领先得太离谱;等到所有人都觉得这事显而易见时,又已经太晚了。
Speaker 104:33 - 04:39
But if you're fifty years ahead, then there's probably no way you can get it to be successful.
Speaker 104:33 - 04:39
但如果你领先了五十年,那它大概就根本没有办法取得成功。
Speaker 204:39 - 04:45
Alright. So speaking of not being too far ahead of your time, it was 2009 and you decided there would be AGI.
Speaker 204:39 - 04:45
好的。那么说到不要比时代超前太多,时间来到 2009 年,你当时认定 AGI 会出现。
Speaker 104:48 - 04:52
Maybe it was only ten years ahead of our time, that time, better than fifty years.
Speaker 104:48 - 04:52
也许那次我们只比时代早了十年,总比早五十年好。
Speaker 204:52 - 05:08
So tell us about, again, room full of founders here, tell us about 'nine. How did you convince the first few brilliant talents? Because you pulled in really high caliber employees, early team members. How do you convince them to believe in what seemed like total sci fi at the time?
Speaker 204:52 - 05:08
那么再跟我们讲讲吧,毕竟这里坐着满屋子的 founders(创业者),讲讲 2009 年。当时你是怎么说服最早那几位杰出人才的?因为你很早就吸引到了非常高水平的员工和早期团队成员。你是怎么让他们相信一件在当时看起来简直像彻底 sci-fi(科幻小说)的事情的?
Speaker 105:09 - 05:43
Well, there were some interesting threads that I think we picked up on. I think we thought we were five years ahead, but maybe we were more like ten. But it was deep learning, it just being invented by Geoff Hinton and colleagues sort of in academia, but almost no one had really realized it was a big deal. We knew a lot about reinforcement learning and we felt there was huge progress to be made by combining those two techniques, which almost had never been mixed together really, certainly not any kind of anything other than toy problems in the academic subjects. There were quite two quite siloed parts of AI.
Speaker 105:09 - 05:43
嗯,我觉得我们抓住了一些很有意思的线索。我们原以为自己只是领先了五年,但也许实际上更像是十年。当时正值 deep learning(深度学习)刚刚由 Geoff Hinton 和同事们在学术界发明出来,但几乎还没有人真正意识到这是一件大事。我们对 reinforcement learning(强化学习)了解很多,也觉得把这两种技术结合起来会带来巨大的进展,而这两者当时几乎从未真正被混合使用过,当然更不用说除学术研究里的 toy problems(玩具问题)之外的任何场景了。它们当时是 AI 里两个彼此相当隔绝的分支。
Speaker 105:44 - 06:39
Then we could see the compute, the GPUs at the time were going to be really useful, of course we use TPUs now, but the accelerated computing industry was going to be very helpful. And then we also felt at the end of my PhD and postdoc and some of the other people I got together were computational neuroscientists that we had enough ideas and principles from the brain that could be useful, including the idea that reinforcement learning can eventually scale to AGI. So we felt we had these ingredients and we almost felt like we were keepers of a secret because no one either in academia or industry really believed that any big progress was possible. In fact, a lot of the people in academia used to roll out, literally roll their eyes up at us when we were sort of suggested we would work on AGI or strong AI, it was sometimes called at the time because it was like, well, we know this doesn't work. So everyone tried it in the 90s.
Speaker 105:44 - 06:39
然后我们也看到了算力的发展,当时的 GPUs 会变得非常有用——当然我们现在用的是 TPUs——但整个 accelerated computing(加速计算)行业都会带来很大帮助。另外,在我 PhD 和 postdoc 快结束的时候,我聚集起来的一些人也是 computational neuroscientists(计算神经科学家),我们觉得自己已经从大脑中提炼出了足够多可能有用的想法和原则,其中也包括这样一个观点:reinforcement learning 最终是可以扩展到 AGI 的。所以我们觉得这些关键要素已经具备了,甚至几乎觉得自己像是在守着一个秘密,因为无论在 academia 还是 industry,几乎都没有人相信还可能取得什么重大进展。事实上,当时学术界很多人一听到我们说要做 AGI,或者那时有时被称作 strong AI,就真的会对我们翻白眼,因为他们会觉得:这个东西行不通,我们早就知道了。毕竟 90 年代大家都试过了。
Speaker 106:39 - 07:24
I did my postdoc at MIT, which was the sort of center point for expert systems and first order logic language systems. I mean, seems amazing to think that now, but I was already feeling that that was a chaotic then. But they they you know, that's still how it was done, both in Cambridge and The UK and also in MIT, these big centers of traditional AI. And it felt like but actually that convinced me even more that we were on to something because at least if we were going to fail, we would fail in a different way than people had failed to get to AGI in the 90s. So that felt like it was worth doing no matter what, even if obviously it was research, we didn't know for sure it would be successful, but at least we would would fail in an original way if it didn't work.
Speaker 106:39 - 07:24
我在 MIT 做 postdoc,而那里当时可以说是 expert systems(专家系统)和 first order logic language systems(一阶逻辑语言系统)的核心重镇。现在回头想,这听起来都很惊人,但我当时已经感觉那一套有些混乱了。不过你知道,在 Cambridge、在 UK,还有在 MIT 这些传统 AI 的大型中心里,主流做法依然是那样。可这反而让我更相信我们是真的找对了方向,因为至少如果我们会失败,我们也会以一种不同于 90 年代那些通往 AGI 的失败方式去失败。所以无论如何,这件事都值得去做。显然那是 research,我们并不确定一定会成功,但至少如果最后不行,我们也会是以一种原创的方式失败。
Speaker 207:25 - 07:33
Was there any common sticking point in that early belief? Was there something that you had to prove either to yourself or to your early followers to get them on board?
