谷歌DeepMind的运作与实验——对话Lila Ibrahim和James Manyika

谷歌DeepMind是如何运作并做出决策的？是什么让谷歌变得更加实验性？

How does Google DeepMind operate and make bets, and what's making Google more experimental?

让我们在接下来与两位谷歌领导者一起探讨这个问题。

Let's talk about it with two Google leaders right after this.

你是否一直在等待最佳时机来升级你的科技设备？

Have you been waiting for the perfect time to upgrade your tech?

好消息。

Good news.

等待结束了。

The wait is over.

戴尔科技日的年度促销活动现已开启，我们正为最优秀的客户带来超值优惠，涵盖搭载英特尔酷睿Ultra处理器的最新笔记本电脑，例如戴尔14 Plus。

Dell Tech Day's annual sales event is here, and we're celebrating our best customers with fantastic deals on the latest PCs, like the Dell 14 plus with Intel Core Ultra processors.

我们还提供诸多超值福利，包括戴尔积分奖励、快速免费配送、高级技术支持、价格匹配保证等。

We've also got incredible perks, like Dell rewards, fast free shipping, premium support, price match guarantee, and more.

在升级你的电脑时，不妨一并升级我们的高端显示器和配件系列，因为目前它们也享有大幅优惠。

And while you're upgrading your PC, you may as well go all out because we're also offering huge deals on our premium suite of monitors and accessories.

你知道这意味着什么。

You know what that means.

没错。

That's right.

你可以以惊人的折扣购置一套全新的设备。

You can get a whole new setup with amazing savings.

很明显，这是一场你不能错过的促销。

Clearly, this is a sale you don't wanna miss.

访问 dell.com/deals。

Visit dell.com/deals.

就是 dell.com/deals。

That's dell.com/deals.

我是迈克尔·刘易斯。

Michael Lewis here.

我的畅销书《大空头》讲述了2008年美国房地产市场泡沫形成与崩塌的故事。

My bestselling book, The Big Short, tells the story of the buildup and birth of The US housing market back in 2008.

十年前，《大空头》被拍成了获得奥斯卡奖的电影，现在我首次将其以有声书的形式呈现给你，由尤鲁斯·特鲁利朗读。

A decade ago, The Big Short was made into an Academy Award winning movie, and now I'm bringing it to you for the first time as an audiobook narrated by Eurus Truly.

《大空头》的故事——关于押注市场反向走势的意义，以及谁真正为失控的金融体系买单——在今天依然和以往一样具有现实意义。

The Big Short story, what it means to bet against the market, and who really pays for an unchecked financial system is as relevant today as it's ever been.

现在就前往 pushkin.fm/audiobook 或在任何有声书平台获取《大空头》。

Get The Big Short now at pushkin.fm/audiobook or wherever audiobooks are sold.

欢迎收听《大科技》播客，这是一档致力于对科技世界及其更广领域进行冷静而深入对话的节目。

Welcome to Big Technology Podcast, a show for cool headed and nuanced conversation of the tech world and beyond.

今天我们有一期精彩节目，因为我们将深入探讨谷歌人工智能与技术研究部门的运作方式。

We have a great show for you today because we're gonna go deep inside the way Google's AI and technology research operations work.

今天我们请到了两位出色的嘉宾。

We have two great guests with us today.

莉拉·易卜拉欣在这里。

Lila Ebrahim is here.

她是谷歌DeepMind的首席运营官。

She is the chief operating officer of Google DeepMind.

莉拉，欢迎。

Lila, welcome.

谢谢。

Thank you.

我们还邀请了詹姆斯·莫尼卡。

And we're also joined by James Monika.

詹姆斯是谷歌研究实验室科技与社会部门的高级副总裁。

James is the SVP of research labs technology and society at Google.

詹姆斯，欢迎。

James, welcome.

谢谢你们邀请我。

Well, thanks for having me.

当然，这是我们达沃斯系列对话的最后一篇，现场还有观众。

And of course, this is our concluding conversation here in our series at Davos, and we do have a live audience.

现场的观众们，来点掌声吧。

Live audience, make some noise.

让他们知道你们在这里

Let them know you're

在这儿。

here.

好的。

Alright.

要谈的内容太多了，时间却不多。

So much to get to, not a lot of time.

我们先从谷歌DeepMind的运作方式说起。

Let's just start with the way that Google DeepMind operates.

谷歌DeepMind的首席执行官德米斯·哈萨比斯最近在节目中将DeepMind描述为现代版的贝尔实验室。

Demises Habes, the CEO of Google DeepMind who's recently on the show has described DeepMind as sort of a modern day Bell Labs.

但这意味着什么，莱拉？

But what does that mean, Laila?

你能给我们讲讲，这个研究机构是如何运作的吗？

Can you tell us a little bit about how the the research, is it a lab, the operation company?

它是如何运作的？

How does it operate?

也许我应该从我们的使命开始，因为我认为一切都基于此，那就是负责任地构建人工智能以造福人类。

Maybe I should start with our mission because I think everything is kind of based off of that, which is to build AI responsibly to benefit humanity.

因此，我们首先会设定非常雄心勃勃的研究议程。

And so the first thing we do is take really ambitious research agendas.

我们以一种方式组织它，即关注重大问题，但不告诉人们该如何去做。

We structure it in a way where we're looking at what are the big problems, but not telling people how to do it.

