OpenAI的超级应用野心，詹森谈工作，贝索斯的千亿美元自动化基金

嗯，这种情况下的平衡非常微妙，比如你有Palantir的模式，即采用技术并派遣前线工程师，我们的人员会深入其中实施技术。

Well, so this is a very delicate balance in turn in these kind of situations in terms of like, you have the Palantir model of take technology and forward deployed engineer, and our people will be in there implementing technology.

而在这个案例中，你有合作伙伴模式。

You have the partnership model in this case.

我认为，Anthropic传统上更偏向于派遣前线工程师的实施模式。

And I think, like, it's interest Anthropic has been traditionally more kind of the Ford deployed engineer implementation model.

他们上周也刚刚推出了大规模的合作计划。

They've they've also launched large partnership efforts very I think just last week as well.

同样，他们正在以这种方式做出反应，模仿Anthropic的做法。

Again, reacting in these kind of ways, like copying what Anthropic is doing in this case.

我认为，如果OpenAI真正把与合作伙伴的关系放在首位，而不是全力投入构建围绕这一模式的业务，这会增加另一层风险。

I I think, like, giving the relationship to the partner if OpenAI, if this is truly the priority rather than just going all in and being like, we're building a business around that, I think it's another it adds another element of risk here.

但，是的，我认为这是一场赌注，实际上将成为衡量他们能否取得成功的重要标准。

But, yeah, I think it's a bet, and it's actually gonna kind of, like, be a big judge of the type of success they do have.

现在，再谈一个关于我们为何看到这种现象的原因。

Now one more idea about why we're seeing this.

这是一些来自RAMP的疯狂数据，RAMP涉及企业支出，Axios为此写了一篇题为《AI支出逆转》的文章。

This is some crazy stuff that came out of RAMP, which is access to enterprise spending, and Axios wrote it up in a story called the AI spending flip.

这是这篇报道。

Here's the story.

Anthropic现在已占据所有首次购买AI工具的公司支出的73%以上。

Anthropic is now capturing over 73% of all spending among companies buying AI tools for the first time.

73%。

73%.

仅仅十周前，Anthropic与OpenAI的支出比例还是五五开，就在去年12月，OpenAI还以六四领先。

Just ten weeks ago, the split with OpenAI was fifty fifty, and it was sixty forty in OpenAI's favor as recently as December.

这是一次难以置信的逆转，你开始看到Anthropic成为首选。

This is an unbelievable flip where you're starting to see Anthropic be the first choice.

显然，这与需要大语言模型技术的公司所使用的Cloud Code有关，他们已经超越了OpenAI。

Obviously, it's related to Cloud Code among companies who need LM technology, and they've surpassed OpenAI.

我想他们——我肯定OpenAI也掌握了这些数据，这很可能推动了我们在节目前半段所看到和讨论的种种现象。

I guess they I'm sure OpenAI has access to this data and it's probably driving a lot of what we're seeing and talked about in the first half of the show.

我只是觉得有点好笑，因为尽管我短暂地抱怨了一下，但依然让我震惊的是，整个行业竟然对Ramp发布这种数据都习以为常，而我的公司恰恰在使用Ramp。

I was just it's funny because I will take my momentary rant that it still shocks me that everyone in the industry is okay that Ramp does release this kind of data as someone whose company uses Ramp.

对我来说，这依然有点奇怪，不管我花多少钱，这些信息都会以匿名的方式被公之于众。

It's still just kind of weird to me that, like, whatever I'm spending on will be able to be, in an anonymized way, still kind of advertised to the entire world.

但我完全支持Ramp的经济实验室，老兄。

But I'm all about that ramp economics lab, man.

那边的Arrow，数据简直不可思议。

Arrow over there, just unbelievable data.

不。

No.

不。

No.

别扫兴啊，拉詹。

It is Don't, end that party, Ranjan.

对不起。

I'm sorry.

这太惊人了，但还是让我觉得奇怪，为什么大家都对此习以为常。

It's amazing, but it's still kind of weird to me that everyone's okay with it.

但这事儿改天再说吧。

But that's for another day.

我觉得挺有意思，比如说如果OpenAI在看这些数据的话，因为Ramp是一家非常特定的公司。

I think it's funny if op let's say OpenAI is looking at this data because Ramp is a really specific company.

我在一家高速成长的科技企业AI初创公司工作，Ryder。

I work at a high growth technology enterprise AI startup, Ryder.

我们使用Ramp。

We use Ramp.

我认识的很多在前沿科技公司工作的人也都用Ramp。

Many people I know who work at cutting edge technology companies use Ramp.

这是一个很棒的产品。

It's an amazing product.

我有。

I have.

作为一个曾经不得不填报销单的人，我特别喜欢它，那真的让人头疼。

I love it as someone who had to file expense reports in the past, and it was a pain in the ass.

但它的用户群体还是太特定了，所以他们的数据肯定会严重偏向最时髦的新技术，从而显示出增长势头。

It's still such a specific profile, so their data is gonna be heavily skewed in whatever the coolest new so it'll show momentum.

但从展现经济整体实际影响的角度来看，大多数大公司并没有使用Ramp。

But, like, in terms of showing actual, like, aggregate impact in the economy, most large companies are not using Ramp.

