BuildSpeak每日 builder 文摘
今日归档生词本关于
🎙 播客Training Data· 2026 年 6 月 11 日· 9,570 词 · 约 48 分钟

Google DeepMind's Logan Kilpatrick: Why the Model Eats the Harness

SPACE 播放 / 暂停·←→ 上一句 / 下一句
Speaker 100:00 - 00:02
So we could edit this set so it looks like we're
Speaker 100:00 - 00:02
所以我们可以把这一段剪一剪,让它看起来像是我们在这里。
Speaker 200:02 - 00:06
here. Okay? Yeah. Yeah. I I want this where where we were talking off camera.
Speaker 200:02 - 00:06
这里。好吗?对。对。我我是想把这个放到我们刚才没开机、镜头外聊天的那个位置。
Speaker 200:06 - 00:36
Like, we should do that for the intro because I think it just, like, makes all this stuff more capable. I've seen these examples of, like, such subtle nuance that make me appreciate that it's like the world understanding playing out. I was giving a talk and was on stage with my friend, Tulsi, who leads the model team. I had mentioned to someone in the crowd to edit the video, and they literally took the picture, edited it with Omni in real time, and this dog came on the stage. In the edited version, the other guests looked down and see the dog.
Speaker 200:06 - 00:36
比如说,我们应该把那个放到开场里,因为我觉得它会让所有这些东西显得更有能力。我见过一些例子,里面那种非常细微的 nuance(细微差别)让我意识到,这就像是 world understanding(世界理解)在真实展开。我当时在做一个演讲,和我的朋友 Tulsi 一起站在台上,他负责 model team(模型团队)。我跟台下的某个人提到,让他编辑一下这段视频,结果他们真的把那张画面拿去,用 Omni 实时编辑,然后一只狗就出现在了台上。在编辑后的版本里,其他嘉宾会低头看到那只狗。
Speaker 200:36 - 00:44
They chuckle a little bit. This is while I'm opining about whatever AI is about your jokes. Yeah. That it was not my jokes. They they laugh at the dog coming up.
Speaker 200:36 - 00:44
他们会轻轻笑一下。那时候我正在台上大谈特谈某些关于 AI 的内容。对。不是因为我的笑话好笑。他们他们是因为那只狗走上来才笑的。
Speaker 200:44 - 00:56
It jumps onto my lap. I sort of, like, acknowledge the dog. I keep talking. I'm, like, petting it or whatever. And just, like, there's, like, so much subtle subtlety in getting that right, and the model crushed it.
Speaker 200:44 - 00:56
它跳到我的腿上。我会有点像是,注意到那只狗,但继续讲话。我一边说,一边摸摸它之类的。这里面要拿捏准确的微妙细节实在太多了,而这个 model(模型)把它完成得非常出色。
Speaker 200:56 - 01:06
And it's just it it very interesting and, like, still trying to, like, absorb and digest, like, what that means for, you know, the way we make content and all these other things.
Speaker 200:56 - 01:06
而且这件事真的非常有意思,我也还在试着慢慢吸收、消化,这对我们制作内容的方式,以及其他所有这些事情,究竟意味着什么。
Speaker 101:06 - 01:33
That's so interesting. I'm delighted to have Logan on the show. Logan runs Google AI Studio and the Gemini API. You spend a lot of your time thinking and building for the next generation of builders.
Speaker 101:06 - 01:33
这太有意思了。很高兴邀请 Logan 来到节目中。Logan 负责 Google AI Studio 和 Gemini API。你花了很多时间在为下一代 builder(构建者/开发者)思考并打造产品。
Speaker 201:33 - 01:34
Yes.
Speaker 201:33 - 01:34
对。
Speaker 101:34 - 01:43
So I'm excited to talk to you about everything from AgenTik AI to AI coding, world models, more today, and right off the heels of Google IO. So what better timing?
Speaker 101:34 - 01:43
所以我很期待今天和你聊聊各种话题,从 AgenTik AI 到 AI coding(AI 编程)、world models(世界模型)等等,而且时间点也正好紧接着 Google IO 之后。还能有比这更好的时机吗?
Speaker 201:43 - 01:45
Yeah. I'm super excited. Thank you for having me.
Speaker 201:43 - 01:45
对,我超级兴奋。感谢你邀请我来。
Speaker 101:45 - 01:55
Wonderful. Let's start with AgenTik AI. So Sundar opened at IO by calling this the AgenTik Gemini era. What does AgenTik AI mean for Google?
Speaker 101:45 - 01:55
太好了。我们先从 AgenTik AI 说起。Sundar 在 IO 开场时把这称为 AgenTik Gemini 时代。AgenTik AI 对 Google 意味着什么?
Speaker 201:55 - 02:33
Yeah. It's a good question. I think and we were, we sort of if you if you followed closely, we did sort of mention some of these things back with, like, Gemini two point o, which I think was, like, a a little bit early. And so I think this era, this, like, Gemini 3.5 era, feels like it's actually becoming true now, we're in the era of agentic coating, or agentic products and everything agents as far as Gemini goes. Think for us, this agentic layer, and I think we announced this actually at IO, sort of being powered by the antigravity agent harness, is this additional through line for Google that sort of connects all of our products that they're sort of based on now.
Speaker 201:55 - 02:33
对,这是个好问题。我想,我们之前其实也算是——如果你当时有仔细关注的话——在 Gemini 2.0 那会儿就稍微提到过其中一些东西,只不过我觉得当时有点偏早了。所以我认为,现在这个时代,这个像是 Gemini 3.5 的时代,感觉现在真的开始成为现实了:我们正处在 agentic coding(智能体式编码)、agentic products(智能体式产品),以及就 Gemini 而言几乎一切都围绕 agents(智能体)的时代。对我们来说,这一层 agentic layer(智能体层),我想我们其实也在 IO 上发布过,某种意义上是由 antigravity agent harness 提供支持,它成了 Google 的一条额外主线,把我们现在所依托的所有产品连接起来。
Speaker 202:33 - 02:49
Historically, prior to Gemini, there actually wasn't a through line for the probably sub 100 number of Google products that we have, the 50 Google products we have. There wasn't a through line. We had Gemini. It became this through line. Everything is now sort of using Gemini in some way.
Speaker 202:33 - 02:49
从历史上看,在 Gemini 之前,Google 旗下大概不到 100 个、或者说我们这 50 个产品之间,其实并没有一条统一主线。那时候没有这样一条主线。后来有了 Gemini,它就成了这条主线。现在所有东西都在以某种方式使用 Gemini。
Speaker 202:49 - 03:05
That's now becoming true for anti gravity as sort of all of the products rebase to become sort of like agentic native products and actually taking action on behalf of users and helping them get things done. You see this new through line emerging, which I think is actually really, really interesting.
Speaker 202:49 - 03:05
现在,对 antigravity 来说,这也正在变成现实:随着所有产品都重新构建,逐步变成某种 agentic native products(原生智能体产品),真正开始代表用户采取行动、帮助他们把事情做成,你会看到一条新的主线正在浮现。我觉得这其实非常非常有意思。
Speaker 103:06 - 03:10
I'm sorry, help me with Antigravity is the IDE, right? Or the Yeah, non
Speaker 103:06 - 03:10
抱歉,帮我确认一下,Antigravity 是那个 IDE(集成开发环境),对吧?还是说——对,不是——
Speaker 203:11 - 03:24
antigravity is a lot of things, which I think is sort of, again, is an opportunity for us. You have sort of a core IDE. You have sort of the agent first experience if you want it on the web. You have a CLI. You have an SDK.
Speaker 203:11 - 03:24
antigravity 包含很多东西,而我觉得这恰恰再次说明了它对我们来说是一个机会。你可以把它理解为有一个核心 IDE,也有一个如果你愿意的话以 agent first(智能体优先)为核心的 web 体验,还有一个 CLI(命令行界面),还有一个 SDK(软件开发工具包)。
Speaker 203:24 - 03:58
So I actually think, and I don't know how much we've framed it this way, but it really is an ecosystem of stuff that we built, and it's designed to sort of, like, meet developers wherever they are. So you could use it through the Gemini API if you want to and you want a managed agent that you don't have to do any of the the sort of infrastructure work for. And then the most interesting bit is, like, it's not just the ecosystem of anti gravity stuff. It's also powering, like, literally, it's the the same harness is actually powering all the other Google products. So anti gravity will be powering a bunch of agent stuff in search, in the Gemini app, across, like, cloud and and AI Studio, really is exciting.
Speaker 203:24 - 03:58
所以我其实觉得——虽然我不确定我们有没有一直这样定义它——它本质上确实是一个我们构建出来的生态系统,设计目标就是在开发者所处的任何位置与他们衔接。你可以通过 Gemini API 来使用它,如果你想要的是一个 managed agent(托管式智能体),不想自己去做那些基础设施工作的话,也完全可以。然后最有意思的部分在于,它不只是 antigravity 自身这套生态;它实际上还在驱动——几乎可以说是字面意义上的——其他所有 Google 产品。也就是说,正是同一个 harness 也在为所有其他 Google 产品提供支持。所以 antigravity 将会为 search、Gemini app,以及 cloud 和 AI Studio 等各处的大量 agent 相关能力提供支持,这一点真的很令人兴奋。
Speaker 103:58 - 04:08
I see. So it used to be the Gemini API. So, like, the language model was the through line in terms of how AI gets baked into every Google product. Yeah. And now it's not only the API, it's the the coding harness
Speaker 103:58 - 04:08
我明白了。所以以前是 Gemini API。也就是说,在 AI 如何被嵌入每一个 Google 产品这件事上,language model(语言模型)一直是那条主线。对。而现在不只是 API 了,还有 coding harness。
Speaker 204:09 - 04:09
Exactly.
Speaker 204:09 - 04:09
完全正确。
Speaker 104:09 - 04:19
That's that's being used in each of these products, and, therefore, it's, like, coding agent itself that's driving more agentic properties inside the products. And I think Fair description?
Speaker 104:09 - 04:19
这些产品里都在使用它,因此,某种意义上说,真正推动产品内部出现更多 agentic(智能体式)特性的,是 coding agent 本身。我这样描述公平吗?
Speaker 204:19 - 04:33
Fair fair description. I think more generically too, it's just, like, it is the agent harness. I think, like, coding as sort of, like, a specialized use case of the agent harness, I think, is is obviously powerful, but it is, like, coding has proved to be the general purpose agent harness in addition to also working really well for coding.
Speaker 204:19 - 04:33
这个描述基本公平。我觉得更泛化地说,它其实就是 agent harness。我认为 coding 可以算是 agent harness 的一种专门 use case(使用场景),这显然很强大;但同时,coding 也已经被证明是一种通用的 agent harness,不仅适用于 coding,而且作为通用用途也很好用。
Speaker 104:33 - 04:37
Are agent harness and coding harness synonymous or not?
Speaker 104:33 - 04:37
agent harness 和 coding harness 是同义词吗,还是不是?
Speaker 204:37 - 05:07
There's definitely nuance. I think there's, like, optimization that you can squeeze out of, like, specializing, and actually, you see this where technically the agent harness that gets used for the way that AI Studio uses it is a little bit specialized for the vibe coding use case, and the way that the Gemini app is using the agent harness is a little bit specialized for the consumer, always on 20 fourseven agent. So I think you have that base harness that probably has 80% of the same stuff, and then you specialize for coding or for whatever the use case is. Interesting.
Speaker 204:37 - 05:07
肯定还是有细微差别的。我觉得,你确实可以通过做 specialize(专门优化)榨出更多优化空间。实际上你已经能看到这种情况:从技术上说,AI Studio 使用的 agent harness,会针对 vibe coding 这个 use case 稍微做一些专门化;而 Gemini app 使用的 agent harness,则会针对面向 consumer(消费者)、always-on(二十四小时在线)的 24/7 agent 稍微做一些专门化。所以我认为,你会有一个基础 harness,其中大概 80% 的东西是相同的,然后再针对 coding 或其他具体 use case 做专门化。很有意思。
Speaker 105:07 - 05:44
How do you think about the cannibalization of the existing business, especially now that you are going much more aggressively into agentic properties? Because I could see, for example, if all you're doing is search or summarization, there's not as much of a cannibalization fear. Whereas if you're actually going through my emails, replying to them for me, am I even going through my email anymore? And so I could imagine that there's actually just fewer human eyeball hours on your products as a result of having more agentic capabilities. Is that fair, or how do you think about the cannibalization?
Speaker 105:07 - 05:44
你怎么看对现有业务的 cannibalization(蚕食)问题,尤其是现在你们更激进地推进 agentic 特性之后?因为我能理解,比如如果你做的只是 search 或 summarization(摘要),那对业务被蚕食的担忧可能没那么大。可如果 AI 真的在替我处理 email、替我回复,那我还会不会自己去看 email?所以我能想象,随着 agentic 能力变多,你们产品上由真人眼球产生的使用时长反而可能会变少。这种看法公平吗?或者说,你们是怎么思考这种 cannibalization 的?
