谁在AI助手战争中胜出？+ 为什么谷歌没有构建Muse？

他们也会说想要消费者市场，但我觉得他们只要能守住企业这块，顺便把OpenAI那些潜在的竞争对手压下去，他们就很满意了，因为正如你所说，OpenAI在这里的布局比Meta更偏向商业。

They'll say they want consumer too, but I think that they'll be happy if they can stay in the- ... in the enterprise lane and sort of snuff out some of the would-be competition from OpenAI in particular, because as you note, OpenAI's play here is much more business-focused than what Meta's going after.

不过，我还是觉得，最终大家都会殊途同归。

Now, I do think, again, they all converge over time.

他们都想通吃一切。

They all want to own it all.

他们都想成为那个唯一的智能助手。

They want to be the one agent.

这里有个悬而未决的问题，这其实可以追溯到几个月前，在所有这些智能体热潮之前，我写过的一篇文章，讲的是“自带AI”这个概念，对吧？

There is this open question, and it sort of dates back to something I wrote several months ago, ahead of all this agentic stuff, which is the notion of bring your own AI, right?

未来的工作环境会不会变成你把自己用的消费级AI带到工作中，或者反过来，你把工作中用的AI带回家用。

Is the workplace environment going to be such that you bring your own consumer AI service into the work, or vice versa, that you bring home the AI that you're already using.

人们会想要两个不同的AI，一个工作用，一个家用吗？

Do people want two different AIs, one for work, one for home?

可能会是这样，也可能最终会有一个AI统领你生活的方方面面，这将是这个领域特别大的一个问号。

That may play out that way, or it may play out that there's the one AI to rule them all of your life, and that's going to be a big question mark for this space in particular.

因为就像你说的，Meta 更直接地瞄准消费者市场，而微软更直接地瞄准企业市场，OpenAI 则有点混合，再加上那些可爱的形象，这就像我在文章里提到的，那种奇怪的二元对立，一边是表面上的样子，一边是他们似乎想让你做的事情，而后者更偏向工作导向。

Because again, as you're noting, while Meta is going more straightforward after consumer and Microsoft goes more straightforward after enterprise, OpenAI sort of has a hybrid thing, again, with these cutesy figures, which is just this weird dichotomy, as I note in the piece, between what that looks like and what it seems like they're trying to get you to do, which is much more work-oriented.

我觉得这反映了整个公司的现状，因为他们显然打造了 ChatGPT 这个面向消费者的巨头，但后来他们不得不踩刹车，意识到“天哪，Anthropic 在商业方面已经超过我们了，估值也超过了”，这是为什么呢？

I just felt like that's indicative of the whole company right now because obviously they built this consumer-facing behemoth in ChatGPT itself, but then they, of course, had to sort of halt the breaks and figure out, like, "Oh my God, Anthropic has passed us in terms of the business side, and then in terms of valuation," and why is that?

也许我们应该关注他们在做什么。

Maybe we should focus on what they're doing.

所以对我来说，这反映了整个公司过去几个月经历的剧烈摇摆，我觉得 Dots 在某种程度上实时展现了这一点。

And so to me, that's sort of indicative of the whiplash the entire company has been under the past several months, and so I feel like Dots is in some ways playing that out in real time.

是的，OpenAI 确实有点卡在两种不同用例之间，而且我觉得，我仍然会说它可能是 AI 领域最好的产品。

Yeah, OpenAI does feel a little bit stuck between the two different use case, and it does feel like, I guess I'd still say it's maybe the best product when it comes to AI.

但它仍然感觉不像是在引领它一直在追随的愿景。

But it still feels like as opposed to setting the vision it has been following to a degree.

你描述这些的方式让我好奇，它们最终会不会融合在一起？

And the way that you describe these, I do wonder, do they all blend in together?

有趣的是，当我谈到计算机使用时，听到你对微软的回答，让我想起微软多年前有一个产品，可以监视你的桌面并帮你完成任务，当时引起了反弹，最后他们……

It's funny, when I was talking about computer use, hearing your answer about Microsoft reminded me that Microsoft years ago had this product that watches your desktop and gets stuff done for you, and there was a backlash there, and they ended up-

对，就是那个Recall。

Right, Recall

……

...

我觉得，那个先放一放。

I think, pausing that.

是的。

Yep.

是的。

Yep.

所以想想看，我们会有这么多不同的玩家参与进来。

So just thinking through, we're going to have all these different players involved.

那你觉得，谁会是最终的赢家，关键区别在哪里？

So what would you say the differentiator is in terms of who's going to win?

会是那个最能巧妙引导你的那个吗？

Is it going to be the one that sort of nudges you the best?

还是说，赢家会是那个更能把事情办成的那个？

Is it going to be the one that's more effective in getting things done versus not?

你觉得这场竞争的界限在哪里，谁会率先冲过终点线？

Where do you think the line of competition is drawn here, and who crosses the finish line?

我确实觉得，这类产品会有巨大的先发优势。

I do feel like there will be a huge first-mover advantage to a lot of these.

谁能吸引最多的人注册并真正使用这些产品，谁就能赢。显然，Meta在这方面有巨大的优势，因为他们拥有数十亿用户，横跨多个产品。

Whoever can get sort of the most people signed up and actually using these products, and obviously Meta has a huge advantage there because they have the multi-billion plus user base of several products.

而OpenAI有ChatGPT，他们从一开始就把这个功能融入了ChatGPT这个超级应用里，所以他们也不是没有底牌。

And OpenAI has ChatGPT, which of course they baked this into the ChatGPT super app from the get-go, and so it's not like they have nothing.

他们有十亿用户，Sam Altman在Dots发布会上说，他们现在每周活跃用户达到了12亿。

They have a billion, Sam Altman said during the Dots unveiling, they're now at 1.2 billion weekly active users.

所以这可不是小数目，但跟Meta那三十多亿用户相比，还是差一些，Meta能把这个功能推广到那么多不同的服务里。

So it's nothing to sneeze at, but it's not quite Meta at the three billion plus user scale that they can get this in front of so many different services.

而且OpenAI只允许Pro订阅用户使用这个功能——

And OpenAI will only let you use it if you're part of the Pro subscription tier-

确实如此。

That's true

……才能使用Dots。

... to be able to use Dots.

说得好。

Good point.

所以每月100美元。

So $100 a month.

我得说，我刚刚从每月20美元升级到100美元，就为了能用Dots。

And I have to say, I just upgraded from 20 a month to 100 a month to be able to use Dots.

我真的很喜欢Dots。

I really like Dots.

我们在周五的节目里稍微聊过这个。

We spoke about this a little bit on the Friday show.

我觉得它很棒，而且我知道你写文章说有多喜欢Muse时，也收到了这样的反馈。

I think it's great, and I know you got this feedback when you wrote about how much you like Muse.

我就是不想让Meta访问我的Gmail和日历，虽然我也说不上完全放心把数据交给OpenAI，但连接ChatGPT的时候，我至少没那么反感。

I just don't have any interest in letting Meta have access to Gmail and Calendar, where for whatever reason, I wouldn't say I feel comfortable giving it to OpenAI, but I feel, I guess, a little bit less of a gag reflex when I connect it to ChatGPT.