Speaker 207:25 - 07:33
在你们早期的这种信念里,有没有某个共同的卡点?有没有什么事情是你必须先向自己、或者向最早追随你的人证明,才能让他们真正加入的?
Speaker 107:33 - 07:56
Well, had, put put it this way, I would have been spent my life on AI no matter what had happened. So as it's turned out, it's gone it's sort of gone on the absolutely amazing side of the optimistic side of what we thought. Still actually within what we were predicting in 2010. We thought it would be a twenty year mission. And I think we're basically exactly on track as a field for that.
Speaker 107:33 - 07:56
嗯,这么说吧,不管后来发生什么,我这辈子都会投身于 AI。结果证明,事情的发展落在了我们当初最乐观设想中那种近乎惊人的一侧。不过其实它仍然在我们 2010 年时的预测范围内。我们当时认为这会是一项二十年的使命,而我觉得就整个领域而言,我们现在基本上正好在这条轨道上。
Speaker 107:56 - 08:18
And obviously, we played our part in that. But even if it hadn't transpired that way and it was still now niche subject, that's what I would still be doing because I felt it was the most important technology ever if it was obvious to me. Our original mission statement at DeepMind was step one, solve intelligence, I. E. Build AGI, step two, use it to solve everything else.
Speaker 107:56 - 08:18
很显然,我们在其中也发挥了自己的作用。但即便事情并没有那样展开,即便直到现在它仍然只是个小众领域,我也还是会在做这件事,因为在我看来,如果这条路成立,它就是有史以来最重要的技术。我们当初在 DeepMind 的 mission statement(使命宣言)是:第一步,solve intelligence,也就是 build AGI;第二步,用它去解决其他一切问题。
Speaker 108:18 - 08:52
So it was always I always thought it was the most important technology that could ever be invented, but also the most interesting one. So as a tool for science, as an interesting artifact in itself and actually as one of the best ways to understand our own minds, you know, like the nature of consciousness, dreaming, creativity, all of these questions I had as a neuroscientist, I felt one of the things that was missing was an analysis tool like AI, but also a comparison that you could do sort of a controlled experiment, study, and compare two different systems against each other.
Speaker 108:18 - 08:52
所以我一直都认为,这可能是有史以来最重要的技术,同时也是最有趣的技术。无论把它看作 science 的工具、一个本身就极其有意思的 artifact(人造产物),还是把它看作理解我们自身心智的最佳方式之一——比如 consciousness(意识)、dreaming(做梦)、creativity(创造力)这些问题,都是我作为 neuroscientist 曾经关心的——我都觉得,其中缺失的一块正是像 AI 这样的分析工具。而且它还提供了一种比较对象,让你能够做某种受控的实验、研究,并把两个不同系统彼此对照比较。
Speaker 208:52 - 09:07
Let's talk about AI for science. You've been early to that, you've been a believer, and you've been really a purist about this. This is the driving mission. What about the way you set up DeepMind and set the culture has positioned it to be on the constant forefront of AI for science?
Speaker 208:52 - 09:07
我们来谈谈 AI for science。你很早就投入这一方向,你一直相信它,而且你在这件事上确实相当纯粹。这就是驱动一切的核心使命。你当初是如何搭建 DeepMind、如何塑造其文化的,这又怎样让它始终处在 AI for science 的前沿?
Speaker 109:08 - 09:46
Well, that was the ultimate goal, at least for me, my personal passion. There's my own drive to build AI, which was to advance science and medicine and our understanding of the world. It's my expression of that mission was to sort of do it in a meta way, right, build the ultimate tool and then come back when that was ready and use it to make breakthroughs in science, things like AlphaFold that we've done and I think many more things. So we've always had that at the heart of what we've been trying to do at DeepMind. Actually we've had an AI for Science group division led by Pushme Kohli that has existed for nearly a decade now.
Speaker 109:08 - 09:46
嗯,那一直都是最终目标,至少对我来说、对我个人的热情而言是这样。推动我去构建 AI 的内在动力,是推进科学、医学以及我们对世界的理解。可以说,我对这一使命的表达方式有点“元”的意味,对吧:先打造终极工具,等它准备好之后,再回过头来用它在科学上实现突破,比如我们已经做出的 AlphaFold 这样的成果,而且我认为还会有更多。所以,这始终是我们在 DeepMind 努力方向的核心。实际上,我们一直有一个由 Pushme Kohli 领导的 AI for Science 团队/部门,至今已经存在将近十年了。
Speaker 109:46 - 10:37
Actually, pretty much the day after we got back from Seoul and the AlphaGo match, is sort of ten years to the month now is when we started formally started the AI for Science efforts because I was waiting for the algorithms to be powerful enough and the ideas to be general enough. And for me, cracking Go was that point, that time that we thought, okay, now we're ready to really apply these ideas to important real world problems starting with these big scientific challenges. So we've always had that in mind as the most beneficial use of AI, like what could be better than using it to cure diseases and give us healthier lifespans and to help with medicine, followed obviously by other really important areas like material science and the environment and energy and these kinds of topics, which I think AI is also going to have a huge part to play in the next few years.