Speaker 205:44 - 06:10
Yeah, it's interesting. I think one observation I have is that at the beginning and I think Sundar has done a great job of talking through this is at the beginning of the current AI era, everyone assumed that AI being able to answer questions for you was going to be negative some for search. And actually, what's ended up happening is it's an incredibly positive sum for search. People are searching more. People are doing more.
Speaker 205:44 - 06:10
对,这很有意思。我的一个观察是,在当前这个 AI 时代刚开始的时候——我觉得 Sundar 在这方面解释得非常好——一开始大家都假设,AI 能替你回答问题这件事,对 search 来说会是负和的。但实际上最后发生的是,它对 search 来说是一个极其正和的结果。人们搜索得更多了,人们做的事情也更多了。
Speaker 106:11 - 06:12
Agents are searching too.
Speaker 106:11 - 06:12
agent 也会搜索。
Speaker 206:12 - 06:50
Yeah. Actually, again, there's this whole market that spawned at the same time that agents are doing more, at the same time that humans are also searching more. Obviously, there's a finite amount of human time in the world, but from my early feelings of how a lot of this is playing out, it does feel like it's very positive from an ecosystem value creation. How the human behavior aspect of it turns out I think is somewhat clear in the next one to two years, much less clear three to five years from now when the technology has improved and the products probably look a little bit different than the way that they do. But ultimately, that is the success of product.
Speaker 206:12 - 06:50
对。其实,还是那句话,恰恰在 agent 能做更多事情的同时,也出现了一个完整的新市场;与此同时,人类自己的搜索也变得更多了。显然,世界上人类时间总量是有限的,但就我目前对这件事发展方式的初步感受来看,它确实像是会从 ecosystem(生态系统)价值创造的角度带来非常积极的结果。至于其中“人类行为”这一面最终会怎么发展,我觉得在未来一到两年里会比较清楚;而三到五年之后就没那么清楚了,因为到那时技术会进一步提升,产品形态可能也会和现在看起来有些不同。但归根结底,这就是产品成功与否的体现。
Speaker 206:50 - 07:18
I think we have a bunch of conversations with Demis all the time, and it's like the point of building the technology is so that it can go and do stuff for you. Success for Google probably doesn't look like maximizing eyeball time in front of our products. It's like maximizing outcome for customers to do the thing that they wanna do so that they can go and live their life and do what they want, and so I feel like you'll probably see us go down the route of maximizing outcomes for customers and not maximizing eyeballs.
Speaker 206:50 - 07:18
我觉得我们一直都会和 Demis 进行很多交流,而核心点在于,构建这项技术的目的,就是让它能替你去做事情。对 Google 来说,成功大概并不意味着尽可能把用户眼球时间最大化地留在我们的产品前;更像是尽可能为客户实现他们想完成的结果,让他们做成自己想做的事,然后去过自己的生活、做自己想做的事情。所以我感觉,你大概会看到我们走向的是“最大化客户结果”,而不是“最大化眼球停留”。
Speaker 107:18 - 07:37
Yeah. I have this term stuck in my head, agent led growth. It seems to me So I'm using coding agents a lot in my personal time, and I just let the agent make all the infrastructure choices for me. I'm like, I don't care which database, you tell me. And the reason I ask is, it's true in coding today.
Speaker 107:18 - 07:37
对,我脑子里一直卡着一个词:agent-led growth(agent 驱动增长)。在我看来——我自己私下里现在就大量在用 coding agents(编程 agent),而且我会直接让 agent 替我做所有基础设施选择。我就会说,我不在乎用哪个 database(数据库),你来告诉我。之所以提这个,是因为这在今天的编程场景里确实已经成立了。
Speaker 107:37 - 07:50
I would imagine it's maybe gonna be generally true for a lot of things, let's say shopping down the line. How do you think that's gonna change how advertising works, how value capture works for the aggregators?
Speaker 107:37 - 07:50
我猜再往后看,这种情况大概也会普遍适用于很多事情,比如 shopping(购物)。你觉得这会怎样改变 advertising(广告)的运作方式,以及 aggregators(聚合平台)的价值获取方式?
Speaker 207:50 - 08:15
It feels like it's a very similar trend. This isn't perfectly true, but a lot of these things are just proxies of each other. The way that SEO works, I think, is directly correlated with the way that, I forgot what the term now for it. It's like GEO is the generative engine optimization or whatever it's called. And so it does feel like there's a lot of correlation between the things.
Speaker 207:50 - 08:15
这感觉像是一个非常相似的趋势。虽然这样说并不完全准确,但这些东西很多时候都只是彼此的 proxy(代理指标)。我认为 SEO 的运作方式,和——我一时忘了那个词现在叫什么了——大概就是 GEO,generative engine optimization(生成式引擎优化)之类的说法,是直接相关的。所以确实会让人感觉,这些事情之间存在很强的相关性。
Speaker 208:15 - 08:24
My guess is it looks like much less of a radical shift than I think maybe what we assume right now, just because these things compound on top of each other.
Speaker 208:15 - 08:24
我的猜测是,它看起来会比我们现在设想的那种情况要激进得少得多,因为这些东西是层层叠加、相互复利发展的。
Speaker 108:24 - 08:34
If you were to grade the scale of agenticness in terms of crawl, walk, run, where are we in terms of how agentic the Google suite of products is?
Speaker 108:24 - 08:34
如果用 crawl、walk、run 这个尺度来给 agenticness(agent 化程度)打分的话,你觉得 Google 整套产品目前处在什么阶段?
Speaker 208:34 - 09:10
Yeah, that's a great question. It's definitely crawl right now, and I think some of this is all of the inherent product tension for Google. You have, what, 13,000,000,000 plus user products, and so I actually think we have some more labs like experiences where you're probably closer to running or walking, but I think most of the product experience today is definitely closer to crawling, and I think that's just the stewardship responsibility we have building a product that's being used by lots of people. I don't think the long tail of customers are ready to have AI running and just doing all the things. They wanna be in the driver's seat.
Speaker 208:34 - 09:10
对,这是个很好的问题。现在肯定还是 crawl 阶段,我觉得这其中一部分原因在于 Google 天生存在的产品张力。你们有——多少来着——超过 13,000,000,000 的用户产品,所以我其实认为,我们也有一些更像 labs(实验室)式的体验,在那里你可能已经更接近 run 或者 walk 了;但我觉得,今天大多数产品体验肯定还是更接近 crawl。我认为这就是我们在构建一款被大量人群使用的产品时所承担的 stewardship(审慎管理)责任。我不觉得长尾客户群体已经准备好让 AI 自己跑起来、把所有事情都做掉。他们还是希望自己坐在驾驶位上。
Speaker 209:10 - 09:28
They're cautiously taking the first step, and I think Google team And Search is maybe the most quintessential example of this. I think they have a lot of responsibility to actually do that in a way that it brings people along and doesn't just change everything of how they interact with the internet and the way they associate with products and stuff like that.
Speaker 209:10 - 09:28
他们正在谨慎地迈出第一步,我觉得 Google team And Search 也许是这方面最典型的例子。我认为他们确实承担着很大的责任,要以一种能把用户一起带上的方式来做这件事,而不是一下子把人们与互联网互动的方式、以及他们与产品建立联系的方式之类的一切都改变掉。
Speaker 109:28 - 09:31
Yeah. Which products do you think are closest to the Walk?
Speaker 109:28 - 09:31
嗯。你觉得哪些产品最接近 Walk?
Speaker 209:31 - 10:12
That's a good question. I think Gemini app is definitely closest to Walk. And so for Spark, I think having a 20 fourseven always agent, literally going and potentially doing a bunch of actions on your behalf, is definitely one of the frontier use cases, and I think you'll see I think antigravity is another one where it's like you could have autonomous coding agents rebuilding operating systems and doing billions of tokens and spending thousands of dollars on your behalf. And think those are, again, more And actually, they're in GDM as well as another angle of this, so I think GDM is taking very much a frontier look at this, where I think the rest of Google's products, I think, are more incrementally getting there, which, again, makes reasonable sense to me.
Speaker 209:31 - 10:12
这是个好问题。我觉得 Gemini app 肯定是最接近 Walk 的。至于 Spark,我认为那种 24/7 常驻的 agent(智能代理),真的会替你行动、并且可能代表你执行一大堆操作,绝对是最前沿的 use case(用例)之一;而且我觉得你还会看到——我认为 antigravity 也是另一个例子,在那里你可能会有 autonomous coding agents(自主编程代理)去重建 operating systems(操作系统),消耗数十亿 token(令牌),并代表你花掉数千美元。我觉得这些再次都更偏向——而且实际上,GDM 里也有这方面的另一个角度,所以我认为 GDM 对这件事采取的是一种非常前沿的视角;而我觉得 Google 的其他产品,则是在更渐进地往那个方向靠近,这一点在我看来同样是合理的。
Speaker 110:12 - 10:20
Yeah. Do you think that Google ends up with one, two, three product surfaces for using AI or thousands?
Speaker 110:12 - 10:20
嗯。你觉得 Google 最终会形成一套、两套、三套用于使用 AI 的产品界面,还是会有成千上万套?
Speaker 210:21 - 11:06
It's tough. I think a lot of this is actually baked in just how humans consume products, and my sense is that there's something nice about having this compartmentalization and this specialization of products where it becomes If you end up with a product that is doing everything for you, inherently, there's more work involved in using that version of the product. Think would be the default say, I think maybe somebody will spin together the truly magic experience that doesn't make that true, but I think the long tail of folks end up having to spend more mental energy and more time to actually get the general purpose product to do the thing that they actually want to do versus there's something nice about, I click my calendar app, it just shows me my calendar. Like, I don't need to worry and deal with anything else.
Speaker 210:21 - 11:06
这很难说。我觉得其中很大一部分其实已经写在了人类消费产品的方式里。我的感觉是,这种产品之间的区隔和专业化其实有某种好处:如果你最后得到的是一个什么都替你做的产品,那么从本质上说,使用那个版本的产品反而需要投入更多工作。我认为那会成为默认情况。也许未来会有人把真正神奇的体验整合出来,让事情不再如此;但我觉得,对绝大多数人来说,要让一个通用型产品真正做出他们想做的事,往往得花更多心智和更多时间。相较之下,有一种体验就很好:我点开 calendar app,它就只给我看我的日历。像这样,我不需要去担心或处理别的任何东西。
Speaker 111:06 - 11:14
This is my hot take for why slide decks have existed for so long of just like, you know, the thing, the piece of information, you want it to be exactly in the same place.
Speaker 111:06 - 11:14
这就是我对为什么 slide decks(幻灯片文稿)能存在这么久的一个 hot take(个人判断):因为你会希望那个东西、那条信息,始终精确地待在同一个位置。
Speaker 211:14 - 11:14
And
Speaker 211:14 - 11:14
还有,
Speaker 111:15 - 11:25
I think we as humans are just actually very used to that as The opposed to idea of a generative interface sounds so cool to me, but it's like, do our brains really Isn't that just more cognitive overhead for us?
Speaker 111:15 - 11:25
我觉得我们人类其实就是已经非常习惯这种方式了。相比之下,generative interface(生成式界面)这个概念听起来对我来说很酷,但问题在于:我们的大脑真的——那难道不是会给我们带来更多认知负担吗?
Speaker 211:25 - 11:51
It definitely is in certain cases. I think somebody needs to Again, there's a lot of incredibly smart people in the world, and so maybe somebody will find the experience that makes it feel more natural. But to me right now, maybe not 10,000 is the extreme version. I'm guessing it looks more like more products going after different And maybe the other answer is, I don't know what it looks like for Google. For the ecosystem, it looks like a lot more products, I think.
Speaker 211:25 - 11:51
在某些情况下,确实是这样。我觉得还是需要有人去——再次强调,世界上有很多极其聪明的人,所以也许会有人找到一种让这件事感觉更自然的体验。但对现在的我来说,也许不是 10,000 那种极端版本。我猜更像是会有更多产品去针对不同的——也许另一个答案是,我不知道这对 Google 来说会是什么样子。但对整个 ecosystem(生态)来说,我认为看起来会是更多得多的产品。
Speaker 211:51 - 12:03
And that's really exciting. I think how Google will end up strategically deciding, do our customers want to deal with us having 10,000 products, or would it be better to only have three? Will come down to a strategic decision for us.
Speaker 211:51 - 12:03
这确实非常令人兴奋。我想,Google 最终会如何从战略上做决定:我们的客户是想和一个拥有 10,000 个产品的我们打交道,还是只保留三个会更好?这最终会归结为我们的一个战略决策。
Speaker 112:03 - 12:17
That totally makes sense. When I talk to companies in the enterprise, they say, Everyone's talking about agentic AI, but the only place they've seen agents really working is coding agents. Do you agree or disagree with that take?