比如，我很难想象那些比较传统的企业会用它。

Like, I can't imagine, especially more kind of, like, old fashioned companies.

所以，我不确定这是否属实。

So so I I don't know if it's true.

如果它反映的只是经济整体状况，而不是因为现在酷 kids 更爱用Claude而不是OpenAI，那我会很惊讶。

It would be surprised to me if it's actually like reflective of the economy at large versus the cool kids are using Claude more than OpenAI right now.

是的。

Yeah.

也许它们是一个领先指标。

Well, maybe they're a leading indicator.

我想这就是问题所在。

I guess that's the question.

如今硅谷初创公司使用的东西，会不会成为整个经济未来走向的指标呢？

Is is what a Silicon Valley startup is using today, is that gonna be an indicator of what's gonna happen to the rest of the economy?

你觉得这是一个好的指标吗？还是说这实际上几乎适得其反，因为它们的用户群体和消费习惯完全不同？

Do you think it's a good one, or do you think it's actually almost like counterproductive because it's such a different personality and consumer?

这是个好问题。

That's a good question.

我的意思是，从类别来看，没错，它确实是一个领先指标，但可能不是针对具体供应商。

I mean, think category, yes, it is a leading indicator, but maybe not specific vendor.

我们今年晚些时候就会知道答案。

And we're gonna find out later this year.

我不知道你有没有看过，连摩根士丹利都警告了——根据《财富》杂志的报道，2026年将出现一场人工智能突破，而世界上大多数人都还没做好准备。

I don't know if you've seen this Morgan Stanley, of all entities, has warned, this is according to Fortune, that an AI breakthrough is coming in 2026, and most of the world isn't ready.

一场巨大的人工智能突破将在2026年到来。

A massive AI breakthrough is coming in the 2026.

我想我们已经走了一半了，而摩根士丹利说，世界上大多数人都还没有准备好。

I guess we're halfway through, and Morgan Stanley says most of the world isn't ready for it.

在一份全面的新报告中，这家投资银行警告称，一场变革性的AI飞跃即将来临，其驱动力是美国顶尖AI实验室前所未有的算力积累。

In a sweeping new report, the investment bank warns that a transformative leap in artificial intelligence is imminent, driven by an unprecedented accumulation of compute at America's top AI labs.

主要AI实验室的高管们正在告诉投资者，要做好迎接令他们震惊的进展的准备。

Executives at major AI labs are telling investors to brace for progress that will shock them.

你觉得他们可能看到了什么？

What do you think they could be seeing?

我的意思是，摩根士丹利知道些什么我们不知道的？

I mean, what does Morgan Stanley know that?

还是说他们知道的，其实我们也知道？

That we don't or we do?

嗯，你提到报告中还有一部分指出，研究人员特别引用了最近对埃隆·马斯克的一次采访，他相信将算力增加十倍用于大语言模型训练，可以使模型的智能翻倍。

Well, you did there's also the part that says researchers specifically highlighted a recent interview with Elon Musk citing his belief that applying 10 x to compute LLM training will double the model's intelligence.

所以，这是他们引用的其中一个观点。

So so that's one of their citations.

我是说，我觉得除此之外，我真的不知道。

Like, I think I beyond that, I don't know.

对啊。

Like yeah.

我真的不知道，而我本人其实非常看好人工智能，同时也认为我们需要谨慎思考这项技术能有多智能、发展有多快。

I genuinely don't know, and this is against as someone who is very bullish and some somewhat thinks we need to be thoughtful about how smart and fast this technology can move.

对我来说，摩根士丹利这种充满好奇心缺口的研究报告还是挺奇怪的，为什么不直接说清楚呢？

I'm still it's still funny to me this kind of curiosity gap style research report from a Morgan Stanley versus just say, what is it?

到底会发生什么？

What is it just what's gonna happen?

下个注吧。

Take a bet.

下个注吧。

Take a bet.

做个预测吧。

Give a prediction.

这太奇怪了。

It was very bizarre.

总之，还有另一个有趣的点。

Anyway, there's another interesting note.

在文章底部提到，为了撰写这篇文章，《财富》杂志的记者使用了生成式AI作为研究工具。

On the bottom of the story, says, for this story, Fortune journalists used generative AI as a research tool.

编辑在发布前核实了信息的准确性。

An editor verified the accuracy of the information before publishing.

我觉得这很有趣，但在我看来，这还不是本周科技或新闻界最奇怪的生成式AI使用方式。

I thought that was interesting, but it wasn't to me the sort of weirdest use of generative AI in tech journalism or journalism this week.

我不知道你有没有看到《名利场》关于达里奥·阿马德的文章。

I don't know if you saw the Vanity Fair story about Dario Amade.

这篇文章最初标题为《达里奥·阿马德感冒了》，似乎这位记者获得了对Anthropic公司的一些良好采访机会，但到了文章末尾，他一直在铺垫整个故事，讲述自己如何得以见到Anthropic的首席执行官达里奥·阿马德，然后写了一整篇他与达里奥的长篇访谈，可事后他却说：‘其实我根本没有采访达里奥，也没有向他提出这些尖锐的问题。’

Was initially titled Dario Amade Has a Cold and it was it seemed like this reporter got some decent access into Anthropic, and then, like, towards the end of the story, he's been building up the whole story about getting to meet Dario Amade, CEO of Anthropic, and then he writes this whole long interview that he had with Dario, and then afterwards he's like, oh, I actually didn't interview Dario and asked these, like, biting questions to him.