Speaker 109:46 - 10:37
实际上,几乎就在我们从 Seoul 和 AlphaGo 比赛回来后的第二天——到现在差不多正好十年——我们就正式启动了 AI for Science 的工作,因为我一直在等待算法变得足够强大、理念变得足够通用。对我来说,攻克 Go 就是那个节点,那个让我们觉得“好,现在已经准备好把这些想法真正应用到重要现实问题上”的时刻,而起点就是这些重大的科学挑战。所以,我们一直都认为,这是 AI 最有益的用途:还有什么能比用它来治愈疾病、让我们拥有更健康的寿命、帮助医学更好发展更好的事呢?当然,紧随其后的还有其他同样非常重要的领域,比如材料科学、环境、能源这类议题,我认为在未来几年里,AI 也会在这些方面发挥巨大作用。
Speaker 210:37 - 10:56
And how does AI breakthrough in biology? You're deeply involved with isomorphic. This is an area of deep passion. You have been a purist on the potential of AI to cure diseases from the very beginning. When do we have the type of moment that we've had in language and coding, but in biology?
Speaker 210:37 - 10:56
那么,AI 在生物学中的突破会是什么样?你深度参与了 isomorphic。这个领域是你极为热爱的方向。从一开始,你就是那种坚信 AI 有潜力治愈疾病的“纯粹派”。那么,生物学领域什么时候会迎来像语言和编程领域那样的时刻?
Speaker 110:56 - 11:42
Yeah, well, mean, I'd argue we've already had one of those moments with AlphaFold. So it's a 58 gram challenge, protein folding and the three d structure of proteins is an incredibly important thing to know about if you want to design medicines or if you want to understand biology. Of course, it's only one part of the drug discovery process, it's an important part, but it's only one part. So isomorphic labs, is our latest spin out, having a lot of fun running that as well is to build adjacent technologies in more biochemistry and chemistry space that can actually design the compounds automatically to kind of fit and bind to the right part of the protein. So we now know the protein, the shape of the protein, we know that what's on the surface of the protein and what we have to target.
Speaker 110:56 - 11:42
对,我会说,其实我们已经有过那样一个时刻了,就是 AlphaFold。它攻克的是一个已有 58 年历史的难题——protein folding(蛋白质折叠);如果你想设计药物,或者想理解生物学,了解蛋白质的 three-dimensional structure(三维结构)是极其重要的。当然,这只是药物发现流程中的一部分;它很重要,但也只是其中一部分。isomorphic labs 是我们最新分拆出去的公司,我也很享受运营它。它的目标是在更偏 biochemistry(生物化学)和 chemistry(化学)的领域构建相邻技术,能够自动设计化合物,使其匹配并结合到蛋白质上正确的位置。所以我们现在已经知道蛋白质了,知道蛋白质的形状,也知道蛋白质表面有什么,以及我们需要瞄准什么。
Speaker 111:42 - 12:41
But now we've got to build the right compound that of course binds strongly to where you want it to bind on the target of interest, but doesn't bind to anything else ideally because that would be a toxic side effect. So the dream is to do almost all the exploration, which is 99% of the work and the time in silico and then save the wet lab step just for the validation step, right? So that would be I think if we can do that and I think we can get there in the next few years, I think we could reduce drug discovery times instead of doubt for taking like an average of ten years down to months, maybe even weeks And and perhaps even days one then I think then all disease could be in reach and I think things like personalized medicine will become possible, personalized variations off of base medicines. So I think the medical area, drug discovery areas is going to be revolutionized in the next few years.
Speaker 111:42 - 12:41
但现在我们还必须构建出正确的化合物:它当然要能强力结合到你希望它在目标位点上结合的位置,但理想情况下又不能和其他任何东西结合,因为那会带来毒性副作用。所以,理想图景是,几乎把所有探索工作——这占了 99% 的工作量和时间——都放到 in silico(计算机模拟)中完成,然后把 wet lab(湿实验室)步骤只留给验证环节,对吧?如果我们能做到这一点,而且我认为未来几年内我们可以做到,那么我想我们就能把药物发现时间从如今平均大约十年,缩短到几个月,甚至几周,或许甚至几天。一旦如此,我认为所有疾病都将变得可触及、可治疗,而且像 personalized medicine(个性化医疗)这样的事情也会成为可能,即在基础药物之上做个性化变体。所以我认为,医疗和药物发现领域将在未来几年被彻底改变。
Speaker 212:41 - 13:00
Brilliant. You talked a lot about AI for science. Do you think that at some point AI will create new sciences, a la industrial revolution and thermodynamics? Will there be something new, taught fundamentally in our education system? And if so, what would it be like?
Speaker 212:41 - 13:00
太精彩了。你谈了很多 AI for Science。你是否认为,在某个时点,AI 会像 industrial revolution(工业革命)与 thermodynamics(热力学)那样,创造出全新的科学?会不会出现某种全新的东西,被作为基础内容纳入我们的教育体系?如果会,那会是什么样子?
Speaker 113:01 - 13:36
Well, think there's several things along that those lines that I think is going to happen. So first of all, the understanding and the analysis of AI systems themselves, I think is going to become a whole science, a kind of engineering science. These are incredible, incredibly interesting artifacts that we are building and they're incredibly complex as well. As complex eventually, they'll be as complex as the human mind and the brain. And so they'll need to be studied so we can understand fully way beyond where we are today, how these systems work.