当你思考我们最初是如何approach这个问题时，实际上是受到贝尔实验室黄金时代、政府项目如阿波罗计划，甚至更近期的皮克斯的启发。

And when you think about how did we first approach that, it's really about taking inspiration from the golden era of Bell Labs, but also government programs like the Apollo program, and even more recently, Pixar.

因此，这一切都围绕着吸引顶尖人才并为他们创造一个能够成功和探索的环境。

So, it's all focused around bringing in really great talent and creating an environment for them to succeed and to explore.

所以，第一件事是设定大型研究议程，告诉人们应该专注于哪个领域，但不干预他们如何完成工作。

So, first thing is that big research agenda, telling people what kind of the area to focus on, but not how to do their job.

第二件事是，由于议程非常广泛，我们希望组建跨学科团队。

The second thing is really because it's such a broad agenda, we want to build interdisciplinary teams.

你如何营造一种文化，让生物伦理学家能与计算机科学家和神经科学家并肩工作？

How do you create a culture where you can have a bioethicist next to a computer scientist and a neuroscientist?

因为我们认为，真正的奇迹就发生在这样的交汇点上，它能释放出巨大的潜力。

Because we think that's really where the magic happens and unlocks the work.

这种做法已经催生了如此非凡的成果。

And this type of approach has resulted in such extraordinary efforts.

同时，我们也不惧于探索，然后适时地问：现在是时候了吗？

And we're also not afraid to explore and then say, when is it time?

我认为德米斯在衡量时间方面有着非凡的方式。

I think Demis has a remarkable way of measuring time.

比如，探索的时间够不够？我们是否设定了真正雄心勃勃的目标？

Like time to explore, are we setting the really ambitious goals?

我们在朝着这些目标前进的过程中进展如何？

How are we doing progress towards that?

同时，我们也敢于坦率地说：现在是时候退一步暂停，或者加倍投入了。

And also not being shy to say, okay, now's the time to take a step back and pause it or double down.

一个很好的例子是，在过去几年里，我们一直在围绕一个科学领域——学习科学——开展大量工作。

Great example of that is over the past few years, we've been doing a lot of work around one science area, learning science.

人们是如何学习的？我们能否改进它？

How do people learn and can we improve it?

今年，德米斯真的说：‘好了，Gemini 已经足够好了。’

And then this year was really, Demis was like, Okay, Gemini is good enough.

是时候把我们在学习科学领域积累的所有成果融入 Gemini 了。

It's time to infuse everything we've done with the industry around learning science into Gemini.

这成为我们重点推进的方向之一，旨在显著提升 Gemini 对学习者的服务能力。

And that was one of our focus areas to really advance how Gemini could be provided for learners.

因此，我认为谷歌深脑在时机把握上有一种非常神奇的特质。

So there's something I think quite magical within Google DeepMind about timing.

好的，GDM，我想我们接下来要面向整个科技行业了。

Okay, GDM, I guess we're going to go everybody in the tech industry.

我差点说漏嘴了

I I almost got myself for saying

它。

it.

所以让我们来谈谈这个。

So let's but let's talk about it.

我想稍微梳理一下这个过程。

So the way I just want to talk through process a little bit.

你刚才描述的这种方式。

The way that you just described that.

德米斯说，Gemini 已经准备好应用于学习领域，然后谷歌深度思维就开始推进了。

Demis said that that Gemini was ready for learning and then Google DeepMind started to pursue it.

谷歌深度思维所从事的工作中，有多少是自上而下的，有多少是自下而上的？

How much of what Google DeepMind works on is, you know, top down versus bottom up?

我听说 OpenAI 描述他们的运作方式，像是一个大公司内部的多个初创团队。

A way that I've heard OpenAI describe the way that it works is like a bunch of different startups within a larger company.

谷歌的运作方式是否也类似，还是更多来自顶层？

Is that a similar way that Google operates or does it come more from the top?

因为我们的使命如此宏大，我们正努力理解人工智能能在哪些重大挑战中帮助我们揭开宇宙的奥秘，并解决人类面临的某些最大难题。

Well, because our mission is so ambitious, we're really trying to understand what are the big challenges where AI can help us unlock our understanding of the universe around us and solve some of humanity's biggest challenges.

这个目标足够宽泛，使我们能够开展诸如天气探测和天气预报预测之类的工作。

And it's broad enough that we can do things like how do we do weather exploration and try to predict weather forecasts?

我们如何利用AlphaFold进行蛋白质结构预测，以更好地理解疾病，从而开发出更有效的治疗方法？

How do we do AlphaFold and protein structure prediction to help us better understand diseases so we can come up with better therapeutics?

生成式人工智能，我们如何持续改进它，以改善人们的生活？

Generative AI, how can we continue to improve that to make people's lives better?

因此，我们采取非常广泛的投资组合视角，但同时为研究人员留出探索的空间。

So again, we take a very broad portfolio perspective, but we allow the space for researchers to explore.

这正是我一开始所说的：我们必须找到合适的人才。

And that's really what I meant in the beginning of like, we've got to find the right talent.

因此，我们需要一种以使命为导向、价值观一致的文化，吸引那些希望参与这种探索、产生重大影响，并能借助谷歌实现规模化的人才。

So mission driven culture and values aligned, people who want to have this type of exploration and a big impact and scale that we can have of being part of Google.

可以说，德米斯在这一领域中的思维方式非常出色，因为他在这个领域已经深耕多年。

So, would say some of this is Demis is quite remarkable in terms of his thinking in this space because he's been doing it for so long.

DeepMind 成立于十六年前。

DeepMind was founded sixteen years ago.