Speaker 112:03 - 12:17
这完全说得通。我和 enterprise(企业)里的公司交流时,他们会说,大家都在谈 agentic AI,但他们真正看到 agent(智能体)有效运作的唯一场景,就是 coding agent(编码智能体)。你同意这种看法,还是不同意?
Speaker 212:17 - 12:52
Yeah. I think it depends what your bar for working is, which I think is a lot of the nuance of this. Like, I think if truly trying to, like, offload very complicated tasks for domains in which the models haven't actually crossed the threshold of quality, then I think that's definitely true. It's not gonna solve the problem, but this is something that I wish we could measure. A good example is OpenRouter, for example, is measuring the total token consumption that's happening, and so you can sort of see these trends play out over time of how much more intelligence is in the world now versus a year ago.
Speaker 212:17 - 12:52
是的。我觉得这取决于你对“有效运作”的标准是什么,而这正是这件事里很微妙的一点。比如,我认为如果你真的想把非常复杂的任务卸载出去,而这些任务所属的领域里,model(模型)其实还没有跨过质量门槛,那么我觉得这种说法肯定是对的。它不会直接解决这个问题,但这是我很希望我们能衡量的一件事。一个很好的例子是,OpenRouter 现在就在衡量正在发生的 total token consumption(总 token 消耗量),所以你能某种程度上看到这些趋势是如何随着时间展开的:和一年前相比,如今世界上多了多少 intelligence(智能)。
Speaker 212:53 - 13:40
In parallel, the thing that I'm actually really interested to measure is how long is the average agent run or the average task actually taking place? I don't think it's something that they publish, but I feel like they probably have interesting data, I'm sure there's others. Because I do think you're seeing these new model capability lands or new model drop, and it's spiking up. And maybe the curve is still very low right now, but you're seeing those early signs of it spiking upward to long running tasks and all the model labs are talking about, We released this new model and it did three days of autonomous work, or whatever it is. That's the extreme, but I think, in practice, you're seeing that trickling up pretty quickly, which is really interesting.
Speaker 212:53 - 13:40
与此同时,我其实还很想衡量的一件事是:平均一次 agent run(智能体运行)或者平均一个任务,实际持续了多长时间。我不觉得这是他们会公开发布的数据,但我感觉他们大概有一些很有意思的数据,我也确定还有别的机构有。因为我确实认为,你会看到每次新的 model capability(模型能力)落地,或者新 model 发布时,这个指标都会突然上升。也许现在这条曲线整体仍然很低,但你已经能看到一些早期迹象:它正在向长时运行任务快速上扬,而所有 model lab(模型实验室)都在说,我们发布了这个新模型,它完成了三天的 autonomous work(自主工作),或者类似这样的事。这当然是极端案例,但我觉得在实践中,你会看到这种变化很快就会逐步渗透上来,这一点非常有意思。
Speaker 213:40 - 13:49
So even if the enterprises haven't felt it outside of coding, they are going to this year as sort of a bunch of those other use cases get much better as well.
Speaker 213:40 - 13:49
所以即便企业界目前除了 coding(编程)之外还没有明显感受到这一点,他们今年也会感受到,因为其他那一批 use case(用例)也会变得好得多。
Speaker 113:49 - 13:56
From the DeepMind perspective, do you think long horizon agents is a KPI that matters? Is it the KPI that matters?
Speaker 113:49 - 13:56
从 DeepMind 的角度看,你觉得 long horizon agents(长时程智能体)是一个重要的 KPI(关键绩效指标)吗?它是最重要的那个 KPI 吗?
Speaker 213:56 - 14:13
It definitely matters. I think for DeepMind, we're doing lots of things, which we can talk more about later. There's a huge portfolio of different bets that are taking place. Long running agents obviously matters a lot. I think also specifically coding agents in that matters a lot.
Speaker 213:56 - 14:13
它当然重要。我觉得对 DeepMind 来说,我们在做很多事情,之后可以再多聊。现在同时在进行的是一个非常庞大的、由不同押注组成的 portfolio(组合)。长时运行的 agent 显然非常重要。我也认为,特别是其中的 coding agent 也非常重要。
Speaker 214:13 - 14:22
It clearly is an accelerant of every other part of your business if you have a great coding model. And so making sure we have that, I think, is super top of mind.
Speaker 214:13 - 14:22
很显然,如果你拥有一个出色的 coding model(编码模型),它会成为你业务其他所有部分的 accelerant(加速器)。所以我认为,确保我们拥有这一点,是当前最重要的关注点之一。
Speaker 114:22 - 14:25
Got it. I'd love to shift gears a little bit and talk about coding.
Speaker 114:22 - 14:25
明白了。我想稍微换个话题,聊聊编程。
Speaker 214:25 - 14:26
Yeah.
Speaker 214:25 - 14:26
对。
Speaker 114:26 - 14:37
Okay. I'm gonna ask a hard question. A lot of my developer friends were using Claude for a long time. OpenAI saw that, declared code red. Codex is now really good.
Speaker 114:26 - 14:37
好。我来问个比较尖锐的问题。我的很多 developer(开发者)朋友很长一段时间里一直在用 Claude。OpenAI 看到了这一点,直接拉响了 code red(红色警报)。现在 Codex 真的非常强。
Speaker 114:37 - 14:49
I would say my friends are maybe split fifty fifty now in using Claude and using Codex. I don't hear a ton of them using Gemini, which has always kind of puzzled me. What's going on with that?
Speaker 114:37 - 14:49
我会说,我这些朋友现在在用 Claude 和用 Codex 之间,大概已经五五开了。我倒是没怎么听说他们很多人在用 Gemini,这一点我一直有点困惑。这里面到底发生了什么?
Speaker 214:49 - 15:13
Yeah. It's a great question. I think there's one there's one part of the story that I'll add, which is which which is makes it even more interesting, which is, December, the narrative was that Google had won. And when we landed Gemini three, I think it was such a profound improvement from a model capability perspective. I think a lot of the narrative was like, Google has taken a huge leap forward and made that happen.
Speaker 214:49 - 15:13
是啊,这是个很好的问题。我觉得这里还有一部分背景可以补充,而且这会让整件事更有意思。那就是在 12 月的时候,外界的叙事是 Google 已经赢了。等到我们推出 Gemini three 的时候,我认为从 model(模型)能力的角度看,那是一次非常深刻的提升。我觉得当时很多人的说法都是,Google 实现了一次巨大的跃进,而且真的做成了。
Speaker 215:13 - 15:34
And I think what was interesting to see as an ecosystem participant is not how quickly that narrative shifted, but just the next wind of the narrative obviously was all the agentic coding stuff that happened over the holidays and then into January and beyond, and that was not that long ago. So it is a
Speaker 215:13 - 15:34
而我觉得,作为 ecosystem(生态系统)里的参与者,一个有意思的观察并不只是这个叙事转向得有多快,而是紧接着到来的下一波叙事,很明显就是假期期间发生、然后延续到 1 月及之后的那一整波 agentic coding(智能体式编程)热潮,而那其实也就是不久之前的事。所以这真的是一种——
Speaker 115:34 - 15:36
I feel like we've been in warp speed ever since.
Speaker 115:34 - 15:36
我感觉从那以后,我们就一直处在 warp speed(曲速)状态。
Speaker 215:36 - 16:13
Yeah, for sure, but it's a matter of reminder of just how fast things can change. I think the observation is not unreasonable. I do think what's happening behind the scenes for us is trying to push the frontier as fast as possible on coding, and so I think antigravity actually is an important part of that. I think one of the takeaways is that it's actually really hard to make a great coding model for this developer use case of really long running, sweet work if you don't actually have a product that does that, and so I think Google realized that. That's why the Windsurf deal happened.
Speaker 215:36 - 16:13
对,绝对是这样,但这也再次提醒我们,事情变化的速度可以有多快。我觉得这个观察并不算不合理。我确实认为,我们在幕后所做的事情,是尽可能快地把 coding(编程)的前沿往前推,所以我觉得 Anti Gravity 实际上是其中一个重要组成部分。我想其中一个结论是:如果你自己没有一个真正做这件事的产品,那么要为 developer(开发者)这种需要超长时间持续运行、完成成套工作的 use case(使用场景)做出一个很棒的 coding model(编程模型),其实是非常难的。我想 Google 也意识到了这一点。这就是 Windsurf deal 发生的原因。
Speaker 216:13 - 17:13
It's why those folks came over and then ultimately built Anti Gravity, and we've been using internally, actually, and Sundar showed this at IO, just the graph of growth of token consumption inside of Google. So you need that engine to spin, and the meta comment, again, is the engine is spinning. It takes time in order to actually make model progress, but I'm super confident. I think the group of folks who we have working on code is like I describe it as the avengers of AI internally, and so it really is some of the best people inside of Google trying to push the rock up the hill on this stuff and taking it super seriously and trying to push, and I think three flash, notwithstanding some of the conversation about the price and stuff like that, is sort of a step towards actually starting to bring a lot of these capabilities and the fruits of that labor paying off. It's a flash model that's better than any pro model we've ever released from a coding standpoint, and the pro models were really good before.
Speaker 216:13 - 17:13
这也是为什么那些人加入了我们,并最终做出了 Anti Gravity。实际上我们一直在内部使用它,Sundar 也在 IO 上展示过这一点——Google 内部 token(令牌)消耗增长的那张曲线图。所以你需要让那台 engine(引擎)真正转起来。而更宏观的一点评论,还是那句话:这台引擎已经转起来了。模型要真正取得进展是需要时间的,但我非常有信心。我觉得我们现在做 code(代码)这一块的这群人,我在内部会把他们形容成 AI 版的 Avengers,所以这真的是 Google 内部最优秀的一批人在非常严肃地推进这件事,努力把这块巨石往山上推。至于 three flash,先不谈一些关于价格之类的讨论,我认为它某种程度上就是一个信号:我们开始把这些能力以及这些努力的成果真正带出来了。作为一个 flash model(Flash 模型),从 coding 的角度看,它比我们过去发布过的任何 pro model(Pro 模型)都更强,而我们以前的 pro models 本来就已经很不错了。
Speaker 217:13 - 17:24
So there's another thread of this also, which is everyone forgets that there's pre training windows. I wonder, somebody should track this online, which would be interesting to see.
Speaker 217:13 - 17:24
这里还有另一条线索,大家也都忘了还有 pre training(预训练)窗口这回事。我在想,应该有人把这个在线追踪起来,那样看着会很有意思。
Speaker 117:24 - 17:27
Meaning the big run, what clusters have been available and
Speaker 117:24 - 17:27
也就是说,在大规模 run(训练运行)方面,哪些 clusters(集群)是可用的,以及
Speaker 217:27 - 18:00
what Exactly. The big runs are an interesting thread of this, and so it might look from an external perspective that, Oh, you're super behind in some way, and actually, you miss all the context of where the big runs are and where the large pre training runs are. So I think that also Obviously, pre training has historically been a massive strength for DeepMind. We have some of the best people in the world, I'm so excited to see the fruits of that labor and everything else that's happened. 3.5 flash was all post training gains, which is really cool.
Speaker 217:27 - 18:00
没错,就是这个。这些大规模 run 本身就是这件事里一个很有意思的线索。所以从外部视角看,可能会觉得“哦,你们在某些方面已经严重落后了”,但实际上,你忽略了一个非常重要的背景:大规模 run 在哪里进行,大规模 pre training run(预训练运行)又在哪里进行。所以我认为这一点也很关键。显然,pre training 历来一直是 DeepMind 的一个巨大强项。我们拥有世界上最优秀的一批人才,我非常期待看到这些努力结出的成果,以及其他所有已经发生的进展。3.5 flash 基本上全都是 post training(后训练)带来的提升,这一点非常酷。
Speaker 218:01 - 18:11
So a huge testament to the work that that team did to actually make the level of gains and surpass the previous pro model literally just with post training, which is awesome.
Speaker 218:01 - 18:11
所以这也极大证明了那个团队所做工作的价值:他们真的把性能提升到了那个层级,并且几乎完全只是靠 post training,就超过了之前的 pro model,这非常棒。
Speaker 118:11 - 18:23
How religious are you all about dogfooding internally? For example, are DeepMind folks still allowed to use other models, or is it like you guys are using the Gemini harness now, we have to make this really, really good?
Speaker 118:11 - 18:23
你们内部对 dogfooding(内部试用自家产品)这件事到底有多坚持?比如说,DeepMind 的人现在还可以用其他 model 吗,还是说你们现在都在用 Gemini harness,所以必须把它做得特别特别好?
Speaker 218:23 - 18:38
Yeah. I think people, it's so healthy to be using other models just because it's sometimes hard to actually grok what's happening in the ecosystem if you're not. So I use all the models. I use all the products. I think folks across the rest of DeepMind are doing the same thing.