Speaker 113:01 - 13:36
嗯,我认为沿着这条思路,会发生好几件事。首先,我认为,对 AI 系统本身的理解与分析将会发展成一整门科学,一种 engineering science(工程科学)。我们正在构建的是极其惊人、也极其有趣的人造产物,而且它们也极其复杂。最终,它们会复杂到接近人类心智和大脑的程度。因此,我们需要研究它们,才能在远远超出今天水平的层面上,真正充分理解这些系统是如何运作的。
Speaker 113:36 - 14:02
So I think there's a whole kind of field, Meckinterp is part of that, but there's a lot more I think that we can do to analyze these systems. So that will be a science. But I think also AI itself will maybe unlock new sciences, which is maybe what you're getting at. The one I'm particularly excited about is AI for simulations. So I love simulations, all the games I wrote not only had AI, but they were simulations.
Speaker 113:36 - 14:02
所以我认为,这里会形成一个完整的领域,Meckinterp 是其中一部分,但我觉得我们还能做更多来分析这些系统。因此,这本身会成为一门科学。但我也认为,AI 本身也许还会解锁新的科学,这可能才是你真正想问的。我尤其兴奋的一个方向是 AI for simulations(面向模拟的 AI)。我非常喜欢 simulations(模拟);我写过的所有游戏里,不仅有 AI,而且它们本身也是 simulations。
Speaker 114:02 - 14:30
And I think simulations is the way we can address some of the what we maybe think of social sciences, like economics and other more humanistic subjects, because it's very difficult to do control studies in that. Why aren't they just sciences like physics today? Because the problem is they're emergent systems just like biology actually. And it's very hard to do repeated controlled experiments. If you raise interest rates by half a percent, you have to do it in the real world and then see what happens.
Speaker 114:02 - 14:30
我认为,simulation(模拟)是我们处理某些通常被视为社会科学的问题的一种方式,比如 economics(经济学)以及其他更偏人文的学科,因为在这些领域里做受控研究非常困难。它们为什么直到今天还不像 physics(物理学)那样成为“科学”?因为问题在于,它们和 biology(生物学)其实一样,都是 emergent systems(涌现系统)。而且要做可重复的受控实验非常难。比如你把利率提高半个百分点,你必须在现实世界里这么做,然后再看会发生什么。
Speaker 114:30 - 14:54
You can have theories, but you can't run it thousands of times. But if you could simulate things really accurately, may be there's sort of new sciences to be done where you can sort of rigorously sample from a very accurate simulator. And then I think that will allow us to make much better decisions in these today, what are very uncertain domains. What will it take
Speaker 114:30 - 14:54
你可以提出理论,但你没法把它运行几千次。但如果你能非常准确地模拟这些事情,也许就能开展某种新的科学:你可以从一个非常精确的 simulator(模拟器)中进行严格采样。这样一来,我认为我们就能在这些如今仍高度不确定的领域里,做出更好的决策。要做到这一点,需要什么
Speaker 214:54 - 15:01
to get to those extremely accurate simulations? World models, what kind of science is necessary and engineering together?
Speaker 214:54 - 15:01
才能实现那些极其精确的模拟?需要 world models(世界模型)吗?需要什么样的科学,再加上什么样的工程?
Speaker 115:01 - 15:33
Yeah, well, look, mean, I'm thinking a lot about that in we do a ton of that work, like learning simulators, basically what it would So these are in domains where you can't we don't know the mathematics of it well enough or it's perhaps too complex. We can't just write directly down a special case simulator. It's just not accurate enough, doesn't capture all the variables. We're doing that we've done it with weather. We have the most accurate kind of weather simulator in the world, WeatherNext, and it's far faster than what the Can
Speaker 115:01 - 15:33
是的,你看,我确实一直在想这个问题,而且我们也做了大量这方面的工作,比如 learning simulators(学习型模拟器),基本上就是——这些通常是在某些领域中,我们对其数学规律了解得还不够充分,或者它本身可能太复杂。我们没法直接写出一个针对特定情况的 simulator(模拟器)。那样不够准确,也无法捕捉所有变量。我们正在做这件事,而且已经在 weather(天气)上做了。我们有世界上最准确的一类 weather simulator(天气模拟器),叫 WeatherNext,而且它比现有的气象
Speaker 215:33 - 15:34
metrology we change the weather yet?
Speaker 215:33 - 15:34
学方法快得多。我们已经能改变天气了吗?
Speaker 115:34 - 15:45
No, we can't. No. And I'm not sure that would be a good idea. But the first step is to understand it better. But then even biology, we're working on a kind of what I call a virtual cell.
Speaker 115:34 - 15:45
不能,还不能。我也不确定那会不会是个好主意。但第一步是先更好地理解它。然后 even biology(甚至生物学)也是一样,我们正在做一种我称之为 virtual cell(虚拟细胞)的东西。
Speaker 115:45 - 16:38
So, a hugely dynamical emergent system. And I think biology is perfect sort of machine learning is perfect description language for biology in the same way math is for physics. Because I think in biology and in lots of these natural systems, have loads of weak signals, weak correlations, tons of data, far too much that any human mind can analyse. But there are connections and correlations and interesting causalities within that massive data. So I think it's sort of it's always struck me that machine learning is the perfect tool to describe those kinds of systems where until today, mathematics hasn't been able to do that, either because we can't manage it as top mathematicians because it's too complex or the expressive power of math is not enough for to understand these sort of highly emergent dynamical systems.