这一直是他毕生的使命。

It's been kind of a lifelong mission of his.

然而，我们有一个充满创造力的团队，他们喜欢在跨学科环境中工作，希望对这个世界产生影响。

And yet, we have an organization full of people who are creative, who like to work in an interdisciplinary environment, who want to have impact in this world.

因此，他们也会提出自己独特的方法。

So they also come up with their own approach to things.

设定目标。

Setting goals.

两者都有一点。

It's a little bit of both.

抱歉。

Pardon me.

是的，有一点

Yeah, little

有一点自上而下来自德米斯，还有一部分

bit of Some top down from Demis and then some

自下而上。

Bottoms up.

嗯，好的。

Which Okay.

这让管理这个组织结构的一部分变得相当

Makes managing part of that organization structure I quite a

我肯定会和你谈谈人才。

will talk with you about talent for sure.

说到这个，事情有什么变化吗？

And on that note, how have things changed?

因为我就要更广泛地谈谈科技行业了。

Because I'm just gonna talk about the tech industry more broadly.

似乎曾经有一段时间，许多科技公司会给予这些有才华的人很大的自由，去探索那些可能不会立即产生成果的事情。

It seems like there used to be a moment where a lot of tech companies gave you know these talented people broad leeway to explore things that might not have immediate results.

然后突然之间，我们进入了这场人工智能竞赛，许多公司把原本从事这些长期项目的研究人员拉得离产品更近了。

Then all of a sudden we got into this AI race and many companies brought their researchers who were working on these long term products much closer or much long term projects much closer to the product.

于是，长期研究必须立即产生产品影响，几乎成了一种必然要求。

And all of a sudden there was a almost imperative for long term research to make immediate product impact.

那么，这种情况随着时间的推移也有变化吗？

So has that changed as well over time?

这种现象在DeepMind内部也同样存在吗？

Is that something that's going on within DeepMind as well?

是的，我八年前加入，我们确实经历了一段旅程。

Yeah, joined about eight years ago and we've definitely been on a journey.

但我认为谷歌DeepMind令人兴奋的地方，也是我们许多员工长期留任的原因，在于我们拥有广泛的产品组合。

But what I think is so exciting about Google DeepMind and I think why so many of our employees stay so long is because we have that breadth of portfolio.

因此，有些人希望继续从事深度研究、前沿人工智能研究，或者更专注于科学领域的工作。

So, there are some people that want to continue the deep research, frontier AI research that they do, or scientific, more focused on the science.

我们为这种探索提供了空间，同时也在生成式AI的进展上取得了成果，比如去年我们在Gemini上取得的所有进步。

And we have the space to do that exploration while also delivering on the advancements around generative AI, such as all the progress we've made last year with Gemini.

好的。

Okay.

让我进一步说明一下。

Let me take that a step further.

谷歌内部转型的方式被描述为：不再让每个产品部门或产品团队各自制定AI方向，而是现在公司内部有一个中央引擎，我认为这就是AI部门，它负责生成AI技术，然后将其分发给各个产品部门。

The way that the transformation within Google has been described is that instead of having every product area or product group chart its own direction on AI there's now this central engine room within the company which is I think the AI division that generate that creates the AI and then farms it out to these product areas.

你能谈谈这个过程以及它是如何运作的吗？

So can you talk a little bit about that process and how that works?

是的。

Yeah.

实际上，我认为过去几年中，将谷歌大脑和DeepMind的最佳AI团队与研究力量整合到一起，是令人兴奋的一件事，这样我们就能探索如此广泛的技术组合。

And actually, I think that's been one of the exciting things over the past few years with a combination of Google Brain and DeepMind of bringing the best of Google's AI teams and research together under one roof where we could explore such a broad portfolio.

因此，正如你提到的，我们一直专注于成为AI创新引擎。

And so we've really been focused on, as you mentioned, becoming the AI innovation engine.

但我不会说我们将技术直接分发给其他谷歌团队。

And then I wouldn't say we farm things out to other Google teams.

我们与产品团队及其客户紧密合作，了解他们的需求，以便从一开始就更好地构建模型，并以协作且负责任的方式进行，这样当模型被应用到不同的谷歌产品时，已经过大量测试，并能针对特定用例进行优化。

We collaborate very closely with the product areas and their customers to understand what the needs are so that we can build the models better from the start and do so in a very collaborative and responsible way such that by the time it goes to different Google products, it's already been through a lot of that testing and can be refined for that specific use case.

好的，最后一个问题是。

Okay, one last question.

这实际上帮助了我们。

And that's actually helped us.

我认为由此产生的一个例子是，我们发布了Gemini 3，随后立即向广大开发者和用户开放。

I think what's resulted in that, for example, is like Gemini three, we launched it and then it was available to a broad group of developers and users right away.

好吧，关于这个话题再问最后一个问题，之后我们转给詹姆斯。

All right, one last question on this and then we're gonna go to James.

詹姆斯，再次感谢你来到这里。

And James, thanks again for being here.

所以我只是想问你一下。

So, me just ask you this.

在我们的节目中，有一个假设认为，桑达尔曾在麦肯锡工作过，这种重组方式——集中化后再与各团队协作——有点像是麦肯锡的风格。

On our show we have this hypothesis that Sundar spent time at McKinsey and this is sort of like a McKinsey style approach to like reorg, centralize and then work with the groups.

这有道理吗？

Is there a truth to that?