我跟你说，我收到了好多类似的反馈。

And I can't tell you how many messages I got along those lines.

现在——

Now-

……

...

我得提醒自己，我们做这行就是靠这个吃饭的。

I want to be mindful that we do what we do for a living.

不管住在哪里，我们都算是待在科技圈的小圈子里。

We're a bit in a tech bubble, regardless of where we live and whatnot.

显然，我们的听众比普通大众更偏向这个方向。

Obviously, the audience is going to skew more in that direction than, say, the general public likely.

但这确实是个真实存在的问题，对吧？

But it is a real thing, right?

确实有个真实的担忧，就是Meta不值得信任，我觉得我们马上就能在好莱坞看到这一幕上演，等那些电影上映的时候。

There is this real concern that Meta is just not to be trusted, and I think we're about to see it play out in Hollywood in real time when all these movies come out.

对。

Yes.

所以……

And so...

也许这就是决定谁赢的关键。

Maybe this is who determines who wins.

我们有一部OpenAI的电影和一部Meta的电影要上映。

We have an OpenAI movie and a Meta movie coming out.

而且——

And-

哪一部能把山姆或者马克·扎克伯格刻画得更像个反派，那家公司的个人助理项目就会更难起步。

And whichever one does a better job at portraying Sam or Mark Zuckerberg as, I don't know, a vile character, they will face a harder time getting their personal assistant off the ground.

我也不知道。

I don't know.

就这么简单吗？

Is it as simple as that?

我觉得，这部叫《社交清算》的新Meta电影，算是《社交网络》的非官方续集，应该这周晚些时候上映。

And I think, so this Social Reckoning, which is the new Meta movie, sort of the unofficial sequel to The Social Network, is coming out, I think later this week.

然后我觉得那部叫《人工智能》的OpenAI电影，今晚应该会首映。

And then I think Artificial, which is the OpenAI movie, I think is having its premiere tonight.

所以我们很快就能知道了。

So we're going to find out soon.

但我听到的消息是，我认识一些好莱坞的人，他们都说扎克伯格和奥特曼都被描绘得极其负面。

But everything I've heard, I have some Hollywood folks I know, and They're both portrayed insanely negative, both Zuckerberg and Altman.

所以我不知道结果会怎样。

So I don't know how this ends up.

顺便说一句，还有一部关于埃隆·马斯克的纪录片，大概四个小时长，马上就要上映了。

And by the way, there's also an Elon Musk documentary, which is like four hours long, coming out imminently.

而且他显然也有自己的AI服务。

And so he has his own AI service obviously as well.

达里奥的电影什么时候上映？

When's the Dario movie coming out?

好奇的人想知道。

Inquiring minds want to know.

我们需要一部关于Anthropic的电影。

We need the Anthropic movie.

顺便说一句，那一定会有的。

That will happen, by the way.

我觉得你提到了SNF那件事。

I think you noted the SNF thing.

百分之百。

100%.

那一定会发生的。

That will happen.

到时候会有关于有效利他主义的电影，还有毛绒玩具之类的。

There's going to be the Effective Altruism, the movie, and stuffed animals and things.

但是——

But-

EA：舞台剧版。

EA: the theater production.

对，那会很有意思……

But yeah, that will be an interesting...

在你们提到的信任问题背景下，这些电影引发的反应会非常值得关注，尤其是在推动这些智能体被采纳的过程中。

The reaction to those movies will be fascinating in the face of trying to get these adoption of these agents for the trust reasons that you're talking about.

我们虽然是在开玩笑，但这确实会影响公众对这些东西的看法。

We joke about it, but that will have some impact on how these things are perceived in public.

谁会愿意注册Meta那个控制你生活的智能体，谁会愿意注册OpenAI的版本，又有谁敢信任Anthropic那个整天说世界末日的版本呢？

And who wants to sign up for the Meta version of an agent that controls your life, versus who wants to sign up for the OpenAI version, versus should anyone trust the Anthropic version with all the stuff that they're saying about the end of the world?

谁敢信任马斯克的那个版本？

Should anyone dare trust the Elon Musk version of this?

我们是回到微软，因为我们应该信任他们，还是选择我们的老朋友谷歌？

Do we default back to Microsoft, because we should trust them, or do we go with our old friend Google?

我知道你接下来想聊这个。

Which I know you want to talk about in a second.

所以，是的。

And so, yeah.

不过，回到你最初的问题，就是哪些会成为关键因素。

To go back to your initial question, though, of what ends up being the key things.

我确实认为这里存在先发优势，这显然对初创公司不利。

So I do think that there's a first-mover advantage here for sure, which obviously hurts the would-be startups.

有Instinct，还有一个叫Towns的，还有其他几个玩家……

There's Instinct, there's another one called Towns, there's a few other players that are...

我觉得很快还会有更多推出。

I think there's a bunch more that are going to launch pretty soon.

所以他们不幸地很难像我们提到的其他玩家那样快速达到规模，仅仅是因为已有的用户基础。

And so they're unfortunately going to have a hard time sort of reaching scale as quickly as some of the other players that we're mentioning, simply because of obviously the baked-in user install base.

但这里也有理由支持一个纯粹的代理服务。

But there is an argument to be made for a pure-play agentic service here.

再说回那个瑞士策略，你就能避开这些公司之间的内斗。

And again, the Switzerland argument, you stay away from all the infighting between these companies.

也许那里有突破口。

Maybe there's an inroads there.

但我确实认为最终会是那些大玩家之一，而且这也是因为你之前提到的那个原因，就是整合在这里太关键了。

But I do think it's probably going to be one of the big players, and I do think it's for, ultimately, also the reason that you hit on earlier, which is that the integrations are so key here.

你需要邮件整合，日历整合，消息整合。

You need the integrations to mail, you need the integrations to calendar, you need the integrations to messaging.

哪个玩家能把这些都做好，而且重要的是，不被封锁，保持完整，我觉得这将是决定最终结果的关键。

And which one of those players can nail all of that and keep them all intact, importantly, without being blocked and everything, I think will be critical in terms of how this ends up playing out.

关于那部电影，我只想说，萨姆·奥尔特曼这周上了CNBC，他们问了他这个问题。

On the movies, I just want to say, Sam Altman was on CNBC this past week, and they asked him about it.

凯特·鲁尼，她那次采访做得很好，问了他这个问题。

Kate Rooney, who did a great job with that interview, asked him about it.

他说：‘嗯，看起来像部烂片。’

And he goes, "Well, it looks like a terrible movie."

这回应真是太搞笑了。

It's just the funniest response.

所以我没法——

So I can't-

他今天还登上了《名利场》封面，配了张超大的——

He is also on the cover of Vanity Fair with some huge-

没错。

That's right

……专访，我觉得这跟那部电影要上映肯定不是巧合。

... profile today, which I do not think is an accident in face of this movie that's coming out.

哦。

Oh.

对，可能那边确实有公关上的协调安排。

Yeah, there might be some coordination and PR effort on that front.