Speaker 218:23 - 18:38
对。我觉得使用其他 model 是很健康的,因为如果你不这么做,有时候就很难真正 grok(真正理解)这个 ecosystem(生态)里在发生什么。所以我会用所有 model,也会用所有产品。我认为 DeepMind 其他很多人也都在这么做。
Speaker 218:38 - 19:11
You definitely have to use the Gemini models, though. It's just great from a feedback flywheel perspective, and it's part of how they get better. DeepMind has, and Google more broadly, has 100,000 plus incredible engineers who are using the models and giving feedback, and it should be a competitive advantage for Google because we have that scale of engineering resources and the depth of the talent and can run AB tests and live experiments and all that stuff. So I think you have to use all the models, but I think for the majority of folks, it's like Gemini as the daily driver, which is great.
Speaker 218:38 - 19:11
不过,你当然还是得用 Gemini model。这从 feedback flywheel(反馈飞轮)的角度看非常重要,而且这本来就是它们持续变好的方式之一。DeepMind,以及更广义上的 Google,拥有超过 100,000 名非常出色的 engineers(工程师)在使用这些 model 并提供反馈;这理应成为 Google 的竞争优势,因为我们有这种规模的工程资源,也有足够深的人才储备,还能做 AB tests(A/B 测试)、live experiments(在线实验)以及所有这些事情。所以我认为所有 model 都得用,但对大多数人来说,Gemini 更像是 daily driver(日常主力工具),而这很好。
Speaker 119:11 - 19:23
Do you believe in this narrative around a soft takeoff of once you have a good enough agentic coding model, then it accelerates the pace of research progress, and it's a self reinforcing cycle.
Speaker 119:11 - 19:23
你是否相信这样一种叙事:一旦你拥有了足够好的 agentic coding model(具备 agent 能力的编程模型),它就会加速研究进展的速度,并形成一个自我强化的循环。
Speaker 219:23 - 19:30
It seems obvious that that's true, but maybe I'm drinking too much Kool Aid that that's the case.
Speaker 219:23 - 19:30
这听起来显而易见是真的,但也许是我 Kool Aid 喝太多了,才会觉得情况就是如此。
Speaker 119:30 - 19:31
Are you seeing the signs of it yet?
Speaker 119:30 - 19:31
你已经开始看到这方面的迹象了吗?
Speaker 219:31 - 20:11
Yeah. You definitely see some signs of this. I think the signs that are still early is doing this from a model perspective, and I think part of the context of that is the resource allocation for some of these larger training runs is just significant, and so you definitely still have a human in the driver's seat of making those decisions because you're not gonna accidentally take 10,000 TPUs to go kick off some job that actually doesn't make that much sense, but from a product perspective, you for sure see it. I think we're seeing this on our team. We've built mobile apps using anti gravity, and we'll launch them to the world faster than I think any team at Google has ever built a mobile app.
Speaker 219:31 - 20:11
看到了。你肯定能看到一些这方面的迹象。我觉得目前仍然偏早期的迹象,是从 model(模型)这个层面来做这件事;而我认为这里的一部分背景是,一些更大规模的训练 run(训练运行)的资源分配实在非常可观,所以在做这些决策时,驾驶位上肯定还是有人类——因为你不可能一不小心就调动 10,000 个 TPU 去启动一个其实并没有太大意义的任务。但从 product(产品)角度看,你绝对已经能看到这种变化了。我觉得我们团队就在经历这个。我们用 anti gravity 构建了 mobile app(移动应用),而且把它们推向全球的速度,比我认为 Google 任何团队过去做出一款移动应用的速度都更快。
Speaker 220:11 - 20:25
Josh's team did this with the Gemini macOS app and sort of like end to end delivered an app sort of faster than any team had ever delivered a Mac app at Google, and it's because of agentic coding, and so it's great from a product perspective.
Speaker 220:11 - 20:25
Josh 的团队也是这样做 Gemini macOS app 的,他们算是端到端地交付了这款 app(应用),速度几乎比 Google 任何团队过去交付一款 Mac app 都更快,而这就是因为 agentic coding(agent 驱动编程),所以从产品角度看,这非常棒。
Speaker 120:25 - 20:36
I think you've said in the past that if you could have a system that could build anything with code, humans can't compete on the same level, and that's narrow superintelligence. Do you think we've reached that point?
Speaker 120:25 - 20:36
我记得你以前说过,如果有一个系统能够用代码构建任何东西,那人类就无法在同一层级上竞争了,而那就是 narrow superintelligence(狭义超级智能)。你觉得我们已经到那个点了吗?
Speaker 220:37 - 21:02
-It is interesting. I think this narrow superintelligence example is interesting to see. Obviously, it kind of feels that way for coding right now, where coding is just so good that it does kind of feel like narrow superintelligence. I don't know. It depends how you actually end up the details of quantifying this, but I think the important thing is, to your point earlier, it works incredibly well for code.
Speaker 220:37 - 21:02
这很有意思。我觉得这个 narrow superintelligence(狭义超级智能)的例子很值得观察。显然,就 coding(编程)而言,现在确实有那种感觉——它的编程能力已经强到,确实会让人感觉像是 narrow superintelligence。我也不确定,这最终还是取决于你怎么把这些细节真正量化;但我认为重要的是,正如你之前说的,它在代码这件事上效果好得惊人。
Speaker 221:03 - 22:01
And so it would be great if it did a bunch of other things, but it's actually just so impactful that it can be great at code. And so I spend a lot of time just letting that fact sort of just wash over me because I think it's Obviously, building AGI is super important and very interesting, but building AGI, if it sort of takes away from the story of the current present capability of the technology, I think is actually kind of a bad sort of trade off, and so I'm trying to always hold these two things in my head equal at the same time, which is we need to build general purpose technology, but obviously, it's so impactful to have this thing. And it feels like it hasn't taken away sort of It's been one of the best positive outcomes is that I feel like it hasn't taken away from human developers. It really does feel like an accelerant of what human I, as a human developer, feel like I have more agency in the world. Feel like I can tackle This is my personal experience.
Speaker 221:03 - 22:01
所以,如果它还能做好很多别的事情,那当然很棒;但实际上,仅仅是它能把代码这件事做到这么好,就已经有巨大的影响了。因此我会花很多时间让这个事实慢慢沉淀在自己心里,因为我觉得,显然,构建 AGI(通用人工智能)极其重要,也非常有趣;但如果构建 AGI 这件事,某种程度上掩盖了这项技术当下现实能力的故事,我觉得那其实是一种不太好的取舍。所以我一直在努力同时、同等地把这两件事放在脑中:我们需要构建通用目的的技术,但显然,拥有现在这个东西本身就已经影响巨大。而且感觉它并没有削弱人类开发者——这反而是最好的积极结果之一。我真的觉得,它更像是对人类开发者能力的加速器。作为一个人类开发者,我感觉自己在这个世界上拥有了更多 agency(行动能力/自主性)。我感觉自己可以去处理……这是我个人的体验。
Speaker 222:01 - 22:41
I feel like I can tackle more ambitious problems. I feel like I used to kick around ideas, they were slightly out of reach, and I would just be like, Ah, wouldn't it be nice? And now I have the opposite problem, which is I'm kicking around an idea, and I'm like, I could probably make this even more ambitious. It adds a different layer of sort of responsibility or a different layer of burden, actually, because I'm like, Oh, I can't just do the sort of MVP of this. I actually need to go 10 steps further because the technology enables me, and resetting my level of ambition, I think, is something that I've also spent a bunch of time thinking about.
Speaker 222:01 - 22:41
我觉得自己可以去解决更有雄心的问题。以前我也会反复琢磨一些想法,但它们总是稍微有点遥不可及,我就会想:啊,要是能做到就好了。现在我反而遇到了相反的问题:我在琢磨一个想法时,会觉得,我大概还能把它做得更有雄心一些。这其实带来了一层不同的责任感,或者说一种不同的负担,因为我会想,哦,我不能只是做一个 MVP(最小可行产品)版本,我实际上得再往前走 10 步,因为技术已经让我做得到。而我觉得,重设自己的雄心水平,也是我花了很多时间思考的一件事。
Speaker 222:41 - 22:58
But I think that will happen in these vertical superintelligence domains, which will be interesting, and it feels like we're gonna get a bunch of those before we've solved like, it's almost like jagged superintelligence, I think, is what we'll end up with.
Speaker 222:41 - 22:58
但我认为这会先发生在这些垂直领域的 superintelligence(超级智能)里,那会很有意思,而且感觉在我们解决某种——几乎可以说是 jagged superintelligence(锯齿状超级智能)——之前,我们会先得到一批这样的东西;我想,最后我们得到的大概就是这种形态。
Speaker 122:58 - 23:01
What verticals do you think we'll get superintelligence at next?
Speaker 122:58 - 23:01
你觉得接下来哪些垂直领域会先出现 superintelligence(超级智能)?
Speaker 223:01 - 23:43
That's a great question. I do spend a lot of my time too much time probably thinking about coding these days, so I'll think for a second of, like, the other domains. I think part of this is things that have better verifiability, obviously, are the ones where you'll see the gains happen more quickly, so things with math and finance. Actually, science could be a really interesting one. It would be fascinating to see some of these domains where there's some level of verifiability actually really start to take off, which would be cool, and I also think an important thing in this broader narrative about just what impact AI is having on the world.
Speaker 223:01 - 23:43
这是个很好的问题。说实话,我最近花了很多时间——可能是太多时间——在想 coding,所以我得稍微想一想其他领域。我觉得其中一部分原因在于,那些更容易 verifiability(可验证)的事情,显然会是你更快看到进展的地方,所以像 math 和 finance 这类领域。其实,science 也可能会是一个非常有意思的方向。看到这些具有一定 verifiability 的领域真正开始起飞,会非常令人着迷;那会很酷。而且我也觉得,就 AI 对世界正在产生什么影响这个更大的叙事而言,这也是件很重要的事。
Speaker 223:43 - 24:08
You almost want that to be the case. In the sequencing of things that work, you want a lot of these really, really good, impactful, positive things for the world to happen as early on as humanly possible so that folks understand what the potential positive impact of the technology is. So I think science could be a really interesting one. Obviously, there's all this stuff happening right now with math proofs and stuff like that, which I'm not a mathematician, so it's somewhat over my head, but
Speaker 223:43 - 24:08
你几乎会希望事情就是这样发展。在那些真正起作用的事物的推进顺序里,你会希望很多这种非常、非常好、影响深远、对世界有积极意义的事情,能尽可能早地发生,这样人们就会明白这项技术潜在的正面影响是什么。所以我觉得 science 可能会是个很有意思的方向。显然,现在还有很多关于 math proofs 之类的事情正在发生;我不是 mathematician,所以这里面有些内容我有点跟不上,不过——
Speaker 124:08 - 24:12
I saw a great tweet the other day. Why did Aerodose have so many problems?
Speaker 124:08 - 24:12
我前几天看到一条很棒的 tweet。为什么 Aerodose 会遇到那么多问题?
Speaker 224:12 - 24:16
Exactly. That's a good one. I like that. That's a good like T shirt.
Speaker 224:12 - 24:16
对,没错。这句很好。我喜欢。很适合印在 T shirt 上。
Speaker 124:17 - 24:25
So funny. Okay. I but sweet speaking of Twitter, I went through your Twitter before this, so I'm gonna read back another tweet at you. The the good thing about Twitter is there's a public record of all your predictions.
Speaker 124:17 - 24:25
太好笑了。好吧。不过,说到 Twitter,我在这之前翻了你的 Twitter,所以我打算再念一条你以前发的 tweet 给你听。Twitter 的好处就在于,那里有你所有预测的公开记录。
Speaker 224:25 - 24:29
So need to turn on that auto tweet deleting feature or whatever it is.
Speaker 224:25 - 24:29
所以我得把那个自动删除 tweet 的功能之类的东西打开了。
Speaker 124:29 - 24:38
Last October, you tweeted, everyone is going to be able to vibe code video games by the end of twenty twenty five. Yeah. Did that end up being true?
Speaker 124:29 - 24:38
去年 10 月,你发推说,到 2025 年年底,所有人都将能够用 vibe code 来做视频游戏。对。那么这最后成真了吗?
Speaker 224:38 - 25:01
It feels close. And I think there's I mean, obviously not AAA games. Like, you're not building, you know, the next Call of Duty or GTA yet. But I think it's it feels closer than it's ever been. And I think a lot actually, a lot of the interesting bit about video games is you actually need to end up building a lot of this, like, other stuff.
Speaker 224:38 - 25:01
感觉已经很接近了。而且我觉得——我的意思是,显然不是 AAA 游戏。比如说,你现在还做不出下一个 Call of Duty 或 GTA。但我觉得,这种感觉比以往任何时候都更接近了。而且我认为,实际上,视频游戏里很多有意思的部分在于,你最终还需要构建很多这类其他东西。
Speaker 225:01 - 25:44
Models, and we were talking off camera before this, Three. Js is a great example of this. Like, Three. Js makes a lot of things possible that weren't before, but there's still all these rough edges that just a coding agent doesn't solve, so you need sprite generation, and the models aren't very good at doing that natively, and so you need some orchestration layer and tooling in order to make that happen. There's a bunch of other things like that that are core to the gaming video game experience that need to have a high degree of reliability that I think it feels like it's within reach, but actually requires a lot of product scaffolding work in order to create experiences that are reusable and replayable and sort of have the level of depth and requires a little bit of taste in there.