嗯，这里有一位前麦肯锡的人，或许能谈谈这个结构。

Well, you have a former McKinsey person here who might be able to address the structure.

詹姆斯？

James?

没有。

No.

我认为你正在经历的是一件非凡的事情。

I think what you've got going on is an is an extraordinary thing.

对吧？

Right?

因为一方面，你有Gemini项目，它支撑着这一切，构建像Gemini、Gemini 2.5、3这样的大规模模型。

Because on the one hand you've got the Gemini program which underlies all of this building the kind of large scale models Gemini itself, Gemini 2.5, three and all of that.

这其实源于三年前，我们把谷歌大脑团队和DeepMind团队合并，创建了Gemini项目。

And this kind of came about back in three years ago when we put together the Google Brain team and the DeepMind team to create the Gemini program.

因此，这个项目现在支撑着公司内所有这些应用。

So that program now underlies all the things across the company.

你可以看到Gemini出现在搜索、Google Workspace中，出现在我们的所有产品、Notebook Column以及这些地方。

So you see Gemini show up in search in Google workspace it shows up in in all our products in notebook column and all of these things.

所以它可以说是基础，正如Lada所说，DeepMind和Gemini项目已经成为核心引擎。

So it's kind of the foundation and that's why as Lada said, know, DeepMind and Gemini program has become the engine room.

但除此之外，你还有其他许多事情正在发生。

But in addition to that, you've got all these other things going on.

公司内部正在进行深入的科学研究。

There's deep science going on in the company.

意思是

Mean

这种理念就是，去应对那些最根本的、能开启多个领域研究与创新的核心问题。

this idea of you know kind of this foundational kind of tackle the biggest root node problems that open up research and innovation in so many areas.

所以，所有这些都在进行中，同时你还有许多其他雄心勃勃的特殊项目，比如Genie，致力于构建世界模型。

So you've got all of that going on too and then you've got all these other special kind of ambitious projects working on things like Genie, build world models.

你们正在为Waymo开发专门的功能，并提升Waymo的模型，从而推动Waymo自动驾驶的发展。

You've got work going on to build special things for Waymo and enhance the models of Waymo's that lead to Waymo the drive the Waymo driver.

所以你们正在同时推进很多这样的项目。

So you've got a lot of these things going on.

因此，我认为这不是自上而下的方式，而是要善用名为Gemini的底层基础，确保每次快速迭代时——你们也看到了，大约每六个月就会推出新一代Gemini，要确保它立即上线。

So I don't think there's a top down as much as let's take advantage of the foundation called the Gemini program, make sure that every time we do these rapid iterations, you've seen Renown cycle of every six ish months there's a new generation of Gemini make sure it shows up immediately.

正如Laleh所描述的，没有延迟和滞后，一旦Gemini的最新版本发布，你就会立刻在搜索中看到它，在Gemini应用本身中看到它，无处不在。

As Laleh described there's no shipping and delay so the minute the latest version of Gemini comes out you're gonna see it in search you're gonna see it in you know in in the Gemini app itself, you're gonna see it everywhere.

所以，过去三年里发生的这一切，真是令人难以置信。

So that's kind of the incredible thing that's that's happened over the last three years.

好的。

Alright.

我想聊聊实验室。

I want to talk about labs.

Google实验室，我们这些早期使用Google产品的人，都经历过Google内部的实验阶段，后来实验室一度消失了。

So Google Labs, a lot of us who use Google products in the early days, you know, we saw this era of experimentation within Google and then Labs went away for a bit.

当然，Labs 并不是公司内唯一的实验性尝试。

Not that Labs was the only bit of experimentation within the company.

但后来 Labs 被重新启用，而且我们似乎开始看到越来越多来自 Google 本部的实验性项目，这是很久以来都没见过的。

But then Labs was revived and and it seems like we're starting to see many more experimental projects come out of Google proper in a way that we hadn't seen in a long time.

那么 Labs 在这其中扮演了多大责任？为什么 Labs 又回来了？

So how responsible is Labs for that and why is Labs back?

哦，Labs 真的特别有趣。

Oh Labs is so much fun.

实际上，三年前，这可以说是 Sundar 的一个灵感时刻，他说：‘让我们重启 Labs’，当时我们正处在 AI 时代，我们该如何探索和实验，打造这些完全以 AI 为核心的全新产品呢？

So what actually happened was three years ago you know this is a kind of an inspired Sundar moment said let's reboot labs and you know we're in this AI moment how do we kind of explore and experiment and build these AI first AI products that are totally AI first.

Labs 的理念就是，选取来自 Google DeepMind、Google 研究院，以及公司内任何其他地方最出色的科研成果，专注于如何构建实验性的、以 AI 为先的产品。

So the idea with labs is let's take the most amazing research coming out of Google DeepMind and Google research and any other place quite frankly in the company where there's incredible research and focus primarily on how do we build experimental AI first products.

我想大多数人最熟悉的可能就是现在所谓的 NotebookLM 了。

I think what most people probably know of the most is what's now you know notebookLM.

顺便说一下，它的起源方式简直不可思议，因为我还记得我第一次接触它的时候

You know the way that started by the way is incredible because I remember when I first encountered

那么，关于NotebookLM，来讲讲这个故事吧。

So what notebookLM and tell the story.

Notebook很有趣，结果它被命名为Tailwind。

Notebook is fun so it's turned out to put a called tailwind.

当时有四五个人在做这个项目，想法是：我们能不能打造一个完全以AI为核心的科研工具，它能基于你输入的内容进行运作。

There were four five people working on it and the idea was you know can we build a very AI native research tool that is grounded on what you put into it.