我觉得他在那次采访里说了些有意思的话，或者可能是政治采访里，他说……等等，他说什么来着？

He did say something interesting in that interview, I think, or this was maybe the political interview, where he was like, he's... Wait, what did he say?

他说他和Anthropic的人——他们……

He said that him and Anthop- They...

OpenAI和Anthropic在AI监管上有根本不同的世界观，而且AI带来的好处值得冒一些风险。

That OpenAI and Anthropic have fundamentally different worldviews on AI regulation, and that the AI benefits justify accepting some risks.

他因为这个被骂惨了。

He's getting killed for that.

然后就没有下文了。

And then there's no follow-up.

这里说的风险到底是什么？

What risks are we talking about here?

我觉得，是的，重要的是……让我们试着模仿他一下。

I think it's, yeah, it's important to... Let's see if we could channel him a little bit.

我想他可能会说，科学和数学领域会有这些突破。

I think what he would probably say is that there's going to be these sort of breakthroughs that will happen in the sciences and maths.

我们已经看到了数学上的突破，比如OpenAI的纳维-斯托克斯方程，而且可能还会有更多，这会帮助我们更好地理解宇宙，活得更久，更高效，减少对工作的依赖，这是一个大问号。

We've already seen the math breakthrough, well, Navier–Stokes from OpenAI, and probably more on the way, and that will help us understand the universe a little bit better, live longer, be more productive, be less tied to work, which is a big question mark.

风险嘛，我也不知道，可能是网络安全吧。

And the risks are, I don't know, cybersecurity probably.

还有什么别的呢？

And what else would it be?

我不清楚。

I don't know.

你觉得他心里想的就是这种成本收益的权衡，还是有别的考虑？

Do you think that's the cost benefit that he's thinking through in his head, or is there something else?

是的。

Yes.

我看到了同样的引述，也有类似的反应。

I saw the same quotes, and had the sort of the same reaction.

我觉得得给他点肯定，我不认为他是在说世界末日是坏处——

I feel like to give him some credit, I don't think that he's suggesting that the end of the world is the downside-

哦，对。

Oh, yeah

……这种情况。

... scenario here.

不，看起来不像。

No, that doesn't seem like it.

我觉得他是在说他想保持现实。

I think that he is saying that he's trying to be realistic.

你可以争论这种回应是否明智，但我确实觉得他是在试图保持现实，承认确实会有负面影响。

And you could argue whether or not that's smart in that type of response, but I do feel like he's trying to be realistic in suggesting that yes, there are going to be downsides.

他以前也说过这个，对吧？

He's talked about it before, right?

说过很多次。

A lot.

显然，工作岗位会减少，学校也会受到影响，生活的各个方面都会受到这种影响，而且不全是正面的。

That obviously job displacements on down, and sort of school displacement, all different sorts of parts of life are going to be impacted by this in ways that are not universally positive.

所以如果他说不是这样，那显然就是在骗人。

And so to suggest otherwise is obviously going to be disingenuous.

但你要是说得这么含糊，那又会继续造成现在围绕AI的各种沟通问题。

But when you leave it that vague, it's going to, again, just lead to the continued messaging problem around everything that's been going on with AI.

好了，M.G.，这个问题我得问问你。

All right, M.G., I got to ask you on this one.

Anthropic在这里扮演什么角色？

Where is Anthropic here?

他们实际上就是……

They were effectively the...

他们引领了这个时刻。

They led this moment.

问得好。

It's a good question.

他们推出了Claude Code和Cowork，这些最初就是“让我把控制权交给AI，看看会发生什么”的工具，而且——

They had Claude Code and Cowork, which were the initial, "Let me give control to an AI and see what happens," and-

人们很快就接受了这些工具，他们因此成了全球最热门的AI公司，因为他们打造了这些工具。

... people took to it, and they became the world's hottest AI company, because they built those tools.

现在也许他们决定不想在这个领域玩了，或者他们跟我们节目上的观点不太一样，认为这不会成为一个受欢迎的产品类别。

And now maybe they've decided that this is not a place they want to play in, or they've decided, sort of in disagreement with us on this show, that this is not going to be a popular product category.

你觉得他们是怎么想的，为什么在这里缺席？

What do you think their perspective is and why they're absent here?

还是说他们其实有参与？

Or are they?

这个问题问得好，原因正如你所说的。

It's a good question for all the reasons that you note.

这也是我一直在心里琢磨的事情。

It's something I've been thinking about in the back of my head as well.

除了提到他们之外，我并没有专门写过，因为他们在这里没有直接的布局。

And I haven't explicitly written about it other than to mention them, that they don't have a straightforward play here.

可能很简单，他们认为Claude Code，尤其是Claude Cowork，随着时间推移增加更多功能后，会自然演变成这个方向。

It might be as simple as they view that basically Claude Code and more specifically Claude Cowork, I guess, will evolve into this as they add more features over time.

但你说得对，他们确实是编程运动中的先行者。

But you're right, they were the first mover with the coding movement.

不过，你也可以说，其实之前也有人给他们展示过一些类似的东西。

Now, you could say that there were others who actually showcased it a little bit to them.

但他们显然进场了，而且正如我们之前聊到的，这让他们公司达到了现在的高度。之后，Cowork、OpenAI，其他所有人也都觉得他们必须对此有所回应。

But they obviously came in, and that's what we were talking about, has risen the company to the levels that they're at right now, and then Cowork, OpenAI, everyone else felt like they had to have an answer for that.

那他们在这里又是什么情况呢？

And so where are they here?

对我来说，这又回到了我之前想到的一点，那时候他们在这个领域还算是弱势方。那就是，尽管他们雇了像迈克·克里格这样的人，但他们一直避免过于面向消费者，总是更注重工作和生产力之类的。

To me, it goes back to what I thought about before when they were still very much the underdog in the space, which is that despite hiring people like Mike Krieger, they've always shied away from being more consumer-oriented and always wanted to be more about work and productivity and whatnot.

所以我觉得他们的回应可能会是：‘听着，我们想做的是帮助我们的用户提高生产力，但我们不一定打算覆盖所有消费者用例。’

And so I feel like their answer to that would probably be like, "Look, we want to do whatever helps our user base be more productive, but we're not necessarily looking to do all of the consumer use cases."

别忘了，他们到现在连一个图像生成产品都没有。

Remember, they still don't even have an image generator product.

他们的语音功能也很有限。

They have minimal voice capabilities.

我确实认为他们得在这方面加强，因为随着时间推移，语音会成为你与这些服务交互的一个更关键的核心部分。

I do think that they'll have to ramp that up more, I think because voice will become a more key integral point of how you interact with these services over time.

但再说一次，他们还没有推出图像产品。

But again, they haven't done an image product.

他们也没有视频产品。

They haven't done a video product.

从某种程度上说，回头看看，你可以说这是明智的，对吧？

And in some ways, looking back, you could say that was smart, right?

他们没有像OpenAI那样被这些事情分散注意力。

They weren't distracted by those things that perhaps OpenAI was.

但我也觉得，图像产品在某个时候可能也是基本配置，对吧？

But I also think an image product is probably table stakes at some point too, right?