Speaker 225:01 - 25:44
模型这件事——我们刚才在镜头外也聊到过——Three. Js 就是一个很好的例子。比如,Three. Js 让很多以前做不到的事情成为可能,但仍然存在很多粗糙边角,这不是一个 coding agent 就能解决的;所以你还需要 sprite generation,而模型原生做这个并不太擅长,因此你需要某种 orchestration layer(编排层)和 tooling(工具链)来让这件事发生。还有很多类似的东西,对游戏这种 video game 体验来说都很核心,而且需要很高程度的可靠性。我觉得这些已经触手可及了,但实际上仍然需要大量的产品 scaffolding(脚手架式支撑)工作,才能创造出可复用、可重玩,并且具有一定深度的体验,而这其中也还需要一点审美判断。
Speaker 125:45 - 25:50
Do you see people making a lot of video games inside AI Studio and the other developer surfaces that you have?
Speaker 125:45 - 25:50
你有没有看到人们在 AI Studio 和你们提供的其他开发者界面里制作很多视频游戏?
Speaker 225:50 - 26:01
Yeah. And so this was actually based on us looking at the early data, and there was something like, in AI Studio at the time, it was like 20% of all apps that folks were making were actually games. People were trying to build games. Lot them
Speaker 225:50 - 26:01
有的。所以这其实是基于我们查看早期数据得出的。当时在 AI Studio 里,大概有类似 20% 的应用,都是大家实际在做的游戏。人们当时都在尝试做游戏,很多很多。
Speaker 126:01 - 26:02
Is that the most popular category?
Speaker 126:01 - 26:02
那是最受欢迎的类别吗?
Speaker 226:02 - 26:11
It's not the most popular category anymore, just because I think the ecosystem has shifted and the user base has shifted, but it is a lot of games.
Speaker 226:02 - 26:11
现在它已经不是最受欢迎的类别了,我觉得主要是因为生态系统变了,用户群也变了,但游戏的数量还是很多。
Speaker 126:11 - 26:13
What is the most popular category?
Speaker 126:11 - 26:13
那最受欢迎的类别是什么?
Speaker 226:13 - 26:18
I think it's like 20% finance related stuff, 20%
Speaker 226:13 - 26:18
我觉得大概有 20% 是和金融相关的东西,20%。
Speaker 126:17 - 26:19
Wow, people like counting their money that much.
Speaker 126:17 - 26:19
哇,人们这么喜欢数自己的钱啊。
Speaker 226:19 - 27:07
I think it's something around crypto, actually, I think is what people are doing a lot of stuff with finance, a lot of personal productivity things and a lot of gen media stuff, actually, because obviously the Google suite of gen media stuff has done a great job. But I also think GDM has sort of like a Obviously, Demis cares a ton about games and sort of started his career in doing AI stuff because of games, and so I think we'll have some interesting swings at this. Our team actually in Kaggle, which is sort of a bunch of the AI benchmarking stuff we do in GDM, sort of works with GDM to build this game arena, which is sort of our way of testing progress towards AGI, using games as a proxy, which, again, is very deeply rooted in GDM's history.
Speaker 226:19 - 27:07
我觉得其实有不少是围绕 crypto 的,我想人们确实在做很多金融相关的东西、很多个人效率相关的东西,以及很多 gen media 相关的东西,因为很显然,Google 那套 gen media 产品做得非常好。但我也觉得 GDM 有点像——很明显,Demis 非常关心游戏,而且他职业生涯一开始做 AI 相关的事情,本来就和游戏有关,所以我觉得我们会在这方面做一些有意思的尝试。我们在 Kaggle 的团队——某种程度上就是我们在 GDM 里做的一批 AI 基准测试相关工作——实际上也在和 GDM 一起搭这个 game arena,这算是我们用来测试朝着 AGI 进展的一种方式:把游戏当作一种代理指标,而这同样也深深植根于 GDM 的历史。
Speaker 127:07 - 27:15
How close do you think we are to Rando off the street, what a good idea, can vibe code a really fun, playable game?
Speaker 127:07 - 27:15
你觉得我们离这样一个状态还有多远:街上随便一个普通人,冒出一个不错的点子,就能用 vibe code 做出一个真正好玩、能玩的游戏?
Speaker 227:15 - 28:00
I wanna say this year Actually, I the model capability makes it I think this is where I've gotten excited on the product side, and again, we were also talking off camera about sort of the startups in this ecosystem because it feels like it's possible. It doesn't feel like there's a gap in model quality. It feels like there's a gap in someone who knows what it takes to build a great game actually putting the scaffolding together in the right way to make that possible. I think there are folks who are doing this right now, and so some of it is a discoverability and awareness thing that people just don't even know that they can do that, And some of it is just maybe certain categories of model capabilities are just slightly off and we're weeks or months away from that chasm being crossed and then it just working for most people.
Speaker 227:15 - 28:00
我想说就是今年。其实我觉得,model capability(模型能力)已经让这件事变得可行了。我认为这也是我在产品侧开始兴奋起来的地方,而且我们刚才在镜头外也聊到这个生态里的 startups,因为这感觉上已经可能了。现在不像是模型质量还有明显缺口,更像是还缺一个真正知道做好游戏需要什么的人,把必要的 scaffolding(脚手架式支撑结构)以正确方式搭起来,让这件事成为现实。我觉得现在已经有人在这么做了,所以一部分问题在于 discoverability 和 awareness——很多人甚至还不知道自己已经能这么做;另一部分则可能只是某些类别的模型能力还稍微差一点,我们距离跨过那道鸿沟也许只差几周或几个月,一旦跨过去,这件事对大多数人来说就会直接可用了。
Speaker 128:00 - 28:17
And so this is a good segue into, I wanna ask about world models next, but do you think vibe coded video games is more likely going to be game engine plus coding agents based, or do you think it's more likely to be world model based?
Speaker 128:00 - 28:17
这正好可以自然过渡到下一个问题。我接下来想问 world models,不过你觉得,用 vibe code 来做视频游戏,更可能会是基于 game engine 加 coding agents 的路线,还是更可能会是基于 world model 的路线?
Speaker 228:17 - 29:16
Yeah. I think the what will end up happening is the definition of world models will blur, which we should which we should talk about with Omni. And it will still I think the coding agent will look like some sort of world model type system, but you actually do need To make world models useful for real things, you need scaffolding. And so I think, again, there's actually a bunch of interesting startups doing work, figuring out what is the scaffolding for world models so that you can take them from these very open ended inherent design of world models, very open ended spaces, and do it in a tangible way so that it's grounded in a use case that you could use in a reoccurring way. Could be somebody maybe will figure out the scaffolding for world models to make games possible, but the inherent nature of world models right now, I think, make it so that it's actually not well suited for games in the current form.
Speaker 228:17 - 29:16
对。我觉得最终会发生的是,world models 的定义会变得模糊起来,这点我们应该结合 Omni 来聊。我认为 coding agent 最终看起来仍会像某种 world model 类型的系统,但实际上,如果你想让 world models 对真实任务有用,你是需要 scaffolding 的。所以我觉得,同样地,现在其实已经有一批很有意思的 startups 在做这方面的工作:他们在摸索,world models 的 scaffolding 到底应该是什么,才能把它们从这种天生非常开放、设计上本就非常开放的空间里,真正落到一个可操作的形式上,让它锚定在某个具体 use case 中,并且能够被反复使用。也许会有人最终找出让 world models 能够用于做游戏的那套 scaffolding,但就 world models 目前的内在性质而言,我觉得它们以现在这种形式其实并不太适合做游戏。
Speaker 229:16 - 29:28
But the progress has been crazy, so who knows? Maybe in two years, the versions will be able to, but at least in the short term, it's like coding agent plus some sort of game engine, I think, is where you'll see way more alpha from a game's perspective.
Speaker 229:16 - 29:28
但进展确实快得惊人,所以谁知道呢?也许两年后,那些版本就能做到了;不过至少在短期内,我觉得从游戏的角度看,真正更可能出现大量 alpha(超额优势/领先结果)的,还是 coding agent 加上某种 game engine 这条路。
Speaker 129:28 - 29:33
That makes sense. Okay, so you said the definitions of world models are blurry. Can we unpack that?
Speaker 129:28 - 29:33
这很有道理。好,所以你说 world models(世界模型)的定义很模糊。我们能展开讲讲吗?
Speaker 229:33 - 30:09
Yeah. I think Omni is an example of this. We launched this at IO. You can sort of take in any input, create any output, and I think Demis sort of framed it to the world, rightfully so, as a world model because of just the level of understanding that it has of the world. I think that technically looks different than And I'm not an architecture expert on the way that we've done world models before, but it is different from an architectural standpoint than what's happened in the past, which I think is positive because it's getting closer to some of the ways in which it might actually be more scalable.
Speaker 229:33 - 30:09
对。我觉得 Omni 就是一个例子。我们在 IO 上发布了它。它基本上可以接收任何输入、生成任何输出,我觉得 Demis 也算是恰如其分地把它向外界定义成一个 world model,因为它对世界的理解程度确实很高。我想从技术上看,这和以前的不太一样——我不是 architecture(架构)方面的专家,不太清楚我们以前做 world models 的具体方式——但从架构角度来说,它确实不同于过去的做法。我觉得这是积极的,因为它更接近一些可能真正更具可扩展性的方向。
Speaker 230:09 - 30:17
I see. Historically, it's been super not scalable. It's very, very expensive to run traditional online world models.
Speaker 230:09 - 30:17
明白。从历史上看,这类东西一直都非常不可扩展。运行传统的在线 world models 成本非常、非常高。
Speaker 130:17 - 30:36
Yeah, like Genie being like a traditional so if you think of traditional world models as being like an action conditioned video model almost, then right now, when we say world model, what we actually mean is a model that has some understanding of the world as opposed to being strictly technically a action conditioned video model.
Speaker 130:17 - 30:36
对,比如说 Genie 就更像是一种传统形式——如果你把传统的 world models 理解为一种由 action(动作)条件控制的视频模型的话,那么现在我们说 world model,实际指的是一个对世界有某种理解的模型,而不再是严格技术意义上的 action conditioned video model(动作条件视频模型)。
Speaker 230:36 - 31:10
Yeah, and so the interesting thing, though, is it has understanding of the world, but then it also has that really great, and that's where the line is blurry to me, where it's like it can do a lot of those same use cases. It's not real time right now, but it can do a lot of those same use cases that you would describe or visually could create with that same exact world model, which I think is what's most interesting to me. So I do feel like this world model, video model thing is gonna change and play out in a different way than was obvious before.
Speaker 230:36 - 31:10
对,不过有意思的是,它不仅对世界有理解,而且它还具备那种非常强的能力——这也是为什么我觉得界线很模糊——因为它可以完成很多同样的 use cases(用例)。它现在还不是 real time(实时)的,但它确实能完成很多那种你原本会用同一个 world model 来描述、或者在视觉上生成出来的用例。我觉得这才是最有意思的地方。所以我确实感觉,world model 和 video model 这件事,接下来会以一种和之前明显不同的方式发生变化并逐步展开。
Speaker 131:10 - 31:17
And how does it work under the hood, like, whatever you're able to share? Is it Gemini plus video models? Is it something different entirely?
Speaker 131:10 - 31:17
那它底层是怎么工作的?在你能分享的范围内说说。它是 Gemini 加上 video models,还是完全不同的东西?
Speaker 231:17 - 31:36
It is a single model, which I think is the important part. Was actually part of the original desire was, like, you were training, like, eight different models to do all of those things historically. It's like you have a text model with the baseline Gemini model. You have audio. You have music models with Lyria.
Speaker 231:17 - 31:36
它是一个单一模型,我觉得这点很关键。最初的动机之一其实就是,以前为了做成所有这些事情,你往往要训练大概八个不同的模型。比如你会有一个文本模型,也就是基础版 Gemini model。你会有 audio(音频)模型。你还会有像 Lyria 那样的 music models(音乐模型)。
Speaker 231:36 - 32:11
You have Nano Banana. You have VO video models. You have a we have a whole suite of audio models, and, like, it would be great for us, our customers, if you just had a single model to do all those things. So it is like a new setup that sort of makes that possible. It's not like routing to a bunch of different models, which you could have imagined we could have done something like that actually before and done a Gemini omni model, but this is a true omni model, and it's starting with, like, the use case that works the best right now, which is the why it's the one that's available, is this, like, video editing capability.