换句话说，你的资料——可能是书籍、论文、草稿，或者任何你想作为基础的内容——都可以放进一个笔记本里，然后与之互动。

So in other words you know your sources you might have books, you might have papers, you might have drafts, you may have whatever your content that you want to ground it on put it in a notebook and be able to engage with it.

这就是这个想法的起源，事实上，Stephen Johnson——一位作家——给了我们额外的推动力，他就是这样一种人：会保存一切东西。

So that was the conception of the idea and in fact in some ways you got additional impetus from Stephen Johnson who's a writer And Stephen Johnson is one of these people who kind of keeps everything.

他保存了90年代的笔记、书籍草稿，还有各种各样的东西。

So he has notes from the 90s and drafts of books and all kinds of things.

他说：我真希望有个产品，能让我把所有这些东西都丢进去，然后和它们互动。

He said, I'd love a product where I can dump all that stuff in and engage with it.

我1997年到底在想些什么？

What was I thinking 1997?

我那个草稿是哪个？

What was that draft I did?

所以，Notebook LM 最终演变成了一款强大的研究工具，它基于你输入的内容，当你与之互动时，它会进行总结或起草，并提供这些引用，这在某种程度上是它的关键功能。

So what Notebook LM has become is this incredible research tool grounded on what you've put in and when you engage with it and it summarizes or drafts something it gives you these citations and that's in some ways was a key feature of it.

所以，如果它说‘亚历克斯，你之前说过这个’或‘你的来源提到这个’并加以总结，它会给出引用。

So if it says Alex you know you said this or your source says this and summarizes in some way It'll give you citations.

如果你想，可以点击这些引用。

If you want you can click on the citations.

它们会带你直接回到原始内容。

They take you all the way back to the original content.

对。

Right.

这非常有用。

So which is incredibly useful.

然后发生了一件有趣的事：既然它已经是一个非常有用工具了，我们就想，其实有时候我更想听我的资料内容，而不仅仅是与之互动，于是我说，好吧，技术已经足够成熟了，我们可以加入AI语音概要功能。

Then a fun thing happened which was well you know so it was already a very useful tool then we said well actually you know what sometimes I want to hear my sources as opposed to just engage with them so I said okay well actually the technology is ready enough we can actually add AI audio overviews.

这实际上就像一个播客。

Which is like effectively a podcast.

你可以让它像有两个主持人一样。

You can have it with like two hosts.

实际上，最初的想法根本不是这样做。

You could have it do actually, the original the idea wasn't even to do that.

最初，我们几个人，比如杰夫·迪恩这些人，那位传奇的杰夫·迪恩说：等等，其实呢？

So initially, the idea was a few of us, you know, Jeff Dean and those, you know, this legendary Jeff Dean said, well, actually, you know what?

我们正在阅读计算机科学领域以惊人速度不断涌现的大量论文。

We're reading all these papers that are coming out of this incredible pace in the computer science field.

如果能在开车上班时听到这些论文的口头摘要，那就太好了。

It'd be nice to be able to hear a summary of them verbally while I'm driving into work or something.

这样我就能先判断出哪些论文值得深入阅读。

So just, you know, and then I can figure out which people I wanna read.

这就是最初的想法。

So that was the original idea.

实际上，当你听到别人谈论并深入讨论时，学习起来会更容易。

There was actually no, you know what, it's easier to learn stuff when you have you hear people talking about it, engaging.

这就是为什么研讨会很有趣，对吧？

That's why seminars are interesting, right?

作为一种学习方式。

As a learning mechanism.

所以这个想法就是这样产生的。

So that's where the idea came from.

于是我们制作了这些音频概要，以播客或两位主持人讨论的形式呈现。

So we did these audio overviews, which, you know, in the form of a podcast or a discussion with two hosts discussing it.

现在它已经发展了，产品也就这样突然火起来了。

And now it's evolved, and that's when the product just kinda took off.

是的。

Yeah.

每当我做关于人工智能的演讲时，我都会在现场观众面前演示如何构建这样一个笔记本，作为我的压轴表演。

Whenever I give a presentation about AI, that's the party trick where I build one of these notebooks in front of the audience.

对。

Right.

然后我播放播客，那些第一次看到的人，都会惊讶得合不拢嘴。

And then I hit play on the podcast and people who haven't seen it before, it's like a jaw drop moment.

事实上，我们在YouTube频道和播客听众中，多次有人问亚历克斯：他们是不是用你的声音训练的？

And in fact, we've had multiple people on our YouTube feed and coming from the podcast, they've asked, Alex, did they train on your voice?

因为听起来特别像你。

Because it sounds a lot like me.

我说：不，听好了，他们总喜欢一开始就说清楚，你得明白，每个播客主持人都这么说。

And I say, no, listen, it's they they always say, let's unpack this at the beginning and you have to understand every podcaster says that.

所以这并不是我。

So it's not me.

顺便说一下，笔记本最有趣的用途之一是，现在你可以导入各种格式的内容。

One of the most fun use cases of notebook by the way is because now you can put in things in all kinds of formats.

可以是论文、YouTube视频，也可以是你硬盘里的任何东西。

There could be papers, could be YouTube videos, there can be whatever's on your hard drive.

我一个有趣的用途是，当我需要处理来自全球上百个国家、不同语言的论文时。

One of my fun use cases was actually when I had to do this thing where I was seeing all these papers from literally over a 100 countries in different languages.