因为你在做演示文稿。

Because you're making a presentation.

你会用另一个服务的图像生成模型来配合它生成的演示文稿吗？

Are you going to use another service's image-generating model to go along with the presentation that it's making?

有些事情他们必须做，而且我确实觉得最终他们会被迫更多地涉足这个领域，即使他们认为这更像是面向消费者的个人AI助手模式。

There's stuff that they will have to do, and I do feel like that eventually they will be probably dragged into this more, even if they view it more consumer-facing personal AI assistant mode.

不过，我想他们会说：“看，我们打算从专业AI助手开始，而且我们已经通过Claude Code做了很多工作，在Cowork方面已经有不少积累了。”

I would imagine, though, they'll say, "Look, we're going to start with the professional AI assistants, and we already have a lot of that with Cowork, thanks to all the work that we've done with Claude Code.

“所以我们会慢慢发展到那一步，然后逐渐转型。”

And so we'll get to that down the road, and we'll morph into that."

他们确实做了一个有意思的举动，你可能看到了，我想就是上周的事，他们把Claude和Cowork的输入框合并了。

They did do an interesting thing, which you probably saw, I think it was just last week, which is that they merged the Claude and Cowork text box.

所以不再是两个不同的标签页来回切换，这一点连OpenAI都还没做到。

So they're no longer the two different tabs that you click between, which even OpenAI has not done yet.

所以我觉得他们正在朝着那个领域的终极目标迈进，也就是一个统一的聊天框，什么都能处理。

And so I feel like they're moving towards the holy grail of that space, which is the unified single chat box for anything.

而且我觉得Dots在某些方面也可能指向这个方向。

And I do feel like Dots, in some ways, might be pointing at this as well.

比如，Dots最终会不会成为ChatGPT的全面前端，它在后台利用自己的技术来判断你问的是个人问题还是工作问题，然后自动帮你路由到相应的处理？

Like is Dots actually going to be the eventual total front end of ChatGPT, and it uses its stuff behind the scenes to basically determine what you're asking for, whether it's a personal question or a work-oriented question, and routes it behind the scenes for you?

对，我能理解。

Yeah, I can see it.

Dots 的一个好处是，你可以不断往里面扔东西，它会在后台帮你排序、处理，然后替你执行。

One of the nice things about Dots is you can continue to throw stuff at it, and then it will basically prioritize and work behind the scenes and then execute for you.

那好吧。

So, all right.

我们在节目里其实已经聊了不少这个话题了。

We've kind of danced around it through the show.

我想说说我的看法，我觉得谁会赢下这场竞争，还有我心中的黑马，我也很好奇你那边有没有什么答案。

I will give you my perspective on who I think will win this battle and my dark horse, and I'd be curious if you have an answer on your end.

好的。

Okay.

我觉得 OpenAI 会赢。

I think OpenAI is going to win.

我就是觉得他们已经证明了他们有做产品的能力。

I just think they have shown that they have the product chops.

他们有 12 亿用户，所以他们不像 Instinct 那样，要拼命抵挡 Meta 那种极其激进的推广攻势。

They have 1.2 billion users, so it's not like they're in instinct where they're trying to fend off Meta's incredibly aggressive distribution attempt.

Meta不仅在WhatsApp和Instagram这些自家产品里推这个，还在NFL橄榄球赛期间打广告。

Meta's not only pushing this in its own products like WhatsApp and Instagram, but it's got ads during NFL football.

我在地铁里，看到的全是Muse。

I'm in the subway, all I see is Muse.

所以我觉得OpenAI确实有那12亿用户，这些人是为了用AI来的，更习惯用它的产品，而且我认为Dots会脱颖而出。

So I think OpenAI does have that 1.2 billion users, people who are there to use AI, people who are more comfortable to use its products, and I think Dots is going to be the one that comes out on top.

当然，我的黑马，要是我们之间不提苹果，那就不算数了。

And my dark horse here, of course, wouldn't be a show between us without bringing up Apple.

我觉得苹果，从一开始就能接触到设备，而且已经表现出对AI产品更积极的兴趣，是能实现这一点的黑马，哪怕它可能是最后一个。

I think Apple, which will have device access from the get-go and has shown an interest in being more aggressive on AI products, is the dark horse to pull this off, even if it might be last.

你怎么看？

What do you think?

我觉得我听起来会像《侏罗纪公园》里的那个场景，我就是那个为吸血律师辩护的人。

I feel like I'm going to sound like that one scene in "Jurassic Park" where I'm the one defending the blood-sucking lawyer here.

就像我在为Meta辩护一样。

Where it's like I'm the one defending Meta.

你这是在挺Meta啊。

You're going Meta.

真没想到我居然在替Meta说话。

Can't believe I'm the one defending Meta.

但我觉得还有一个我们还没怎么提到的点，挺有意思的，就是Meta做这一切，背后甚至没有一个他们自己认为的顶尖模型在支撑，对吧？

But I feel like one other element that we haven't really even hit on, which is interesting, is that Meta is doing all of this without what they even consider to be a frontier model behind the scenes powering this, right?

显然，Muse Spark，我记得现在是1.3，可能快1.4了。

Obviously, Muse Spark, I think it's 1.3 now, might be 1.4.

我觉得1.3就挺好的。

I think it's 1.3, is good.

而且现在就是它在驱动这一切，驱动Muse。

And it is powering this right now, powering Muse right now.

但他们说随时都可能推出那个所谓的“西瓜”，就是他们真正的重磅、顶尖版本的Spark模型，或者抱歉，是他们的Muse模型，Spark只是其中一个变体，很快就要来了。

But they say any day now that, quote-unquote, "Watermelon," their actual big time, big frontier version of their Spark models, or sorry, their Muse models, Spark is one variant of it, is coming soon.

所以一旦那个来了，会有什么变化呢？

And so when that happens, what does that change?

这会改变什么吗？

Does that change anything?

按理说，它会好很多，做某些任务时功能也会强很多，但也许他们构建Muse的方式就是让它即使在不那么顶尖的模型上也能运行得很好。

Presumably, it will be much better, much more functional at doing certain tasks, but maybe they built Muse in such a way that it is fine just running on not really good, but not quite frontier.

也许你并不需要它来处理所有这些代理任务。

And maybe you don't need it for all of these agentic tasks.

也许我们正在进入一个世界，在那里，绝对前沿的模型对科学、对所有这些真正困难的发现和大数据处理至关重要。

And maybe we're starting to break into a world where the absolute frontier matters for science and for all of these really hard discovery and big data stuff.

但日常操作可以依赖这些模型，当然，大家最关心的是它们运行成本更低，提供起来也更便宜。

But the day-to-day operations can fall back to these models, which, of course, on top of everyone's mind is that they can run cheaper, they can be offered cheaper.

所以我们就是在这样的世界里运作的。

And so that's the world in which we're playing.