Speaker 115:45 - 16:38
所以,这是一个高度动态的 emergent system(涌现系统)。而我认为 biology(生物学)非常适合用 machine learning(机器学习)来描述,就像 math(数学)之于 physics(物理学)一样。因为我觉得,在 biology(生物学)以及很多这类自然系统中,存在大量 weak signals(弱信号)、弱相关性和海量数据,多到任何人类大脑都无法分析。但在这些海量数据内部,确实存在联系、相关性,以及有趣的 causalities(因果关系)。所以我一直觉得,machine learning(机器学习)是描述这类系统的完美工具;直到今天,mathematics(数学)还做不到这一点,要么是因为即便是顶尖数学家也无法驾驭这种复杂性,要么是因为 math(数学)的表达能力还不足以理解这类高度涌现、动态变化的系统。
Speaker 216:38 - 16:41
Is it also because of the messiness and stochastic nature?
Speaker 216:38 - 16:41
这是否也和它的混乱性以及 stochastic(随机)特性有关?
Speaker 116:41 - 17:07
Yeah, sure. I mean, eventually you could, by the way, once you learn these simulators, it may be there's another branch of new branch of science. You could maybe extract some equations from the once you have the simulator. So you have this sort of implicit simulator, intuitive simulator, and then maybe you could extract explicit equations from that partly because you could also sample it as many times as you want.
Speaker 116:41 - 17:07
是的,当然。我的意思是,最终你也许可以——顺便说一句,一旦你学会使用这些 simulator(模拟器),也许会出现科学的另一个新分支。也许一旦你有了这个 simulator,你就可以从中提取出一些方程。所以你会先有这种隐式的 simulator、直觉式的 simulator,然后也许还能从里面提取出显式方程,部分原因是你也可以按你想要的次数反复对它进行采样。
Speaker 217:07 - 17:10
Fundamental as Maxwell's or something Maybe. Like
Speaker 217:07 - 17:10
像 Maxwell 那样基础的东西,或者类似那样的,也许吧。就像——
Speaker 117:10 - 17:17
I don't know if that exists for such emergent systems, but if they do exist, I don't see why we won't be able to find them with these methods.
Speaker 117:10 - 17:17
我不知道对于这类 emergent systems(涌现系统)来说,这样的东西是否存在;但如果它们确实存在,我看不出为什么我们不能用这些方法把它们找出来。
Speaker 217:18 - 17:33
That would be amazing. You've talked about this theory that the basic building block of everything in the universe could be information like, this is more theoretical. How do you think about that and what does that mean for a traditional classical Turing computer?
Speaker 217:18 - 17:33
那会非常惊人。你谈到过这样一种理论:宇宙中一切事物最基本的构件可能是 information(信息)——这是一个更偏理论的问题。你是怎么思考这件事的?它对于传统的经典 Turing computer(图灵计算机)又意味着什么?
Speaker 117:33 - 18:12
Well look, think you can, of course all the famous E equals MC squared and all the stuff Einstein did and energy and matter are kind of equivalent. But I actually think information has a kind of equivalency in the same way. So you can think of the organization of matter and structure and especially things like biology that are resisting entropy as basically information processing systems at their heart. So I think one can convert all of those three kind of quantities into each other, but I have this feeling information is most fundamental. So it's a little bit the opposite way around to the classic physicist thought in the 1920s and things where it's sort of energy and matter primary.
Speaker 117:33 - 18:12
你看,我觉得当然可以这样想:那些著名的 E=MC²,以及 Einstein 关于能量和物质所做的一切,都说明能量和物质在某种意义上是等价的。但我实际上认为,information 也以同样的方式具有某种等价性。所以你可以把物质的组织形式、结构,尤其是 biology(生物学)这类在对抗 entropy(熵)的东西,理解为其核心本质上是 information processing(信息处理)系统。所以我认为,这三类量某种程度上是可以彼此转化的,但我有一种感觉:information 是最根本的。因此,这在某种程度上与 1920 年代经典物理学家的想法相反;在那种想法里,能量和物质才是首要的。
Speaker 118:13 - 18:43
I actually think it's a better way to understand the world, the universe is to think about it as information first. And if that's true, and I think there's quite a lot of evidence for that, then of course AI is even more sort of profound in a sense Totally. Than we think. And it's already pretty profound because it's also about organizing information and understanding information and constructing informational objects. So AI, in my opinion, is all about information processing.
Speaker 118:13 - 18:43
我实际上认为,理解世界、理解宇宙的更好方式,是首先把它看作 information。如果这是真的——而且我认为有相当多证据支持这一点——那么当然,AI 在某种意义上就比我们以为的还要深刻得多。它本身已经非常深刻了,因为它同样关乎组织 information、理解 information,以及构造 informational objects(信息对象)。所以在我看来,AI 的核心就是 information processing。
Speaker 118:44 - 18:53
So I think there's something sort of very deeply connected with these different areas if you think of it through the lens of information processing as the primary way to think about it.
Speaker 118:44 - 18:53
所以我认为,如果你通过 information processing 这个视角,把它看作首要的思考方式,那么这些不同领域之间就存在某种非常深层的联系。
Speaker 218:54 - 18:58
Do you think a classical Turing machine will be able to compute everything?