我其实是把达里奥的很多演讲内容上传到Claude中，然后通过它来‘采访’了他，你们刚刚读到的就是这么来的。

I, like, uploaded a lot of Dario's talks into Claude and I interviewed him that way and that's what you've just read.

是的

Yeah.

你会这么做还称之为采访吗？

This was would you do that and call it an interview?

不会。

No.

好吧。

Okay.

我不会考虑这么做。

Not I wouldn't think about doing that.

这太不尊重人了。

It's so disrespectful.

对读者、对你们所报道的公司，尤其是对读者来说。

I To the reader, to the companies you're it's especially to the reader.

我的天，这太糟糕了。

I mean, that's awful.

是的。

Yeah.

我觉得，我的意思是，是的。

I think, I mean, yeah.

这更像是行为艺术之类的东西吗？

It's more was it like performance art or something?

我希望如此。

I hope that.

这是最合理的解释。

That's the best possible explanation.

我的意思是，我觉得你

I mean, I think you

应该，这不仅仅是一次采访。

should It wasn't know just the interview.

并不是说，这个人似乎真的做了实地报道，然后只是在最后加了一个假的采访。

It wasn't like, this person seems like they actually did real reporting and then just wrote the put the fake interview at the end.

真是一个令人费解的情况。

Truly a puzzling situation.

我的意思是，我们本来没打算谈这个，但关于什么是真实的、什么不是真实的，我相信你这周一直关注着老本·内塔尼亚胡和这些视频。

I mean, like, I guess we we hadn't planned on talking about it, but in terms of what is real and what's not anymore, I'm sure you've been following good old b b Netanyahu and these videos this week.

我有关注。

I have.

是的。

Yeah.

那你就说吧。

So the go ahead.

你可以先介绍一下。

You can introduce it.

我的意思是，我几乎可以想象每个听众都遇到过这种泛滥的猜测，即本雅明·内塔尼亚胡已经去世，而他一直在发布自己在咖啡馆的AI视频。

I mean, I almost have to imagine every listener would have crossed paths with the rampant speculations that Benjamin Netanyahu is deceased and is been has been putting out AI videos of himself at a coffee shop.

甚至今天还举行了记者会，但仍然有无尽的猜测认为那是AI生成的。

And then even today, there was a a press conference, but still endless speculation that it was AI generated.

而且说实话，这个达里奥的采访，还有那些关于内塔尼亚胡的AI视频，真的让我对AI和视频技术的影响感到前所未有的恐惧——人们竟然开始怀疑什么是真实的。想想看，现在世界领导人居然通过发布一些奇怪的咖啡店视频来代替亲自现身，和一大群人一起露面，这简直荒谬至极。

And, honestly, like, this like, again, this kinda, like, Dario interview and then it was AI, like, the Netanyahu stuff is actually, I think, the most scared I have been around the impact of, like, AI and video and people trying to understand what is true that, like, just how absurd and crazy it is that if we're actually living in an era where world leaders somehow and are they is he trying to troll us by putting out a coffee weird coffee shop video rather than just showing up live with, like, a bunch of people?

但没错，这周这件事真的让我很不安。

But, yeah, this one has got me pretty rattled this week.

我不知道。

I don't know.

我觉得你太沉迷于Blue Sky了。

I think you're on Blue Sky too much.

我看过那些视频。

I I saw those videos.

我对它们的真实性没有任何疑问。

I didn't have any question about the veracity of them.

也许我太笨了，不，不是。

Maybe I'm dumb, No.

但是

But

不。

No.

不。

No.

不。

No.

这是x。

This is x.

这确实是，我觉得你对x的两方面都了解了。

This is this is I feel you're getting both sides on x.

是的。

Yeah.

没错。

That's true.

我觉得这是所有事情中一个共同点

I feel like this is one of the unifying things on all

这就是我们的马蹄理论，热爱AI生成的内塔尼亚胡。

This is our horseshoe theory loves AI generated Netanyahu.

没错。

That's true.

这简直就是它的典范。

It's like the epitome of it.

是的。

Yeah.

好的。

Alright.

我们得在事情彻底失控前先休息一下。

We we need let's let's go to break before this really goes off the rails.

但如果你对这个政治故事感兴趣，各位，参议员马克·沃纳周三将做客节目。

But if you're interested in the political story, folks, Senator Mark Warner is gonna be on the show on Wednesday.

我们会讨论AI导致的失业问题，讨论Anthropic和五角大楼的事情，当然，我最感兴趣的话题之一——为什么国会议员们持续进行看似内幕交易的行为，而且他们就是停不下来。

We're gonna talk about AI job loss, we're gonna talk about the Anthropic and the Pentagon thing, and yes, we'll talk about, one of my favorite topics, which is why do members of Congress continue to conduct seemingly insider trading and they they will not stop.

他们停不下来。

They can't stop.

星期三，我将和沃纳参议员讨论这个问题。

And senator Warner and I will discuss that next week on Wednesday.

好的。

Alright.