但再说一次，我想指出的是，Meta在没有绝对前沿模型的情况下做到了这一点，而显然OpenAI现在有，Anthropic，正如我们讨论过的，目前没有参与这个领域，但它有那样的模型。

But again, I just want to put that out there that Meta has been able to do this without having the absolute frontier model, whereas obviously OpenAI does right now And Anthropic, as we talked about, is not playing in the space right now, but has that model.

所以我很好奇，当他们推出Watermelon时，会有什么变化，如果有的话。

And so I'm curious to see what, if anything, changes when they launch Watermelon.

他们会不会给Muse推出一些新功能，让人眼前一亮？

Do they launch with new functionality for Muse that just blows people's minds?

也许不会。

Maybe not.

也许不会。

Maybe it's not.

那是Meta即将推出的大模型，我们正在……

That's Meta's big upcoming model that we're-

……等待它。

... waiting for.

然后你提到了苹果。

And then you bring up Apple.

所以，我在WWDC之后写了一篇文章，当时他们正式发布了新的Siri AI，至少是揭晓了，现在它已经面世了。

So again, I wrote a post right after WWDC, after they launched the new Siri AI officially, at least unveiled it, and now it's out there in the world.

我仍然觉得那篇文章的观点站得住脚，就是说苹果可能会在AI领域胜出，而且很可能会默认胜出，因为，再说一次，它已经内置在数十亿台设备中了。

I still feel like that post holds up, where it's like Apple could win, and probably will win, AI by default, just because, again, it's baked into billions of devices automatically.

我确实觉得有一个很有意思的点，我在那篇文章里稍微提到过，而且我知道其他重要的苹果观察者，比如约翰·格鲁伯他们，最近也在聊这个，那就是Siri AI最大的问题，杰森·斯内尔我知道也聊过一点，就是你总是忘了它现在其实很好用，所以你就忘了去用它，对吧？

I do think that there's an interesting element, which I think I hit upon a little bit back in that piece, and I know that other key Apple watchers, like John Gruber and others, have been talking about more recently, which is that the biggest problem with Siri AI, and Jason Snell I know has talked about this a bit, too, is that you just forget that it's actually good now, and so you forget to use it, right?

即使它就在你的手机里，就在那儿。

Even though it's baked into your phone, it's like it's right there.

每个人都有。

Everyone has it.

你不需要下载新应用，但你还是在打开ChatGPT，打开Claude，去问那些你完全可以用内置版Siri轻松搞定的事情，你就是忘了现在可以信任它了。

You don't need to download a new app, but you're still opening up ChatGPT and opening up Claude for queries that you could easily do in this baked-in version of Siri, and you just forget that you can trust it now.

所以从某种程度上说，它和Meta那边的情况正好相反，对吧？

And so in some ways, it has the opposite trust of what we were talking about with Meta, right?

就是苹果这边，你不信任Siri能正常工作，因为它十多年来一直不好用，现在它确实好用了，这当然是好事，但现在这仍然是个问题，因为他们以另一种奇怪的方式存在信任赤字。

It's just with Apple, you don't trust that Siri's going to work because it hasn't for over a decade, and now the fact that it does is good, obviously, but it's still a problem now because they have that trust deficit in another weird way.

所以我确实认为时间会缓解这个问题，尤其是当他们发布新设备的时候。

And so I do think time will alleviate that, and especially as they release new devices.

有传言说他们会推出第一款家庭设备，可能会很大程度上围绕Siri AI来打造。

The rumors are that they're going to have their first home device that could be largely predicated around Siri AI.

听起来他们正在加紧推出新的HomePod，我觉得这确实能解决很多问题。

It sounds like they're working on getting new HomePods out there imminently, and so I do feel like that will help a lot of these problems.

你说得对。

But you're right.

苹果最大的优势，别人都没有的，就是他们这些设备。

The key thing that Apple has, which no one else has, is, again, they're on these devices.

他们天生就内置了对邮件、短信、日历、照片的支持，对吧？

They have inherently baked in support for mail, messaging, your calendar, your pictures, right?

他们能看到你屏幕上的所有内容。

They see everything on your screen.

没错。

Exactly.

就在设备上。

It's on the device.

你可以说：“嘿，Siri，我现在在看什么？”

You can say, "Hey, Siri, what am I looking at right now?"

其他服务做不到这一点，除非你费尽心思去同意所有这些提示，而正如你之前提到的，苹果正打算在未来进一步限制这一点，至少在Mac上是这样。

You can't do that with the other services, unless you move heaven and earth to basically agree to all of these prompts, which, as I think you alluded to earlier, Apple is moving to potentially lock that down more, at least on the Mac, in the future.

他们还没有明确说明具体要做什么，但显然会设置更多警告，即使是Mac——历史上一直对用户开放，你想怎么折腾都行——现在也可能出现这样的提示，比如“嘿，也许你不想让Meta访问你电脑上的所有信息。”

They haven't exactly explicitly stated what they're going to do, but they're clearly going to put up more warnings that even the Mac, which historically has been open to whatever you as a user want to do to it, will potentially now be getting these prompts like, "Hey, maybe you don't want to give Meta access to all of the information on your machine."

我在想，如果他们直接说“也许你不想给Meta这么多权限，或者给马克·扎克伯格”，会不会更有效。

I wonder if it'd be more effective if they would've said maybe you don't want to give Meta all this access, or Mark Zuckerberg.

我觉得也许——

I feel like maybe-

……再说吧，等电影上映后看看。

... again, after the movie, we'll see.

你看过那部电影了吗？

Did you see the movie?

你说什么？

Did you what?

你知道这个人想要什么。

You know what this person wants.

现在在Apple TV+上免费观看。

Now free on Apple TV+.

没错。

Exactly.

这简直就是个大型流媒体广告。

It's just a giant ad for streaming.

得问问你，你买了Fold吗？

Got to ask, do you have the Fold?

你买了Fold还是Duo？

Did you buy the Fold, the Duo?

我还没买。

I have not.

周五开始预售。

It's up for pre-order on Friday.

哦，好的。

Oh, okay.

所以我还没有拿到。

So I don't have it just yet.

太早了。

Too early.

我会预购的。

I will pre-order it.

到时候我会拿到，我觉得大概一周后吧。

I will get it then, I think it's a week later.

所以我忍不住买了一台18 Pro，就为了有个东西——

So I broke down and bought an 18 Pro just to have something-

好的

Okay

……在我等待的时候用。

... while I wait.

好吧。

All right.

我们会在十一月的节目里给你做评测。

We'll get your review in our November show.

好的。

Okay.

好，那我们休息一下，回来再聊这个我一直期待的话题。我其实一直憋着没问，为什么谷歌没有加入这场竞争？

Okay, so let's take a break and come back and talk about this whole conversation I've been looking forward to, but I've kind of been chomping at the bit and barely holding back from asking you why has Google not gotten into this battle?

那我们回来之后再说这个，广告之后见。

So let's do that when we come back right after this.

本期节目由AvePoint赞助播出。

This episode is brought to you by AvePoint.

现在大家都在急着推出AI。

Everyone's racing to roll out AI right now.

副驾驶、聊天机器人、还有干实事的智能体。

Copilots, chatbots, agents doing real work.

但有个部分没人爱提。

But here's the part nobody loves talking about.