Speaker 218:54 - 18:58
你认为经典 Turing machine(图灵机)将能够计算一切吗?
Speaker 118:58 - 19:35
Well, I sometimes think about what we're doing and refer to ourselves as Turing's champion because Turing machines, I think Alan Turing is one of my all time favorite scientific heroes. I think what he did obviously laid the foundations for computer science, but also AI. I think one of the most profound results ever is the Turing machine result. Everything that is computable can be computed by a relatively simple description of a machine. So I think our brains are likely to be approximate Turing machines.
Speaker 118:58 - 19:35
嗯,我有时会思考我们正在做的事情,并把我们自己称作 Turing 的 champion(捍卫者/传人),因为 Turing machine(图灵机)——我认为 Alan Turing 是我一直以来最喜欢的科学英雄之一。我认为他所做的工作显然奠定了 computer science 的基础,也奠定了 AI 的基础。我觉得有史以来最深刻的结果之一,就是 Turing machine 的结论:凡是可计算的东西,都可以由一种相对简单的机器描述来计算出来。所以我认为,我们的大脑很可能是近似的 Turing machines。
Speaker 119:35 - 20:29
And I think it's interesting to think about the connection between Turing machines and quantum computers and quantum systems. But I think at least what we've shown with things like AlphaGo and especially AlphaFold is that a classical Turing machine, obviously in the guise of a modern neural network, it can model what was thought to be in the case of protein folding, it's a quantum system. At some sense, it's very it's dealing with very small particles and one might think you'd have to take into account all the quantum effects of the water bonds and all sorts of things. But it turns out you can get to an approximate optimal sort of solution on a classical system. So it may turn out there are a lot of things that we think that would need a quantum system to model or run might be modellable on a classical system if thought about in the right way.
Speaker 119:35 - 20:29
我觉得,思考 Turing machines 与 quantum computers 和 quantum systems 之间的联系很有意思。但我认为,至少我们通过 AlphaGo,尤其是 AlphaFold 这类成果已经表明,一个 classical Turing machine,当然这里是以现代 neural network(神经网络)的形式出现,能够对某些原本被认为——以 protein folding(蛋白质折叠)为例——属于 quantum system 的问题进行建模。从某种意义上说,它确实处理的是非常微小的粒子,人们可能会觉得你必须把 water bonds 的各种 quantum effects,以及各种其他因素,全都考虑进去。但事实证明,你可以在 classical system 上得到一种近似最优的解。所以最终很可能会发现,有很多我们原以为需要 quantum system 才能建模或运行的东西,如果用正确的方式思考,其实是可以在 classical system 上建模的。
Speaker 220:29 - 20:35
So you've talked about AI consistently as a tool, like a telescope or
Speaker 220:29 - 20:35
所以你一直把 AI 说成是一种工具,像 telescope(望远镜)或者
Speaker 120:35 - 20:35
a
Speaker 120:35 - 20:35
microscope(显微镜)、astrolabe(星盘)那样,贯穿几个世纪。
Speaker 220:35 - 20:50
microscope, astrolabe through the centuries. But when you think about a machine that can model almost anything, let's say it can't even model quantum systems, like you pointed out, when does it stop becoming a tool and will that ever happen?
Speaker 220:35 - 20:50
但当你想到一种几乎可以为任何事物建模的机器时——就算它甚至不能像你指出的那样对 quantum systems 建模——它会在什么时候不再只是一个工具?这种情况真的会发生吗?
Speaker 120:51 - 21:28
Well, my strong feeling is we should in this sort of mission and journey to build AGI, those of us on that journey, many of people in this room, I feel like it would be best to build a tool first, an incredibly intelligent and useful and precise tool and then cross the next sort of Rubicon. That's already profound enough and has of course, the tool could start becoming more and more autonomous and agent like that we're all seeing. We're in the midst of that, the agent era now. But then there's a further step of like, does it have agency? Is it conscious?
Speaker 120:51 - 21:28
嗯,我非常强烈的感觉是,在这种构建 AGI 的使命和旅程中——我们这些走在这条路上的人,在这个房间里的很多人都是如此——我觉得最好的做法是先打造一个工具,一个极其智能、极其有用、极其精确的工具,然后再跨越下一道 Rubicon(卢比孔河,意指关键门槛)。这本身就已经足够深刻了;当然,这个工具也可能开始变得越来越 autonomous(自主),越来越像 agent,而我们现在都正在看到这一点。我们此刻就处在 agent era 之中。但接下来还有更进一步的一步,比如:它是否拥有 agency(自主行动能力)?它是否有 consciousness(意识)?
Speaker 121:28 - 21:52
These sorts of questions, which are also going to be questions we're going to need to address. But I would recommend we do that as a second step, perhaps using the tool in the first step to help us with those next profound questions and ideally also we could understand our own brain and minds better and define things like consciousness a lot more precisely than we can today.
Speaker 121:28 - 21:52
这类问题也将是我们必须面对的问题。但我的建议是,把那作为第二步来做;也许可以在第一步中先利用这个工具,帮助我们处理那些更深刻的后续问题。理想情况下,我们也可以因此更好地理解我们自己的大脑和心智,并且比今天更精确地定义 consciousness 这类概念。
Speaker 221:53 - 21:56
Do you have estimations of what that definition of consciousness might look like?