拉詹和我将在广告后回来，讨论黄仁勋关于AI裁员的评论，稍微谈一点元宇宙，当然，还有一个非常有趣的话题——谁

Ranjan and I will be back right after this to talk about Jensen Huang's comments on AI layoffs, a little bit about the Metaverse, and then, of course, it's really interesting Who

你刚才说谁会来节目讨论内幕交易和

did you did you say is gonna be on the show for discussing insider trading and

马克·沃纳。

Mark Warner.

天啊。

Jesus.

你请来的嘉宾真是太棒了，亚历克斯，从不提前通知，就在我们即将休息前随意抛出这些名字。

The guests you get, I love for listeners, I love that Alex, with no heads up, just casually drops these names right before we go to break.

抱歉。

Sorry.

好的。

Okay.

我我得说，那真是

I I have to That was

一个赞许。

a compliment.

这是个赞许。

That a compliment.

是的。

Yeah.

好吧。

Alright.

我想我们就保留这段吧，拉詹。

I guess I guess we're leaving this in, Ranjan.

我只是在为第二件事做我的陈述。

I was just making my my pitch for the second.

好的。

Alright.

广告后我们将回来，继续讨论詹森·黄关于人工智能、詹森·黄对人工智能裁员的评论、元宇宙，以及杰夫·贝佐斯的基金。

We'll be back after this with a conversation about Jensen Huang's AI, Jensen Huang's comments on AI layoffs, the Metaverse, and then Jeff Bezos' fund.

广告后我们马上回来。

We'll be back right after this.

在这档节目中，我采访过许多优秀的科技创始人，其中有一个出人意料的普遍挑战反复出现：找到合适的域名。

I've interviewed a lot of great tech founders on this show, and one surprisingly universal challenge comes up again and again, finding the right domain name.

我在推出大型科技项目时也遇到过这个问题。

It's something I ran into myself when launching big technology.

你想要的名字往往已经被注册了，很容易就想随便选一个算了。

The names you want are often taken, and it's tempting just to settle and move on.

但我最敬重的那些创始人，在基础问题上从不妥协，而你的域名就是其中之一。

But the founders I respect most don't settle on fundamentals, and your name is one of them.

它应该立即传达出你实际构建的东西。

It should immediately signal what you actually built.

这正是我欣赏 dot tech 域名的地方。

That's what I appreciate about dot tech domains.

这再合理不过了。

It just makes sense.

它向世界、你的客户、你的投资者以及任何搜索你的人明确传达：你正在打造科技产品，简洁直接，无需任何修饰。

It tells the world, your customers, your investors, and anyone googling you that you're building technology, clean, direct, and no qualifiers.

我看到越来越多认真的初创公司开始采用这种域名。

And I'm seeing more serious startups leading into it.

Nothing dot tech，one x。

Nothing dot tech, one x.

Tech、aurora.tech、ces.tech、pie.tech，还有更多。

Tech, aurora.tech, ces.tech, pie.tech, and so many more.

如果你正在打造以科技为核心的产品，别将就。

If you're building something tech first, don't settle.

从您选择的任何注册商处注册您的.dot-tech域名，从第一天起就明确您的定位。

Secure your dot tech domain from any registrar of your choice and make your positioning obvious from day one.

开始新事物不仅困难。

Starting something new isn't just hard.

它令人恐惧。

It's terrifying.

你为这件事投入了如此多的努力，却仍不确定它是否能成功，要迈出这一步确实很难。

So much work goes into this thing that you're not entirely sure will work out, and it can be hard to make that leap of faith.

当我开始这个播客时，我不确定是否有人会听。

When I started this podcast, I wasn't sure if anyone would listen.

现在我知道这是正确的选择。

Now I know it was the right choice.

当你有Shopify这样的合作伙伴支持时，也会更有帮助。

It also helps when you have a partner like Shopify on your side to help.

Shopify是全球数百万企业的电商平台，占美国所有电子商务的10%。

Shopify is the commerce platform behind millions of businesses around the world and 10% of all ecommerce in The US.

从Allbirds和Cotopaxi这样的知名品牌，到刚刚起步的新品牌。

From household names like Allbirds and Cotopaxi to brands just getting started.

凭借数百个即用型模板，Shopify帮助你打造一个与品牌风格一致的精美在线商店。

With hundreds of ready to use templates, Shopify helps you build a beautiful online store that matches your brand style.

让你的宣传效果仿佛背后有一个营销团队在支持。

Get the word out like you have a marketing team behind you.

轻松创建电子邮件和社交媒体活动，覆盖你的客户正在浏览的各个平台。

Easily create email and social campaigns wherever your customers are scrolling or scrolling.

是时候用Shopify将那些‘如果’变为现实了。

It's time to turn those what ifs into with Shopify today.

今天就前往shopify.com/bigtech，注册每月1美元的试用。

Sign up for your $1 per month trial today at shopify.com/bigtech.

前往shopify.com/bigtech。

Go to shopify.com/bigtech.

就是shopify.com/bigtech。

That's shopify.com/bigtech.

我们回到《大科技播客》周五版了。

And we're back here on Big Technology Podcast, Friday edition.

是的。

Yes.

正如你在广告前听到的，华纳参议员下周将做客我们的节目。

As you heard before the break, Senator Mark Warner is gonna be on the show next week.