Speaker 231:36 - 32:11
你还有 Nano Banana。你还有 VO video models。还有——我们其实有整套 audio models,而且无论是对我们自己还是对客户来说,如果只用一个模型就能完成所有这些事,那会非常好。所以这确实是一种新的 setup(配置/架构),使这件事成为可能。它并不是把请求路由到一堆不同模型上;你也可以想象,其实我们以前本来是可以做类似那种方案的,做一个 Gemini omni model,但这个才是真正的 omni model。它现在是从当前效果最好的 use case 开始的,也就是为什么目前对外可用的是这种 video editing(视频编辑)能力。
Speaker 232:12 - 32:35
Technically, it's, like, functional with the other things. It's just, like, the quality isn't isn't, like, perfect and is not state of the art, so we haven't rolled that out yet. It's also just the first crank of the model turn on Omni. It's the Omni Flash model, the first iteration, and so we'll have much, much more capable, powerful versions, which will be exciting to see. So
Speaker 232:12 - 32:35
从技术上说,它算是能和其他东西配合工作的。只是质量还不算完美,也还不是 state of the art(当前最先进水平),所以我们还没有把它正式推出。这也只是模型在 Omni 上启动后的第一次尝试而已。它是 Omni Flash model 的第一版,所以之后我们还会有能力强得多、功能强大得多的版本,这会很让人期待。所以——
Speaker 132:35 - 32:37
we could edit this set so it looks like we're
Speaker 132:35 - 32:37
我们可以把这个场景编辑一下,让它看起来像我们——
Speaker 232:37 - 32:41
in Yeah. Here. Yeah. I want this. Again, we were talking off camera.
Speaker 232:37 - 32:41
在——对,这里。对,我想要这个。还是那句话,我们刚才在镜头外也聊到过这个。
Speaker 232:41 - 33:14
We should do that for the intro because I think it just makes all this stuff more capable. And I've seen these examples of such subtle nuance that make me appreciate that it's like the world understanding playing out. I was giving a talk and was on stage with my friend Tulsi, who leads the model team, who I don't know if you've ever had on before, but she's amazing. I love Tulsi. I had mentioned to someone in the crowd to edit the video, and they literally took the picture, edited it with Omni in real time, and this dog came on the stage.
Speaker 232:41 - 33:14
我们应该把这个用在开场里,因为我觉得它会让所有这些东西变得更有能力。我还看过一些例子,里面那种非常细微的 nuance(细腻差别)让我意识到,这就像是 world understanding(世界理解)在真实展现出来。当时我在做一个演讲,和我的朋友 Tulsi 一起在台上,她负责 model team,我不知道你以前有没有采访过她,但她非常厉害。我很喜欢 Tulsi。我当时提到,让台下某个人去编辑那段视频,然后他们真的就把那张画面拿去,用 Omni 实时编辑,结果一只狗就出现在了台上。
Speaker 233:15 - 33:23
In the edited version, the other guests look down and see the dog. They chuckle a little bit. This is while I'm opining about whatever AI it -They even laugh
Speaker 233:15 - 33:23
在编辑后的版本里,其他嘉宾低头看到了那只狗,还轻轻笑了笑。而这时我还在台上高谈阔论,不管是在讲什么 AI——他们甚至还会笑——
Speaker 133:23 - 33:24
about your jokes.
Speaker 133:23 - 33:24
笑你的笑话。
Speaker 233:24 - 33:31
-Yeah, it was not my jokes. They laugh at the dog coming up. It jumps onto my lap. I sort of acknowledge the dog. I keep talking.
Speaker 233:24 - 33:31
——对,但不是因为我的笑话。他们是在笑那只狗走上来。它跳到我的腿上,我稍微回应了一下那只狗,然后继续讲话。
Speaker 233:31 - 33:49
I'm petting it or whatever, And just like, there's so much subtlety in getting that right, and the model crushed it. And it's just very interesting and still trying to absorb and digest what that means for the way we make content and all these other things.
Speaker 233:31 - 33:49
我一边摸它之类的。这里面有太多微妙之处,要把这些都做对非常难,但这个模型把它完成得非常出色。这真的很有意思,我现在也还在试着吸收、消化这件事对我们制作内容方式以及所有其他事情意味着什么。
Speaker 133:49 - 34:00
That's so interesting. Yeah. I'm the biggest bull on generative media and what it means. I mean, one of the things we've thought about for our podcast is the visuals matter as much as the content.
Speaker 133:49 - 34:00
这太有意思了。对。我是最看好 generative media(生成式媒体)及其意义的人之一。我的意思是,我们一直在想的一件事是:对我们的 podcast 来说,视觉效果和内容本身同样重要。
Speaker 234:00 - 34:00
For sure.
Speaker 234:00 - 34:00
当然。
Speaker 134:00 - 34:06
That's how you catch people's attention in the first place. Yeah. And so, okay, I'm excited to play with Omni.
Speaker 134:00 - 34:06
人们最开始就是这样被吸引住注意力的。对。所以,好吧,我很期待试试 Omni。
Speaker 234:06 - 34:30
I'm excited too, and I think you probably feel this way as somebody who makes content, but I've historically been very For myself personally, I don't use AI to make any content that I produce. It's all my words. It's always my voice. It's always my image and picture showing up. I feel like there's just so much alpha and authenticity, and so I would much rather it be me than some AI version of me.
Speaker 234:06 - 34:30
我也很期待,而且我觉得,作为一个做内容的人,你大概也会有这种感觉:一直以来,就我个人而言,我不会用 AI 来制作任何我产出的内容。那都是我自己写的话。一直都是我的声音。出镜的也一直都是我自己的形象和照片。我觉得这里面有太多 alpha 和真实性了,所以比起某个 AI 版的我,我宁愿那就是我本人。
Speaker 234:31 - 35:01
What I like so much about Omni is that it's not changing me. It is changing a bunch of these other bits, which are not me. I didn't choose any of the set around us or the coffee table. So our words can stay the same, and you can change these bits that are not personal and do something more interesting with them, which I think is really, really cool and feels like the version of what I want sort of Genmedia to be, which is not a bunch of AI avatars.
Speaker 234:31 - 35:01
我特别喜欢 Omni 的一点在于,它并没有改变我。它改变的是周围这些其他部分,而这些并不是我。比如我们周围的布景,或者这个咖啡桌,都不是我选的。所以我们的表达可以保持不变,而你可以去改变这些不属于个人本身的部分,把它们处理得更有意思,我觉得这真的、真的很酷,也很像我所希望的那种 Genmedia 的样子——不是搞出一堆 AI avatars(AI 虚拟化身)。
Speaker 135:01 - 35:03
No Fruit Island videos?
Speaker 135:01 - 35:03
没有 Fruit Island 视频吗?
Speaker 235:03 - 35:11
Exactly, truly. It really is like, it's the original content. The person. It's like the personhood is there. It's just different and amplified.
Speaker 235:03 - 35:11
没错,真的。它确实是那种感觉:内容本身还是原始内容,还是那个人。就像人的主体性还在那里,只是变得不同了,也被放大了。
Speaker 135:11 - 35:14
Super interesting. Okay. I'm excited to play with it.
Speaker 135:11 - 35:14
特别有意思。好,我很期待上手玩玩。
Speaker 235:14 - 35:17
Yeah. We should we should send some prompts right after this and try
Speaker 235:14 - 35:17
是的。我们应该在这之后马上发一些 prompts(提示词)试试看。
Speaker 135:17 - 35:20
some of I mind the fruit videos, though. I'm I'm I'm happy for
Speaker 135:17 - 35:20
不过我确实有点介意那些水果视频。我我我还是很乐见一个
Speaker 235:20 - 35:20
a world
Speaker 235:20 - 35:20
两者并存的世界。
Speaker 135:20 - 35:27
of of both. On the coding side, you launched the ability in AI Studio for people to vibe code Android apps.
Speaker 135:20 - 35:27
在 coding(编程)这边,你们在 AI Studio 里上线了让人们可以用 vibe code(氛围式编程)来开发 Android apps 的能力。
Speaker 235:27 - 35:28
Yeah. Yeah.
Speaker 235:27 - 35:28
对。对。
Speaker 135:28 - 35:33
I'd love to hear how that's going so far and and where you wanna take that.
Speaker 135:28 - 35:33
我很想听听,到目前为止这件事进展得怎么样,以及你接下来想把它带到什么方向。
Speaker 235:33 - 36:06
Yeah. It's super exciting. I think one of the strategic things for AI Studio, and actually this is based on a lot of the feedback from the ecosystem and actually from developers, from others, is there's so many Google products. There's so many different ways in which you touch Google through all these different journeys of building a startup or bringing an idea to your life. And so we have this first class principle of how do we bring things into AI Studio that make it so that you are exposed to other parts of the Google ecosystem without having to go through nine different UIs across Google.
Speaker 235:33 - 36:06
是的。这非常令人兴奋。我觉得对 AI Studio 来说,其中一个战略层面的事情——实际上这也来自 ecosystem(生态系统)以及 developers(开发者)和其他很多人的大量反馈——就是 Google 的产品实在太多了。你在创建 startup(创业公司)或把一个想法变成现实的各种不同旅程中,会以很多不同方式接触到 Google。所以我们的一个 first-class principle(首要原则)是:我们怎样把一些东西带进 AI Studio,让你能够接触到 Google ecosystem(Google 生态系统)的其他部分,而不需要在 Google 里来回切换九种不同的 UI(用户界面)。
Speaker 236:06 - 36:20
And so Androids are a great example, not only of that, but also of enabling people who wouldn't have otherwise built an Android app. And so I literally built my first Android app in AI Studio. Nice. Very cool to see. It's
Speaker 236:06 - 36:20
所以 Android 就是一个很好的例子,它不仅体现了这一点,也让那些原本不会去开发 Android app 的人也能做出来。所以我真的是在 AI Studio 里做出了我的第一个 Android app。不错。看到这个很酷。它是
Speaker 136:20 - 36:20
What is
Speaker 136:20 - 36:20
是什么
Speaker 236:20 - 36:26
it? I just did like Crypto app? Not a crypto app, just plant one. I was planting trees in my backyard.
Speaker 236:20 - 36:26
那个吗?我刚做了个类似 Crypto app 的东西?不是 crypto app,就是种植物的。我当时是在自家后院种树。
Speaker 136:26 - 36:27
Oh, like gardening app.
Speaker 136:26 - 36:27
哦,像园艺 app。
Speaker 236:27 - 36:39
Yeah, and so it was just playing around with a gardening app as I was kicking the tires. I haven't had my breakthrough idea yet of what I want for a mobile app, but I'm gonna come up with something and go compete on the App Store.
Speaker 236:27 - 36:39
对,所以我其实就是在随便玩一个园艺 app,顺便试试手感(kicking the tires,意为先上手体验一下)。不过我还没找到自己真正想做的那个 mobile app 的突破性点子,但我肯定会想到一个,然后去 App Store 上竞争。
Speaker 136:39 - 36:42
Have you seen anything vibe coded really fly in the App Store yet?
Speaker 136:39 - 36:42
你有见过什么 vibe coded 的东西真的在 App Store 上火起来吗?
Speaker 236:43 - 37:12
It'd actually be interesting to see some analysis. I don't know. I'm sure it's accelerating a lot of things on the App Store, but I don't know how much. I don't know anyone personally who's done that. It is interesting, and I was gonna make the observation too that I think the last time I checked the numbers, we were viewing it this morning, it was like 350,000 Android apps built in AI Studio since last week, which is crazy, and excitingly, it's like 350,000 apps that probably no one was going to build before.
Speaker 236:43 - 37:12
其实如果能看到一些分析会挺有意思的。我也不清楚。我相信它肯定正在加速 App Store 上很多事情的发展,但到底加速了多少,我不知道。我身边也不认识有人亲自这么做过。不过这确实很有意思,我本来也想提一个观察:我记得我上次看数据——我们今天早上还在看——好像是自上周以来,已经有 350,000 个 Android app 在 AI Studio 里被构建出来了,这太疯狂了。而更令人兴奋的是,这 350,000 个 app 里,可能大多数本来根本不会有人去做。
Speaker 237:13 - 37:48
A lot of these are personal too, and so this is where I think this Maybe GenUI is farther out there, but I think the idea of you building software to solve your personal problem is very real right now, and people are doing that. It's one of the most common use cases of a lot of these products. And being able to unlock a bunch of the native capabilities of the phone, I think, is also really interesting because you just have so much context that's in different places. So I'm getting very excited about sort of that opportunity, and Android feels like it's becoming the platform for builders.
Speaker 237:13 - 37:48
这里面很多也都是给个人自己用的,所以我觉得 Maybe GenUI 这个方向可能还更偏前沿一些,但我认为,人们为了真正解决自己的个人问题而去构建软件,这件事现在已经非常真实了,而且大家确实在这么做。这也是很多这类产品最常见的 use case(使用场景)之一。而且我觉得,能够解锁手机上一大堆 native capabilities(原生能力)也特别有意思,因为你的大量 context(上下文)其实分散在不同地方。所以我现在对这类机会越来越兴奋了,而且 Android 感觉正在变成 builders(开发者 / 构建者)的平台。
Speaker 137:48 - 37:52
Does it matter that something is an app versus just like the web is so powerful now?