所有这些AI都运行在你的数据上，而大多数团队没有统一的方法来查看、保护并证明这些数据处于可控状态。

All that AI runs on your data, and most teams have no single way to see it, secure it, and prove it's under control.

这正是AvePoint所做的。

That's exactly what AvePoint does.

25年来，他们一直是全球要求最严苛的数据背后的可信层，现在已扩展到你的整个AI资产。

For 25 years, they've been the trusted layer beneath the world's most demanding data, now extended across your entire AI estate.

你的数据、你的云，以及代表你行动的智能体。

Your data, your cloud, and the agents acting on your behalf.

这就是超过28,000家组织能够自信部署AI的方式，让创新规模化而不扩大风险。

It's how more than 28,000 organizations deploy AI with confidence, so innovation scales without scaling risk.

AvePoint，AI的统一信任层。

AvePoint, the unifying trust layer for AI.

访问avepoint.com，了解企业如何自信部署AI。

Head to avepoint.com to see how enterprises deploy AI with confidence.

更多信息，请访问avpt.co/bigtechnologypodcast。

Learn more at avpt.co/bigtechnologypodcast.

网址是 avpt.co/bigtechnologypodcast。

That's avpt.co/bigtechnologypodcast.

你知道衣柜里那些你总是不停穿的衣服吗？

You know those pieces in your closet that you find yourself wearing on repeat?

那件百搭的毛衣、那条完美的裤子，或者那件你总是第一个拿起的T恤。

The sweater that goes with everything, the perfect pair of pants, or the tee you somehow always reach for first.

这些衣服让穿衣变得简单，而这正是 Quince 最擅长的。

Those are the pieces that make getting dressed easy, and that's what Quince does best.

Quince 专注于高品质的基础款服饰，采用优质材料，比如100%蒙古羊绒、有机棉和美利奴羊毛。

Quince focuses on high-quality wardrobe staples made with premium materials like 100% Mongolian cashmere, organic cotton, and Merino wool.

这些款式经典、百搭，设计初衷就是成为你日复一日依赖的可靠单品。

The styles are timeless, versatile, and designed to become the dependable pieces you reach for day after day.

我有点迷上我的 Quince 黑色拉链卫衣了。

I am kind of obsessed with my black zip-down sweatshirt from Quince.

它非常适合坐飞机和逛商场的时候穿。

It's great for planes and malls.

天热的时候，我很容易把它系在腰上，然后风一吹，就又穿回去了。

It's easy to tie around my waist whenever it's warm out, and then when the breeze comes in, back on it goes.

在Quince找到你秋季的新宠吧。

Find your next fall favorites at Quince.

下载Quince应用，享受应用专属优惠，或者访问quince.com/bigtech。

Download the Quince app for app-exclusive offers, or go to quince.com/bigtech.

订单免运费，还有365天退货服务。

Get free shipping on your order and 365-day returns.

现在加拿大和英国也有售了。

Now available in Canada and the UK, too.

网址是Q-U-I-N-C-E.com/bigtech。

That's Q-U-I-N-C-E.com/bigtech.

嗨，我是Rick Ruback。

Hi, I'm Rick Ruback.

我是Royce Yudkoff。

And I'm Royce Yudkoff.

我们是哈佛商学院的教授，也是热门播客《Think Big, Buy Small》的联合主持人。

We're professors at Harvard Business School and co-hosts of the chart-topping podcast, Think Big, Buy Small.

如果你想成为企业家，从零开始创业并不是唯一的选择。

If you want to become an entrepreneur, starting a company from scratch isn't your only option.

你可以收购一家成熟且盈利的小企业，以CEO的身份加入，并在现有基础上继续发展。

You can buy an established, profitable small business, step in as CEO, and build on what already works.

这条路被称为收购型创业，我们花了超过15年时间帮助哈佛商学院的学生探索这一路径。

This path is known as acquisition entrepreneurship, and we've spent over 15 years helping Harvard Business School students explore it.

在《Think Big, Buy Small》的新一季中，我们通过与企业家、投资者和行业专家的坦诚对话，将这些经验分享给你。

On the new season of Think Big, Buy Small, we bring those lessons to you through candid conversations with entrepreneurs, investors, and industry experts.

我们一起探讨旅程的每个阶段，从寻找和融资合适的业务，到完成交易、领导公司，再到最终出售。

Together, we explore every stage of the journey, from finding and financing the right business, to closing the deal, leading the company, and eventually selling it.

无论你是积极考虑拥有企业，还是只是对创业的另一条路径感到好奇，《Think Big, Buy Small》都能帮你迈出第一步。

Whether you're actively considering business ownership or simply curious about a different route to entrepreneurship, Think Big, Buy Small can help you take the first step.

在Apple Podcasts、Spotify或任何你收听播客的平台上关注《Think Big, Buy Small》。

Follow Think Big, Buy Small on Apple Podcasts, Spotify, or wherever you get your podcasts

欢迎回到《大科技》播客，今天我们请到了Spyglass的M.G. Siegler。

And we're back here on "Big Technology" podcast with M.G. Siegler of Spyglass.

你可以在spyglass.org找到Spyglass。

You can find Spyglass at spyglass.org.

强烈推荐你订阅他们的通讯，那是我最喜欢的之一。

Highly recommend you sign up for the newsletter, one of my favorites.

好了，M.G.，我们来聊聊谷歌吧，怎么样？

All right, M.G., so let's talk a little bit about Google, shall we?

好，开始吧。

Let's do it.

谷歌在这些产品发布上似乎总是慢半拍，这次也不例外。

Google always seems to be a little bit far behind on these product releases, and that is also the case here.

他们有个叫Spark的AI助手，但显然和其他家的不在一个讨论层面上。

They have this AI assistant called Spark, but certainly not in the conversation with the others.

就像我们在前半段提到的，因为时间不多了，他们还有Gmail、日历和Talent。

And like we mentioned in the first half, or in the beginning, because we only have a couple more minutes left, they have Gmail, they have Calendar, they have Talent.

他们没有Muse，没有Dots，也没有Instinct。

They don't have a Muse, they don't have a Dots, they don't have an Instinct.

这到底是怎么回事？

What is going on there?

现在我要再提一部电影《摇摆者》，你还记得那个场景吗？就是“你是熊。”

And now I'll do another movie callback of "Swingers," if you remember the scene where it's like, "You're the bear.

你是熊，你有这些爪子，你拥有一切。

You're the bear and you got these claws, and you got everything.

你只需要用这些爪子，兄弟。”

You just got to use these claws, man."

那就是谷歌。

That is Google.

它就是那只熊。

It's the bear.

不过没有爪子。

No claws, though.

怎么回事？

What's going on?

对，它们被拔了牙。

Yeah, they're defanged.

问题是，而且我像往常一样先声明一下，我在那里工作了十多年，但到现在已经好几年没去了，而且我也不是专门为公司做AI的。

The problem is, and again, I should disclose like I normally do, I worked there for over a decade, but I haven't been there in a couple of years at this point, and nor did I work on AI specifically for the company.