于是我将所有这些论文都导入进来，直接用其他语言的内容进行互动，因为Notebook LM支持多种语言，现在你还可以上传用户的视频，但我认为

So I put them all in and just engage with content in most of other languages because notebook LM can handle multiple languages and now you can do video of a user so but I think

它可以生成一种带图形的视频，不是动画视频，而是带有图形的视频

it can make like an animated not an animated video but a video with like graphics

和幻灯片。

and slides.

但我认为，这正是实验室里正在发生的事情——我们试图将Laila和同事以及其他人在谷歌深脑和谷歌研究中所做的出色研究，转化为如何打造卓越的AI优先产品。

But I think this is an example of the kind of thing that happens in labs where we try to take this incredible research that Laila and colleagues and others are doing at Google deep buying and Google research to say how do we build amazing AI first products.

Flow是另一个例子。

Flow is another example.

如果你玩过

And if you've played

我来给你讲个关于Flow的故事，然后你再继续多说一点。

with I just so I I'll tell you a story about Flow then I'll let you talk a little bit more about it.

我刚刚完成了我的第一次也是最后一次登山。

I just did a a my first and last mountain climb.

地点是厄瓜多尔的科托帕希火山。

And it was Cotopaxi in in Ecuador.

我想拍一段视频来记录那一刻。

And I wanted to make a video sort of capturing the moment.

但发生了一些事情，我根本没拍下来。

But there were a couple things that happened that I just could not that I didn't videotape.

因为我决定专心登山，而不是做YouTube视频。

Because I decided to spend the climb actually climbing as opposed to YouTubing.

据我所知，这在当今可是相当罕见的。

Which is apparently from what I hear rare these days.

有一刻，我的水瓶从背包里掉出来，顺着冰川滚了下去，然后消失在黑暗中。

And there was a moment where my water bottle fell out of my backpack and rolled up down the glacier and then kind of disappeared into the darkness.

我想用画面来表现这一幕。

And I wanted to illustrate that.

所以我去了Flow，一个谷歌的视频生成器，我说我想制作一个动画纪录片风格的视频来展示这一幕，并将它插入到视频中。

So I went to Flow, a Google video generator and I said I want to make an animation documentary style to show this and slotted that into the video.

现在你可以了，而以前我得雇一位动画师。

So now you can and before I would have to hire an animator.

现在你可以做到了。

Now you can do it.

是的。

Yeah.

不。

No.

这简直太不可思议了。

It's it's it's incredible.

但我认为，你知道，Flow就是实验室中发生奇迹的一个例子。

But I think, well, you know, Flow is an example of the magic that happens in labs.

我记得我们一群人曾聚在一起。

So I remember a bunch of us got together.

所以乔什负责实验室，你知道德米斯和我们几个人在想，如果我们把这些现在拥有的工具整合成真正有用的东西会怎样？

So Josh who runs labs and you know Demis and a few of us say what if we put all these tools we now have together into something that's actually useful.

事实上，我们最初的版本在某些方面还挺笨拙的。

And in fact the initial version of it that we have you know in some ways was clunky.

然后我们想，不如直接去和一些真正的电影制作人聊聊，听听他们的意见。

Then we said well actually let's just talk to some actual filmmakers and get their inputs.

顺便说一下，实验室里经常发生的事就是我们会积极与创作者和其他人互动，帮助我们思考如何构建这些工具。

Of the things that happens in labs by the way is we try to engage a lot with creatives and others to help us think about how we build these tools.

所以，这就是Flow的由来。

So anyway that's how Flow came about.

是的。

Yeah.

你可以逐场景地通过提示来生成视频。

It's you can build scene by scene prompting into video.

是的。

Yeah.

而且你可以实现连续生成。

And you can have continuation.

我想这大概就是名字的由来。

I think that's probably where the name comes.

它可以流畅地进行。

It can Flow.

你刚才说的这一点，其实是来自电影制作人的洞察。

And and what you just said was an insight that came from filmmakers.

事实上，最初的版本——我不知道你那里有什么——其实没什么用。

In fact, the initial version was I don't know what you've got is actually not very useful.

我希望能够逐场景地构建，并将它们拼接在一起，实现这样的功能。

I'd like to be able to build things scene by scene and be able to stitch them together, be able to do this.

所以，这确实很有帮助。

So, you know, so that's why it's been helpful.

那么，如果你说，实验室究竟是什么？

So if you say, what is labs?

这是一个我们尝试探索各种想法的地方。

It's a place where we try to experiment with all these things.

在任何给定时间，我们大概都有大约30个实验在进行中。

At any one time, we probably have about 30 experiments cooking.

所以如果你访问谷歌实验室网站，是的。

So if you go to the Google Lab site Yes.

你可能会看到大约30个不同的项目。

You'll probably see about 30 different things.

但我

But I

我有个请求给你。

have a request for you.

扩大访问权限吧，因为里面有很多产品，你知道的，有很多项目看起来非常有趣，值得使用。

Broaden the access because there's a lot of product, you know, a lot of projects in there that seem really interesting to use.

但每次我去的时候，都要排队等候。

But every time I'm there it's a wait list.

我们会处理这个问题。

We'll work on that.

我们会处理这个问题。

We'll work on that.

有时候，我们会惊讶于人们发现哪些功能是有用的。

Mean one of the other ones sometimes we're surprised what people find useful.

我给你举个例子。

I'll give you an example.