我们即将迎来一系列周三节目，阵容简直太棒了，老实说，这可能是我们长期以来，甚至可能是节目历史上最强的一批嘉宾，敬请期待。

We all we have an amazing, honestly, I think the strongest lineup we've had in a very long time, if not in the show's history coming up on some of these Wednesday shows, so so stay tuned for that.

好的。

Alright.

也许黄仁勋会来参加。

Maybe Jensen Huang will come on.

也许不会。

Maybe not.

他最近确实发表过言论

He certainly was speaking

是同一个人。

with same guy.

是的。

Yeah.

哦，对的。

Oh, yeah.

所以，他确实进行了这次GTC巡回活动，正如我们之前所料，主要是为人工智能辩护。

So this was so he did do this this, tour around GTC similar to what we thought, which was basically making the case for artificial intelligence.

其中一部分是吉姆·克莱默在CNBC上问他，对AI是否会促使企业继续裁员有何看法。

And Jim and part of it was Jim Kramer asking him on CNBC about his about his impression on whether companies will continue to, lay people off with AI.

我认为，詹森很好地表达了我对此的感受。

And and Jensen, I think, captured what I was feeling about this, pretty well.

以下是他的回应。

Here was what he's what he said in response.

对于有想象力的公司，你会做得更多。

For companies with imagination, you'll do more.

对于那些领导层已经毫无创意的公司来说，他们别无他法。

For companies where the leadership is just out of ideas, they have nothing else to do.

他们没有理由去想象超越现状的可能性。

They have no reason to imagine greater than they are.

当他们拥有更多能力时，却并不会去做更多的事。

Then when they have more capability, they don't do more.

这让我对整个AI裁员问题有了自己的看法：是的，如果你没有想象力，你就只会裁员以获取利润。

And this has kinda gone to my my thought on the whole AI layoff thing, which is that, yeah, if you have no imagination, you'll just lay off and take profit.

但如果你确实有想象力，你会用这些工具去做更多的事。

But if you do have imagination, you're gonna do more with these tools.

许多有抱负的首席执行官都想做很多事情，因此每当我听到‘AI将导致大规模失业’这类说法时，我都持一点怀疑态度——就像我们之前听到的那样，你知道的，我们在多个地方都听过类似的说法。

And there's so many things that ambitious CEOs want to do, and that's why I'm a little bit skeptical whenever I hear this AI will cause mass unemployment, you know, sort of line like we heard, you know, I don't know, we've heard from multiple places.

我并不是说政府不应该做好准备，这正是我将与华纳参议员讨论的内容，但我们也应该以更全面、更细致的视角来看待这个问题，就像我们在周三与安德鲁·罗斯·奥金讨论时所做的那样。

I don't it doesn't mean that the government we shouldn't be prepared, that's what I'm gonna speak with senator Warner about, But it also means, like, let's look at this in context and with some nuance, like we tried to do on Wednesday with Andrew Ross Orkin.

你觉得詹森的这句话怎么样？

So what do you think about this line from Jensen?

我觉得这句话说得太到位了。

I think it captures it so well.

我认为詹森正在像我们上周节目里讨论的那样，逐渐成为人工智能领域的正面人物。

I think Jensen is slowly cementing himself, as we talked about on last week's episode, as the good guy of AI.

他有机会成功。

And he he's got a shot.

人们会喜欢他穿皮夹克、喝啤酒、吃炸鸡的样子。

He's people are gonna like the leather jacket, the drinking beer, and the fried chicken.

我觉得这句话说得很好。

I think he's this is a good line.

当你有想象力时，你可以做更多事情，可以用它来尝试新事物，更快地扩展规模，是的。

When you have imagination, you can do more, that you cannot use this to just try new things and, like, scale more quickly and just yeah.

它只是打开了更多可能性。

It just opens up.

我看到一件事，我觉得我开始认同了。

I saw one thing I think I'm starting to buy into.

这意味着SaaS巨头可能会减少，但也可能涌现出更多中型SaaS企业，团队规模更小，但软件的创造和新技术的应用方式会更加多样化。

It's like, it could mean fewer SaaS giants, but it could mean many, many more medium sized SaaS businesses of, like, smaller teams, but still just more software is created, more technology is used in new kinds of ways.

我觉得这绝对是乐观的情景。

Like, I think that's definitely the the optimistic scenario.

我想，就像我们之前看到Block公司的情况一样，我们已经多次见证过，接下来还会谈到传闻中Meta可能裁员的事情。

I think where again, we saw with Block, we've seen a lot of time, and we're gonna get into the rumored the reported potential Meta layoffs.

人们会把这些裁员直接归因于人工智能。

Like, people are gonna be attributing these to AI directly.

人们甚至可能把AI当作借口，就像我们看到杰克·多西那样。

People might even use AI as the excuse like we saw with Jack Dorsey.

但在很多公司里，也许他们本来就没多少事可做，只是招人过多、机构臃肿，即使不裁员也可能原本就不需要削减，而且他们也无法把这些员工重新投入到更有意思的项目中。

But in a lot of these companies, maybe they don't have much to do, they over hired, they were bloated, and they potentially wouldn't need to cut anyway, and they couldn't just reinvest those people into more interesting things.