Speaker 221:53 - 21:56
你是否对 consciousness 的那种定义最终会是什么样子,有一些估计吗?
Speaker 121:56 - 22:22
No, I mean, haven't got much to add beyond that thousands of years of philosophy hasn't said already. But I mean, it's very clear to me that it's obvious some components are going to be needed. They're probably necessary but not sufficient. Things like self awareness and the idea of self and other, some kind of continuity over time. So some of these things are clearly needed for anything that might look like consciousness.
Speaker 121:56 - 22:22
不,我是说,除了几千年的哲学早已说过的那些之外,我也没有太多可补充的了。不过我的意思是,对我来说很清楚的一点是:显然需要一些组成要素。它们很可能是必要条件,但不是充分条件。比如 self-awareness(自我意识)、self 和 other 的概念、某种跨时间的连续性。所以,这里面有些东西显然是任何可能看起来像 consciousness(意识)的事物所需要的。
Speaker 122:22 - 22:54
But I mean, obviously it's an open question as what the full definition is. I've talked to many of the great philosophers about that, Daniel Dennett, obviously sadly passed away recently, but we had a long conversation a few years back about this and I think one of the issues is how does a system behave? Does it behave like a conscious system? So that's you could argue some of the AI systems might end up being able to do that as they get close to AGI. But then there's still the question of why do we think each other are conscious?
Speaker 122:22 - 22:54
不过我的意思是,很显然,完整定义到底是什么,仍然是个开放问题。我和很多伟大的哲学家都谈过这个问题,比如 Daniel Dennett——很遗憾他最近去世了——但我们几年前曾就此长谈过。我认为其中一个问题是:一个系统是如何表现的?它会不会表现得像一个有意识的系统?所以你可以说,一些 AI 系统在接近 AGI 时,最终可能会具备这样的能力。但随后仍然有一个问题:为什么我们会认为彼此是有意识的?
Speaker 122:54 - 23:22
One is the way we're behaving, we're behaving like conscious beings. But the other thing is we're running on the same substrate. So I think if both those things are true, then it's parsimonious to imagine you're experiencing the same thing I'm experiencing, which is why we don't have that debate about normally about are each other conscious. But I think we'll obviously we'll never have the substrate equivalence with an artificial system. So there'll always be I think it will be hard to completely close that gap.
Speaker 122:54 - 23:22
其中一个原因是我们的行为方式——我们表现得像有意识的存在。另一个原因是,我们运行在相同的 substrate(底层基质)上。所以我认为,如果这两点都成立,那么设想你正在体验与我相同的东西,就是一种 parsimonious(更简约、假设更少)的看法;这也就是为什么我们平常通常不会去争论彼此是否有意识。但我认为,对于 artificial system(人工系统),我们显然永远不会拥有 substrate 的等价性。所以我想,这个鸿沟始终都会很难被彻底弥合。
Speaker 123:22 - 23:33
So you can look at it behaviorally, but what about experientially? There are probably some ways to do that post AGI, but maybe it's a bit out of scope today, even for AI for science discussion.
Speaker 123:22 - 23:33
所以你可以从行为层面去看,但从 experience(主观体验)层面呢?在 post-AGI 之后,可能会有一些办法来处理这个问题,但也许这在今天有点超出讨论范围了,哪怕是对于 AI for science 的讨论也是如此。
Speaker 223:33 - 23:50
Brilliant. So we're going to open the room to questions in just a moment, get your questions ready. But you brought up philosophers. You've mentioned Kant and Spinoza as two of your favorite philosophers. Kant is this deontological, highly duty driven philosopher.
Speaker 223:33 - 23:50
太精彩了。那么我们马上把现场开放给大家提问,请先准备好你们的问题。不过你提到了哲学家。你说过 Kant 和 Spinoza 是你最喜欢的两位哲学家。Kant 是那种 deontological(义务论的)、高度受 duty(责任/义务)驱动的哲学家。
Speaker 223:50 - 23:59
Spinoza almost has this deterministic view of the universe. How do you kind of connect those two beliefs and where is your thinking of how the world works?
Speaker 223:50 - 23:59
而 Spinoza 几乎持有一种关于宇宙的 deterministic(决定论)观点。你是如何把这两种信念连接起来的?以及你对于世界如何运作的思考处在什么位置?
Speaker 123:59 - 24:22
Well, the reason I like those two, they stuck out for me is that I think Kant, when I was doing my PhD in neuroscience, his sort of statements about the mind creates reality, right? I think that's basically true. And so another reason to study the mind, right? And how the brain works. And I'm interested ultimately in the nature of reality.
Speaker 123:59 - 24:22
嗯,我喜欢这两位、他们之所以让我印象特别深,是因为我觉得 Kant——在我读 neuroscience(神经科学)PhD 的时候——他那种关于 mind(心智)创造 reality(现实)的说法,对吧?我认为这基本上是真的。所以这也是另一个去研究 mind 的理由,对吧?以及研究 brain(大脑)是如何运作的。而我最终感兴趣的是 reality(现实)的本质。
Speaker 124:22 - 25:02
So we have to understand how interpreting that. And so I think that's for me what I took from Kant. Then Spinoza, it's more about the you could almost call spiritual dimension of like, well, if you're trying to understand the universe using science in my case as the tool, you're sort of understanding some deep mystery about how the universe works, right, in really kind of deep way. And that's what I feel we're doing and I'm doing when I do my science and we work on AI and we're building these tools is somehow we're kind of reading the language of the universe.