Speaker 137:48 - 37:52
一个东西是 app,还是说其实 web 现在已经这么强大了,这件事还重要吗?
Speaker 237:52 - 38:31
Yeah, it's also very interesting to see that play out. Web is definitely powerful. There are certain things that the operating systems have that you just can't unlock, lots of native richness that actually make experiences feel so much richer. I think about this for text messaging, actually, that the text messaging experience in all the main operating systems feel way richer to me than any AI chat app that I've ever used. If I could just talk to AI in whatever texting app I use, I would be way happier than having to go to some other app because I think we're also just conditioned on the operating systems.
Speaker 237:52 - 38:31
对,这种变化如何展开也非常有意思。Web 当然很强大,但 operating systems(操作系统)里有些东西你就是没法解锁,那种大量 native richness(原生层面的丰富能力)确实会让整体体验显得丰富得多。我其实会拿短信这件事来想:在所有主流 operating systems 里,text messaging(短信 / 即时消息)的体验对我来说都比我用过的任何 AI chat app 要丰富得多。如果我能直接在自己平时用的任何 texting app 里和 AI 对话,而不是还得专门去另一个 app,我会开心得多,因为我觉得我们其实也已经被 operating systems 的使用方式训练出来了。
Speaker 138:31 - 38:38
Yeah, makes sense. Okay, I wanna ask about the model eats the harness or the model eats the scaffolding. What are your thoughts?
Speaker 138:31 - 38:38
嗯,有道理。好,我想问一下关于“model 吃掉 harness”或者“model 吃掉 scaffolding(脚手架)”这件事。你怎么看?
Speaker 238:38 - 39:07
Yeah, I think it's true, and I think part of this is what we have historically thought of as the model is not the model anymore. I think two years ago, when LLMs were popular, it was like the model was actually just a set of weights. It was a set of weights, it was really like, How can you, as simple as possible, send tokens in and get tokens out? And I think we've just progressively, step by step by step We still call it the model. We still call it Gemini 3.5.
Speaker 238:38 - 39:07
对,我觉得这是真的,而且我认为其中一部分原因是:我们过去在历史上所理解的“model”,现在已经不再是那个“model”了。我觉得两年前,LLM(大语言模型)刚火起来的时候,大家会觉得 model 实际上就只是一组 weights(权重)。它就是一组 weights,本质上很像是:你怎样尽可能简单地把 token(词元)送进去,再把 token 输出出来?而我觉得我们只是一步一步、渐进地演变到了今天。我们现在仍然叫它 model,仍然叫它 Gemini 3.5。
Speaker 239:07 - 39:54
We still call it GPT whatever and Claude whatever, but it's actually not just the weights anymore. It's an entire expanding sprawling system that's built around the weights that enable a lot of these next generation experiences from agentic tool calling to all these hosted tools, search, code execution, etcetera. The models are now being spun up in containers and sort of have an agent harness and all that stuff. So the scaffolding is oftentimes a couple of steps ahead of what is baked directly into the model, and then what ends up happening is the model eats that scaffolding and it becomes part of the native model system. There's still value in having the external scaffolding in certain cases, and search maybe is an example of this.
Speaker 239:07 - 39:54
我们还是会叫它 GPT whatever、Claude whatever,但它实际上已经不只是 weights 了。它已经变成了一个围绕 weights 搭建起来、不断扩张和蔓延的完整系统,使很多下一代体验成为可能,从 agentic tool calling(agent 式工具调用)到各种 hosted tools(托管工具)、search、code execution(代码执行)等等。现在这些 model 会在 container(容器)里被启动,并且某种程度上还带着一个 agent harness(agent 编排框架)以及诸如此类的东西。所以,scaffolding 往往会比直接 baked into(内建进)model 的能力领先几步;然后最终发生的事情就是,model 会把这些 scaffolding“吃掉”,它们会成为原生 model system 的一部分。当然,在某些场景下,外部 scaffolding 仍然有价值,search 也许就是一个例子。
Speaker 239:54 - 40:26
There's lots of folks who use different search providers, and there's different use cases that you want, and so sure, maybe the model can natively use search, but you also want something else. Code execution, another example of that. But it does feel like maybe the agent harness is the quintessential example of this right now, where everyone's like, Ah, we gotta go build a harness, and the harness is where the alpha is. I think that perhaps won't be true, at least in the way that we think of the harness today in twelve months. I think the models will have sort of just digested a bunch of that.
Speaker 239:54 - 40:26
有很多人在使用不同的 search provider(搜索提供商),而且你会有各种不同的使用场景,所以当然,也许 model 原生就能使用 search,但你可能还想要别的东西。code execution 也是另一个例子。但现在的感觉是,也许 agent harness 正是这件事最典型的例子:所有人都在说,“啊,我们得去做一个 harness,alpha(超额优势)就在 harness 里。”我觉得,至少按照我们今天理解 harness 的这种方式来看,十二个月后这件事也许就不成立了。我认为 model 会把其中相当一部分直接“消化掉”。
Speaker 240:26 - 40:35
It'll be upstreamed into the model, and the alpha will be somewhere else now. It won't be in sort of trying to spin your own harness because the model just does it natively.
Speaker 240:26 - 40:35
它会被 upstreamed(上游合并)进 model 里,而 alpha 就会转移到别的地方去。它不会再体现在“你自己想办法搭一个 harness”这件事上,因为 model 本身就会原生完成这些能力。
Speaker 140:35 - 40:45
-But I thought that part of the reason why people are building their own harnesses is because if you use a harness from any given model provider, you're locked in, right? So a lot of the application companies want flexibility, which is why they're building their own harnesses.
Speaker 140:35 - 40:45
——但我原本以为,人们之所以要自己做 harness,部分原因在于:如果你用某个 model provider 提供的 harness,你就会被 lock in(锁定)了,对吧?所以很多应用公司想要灵活性,这也是为什么它们要自己做 harness。
Speaker 240:45 - 41:35
-Yeah, and I think that's part of the scaffolding story is that starts out perhaps true, but then as the model capability improves, becomes less true over time, actually. I think the model, though, you don't have a generalized model if it can't use another harness, And so it is important, and I mentioned this in another conversation with someone a few weeks ago, but we need something like harness bench, which is actually measuring how good are all these different models at adapting to all the different harnesses. I feel like that seems like a reasonable thing we should measure as an ecosystem. And I'd be curious to see what models are actually best. But I think over time, you expect they'd be able to use every harness unless you're completely out of distribution, which in that case, you're still gonna be completely out of distribution even if you're using your own harness, so not sure it matters much.
Speaker 240:45 - 41:35
——对,我觉得这正是 scaffolding 这个故事的一部分:一开始这也许是对的,但随着 model 能力提升,随着时间推移,这一点其实会变得越来越不成立。不过我认为,如果一个 model 连另一个 harness 都用不了,那它就不算是一个 generalized model(通用模型)。所以这件事仍然很重要。几周前我和别人聊的时候也提到过,我们需要某种类似 harness bench 的东西,真正去衡量这些不同 model 适配各种不同 harness 的能力到底有多强。我觉得这看起来像是整个 ecosystem(生态)里一个很合理、值得测量的指标。我也很好奇,究竟哪些 model 实际上表现最好。但我认为,长期来看,你会预期它们应该能够使用每一种 harness;除非你完全 out of distribution(超出分布)了,而如果是那种情况,就算你用的是自己的 harness,你也依然还是完全 out of distribution,所以我不太确定这件事是否真有那么重要。
Speaker 141:35 - 41:48
Fair enough. What about the application layer? How do you think about where independent companies can have a hope of surviving when the model eats the harness and eats the stuff around it.
Speaker 141:35 - 41:48
有道理。那么 application layer(应用层)呢?当 model 把 harness 和它周边的那些东西都“吃掉”之后,你怎么看独立公司还能在哪些地方有机会生存下去?
Speaker 241:48 - 42:08
Yeah. It's an interesting story that both of these things feel true. Both on one hand, everywhere I look, I'm like, There's never been more opportunity to go and build something. At the same time, obviously, the models are doing more than they've ever done before. I think there's that thread of capability overhang, which I think there's a huge amount of alpha in.
Speaker 241:48 - 42:08
是的。这很有意思,因为这两件事似乎同时都是真的。一方面,我放眼望去,会觉得现在从未有过这么多机会去做点什么、去构建点什么。与此同时,显然,模型现在能做到的事也比以往任何时候都更多。我觉得这里有一条“capability overhang(能力溢出)”的线索,而我认为这里面有大量的 alpha(超额收益)可挖。
Speaker 242:08 - 42:51
There's the thread of the model companies are going after these very general problems, and there's just so much value in these verticalized domains. If you have expertise in that domain, you sort of know the customers, you know the ecosystem, you can really run laps around even the best model labs because focus is the superpower of startups. If you can focus, you can do anything, and if you look at all of the companies that are big or doing lots of stuff, there's just not a lot of focus, and for some reasons rightfully so because maybe I'm overly justifying Google strategy, but we just have a lot of products. We have a lot of users. We have a lot of different things going on, and so we actually can't focus in one domain.
Speaker 242:08 - 42:51
另一条线索是,模型公司正在追逐这些非常通用的问题,而在那些垂直化 domain(领域)里,其实蕴藏着巨大的价值。如果你在那个 domain 里有专业能力,你就会了解客户,了解整个 ecosystem(生态系统),你就真的可以把哪怕最顶尖的模型实验室都甩在后面,因为专注就是 startup(初创公司)的超能力。只要你能专注,你几乎什么都能做;而如果你看那些体量很大或者在做很多事情的公司,它们往往就没有那么多专注度。某种程度上这也情有可原,也许我是在过度为 Google 的战略辩护,但我们的确有很多产品。我们有很多用户。我们有很多不同的事情同时在发生,所以我们实际上没法只专注在某一个 domain 里。
Speaker 242:51 - 43:14
We have an obligation to do a bunch of things as a big company. I think that's not true for startups. And so I think twenty four months ago, we were all asking ourselves, Oh, wow. It seems like the opportunity space is shifting, and maybe it's possible one of the outcomes is there's less opportunity for startups in the future. That feels like so far, in a way, not what has ended up playing out, which is really positive.
Speaker 242:51 - 43:14
作为一家大公司,我们有义务去做很多事情。我觉得这对 startup 并不成立。所以我想,二十四个月前,我们都在问自己:哦,哇,看起来机会空间正在变化,也许其中一种结果会是,未来留给 startup 的机会变少了。但到目前为止,实际发展出来的情况似乎并不是那样,这真的很积极。
Speaker 243:15 - 43:51
If anything, it feels like there's just even more opportunity than there was. Now coding has helped you close the gap on larger companies that have established code bases and all this other stuff because you can just run way faster and write software quicker. The agentic primitive is a new category that you can sort of build products around that actually, in a lot of cases, to the conversation about the risks involved with building There's risk involved. The risk appetite of different companies is different. So if you're willing to take more risk in some domains, you can win a user cohort who's interested in also taking risk.
Speaker 243:15 - 43:51
如果真要说的话,现在感觉机会甚至比以前更多了。如今 coding(编程)已经帮助你缩小了与那些拥有既有代码库和各种历史包袱的大公司之间的差距,因为你就是能跑得快得多、写软件也快得多。agentic primitive(agent 式基础能力)是一种新的类别,你可以围绕它来构建产品;而且实际上,在很多情况下,回到刚才关于构建这类东西所涉及风险的讨论——确实是有风险的。不同公司的风险偏好不同。所以如果你愿意在某些 domain 里承担更多风险,你就能赢得一群同样也愿意承担风险的用户群体。
Speaker 243:51 - 43:52
There's so much opportunity.
Speaker 243:51 - 43:52
机会实在太多了。
Speaker 143:53 - 44:05
Awesome. I'd love to talk about Google DeepMind's culture. And I'm curious, what does it feel like to be inside GDM right now? We had Demis at AISense. He was so inspiring.
Speaker 143:53 - 44:05
太棒了。我很想聊聊 Google DeepMind 的文化。我也很好奇,现在身处 GDM 里面是一种什么感觉?我们之前在 AISense 请到了 Demis,他真的非常鼓舞人心。
Speaker 144:05 - 44:12
I've heard Sergei's back. You guys have Noam Shazir back. Walk me through what it's like to be at GDM right now.