一个是Pomelli，这是一个面向中小企业的工具。想象一下，它并不是那种典型的科技创业型中小企业，而是更偏向传统型中小企业，它们希望建立网络存在感。

One is Pomelli which is the it's a tool for SMBs to imagine it is not your typical kind of techie startup SMB, but kind of more kind of traditional SMB where they want to build a web presence.

因此，作为中小企业，你可以直接与Pomelli互动，以极具创意的方式构建网络存在感。

And so you can literally engage with Pomelli and as an SMB and be able to build literally a web presence in incredibly imaginative ways.

所以，我们实验室里总是有这么多项目在进行中。

So we always have all these things cooking in labs.

AI Studio是另一个例子，这类工具是为开发者设计的。

AI Studio is another example of the kinds of things this is for developers.

所以我们正在思考所有这些了不起的创作者，无论是开发者、艺术家、电影制作人还是音乐人，他们都在创造这些令人惊叹的AI优先工具。

So we're trying to think of all these incredible creators whether they're developers, artists, filmmakers, musicians create these incredible AI first tools.

是的。

Yeah.

我特别想体验两个，我认为它们可能会成为爆款。

There's two that I really want to get access to and I think are potentially gonna be big.

也许是下一个NotebookLM。

Maybe the next NotebookLM.

还有一个是CC，它是谷歌内部一个实验性的生产力代理，看起来很棒。

There's CC which is an experimental productivity agent within Google which looks great.

然后是Disco。

And then Disco.

哦，Disco是

Oh Disco's You

你可以基于链接来构建一个网页应用。

can build a web app based basically based on links.

所以如果你正在考虑周末做点什么，你可以直接打开一堆标签页，然后它会帮你决定该创建什么样的应用。

So if you're like thinking about doing something for the weekend you can just like open a bunch of tabs and then it will figure out what type of app to make for you.

比如生成一个自定义地图，用点标记每个潜在活动，你可以选择你希望身处那个地方的日期，然后它会突出显示届时有哪些活动可用。

So maybe a custom map with dots for each potential event and you can pick the dates that you want to actually be in the place that you're thinking about and then it will sort of highlight what's going to be available then.

所以这一点想对你们两位说。

So this is to both of you.

以前谷歌有一个叫做20%时间的概念。

Back in the day Google had this concept called 20% time.

谷歌员工被允许将20%的工作时间用于与自己主要职责无关的项目。

Google employees were basically empowered to spend 20% of their time on something that's not that wasn't core to their job description.

很多重要的谷歌产品都是从这个机制中诞生的。

And that's where a lot of big Google products came out of.

我认为Gmail就是其中之一。

I think Gmail Yeah.

它是一个20%时间的项目。

Was a 20% project.

所以我想问问你们俩关于这些实验性项目的事。

So I I wanna ask you both about about these experimental projects.

谁来构建这些项目？

Who builds them?

20%时间的制度有没有回来？

And is a version of 20% time back?

或者你是怎么理解的，显然有很多很酷的实验。

Or how does this you know, obviously, a lot of cool experiments.

这些实验在公司内部是怎么发生的？

How is it happening inside the company?

我很乐意先说一下。

Well, I'm happy to start.

我认为这个机制实际上仍然在运行。

So I think the the the effectively that's still alive.

所以我回到实验室去看看。

So I go back to labs.

如果你想想实验室里的那些项目，我会说大约80%都来自实验室团队的成员。

So if you think about the things that are in labs, I would say something like maybe 80% of them came out of people actually in the labs team.

另外20%则来自20%时间的项目。

The other 20% came from 20% stuff.

我给你举个关于这个话题的好例子。

I'll give you a good example on a topic that

20%时间在谷歌内部依然存在。

20% time still lives within Google.

我们鼓励人们提出这些想法。

We encourage people to come up with those things.

我来举个具体的例子，拉达和我都非常关注的一个领域——教育和学习。

So I'll give you a good example that an area that Lada and I care deeply about which is education and learning.

谷歌研究团队的一位成员提出了一个想法：他们原本在做其他项目，但突然想到，如果我们能创造一种方式，让人们按照自己的方式、以任何他们喜欢的方式学习，那会怎样？因为目前根本不可能有工具能以多种方式帮助你学习。

So somebody in Google research came up with the idea that oh they're working on something else but they came out the idea what if we created a way for somebody to learn something their way however they want to learn because it's not possible to get these tools to help you learn in any number of ways.

最终，这个想法演变成了‘按你的方法学习’，这是一个你可以在谷歌实验室找到的实验性产品。

So that eventually became learn your way which is an experimental product you'll find in Google Labs.

这个想法并不是来自实验室的人，而是公司其他部门的某个人提出的。

That was not done by somebody in labs, somebody else in another part of the company had come up with the idea.

所以我们不断从谷歌各个部门收到这些了不起的创意。

So we constantly are getting all of these ideas from across Google about these incredible things.

另一个例子来自谷歌深度思维和谷歌研究团队的CoScientist，这些团队开发了这个工具，帮助科学家进行真正的科学发现。

Another example that actually came out of Google DeepMind and Google Research is CoScientist, which those teams worked on which is a tool for scientists to do actual scientific discovery.

现在你会看到它在实验室中作为测试工具出现，让更多人参与其中，但它并不是在实验室内部构建的。

Now you're gonna see that show up in labs as a way to test it, get other people to work on it but it wasn't as it were built inside labs.

来自公司各个部门的人员提出创意这一理念依然充满活力，由此催生了许多令人兴奋的创新。

The notion of people generating ideas from across the company is very much alive and you get some exciting innovations from that.