Speaker 124:22 - 25:02
所以我们必须理解我们是如何在解释它的。因此我觉得,这就是我从 Kant 那里得到的东西。至于 Spinoza,更多是某种几乎可以称为 spiritual(精神性)的维度:比如说,如果你试图理解宇宙——在我的情况下,是把 science(科学)当作工具——那么你其实是在理解一个关于宇宙如何运作的深层奥秘,对吧,而且是以一种非常深刻的方式。我觉得这正是我们在做的事,也是我在做科学、我们研究 AI、我们构建这些工具时所做的事:在某种意义上,我们像是在阅读宇宙的语言。
Speaker 225:03 - 25:14
Beautiful. What a beautiful way to say what you do every day. Demist, scientist, orator, and philosopher. We will, before we wrap, do a couple of rapid fire questions.
Speaker 225:03 - 25:14
真精彩。用这样优美的方式来描述你每天在做的事。demist、scientist、orator 和 philosopher。我们在结束前,会再来几个 rapid fire questions(快问快答)。
Speaker 125:14 - 25:14
Okay.
Speaker 125:14 - 25:14
好的。
Speaker 225:14 - 25:22
Thank you for finishing. He's not seen these yet. Sure. Over under on distribution year of AGI. Oh, wow.
Speaker 225:14 - 25:22
谢谢你把这些准备完。他还没看过这些。可以。对于 AGI 的落地年份,你的判断是 over under(高于还是低于某个年份)?哦,哇。
Speaker 225:23 - 25:26
Or reject premise No. Of 2030.
Speaker 225:23 - 25:26
或者你也可以不接受这个前提。不。2030。
Speaker 125:26 - 25:28
I've been pretty consistent about that.
Speaker 125:26 - 25:28
我对这个判断一直都挺一致的。
Speaker 225:28 - 25:35
Okay. 2030. Yeah. Must read book, poem, or paper for when we achieve AGI? Oh, wow.
Speaker 225:28 - 25:35
好。2030。对。等我们实现 AGI 时,有没有一本必读的书、诗或者 paper(论文)?哦,哇。
Speaker 125:37 - 25:50
For when we achieve once we achieve it. Well, my favorite book is The Fabric of Reality by David Deutsch, so I think that still holds. I'd hope to answer the questions in that book with with the AGI. That's my post AGI work.
Speaker 125:37 - 25:50
是说等我们实现的时候,一旦我们实现了之后。嗯,我最喜欢的书是 David Deutsch 的 The Fabric of Reality,所以我觉得这个答案依然成立。我希望能借助 AGI 来回答那本书里提出的问题。那就是我在 post-AGI 之后要做的工作。
Speaker 225:50 - 25:54
Brilliant. Yeah. Proudest moment so far in DeepMind?
Speaker 225:50 - 25:54
太精彩了。对。到目前为止,你在 DeepMind 最自豪的时刻是什么?
Speaker 125:55 - 25:59
Oh, wow. We've been lucky to have a lot. I mean, probably AlphaFold.
Speaker 125:55 - 25:59
哦,哇。我们很幸运,已经有过很多这样的时刻了。要说的话,大概是 AlphaFold。
Speaker 225:59 - 26:20
Okay. Yeah. Now a couple games questions. If you were engaged in a high stake strategy game, turn based strategy game, Civ, Polytopia series games, you could select one scientist from history, we're thinking the Einsteins, the Turing's, Newtons, Who would you select to be on your team?
Speaker 225:59 - 26:20
好的。是的。现在问几个关于游戏的问题。如果你在参与一场高风险的策略游戏、回合制策略游戏,比如 Civ、Polytopia 系列游戏,而你可以从历史上选择一位科学家——我们想到的是 Einstein、Turing、Newton 这些人——你会选谁加入你的队伍?
Speaker 126:20 - 26:32
On my team? On your team. Oh gosh. Probably Von Neumann, I think. Great I mean he's Yeah, you want a game theorist, I think.
Speaker 126:20 - 26:32
加入我的队伍?加入你的队伍。哦,天哪。我想大概会选 Von Neumann。很好,我是说,他确实——是啊,我觉得你会想要一位 game theorist(博弈论学者)。
Speaker 126:32 - 26:39
I think I think I think I think he's the best. Yeah. Makes sense. Yeah. That's really teammate.
Speaker 126:32 - 26:39
我觉得,我觉得,我觉得,我觉得他是最强的。是的。说得通。是的。那确实是个很棒的队友。
Speaker 126:39 - 26:41
Yeah. Alright.
Speaker 126:39 - 26:41
好的。
Speaker 226:41 - 26:43
Yeah. Well, Demis, you do it all. Thank you so much for being with
Speaker 226:41 - 26:43
是的。那么,Demis,你真是什么都在做。非常感谢你来到这里,
Speaker 126:43 - 26:44
us. Great.
Speaker 126:43 - 26:44
和我们交流。太好了。
Speaker 226:44 - 26:46
Please join me in thanking Demis. Thanks,
Speaker 226:44 - 26:46
请和我一起感谢 Demis。谢谢。
原文 ↗https://www.youtube.com/playlist?list=PLOhHNjZItNnMm5tdW61JpnyxeYH5NDDx8
BuildSpeak — 关于本项目BUILT IN PUBLIC · 跟随 builders 而非 influencers