所以除了我读到的和思考的，我并没有真正的内部消息，而这些就是我们过去几年一直在聊的。

So I don't have any real insider insight into this beyond everything that I read and think about, and that is what we've talked about over the past couple of years.

他们落后了。

They were behind.

感觉他们费了九牛二虎之力想追上来，我觉得他们确实追了一段时间。

They moved heaven and earth, it felt like, to sort of catch back up, and I think that they did that for a little bit.

然后你看，好像又落后了，接着AI那边，或者说DeepMind那边，公司又进行了大调整。

And then lo and behold, it seemingly fell behind again, and then you had the new shakeup of the company on the AI side, or the DeepMind side.

这些都有关系吗？

And is this all related?

我觉得，从某种程度上说，确实有一些方面是有关联的。

I mean, to some extent, for sure, there are elements of it that are definitely related.

在我看来，我也写过一篇文章，这正好说明，几个月前的I/O大会上，他们上台后其实没什么真正准备好的东西。

In my head, and I wrote a piece about it, it just goes to the idea that clearly, this past I/O a few months ago, they go out on stage and they just don't have anything really ready to go.

他们有一些零零散散的小产品，也愿意谈论未来要推出的东西，但他们因为提前宣布这些产品而坑了自己，包括著名的3.5 Pro，他们即将推出的旗舰AI模型，结果一直没发布，现在看起来永远不会发布了。

They have some little products here and there, and they're willing to talk about these things that are coming in the future, but they sort of screwed themselves by pre-announcing these things, including famously 3.5 Pro, their forthcoming flagship AI model, which never came, and now seems like it's never going to come.

我觉得他们现在其实已经承认了，因为他们已经发布了第一个版本的Gemini 4，用于安全测试等等。

I think they've actually acknowledged that now because they have their first version of Gemini 4 out in the wild for security testing and whatnot.

所以我觉得，这部分可能也是因为他们在I/O大会上还提到了一款产品，也叫Spark，挺让人困惑的，但别和Meta的Spark搞混了，这是另一个AI Spark，叫Gemini Spark，是他们针对智能体和助手场景的版本，他们进行了beta测试，但范围非常有限。

And so I think that part of it was probably they also at I/O talked about a product which is also confusingly called Spark, but not to be confused with Meta's Spark, but it's another AI Spark, Gemini Spark, which is their version of agentic and assistant use cases, and they had it in beta testing, but it was very limited.

我记得一开始好像只有Ultra级别的用户才能用。

It was only, I think, at the Ultra tier at one point.

我觉得现在已经开放了。

I think it's come down.

我试过。

I've tried it.

我已经有一段时间能访问它了。

I've had access to it for a little while.

它还行，但不是什么让人惊艳到掉下巴的AI。

It's fine, but it's not blow away, blow your socks off AI.

它某些方面做得不错，因为就像你说的，他们跟Gmail、Google日历、还有Google搜索等等都有集成。

It does certain things well because, as you note, they have the integrations to Gmail, to Google Calendar, to everything else, to Google Search.

他们本该是能做到的。

They should be the ones.

他们手里有这些资源。

They have the claws.

他们本该是能做到的，但就是抓不住时机。

They should be the ones to do this, and yet they cannot seemingly time it right.

他们总是自己给自己使绊子。

They can't get out of their own way.

我一直觉得谷歌在产品方面有问题，他们很多事情的实现方式都不太对劲。

I've always had problems with Google from a product perspective, the way that they implement a lot of things.

当然，很难反驳那些拥有超过十亿用户的产品，但我还是会对其中很多产品提出质疑。

Obviously, it's hard to argue with multiple products that have over a billion users, but I would still argue with a lot of them.

我觉得它们在功能和用户界面的直观性方面都可以做得更好，而且我认为在AI个人助手这个领域，这些在早期阶段非常关键。

I think that they could be a lot better, in terms of functionality and in terms of user interface, intuitiveness, and I feel like it's pretty key in these early days for a product like this in the AI personal assistant space.

再说，Gemini Spark对我来说没什么吸引力。

And again, Gemini Spark, it's not doing much for me.

我觉得对很多人来说也没什么吸引力，尤其是当你把它和Muse、现在的Dots、Instinct以及其他所有同类产品相比时。

I think it's not doing much for a lot of people, especially when you compare it to Muse, and now Dots, and Instinct, and everything else that's out there.

对，你说得对。

Yeah, you're right.

Muse从头像开始就充满了趣味和奇思妙想。

Muse is full of fun and whimsy from the avatar on down.

Gemini Spark缺少那种……我在找什么词来着？

Gemini Spark lacks that, what's the word I'm looking for here?

是的，有点直白，但就是那个意思。

Yes, it's a little on the nose, but it's right there.

不过这就是谷歌的风格。

It's like that's Google, though.

我还是得反复强调这一点。

Again, I'll keep reiterating it.

他们手里有资源。

They have the claws.

他们什么都不缺。

They have everything they need.

他们肯定有人才。

They have the talent, for sure.

收购DeepMind将会是史上最成功的收购之一，他们占尽了天时地利。

The DeepMind purchase will be one of the greatest acquisitions of all times, and they have everything that they needed to be in the right place, right time.

即使他们落后了，感觉他们还是做对了所有事情，能够迎头赶上。

And even when they fell behind, it felt like they, again, did everything right to come back and catch up and be there.

还记得我们之前讨论过，是不是已经结束了？

Remember we were having the discussion, is it over?

谷歌是不是要一骑绝尘了？

Is Google going to run away with the game?

然后突然之间，我们又回到了相反的讨论，就像——

And then all of a sudden, now we're having the opposite discussion again, and it's just like-

感觉自从那次对话之后，他们什么都没做。

It's like they've done nothing since that conversation.

就像《土拨鼠之日》一样。

It's like Groundhog Day.

太疯狂了。

It's crazy.

他们做这些事都有理由，而且现在他们甚至会在公开场合说一些，因为他们不得不这么做。

And they have reasons for all of it, and they'll say some of it externally even now because they sort of have to.

我记得桑达尔说过，"是的，我们有点……"

I think Sundar said, "Yeah, we sort of..."

我不知道他有没有说"在编程上掉链子了"，但他确实说他们之前没有像应该做的那样专注于此。

I don't know if he said, "Dropped the ball on coding," but he said that they weren't focused on it as they should've been appropriately.

所以那显然在一段时间内是个问题。

And then so that was obviously a problem for a while.

然后呢，感觉又有了类似的想法。

And then again, it feels like there was a similar idea.

他们就是有太多事情在同时进行。

They just have way too much going on.

对我来说，问题总是归结到这一点。

This is what it always comes back to for me.

他们就是人多手杂。

They have too many cooks in the kitchen.

他们那里有太多人，都想做同一件事，但各有各的理由，又不想在同一个团队里合作。

They have way too many people that work there that all want to work on the same thing, but all for different reasons, don't want to work on the same team.

所以这些人各有各的算盘，动机也互相冲突，但并不是说都是恶意的或者小心眼。

So they have all these conflicting cooks, all these conflicting motives, and it's not all nefarious or mean-spirited.