莉拉，深度思维的研究人员如果想开发一个实验性产品，是有能力这么做的。

Lila, DeepMind researchers have the ability if they want to build an experimental product to maybe do that.

我认为这实际上是我们的文化的一部分。

I think this is actually just part of our culture.

这真正关乎给予人们探索的机会，同时采取非常跨学科的方法。

And that's really about giving people the chance to explore and also taking a very interdisciplinary approach.

所以这并不仅仅局限于研究人员，这一点非常令人兴奋。

So it's actually not just limited to researchers, which has been quite exciting.

实际上，这是能够整合不同视角，尝试解决真实挑战的能力。

It's actually being able to pull together different perspectives and trying to solve real challenges.

有时甚至会使用AI工具来帮助我们加速工作进程。

And sometimes that's even actually AI tools to help us accelerate how we're working.

我们的法务团队如何加快对研究论文的审核并提供反馈？

How does our legal team make the review of research papers faster and be able to provide feedback?

我们如何为责任团队实现更自动化的红队测试？

How do we do red teaming for more automated red teaming for our responsibility team?

或者我们如何理解古代文献？

Or how do we understand ancient texts?

我们有一个项目，实际上是由一位研究人员决定去探索的。

We have a project that actually one of our researchers decided to, he wanted to explore.

这不仅仅关乎今天的智能，更关乎我们可能还不了解的过去的知识。

It's not just about intelligence today, but what is it about knowledge from the past that we might not know about?

所以他致力于开展一个项目，不仅是为了给泥板断代，还要填补缺失的部分并进行翻译。

So he worked to come up with a project that was not just to be able to date a tablet, but also to fill in missing gaps to translate it.

因此，我们现在有了名为‘埃涅阿斯’的项目，专注于古代文本。

And so we now have Project Aeneas, which is all about ancient texts.

正如詹姆斯所言，谷歌拥有一群非常聪明、充满好奇心的人，以及支持这种探索的文化。

So there are, to James's point, one of the things that we have at Google is really smart, curious people and a culture that supports that exploration.

是的，在这个环节结束之前，谈谈我为什么对它如此感兴趣。

Yeah, as we close this segment, talk a little bit about why I'm so interested in it.

我认为，上个世纪一家公司一旦进入标普500指数，平均能维持67年。

I think the average company last century was on the S and P 500 once they reached for sixty seven years.

而现在，这个时间大约只有15年了。

Now it's like fifteen years right now.

随着这个AI时代的到来，你知道，谷歌已经亲身体验到了这一点。

And as this AI moment happens, you know, it's going to mean Google's seen this firsthand.

对吧？

Right?

事情将会变得更快。

It's going to be things will be moving even faster.

而创意的来源、实验和创建新项目的紧迫性至关重要。

And and where ideas come from, the imperative to experiment and, create new projects.

我认为这对任何公司的长期可持续性都至关重要。

I think that's key to any company's long term sustainability.

所以听到它在谷歌内部是如何运作的，非常有趣。

So it's very interesting to hear how it operates within Google.

我本来想评论一下。

I was going to comment.

我职业生涯中有一段时间在风险投资领域工作。

Spent some of my career in venture capital.

我过去常说，那是最了不起的地方，因为你总会遇到那些怀有雄心壮志、想要实现想法的创业者。

I used to say that that was the most remarkable place to be because you'd have these entrepreneurs with audacious ideas who wanted to build ideas.

而让我在谷歌的经历感到不可思议的是，这已经成为日常文化的一部分，并且在组织的各个部分都在发生。

And I think what's crazy about my experience at Google is this is just part of everyday culture And it happens in all parts of the organization.

我认为它的实现方式非常不同。

I think how it comes to life is quite different.

在谷歌深脑中可能与其他部分大不相同，但关键在于整个组织都在支持这一点。

Might be quite different in Google DeepMind than other parts of Google, but the fact that it's supported across the entire organization.

是的，如果我可以补充一点，亚历克斯，我认为谷歌研究文化中真正独特的一点是——我回想起你最初关于贝尔实验室的问题——这一点在谷歌、深脑和谷歌研究中都存在，那就是我们必须将研究转化为现实。

Yeah, if I could add one other piece on this, Alex, I think one of the things that I think is really quite unique about the research culture at Google, and I'm including back to your original Bell Labs question, between, and this happens in Google, deep mining, Google research is this idea that we've got to go from research to reality.

我认为你经常看到这些源自研究的突破性想法，会迅速转化为实际影响。

And I think what you see a lot of these kind of research or originated breakthrough ideas then very quickly transition into real world impact.

我的意思是，AlphaFold 是一个很好的例子，它是一项了不起的突破，堪称诺贝尔奖级别。

I mean Alpha four is a good example, Which is incredible breakthrough, Nobel Prize worthy and all of that.

但看看此后发生了什么，现在已有三百五十万研究人员在超过190个国家使用它。

But look at what's happened since then, You now have three and a half million researchers accessing it in over 190 countries.

你再看看天气和预测方面的突破性成果，它们现在实际上已经在现实世界中得到应用。

You take some of the breakthroughs in weather and forecasting and prediction, they're now actually being used in the real world.

我们现在进行洪水预测，这原本是一个非常了不起的研究课题，如今已覆盖150个国家，惠及20亿人口。

We now do flood forecasting which is a very incredible kind of research question but now it's covering 150 countries with 2,000,000,000 people.