Speaker 144:05 - 44:12
我听说 Sergei 回来了。你们也把 Noam Shazir 请回来了。跟我讲讲,现在待在 GDM 里到底是什么感觉。
Speaker 244:12 - 45:01
It's incredible. I do try to take it all in because it's like a moment to I try to reflect as much as possible in the chaos of all the things that are happening just because there's so much cool stuff going on. GDM's culture is interesting in maybe three observations. One, back to this thread of focus, we're doing a lot of things, and so I think see sort of I think about this a lot. From a portfolio perspective, I think we have one of the strongest portfolios, which is really exciting, but you do see these moments where another lab or another company, whatever it is, will pull ahead in a certain area where we underinvested, just hadn't been focused enough in that domain, and it's cool to see the way we go about trying to close that gap.
Speaker 244:12 - 45:01
太不可思议了。我确实会努力把这一切都感受进去,因为这就像是一个特别的时刻——我会尽可能在所有事情同时发生的混乱之中去做一些反思,因为正在发生的酷事真的太多了。GDM 的文化很有意思,大概可以从三点来观察。第一,回到刚才说的“专注”这条线索,我们在做很多事情,所以我会经常从这个角度去想。从 portfolio(项目组合)的视角看,我觉得我们拥有最强的组合之一,这一点非常令人兴奋;但你也确实会看到这样的时刻:另一个实验室或另一家公司,不管是谁,会在某个我们投入不足、或者说在那个 domain 里专注不够的领域暂时跑到前面去。而很酷的一点是,你能看到我们是如何努力去缩小这个差距的。
Speaker 245:01 - 45:48
I very much appreciate it. I think I've watched the Demis Thinking Game documentary a few times, and you see a lot of details of that original culture and just the way that strikes work and all this stuff, which is actually really similar today. You just get a bunch of smart people together and go solve the problem, and I love that, and it's very cool to be a part of. Another one is this, I think you see the culture permeate from who the leaders are. And as I Maybe this isn't a perfect characterization of the ecosystem, but Demis is a Nobel Prize scientist and the OG of a lot of this stuff, and you feel that in the DeepMind culture.
Speaker 245:01 - 45:48
我非常感激这一点。我想我把那部关于 Demis 的《Thinking Game》纪录片看了好几遍,你能从中看到很多那种最初文化的细节,以及那种攻关方式是怎么运作的,诸如此类,而这些其实和今天非常相似。就是把一群聪明人聚在一起,然后去解决问题;我很喜欢这一点,能成为其中一部分也非常酷。还有一点是,我觉得你能看到文化会从领导者是谁这件事中渗透出来。也许这不算是对整个生态的完美概括,但 Demis 是一位诺奖级科学家,也是这其中很多事情的 OG(元老),而你能在 DeepMind 的文化里感受到这一点。
Speaker 245:48 - 46:50
I think Sam is maybe one of the world's best businessmen ever, and you sort of see that in the OpenAI culture and the way that they go about the world. I don't have a strong sense of who Dario is, but I think Anthropic is a very interesting place, and you sort of, least as an external observer, he seems like an interesting guy and so somewhat esoteric, and so it seems like they're sort of like that in the DNA and the culture of the company. The other labs are interesting, but I like this very scientific approach to the world, and the way that, like, Demis looks at, like, reason he's doing this, and the reason they started this mission was, like, literally to, like, solve disease and all these things. It's, like, so easy to get and again, I'm always trying to pull myself out of the moment, but it's so easy to get lost in this competitive race of who's pushing a number higher on SWE bench or whatever it is. It's very easy to lose sight of the reason we're doing that is so that we can solve problems that humans actually have.
Speaker 245:48 - 46:50
我觉得 Sam 可能是世界上最厉害的商人之一,甚至可以说是有史以来最强之一,而你也能在 OpenAI 的文化以及他们应对世界的方式中看到这一点。我对 Dario 是怎样的人没有特别强的把握,但我觉得 Anthropic 是个非常有意思的地方,而且至少作为外部观察者来看,他似乎是个很有意思的人,也带点神秘、偏学院派的气质,所以感觉这种特质也某种程度写进了他们公司的 DNA 和文化里。其他 labs(实验室)也很有意思,但我喜欢这种非常科学的看待世界的方式,以及像 Demis 看待这件事的方式——比如他做这件事的原因、他们当初开启这个使命的原因,真的是为了攻克疾病以及所有这些问题。人很容易陷进去——我也总是在努力把自己从当下的情绪里抽离出来——但在这种竞争竞赛里,真的太容易迷失了,比如谁又把 SWE bench 上的某个数字刷得更高,或者别的什么。人很容易忘记,我们之所以做这些,是为了去解决人类真正面临的问题。
Speaker 246:50 - 47:00
My favorite quote from all of Silicon Valley is something like, We can't let other people make the world a better place more than we can, which is what this moment feels like.
Speaker 246:50 - 47:00
我最喜欢的一句 Silicon Valley 里的台词大概是:“我们不能让别人把世界变得比我们更美好。”这句话特别像当下这个时刻的感觉。
Speaker 147:00 - 47:01
-The Gavin Belsen quote.
Speaker 147:00 - 47:01
——Gavin Belsen 的那句台词。
Speaker 247:01 - 47:29
-The Gavin Belsen quote, and I think about that all the time. It's like we're all fighting over who can make the world better more than the other person, which just, when you frame it like that, it seems really goofy to me. And so it's very much not zero sum, and I think that's a way of looking at the world. I think the last thing about DeepMind's culture is we're very It's sort of the engine room of Google, which I think is literally the Twitter bio now of the DeepMind Twitter account, which I love.
Speaker 247:01 - 47:29
——就是 Gavin Belsen 的那句台词,我一直都会想到它。感觉就像我们都在争,到底谁能比另一个人把世界变得更好;可你如果这么去表述,它在我看来就特别滑稽。所以这完全不是 zero-sum(零和)的,我觉得这也是一种看待世界的方式。我觉得关于 DeepMind 文化最后一点是,我们很像 Google 的 engine room(引擎室);我觉得这甚至现在就是 DeepMind Twitter 账号简介里的原话了,我很喜欢这个说法。
Speaker 147:29 - 47:31
Do you man the DeepMind Twitter account?
Speaker 147:29 - 47:31
你在负责 DeepMind 的 Twitter 账号吗?
Speaker 247:31 - 47:55
I don't. I don't want any responsibility manning other people's accounts online. Too much responsibility to do that, but it does feel like that too. So it's like, on one hand, you have sort of the deep rooted lab culture. On the other hand, you have sort of all of these partners across the Google ecosystem that we're collaborating with, everybody from Android that we talked about earlier, to Google Cloud, Gmail to Workspace, etcetera, etcetera.
Speaker 247:31 - 47:55
不是。我可不想承担替别人运营线上账号的责任。那责任太大了,不过那种感觉确实也很贴切。所以,一方面,你有这种根基很深的实验室文化;另一方面,你又有整个 Google 生态里所有这些一起协作的合作伙伴,从我们前面聊到的 Android,到 Google Cloud、Gmail、Workspace,等等等等。
Speaker 247:55 - 48:29
And so it's an interesting blend of I think there's lots of research work happening, but there's tons of applied work that's happening to actually work with some of the forefront customers. Deploying Gemini to billion user products is a problem that only two companies in the world have, and we have 13 of those products. Google goes through this all the time now, and it's such an interesting place to see that happen and see the innovation that takes place in order to make that actually possible, and I feel like you can only do that inside of Google, which is really cool.
Speaker 247:55 - 48:29
所以这是一种很有意思的混合体:我觉得这里有很多 research(研究)工作在发生,但也有大量 applied(应用)工作在发生,真正去和一些最前沿的客户合作。把 Gemini 部署到拥有十亿用户的产品里,这是世界上只有两家公司会遇到的问题,而我们有 13 个这样的产品。Google 现在一直在反复经历这种事,而能在这里看到这一切发生、看到为了让这件事真正可行而产生的创新,是件非常有意思的事;而我感觉,这种事只有在 Google 内部才能做到,这真的很酷。
Speaker 148:30 - 48:35
Beautifully said. Did it give them a lot of heartburn when you joined and were tweeting a lot?
Speaker 148:30 - 48:35
说得真好。你加入之后又经常发推文(tweet),这件事有没有让他们很头疼?
Speaker 248:36 - 48:37
That's a good question. Did he
Speaker 248:36 - 48:37
这是个好问题。他有没有
Speaker 148:37 - 48:40
get sign off from One
Speaker 148:37 - 48:40
从 One 那边拿到批准?
Speaker 248:42 - 49:46
of the silver linings to my Google experience has been just how great that group of folks across marketing comms are to work with, and I think their job is protect Google, make sure we tell the right story, make sure a bunch of bad things don't happen, and so I have a ton of appreciation and partnership with them, but it's been an incredible experience to be able to go try to tell the story that resonates with developers in a way that feels authentic and not have a huge amount of I don't have to get my tweets approved all the time and all this stuff. It's a very positive culture, and I think, hopefully, am always trying to walk the line of not burning the trust and goodwill that I've accumulated with those folks, but it's been super positive because ultimately, I think it's really hard for Google to tell this authentic story. A big company. There's a lot of people. There's a lot of opinions, and so you take the magic of Google and you water it down through a lot of people and a lot of process, and -Yeah.
Speaker 248:42 - 49:46
我在 Google 的这段经历里,其中一个意外收获就是,我深刻体会到 marketing comms(市场传播)那群人合作起来有多棒。我觉得他们的工作就是保护 Google,确保我们讲述的是正确的故事,确保一堆糟糕的事情不要发生。所以我对他们充满感激,也和他们有很强的伙伴关系。但与此同时,能够去尝试用一种能引起 developers(开发者)共鸣、而且感觉真实自然的方式讲故事,这也是一段非常不可思议的经历;而且我并不需要经历大量那种——我不用老是让别人审批我的 tweets(推文)以及诸如此类的事情。这是一种非常积极正向的文化。我想,也希望自己一直都在努力把握分寸,不去耗损我在他们那里积累起来的信任和善意;但整体体验一直都特别正面,因为归根结底,我觉得 Google 很难讲出这种真实的故事。它是一家大公司,人很多,意见也很多。于是 Google 的那种魔力,经过很多人和很多流程之后,就会被不断稀释。——对。
Speaker 249:46 - 50:00
You actually -You miss the beautiful story, which is Google's doing the most interesting technology in the world and helping our users with some of the hardest problems in the world. And it's a privilege to get to help tell that story, so it's a lot of fun. I enjoy it.
Speaker 249:46 - 50:00
你其实会错过那个美妙的故事:Google 正在做全世界最有意思的技术,也在帮助我们的用户解决这个世界上一些最难的问题。能参与讲述这个故事是一种荣幸,所以这件事非常有趣。我很享受。
Speaker 150:00 - 50:09
I love what you're doing. I love what Josh is doing. Think you guys have put a really sincere human touch on, as you put it, the most important problem of our time.
Speaker 150:00 - 50:09
我很喜欢你在做的事。我也很喜欢 Josh 在做的事。我觉得你们确实为——用你的话说——我们这个时代最重要的问题,注入了一种非常真诚、非常有人味的表达。
Speaker 250:09 - 50:10
Thank you.
Speaker 250:09 - 50:10
谢谢。
Speaker 150:10 - 50:24
Well, wonderful. Logan, thank you so much for joining me today. This is a very far ranging conversation, everything from agents and coding to world models and harnesses and GDM culture and lots lots of nuggets here. Thank you for for joining me today.
Speaker 150:10 - 50:24
太好了。Logan,非常感谢你今天来和我聊。这是一场覆盖面非常广的对话,从 agents(智能体)和 coding(编程),到 world models(世界模型)、harnesses,以及 GDM culture(GDM 文化),里面有非常非常多值得回味的 nuggets(干货/亮点)。感谢你今天加入我。
Speaker 250:24 - 50:31
This is a ton of fun. Thank you for having me, and I'm excited to see what the, folks cook up or where we've been sitting this whole time, maybe in front of us.
Speaker 250:24 - 50:31
这次聊天特别有意思。谢谢你邀请我,我也很期待看看大家会拿出什么成果,或者说,看看我们这段时间一直坐着的这个地方——也许答案其实就在我们面前。
Speaker 150:31 - 50:32
And maybe there'll be a
Speaker 150:31 - 50:32
也许还会有一只
Speaker 250:32 - 50:33
A dog.
Speaker 250:32 - 50:33
狗。
Speaker 150:33 - 50:34
You can make my dog dreams come through.
Speaker 150:33 - 50:34
你可以让我关于狗的梦想成真。
Speaker 250:35 - 50:35
I love it.
Speaker 250:35 - 50:35
我喜欢这个。
Speaker 150:35 - 50:37
Awesome. Thanks, Logan.
Speaker 150:35 - 50:37
太棒了。谢谢你,Logan。
Speaker 250:37 - 50:37
Of course.
Speaker 250:37 - 50:37
当然。
原文 ↗https://www.youtube.com/watch?v=cMAs8z2dehs
BuildSpeak — 关于本项目BUILT IN PUBLIC · 跟随 builders 而非 influencers