我只是觉得他们有不同的团队。

I just feel like there's different teams.

这是一家庞大的公司。

It's a massive company.

谷歌云团队和安卓团队的激励不同，对吧，而谷歌搜索团队的激励又和他们都不一样。

The Google Cloud team has different incentives, right, than the Android team, and the Google Search team has a different incentive than all of them.

所以很难让所有这些人都齐心协力，达成共识。

And so it's very hard to get all those people to work together to get on the same page.

我虽然这么说，但我原本以为在谷歌大脑和DeepMind合并之后，我们已经解决了这个问题，那本该是个转折点，而且当时看起来也确实在起作用。

And I say all that, but I thought we were over that when we had the unification of Google Brain and DeepMind, and that that was going to be the time, and it felt like it was working.

但突然之间，这些产品又冒出来了。我在那篇文章里提到过，有个叫CC的产品，听上去确实挺有意思的，对吧？

And then all of a sudden, though, these products creep up, and I talked about in that piece, there's a product called CC, which by all accounts sounds interesting, right?

它是为家庭设计的。

It's meant for families.

但当你有了那个产品——

But when you have that product-

这是另一个助手。

It's another assistant

……而且你正试图做一个面向家庭的助手，当你有那个产品，然后你又有Gemini Spark——

... and you're trying to be a family-based assistant, when you have that product and then you have Gemini Spark-

你有CC，你有Spark，而且你还指出搜索最终可能会有AI模式。

You have CC, you have Spark, and you also point out Search might end up with AI mode.

搜索也有自己的。

And Search has its own.

对，所有内置的AI代理功能。

Right, all the AI agentic stuff built into it.

如果你什么都有，人们就什么都不会用，因为他们不知道该用哪个来做这个。

If you have everything, people are going to use nothing because they don't know which one to go to for this.

所以你可以为每一个都辩护说这是正确的决定，或者它们都是好产品，各有各的功能，但它们需要统一起来。

And so you can make an individual case that all these are the right calls, or they're nice products or they do this and they do that, but they need to get on the same page.

而且再次，感觉他们当时正在这么做。

And again, it felt like they were doing that.

你听到所有报道，谢尔盖·布林参与其中，他们真的在集中精力，但不知为何他们就是无法摆脱自己的问题。

You hear all the reports, Sergey Brin involved, that they're really honing in, and they just can't get out of their own way for whatever reason.

我们聊到这儿的时候，我就在想，"好吧。"

And this is sort of the point where we have these conversations and I'm like, "Okay."

我告诉自己，"记住，可能再过一个月，他们又会重回巅峰了，"毕竟事情一直都是这么发展的。

Just like, I'm telling myself, "Remember that it'll probably just be another month and they'll be on top again," the way that this stuff has been moving.

这很可能没错，对吧？

That's probably true, right?

他们最终会推出面向消费者的Gemini 4，然后大概会试着把它和……联系起来。

They'll have Gemini 4 out there for consumers eventually, and then they'll presumably probably try to tie that to...

希望他们会重新命名，但Gemini Spark全面上线，然后他们会说，"嘿，你看到了用Muse和Dots能做的那些酷事。"

Hopefully they'll rebrand it, but the Gemini Spark full-on launch, and they'll go for it and say like, "Hey, you saw that cool stuff you could do with Muse and Dots.

再看看你能用你已经在用的所有谷歌服务做什么。"

Well, look at what you could do with all the Google services that you already use."

基本上就用这种方式来对抗。

And basically combat it that way.

这可能会吸引人一阵子，但现在你又有了另一个。

And it will probably be compelling for a little bit, but then now you have another.

我们又回到了信任这个话题。

We go back to the trust notion.

你会不会担心，如果把自己的整个智能体生活和私人助理都交给谷歌，而他们六个月后又会落后？

You have the trust notion of should I put my entire agentic life and the personal assistant in Google's hands when they're just going to fall behind again in six months?

我知道这是个老生常谈的话题，已经说了很久了。

And I know this is a sort of beaten-down trope that's been going on for a while.

他们真的需要收购一家创意机构，然后在品牌宣传和产品命名上好好整合一下。

They need to just acquire a creative agency and then get this together in terms of their messaging and product names.

因为我不明白为什么这个问题能持续这么久。

Because I don't understand how you have this problem persist for so long.

真是让人匪夷所思。

It's mind-boggling.

说到收购，也许可以回到我们最开始的话题——我预计会有人收购Instinct，有人可能会收购Townes的产品，还有人会收购其他这些产品。

And speaking of acquisitions, maybe to tie it all back into the beginning point is, I would expect that someone will acquire Instinct, someone will acquire maybe the Townes product, someone will acquire all these other things.

基本上是因为他们会觉得，看，我们看到了Meta做了什么。

Basically because they'll feel like, "Look, we're seeing what Meta has done.

我们也看到了OpenAI的成果。

We're seeing what OpenAI has done.

我们得赶紧跟上。

We need to be in this fast."

像微软，我们看看会怎么样。

Like Microsoft, we'll see how that goes.

他们显然已经推出了自己的产品。

They obviously launched theirs.

那亚马逊呢，对吧？

Where's Amazon, right?

亚马逊的Alexa Plus已经推出有一阵子了。

Amazon has had Alexa Plus out there for a while now.

他们在这个领域进展如何？

Where are they in this space?

有传言说他们一直在幕后开发更智能的工作流程，用于这类应用场景。

There's been rumors that they have work behind the scenes on a more agentic workflows for this kind of use cases.

那他们在这方面的进展如何？

And so where are they in this?

那SpaceX呢？现在好像叫SpaceXSI了吧？

What does SpaceX, SpaceXSI now it's called, I guess?

天哪。

Oh my goodness.

别让我说这个。

Don't get me.

我们还有一小时吗？

Do we have another hour?

Spacexy。

Spacexy.

Spacexy。

Spacexy.

我的天哪。

Oh my God.

他们把GrokBot弄到哪儿去了？

Where did they take GrokBot?

因为GrokBot在外面挺受欢迎的，对吧？

Because GrokBot's gotten some love out there, right?

我们都没提到GrokBot。

We didn't even mention GrokBot.

我知道。

I know.

它确实在做一些聪明的事情。

It's doing some smart things.

我觉得他们在团队协作方面做了一些巧妙的事，其他一些公司也不得不应对这个。

I think they're doing some clever things with teams and some of the others are having to combat that.

所以他们做得也不错。

And so they've done good work, too.

对。

So yeah.

这又是一场战斗。

It's yet another battle.

网站是 spyglass.org。

The website's spyglass.org.

M.G. Siegler 每个月第一个星期一都会来我们这里。

M.G. Siegler joins us on the first Monday of every month.

M.G.，和你聊天总是很开心。

M.G., always great to speak with you.

非常感谢你来做客。

Thanks so much for coming on.

谢谢，Alex。

Thanks, Alex.

回头聊。

Talk soon.

好了，各位。

All right, everybody.

非常感谢大家的收听和观看，我们下期再见，这里是《大科技播客》。

Thanks so much for listening and watching, and we'll see you next time here on Big Technology Podcast.