是的，完全正确。

Yeah, exactly.

我真的相信这里乐观的一面。并不是说不会有任何冲击，但这也意味着，我不会忽视这样一个可能性：他们的某些业务确实可能变糟，所以你必须为此做好准备。

I I really I I'm a believer in the bull case here, Not to say there won't be any disruption, but it also means that, like, I I won't discount the fact that there's a percentage chance that, like, their things will go bad and that's why you have to plan for that.

但我认为詹森真正抓住了重点。

But I think Jensen really captures it.

如果你有想象力，你会做更多。

If you have imagination, you're gonna do more.

如果你没有想象力，你就只会裁员。

If you don't have imagination, you're gonna lay off.

你觉得到夏天的时候，詹森是否已经确立了自己作为AI领域好人形象的地位？

Do you think by the summer is Jensen has he cemented his role as the good guy of AI?

很明显，他正在积极推动。

It's clear he's he's pushing.

我不确定‘好人’这种说法是不是我會使用的表述，但我认为

Well, I don't know if good guy is the the sort of framing I would use, but I I think

他本人。

he's face.

他有机会成为AI领域的史蒂夫·乔布斯，也许他现在已经做到了。

He has a chance to be sort of the Steve Jobs of AI, and maybe he's already there.

但你知道，他可以成为那个 visionary，去解释并说明——就像我们上周说的，总得有人为这项技术辩护，并且要做好，因为民调数据很糟糕。

But, you know, he he can be the visionary that explains and makes the like we were saying last week, someone's gotta make a case for this technology and do it well because the polling numbers are bad.

所以我认为，如果史蒂夫·乔布斯是个伟大的营销者和销售员，那么黄仁勋同样是个伟大的营销者和销售员。

So I think he could be if maybe that Steve Jobs was a great marketer and salesman, Jensen is a great marketer and salesman.

产品不同，但我相信他绝对能胜任这个营销和销售的角色。

Different products, but I think that he can fill that marketer salesman role for sure.

不。

No.

这是个很好的观点。

That's a good point.

好吧。

Okay.

所以也许你并不需要成为一个友善的面孔或好人。

So maybe it's not like you don't have to be even friendly face or good guy.

你只需要努力为这个行业传递乐观的立场。

You just have to try to, like, make the optimist case for the industry.

而且again，我觉得很多时候，即使你试图用乐观的方式来表达，人们依然会觉得：你必须面对它。

And and again, like, I feel there's a lot of times where it's almost like the there's this tone of even when you're trying to couch it as optimism, like, people still it's like, you have to deal with it.

否则，你就显得愚蠢，或者这种信息还是被强加给你，而不是让人真正对可能实现的事情感到兴奋。

Otherwise, you're dumb or you you know, like, it's it's still being shoved down your throat rather than making people want to actually just be excited about what's possible.

而我们确实需要这种态度。

And, we we certainly need that.

让我们看看，因为AI未来发展的潜在影响是……我想你也是那种不会完全相信AI迷思的人，因为如果你全盘接受，途中很可能会遭遇冲击和失望，尤其是当事情没有朝着积极方向发展时。

Let's see because the implications of where this AI thing is going are are you know, I think you're you're also someone who's like, let's not believe, you know, fully in drink the AI Kool Aid on this because you if you go all the way in, you can end up having, you know, shocks and disappointments on the way, you know, if things don't go positively.

对吧？

Right?

嗯，有两种类型的冲击可能发生。

Well, there's two types of shocks you can have.

一种是它没有如宣传的那样发挥作用，这本身就是一种冲击和失望；另一种是它以非常可怕的方式发挥作用，造成大规模的颠覆——正如许多AI圈内人士所讨论的那样，这也是一种独特的冲击。

One is it it doesn't work as advertised, and that's just its own kind of shock and disappointment, or it works just in a very scary way and just causes mass disruption as many in the AI community talk about and its own its own kind of shock.

所以我认为这两种情况都有可能，但确实如此。

So I think it could be either of those, but there yeah.

在这两者之间的狭窄路径上，没有人以任何一种像样的方式阐述过。

That narrow path through the middle of those two, no one has outlined in any kind of, like, decent way.

我的意思是，是的，数据本身就能说明问题。

Well, I mean, yeah, the numbers the numbers tell the story.

对吧？

Right?

那是什么？

What was it?

哦，它们的数值低于AI，比内燃机还低。

Oh, they they were below AI was pulling lower than ICE.

是的。

Yeah.

而且，是的。

And, yeah.

对。

Yeah.

民主党中唯一高于伊朗的。

The only above Iran in the Democratic Party.

是的。

Yeah.

只是现在人们并不太关注这些事情，对吧。

Just not not exactly the things that people are most into, right now.

顺便说一下，我再提一下沃纳的访谈。

By the way, I, you know, I just one more plug on the Warner interview.

我和他讨论过，这种负面民调是否会延迟数据中心的建设。

I do speak with him about whether this negative polling can lead to, delays in data centers.

作为拥有美国最多数据中心的弗吉尼亚州的人，他当然对此了如指掌。

And coming from Virginia, which has the most data centers in The US, certainly knows a thing or two.

所以值得关注一下。

So something to look out for.

好的。

Alright.

Meta 和元宇宙。

Meta and the Metaverse.

我们该不该把这个环节称为薛定谔的元宇宙？

Should we call this segment, is it Schrodinger's Metaverse?

它是活着的，还是已经死了？

Is it alive or is it dead?

这来自 CNBC。

This is from, let's see, CNBC.

Meta 正在关闭 VR 社交平台 Horizon Worlds，并进一步远离元宇宙。

Meta is shutting down VR social platform Horizon Worlds and further pivot away from the Metaverse.

Meta 周二宣布，将关闭 Horizon Worlds——这款曾作为元宇宙战略核心的 Quest VR 头显虚拟现实社交网络。

Meta announced Tuesday that it was shutting down Horizon Worlds, the virtual reality social network, for Quest VR headsets that was once a key piece of the pivot to the metaverse.

顺便说一下，Horizon Worlds 是一个你可以和别人一起消磨时光的 VR 世界。

Horizon Worlds, by the way, was this kind of VR world where you would start hanging out with people.

但根本没人用它。

No one really used it.

也许有些人用过，但没多少人。

Maybe some people did, but not not many.

接下来是来自Mashable的另一则消息。

Then here's the next story from Mashable.

Meta不是，呃，对不起。

Meta isn't oh, I'm sorry.

Engadget。

Engadget.

Meta根本没打算关闭其VR元宇宙。

Meta isn't shutting down its VR metaverse after all.

Meta正在撤回其关闭VR元宇宙版本的计划。

Meta is backtracking on its plans to shut the VR meta, version of its metaverse.

该公司目前计划在可预见的未来继续支持VR端的Horizon Worlds。

The company now plans to support Horizon Worlds in VR for the foreseeable future.

根据Meta首席技术官安德鲁·博斯沃思的说法，我们将继续为VR中的Horizon Worlds提供支持，以回应那些已经联系我们粉丝的需求。

According to Andrew Bosworth, the CTO of Meta, we will keep it working for VR in VR for existing games to support the fans who've reached out.

拉詹，元宇宙是活着的还是已经死了？

Ranjan, is is the Metaverse alive or the Metaverse dead?

我要从薛定谔的元宇宙这个话题转移一下，因为我觉得它已经死了。

I I'm gonna pivot from Schrodinger's Metaverse because it's dead.

对我来说，这并不是一个非死即活的问题。

It's a this isn't a dead or alive question to me.

我想转到一个非常怀旧、略带感伤的角度，回忆一下我曾经看到人们分享的那些荒谬的场景，比如马克·扎克伯格在元宇宙中采访盖尔·金。

I would like to pivot to a very nostalgic kind of, like, wistful look at I I I liked I was seeing people share just those ridiculous like, I think it was Mark Zuckerberg interviewing Gail King in the Metaverse.

还记得那时候吗？

There's like just just remember that time.

我的意思是，我并不

I mean I don't

知道是否

know if

这是否属实，但我看到有人分享说，有人花了数十万美元在元宇宙中买下邻近斯努普·道格的房产。

this is true or not, but I saw someone share that someone paid hundreds of thousands of dollars to live next to Snoop Dogg in the Metaverse.

等等。

Wait.

还记得元宇宙房地产吗？

Is it remember Metaverse real estate?

元宇宙？

Metaverse?

那真是个时代。

What a time.

那真是个了不起的时代。

What a time that was.

这来自《纽约时报》。

This is from the New York Times.

他们提到了其中一些疯狂的举动。

They highlight some of it, some of the madness around it.

迪士尼、Crate and Barrel 等公司迅速任命了首席元宇宙官。

Disney and Crate and Barrel and other companies were quick to appoint chief Metaverse officers.

这是来自麦肯锡的。

And this is from McKinsey.

哦，麦肯锡。

Oh, McKinsey.

凭借到2030年创造高达5万亿美元价值的潜力，元宇宙大到企业无法忽视。

Potential with its potential to generate up to 5,000,000,000,000 in value by 2030, the metaverse is too big for companies to ignore.

所以有个问题。

So question.

你曾经进入过元宇宙吗？

Did you ever go into the Metaverse?

我想我进去过。

I think I did.

我的意思是，你认为什么是元宇宙？你买过Oculus Quest吗？或者至少你试过Vision Pro吧。

Like, what what do we consider the meta did you own an Oculus Quest or I guess you try I mean, you tried the Vision Pro.

我甚至不确定Vision Pro算不算元宇宙。

I don't even know if that counts as Metaverse.

不是。

Like No.

那是空间计算。

That was spatial computing.

不是。

No.

我从未拥有过。

I never owned one.

哦，我确实有一个Oculus。

Oh, I did I did have an Oculus.

你知道，我以前在BuzzFeed工作时，有一个测试用的Oculus，后来因为用得不多，就借给了一位同事。

You know, I had an Oculus when I was, it was a test device when I was at, BuzzFeed back in the day, and then I lent it to a coworker because I wasn't using it very much.

然后疫情来了，我再也没见过那个人。

And then COVID happened, and I never saw that person again.

字幕。

Subtitling.

我当时真希望在疫情期间能拥有一个元宇宙，顺便说一句，它最后几乎成了一场疫情下的幻梦。

I would've I would've loved to have that, you know, a Metaverse with me in COVID, which by the way, it ended up being mostly a COVID fever dream.

我觉得吧，我确实玩过Oculus Quest上的游戏，但没怎么和其他人互动。

Well, I think because so I played, like, games with the Oculus Quest, but I didn't interact with other people also.

更别提什么Horizons世界之类的东西了。

Certainly never like a Horizons world or anything like that.

是的。

Yeah.

没有。

No.

我觉得我从来没真正进入过元宇宙。

I I don't think I ever made it to the Metaverse.

现在回想起来，我有点后悔了。

I'd I kind of am regretting it now.

也许我们该去Facebook市场找找一些Quest设备。

Maybe we should go find ourselves some quests on Facebook marketplace.

很便宜。

Pretty cheap.

去看看现在谁在Horizon世界里闲逛。

And just go see who's hanging out on Horizon's world right now.

如果有听众在，告诉我们一声。

If any listeners are, let us know.

我们去找你。

We'll come find you.

我们一起去玩。

We'll hang out.

如果你是那个要去Facebook市场买Quest、去Metaverse见人的人，他们只会用撬棍敲你脑袋，然后把你的东西抢走

If you're the person that gonna is gonna go meet someone to pick up a Quest on, to go to the Metaverse on Facebook marketplace, they're just gonna hit you in the head with a crowbar and take your

你应该是错的。

You should be wrong.

抱歉。

Sorry.

抱歉。

Sorry.

拉詹。

Ranjan.

礼仪。

Decorum.

不，不。

No, no.

但说正经的，也许元宇宙从来就不仅仅是虚拟现实。

But on a serious note, maybe the metaverse was never just virtual reality.

我觉得这是来自马修·鲍尔的说法。

I think this is from Matthew Ball.

元宇宙被错误地描述为虚拟现实。

The metaverse is misdescribed as virtual reality.

事实上，虚拟现实只是体验元宇宙的一种方式。

In truth, virtual reality is merely a way to experience the metaverse.

说VR就是元宇宙，就像说移动互联网就是一个应用一样。

To say VR is the Metaverse is like saying the mobile Internet is an app.

还要注意，数亿人每天已经在虚拟世界中参与并花费数十亿美元，而他们并未使用VR、AR、MR或XR设备。

Note too that hundreds of millions are already participating in virtual worlds on a daily basis and spending billions of them without VR, AR, MR, or XR devices.

他谈的是Roblox。

He's talking about roll Roblox.

由此推论，VR头显之于元宇宙，就像智能手机之于移动互联网一样，并非等同。

As a corollary to the above, VR headsets aren't the Metaverse anymore than smartphones are the mobile Internet.

所以，也许好吧。

So maybe Okay.

元宇宙，其实你知道吗？

Metaverse Actually, you know what?

这是我的回应。

Here's the comeback.

所以，世界模型，我认为将会像AI模型一样，不再仅仅基于语言，而是基于对物理世界的真正理解，我们接下来要谈谈贝索斯的新基金。

So world models, I think, are gonna be like the you know, like AI models that instead of are being based on language or being based on actually understanding the physical world, and we're gonna get into Bezos' new fund.

我认为他们让元宇宙和现实实验室保持活力，然后慢慢悄悄地转向整个世界模型领域。

I think I think they keep the Metaverse alive and Reality Labs, and then slowly, quietly pivot to the whole world model space.

然后突然间，一切又回来了，扎克伯格从头到尾都是对的。

And then suddenly, it all comes back, and Zuck was right the whole time.

当他最终搞清楚基于AI和大语言模型的广告，并带着世界模型重新归来，骑着他的悬浮板、挥舞着美国国旗，再次获胜时。

As he Anyway figures out AI ad LLM based advertising and comes back with world models and gets on his hoverboard thing with an American flag and just wins again.

是的。

Yeah.

但我顺便说一下，那东西叫箔片。

But I oh, by the way, that thing is called a foil.

我叫它刮削器。

I called it a skimmer.

箔片水母。

Foil hydra.

对。

Yeah.

是的。

Yeah.

对听众们说。

To listeners.

我们的确应该，是的。

Like, we should have Yeah.

就是那一个。

That one.

我们对这个错误表示遗憾。

We we regret the error.

但有趣的是，如果他们真的转向世界模型，谁会是最适合请回来负责的人呢

But the funny thing would be if they do pivot to world models, guess who the perfect person to bring back, to run

那会是莱昂·拉孔。

that would be Leon Lakoun.

那将会是，哦，等等。

That's that would oh, wait.

那不是他的新创业公司吗？

Isn't that his new startup?

我认为很大一部分在于理解世界。

I think it's like Is a big part of it is understanding the world.

是的。

Yes.

对。

Yeah.

对。

Yeah.

世界模型。

World models.

但确实如此。

But it it is.

我觉得这很有趣，因为你看看《堡垒之夜》和《Roblox》，人们仍然在这些游戏的皮肤上花费巨额资金，并在各个地方互动。

I think it is interesting to me because like you take your Fortnites and Robloxes and people are still spending ungodly amounts of money for like skins in those games and interacting everywhere.

这些都是虚拟世界体验。

And those are those are virtual world experiences.

所以，是的。

So so yeah.

元宇宙是活生生的，只是不是戴着虚拟现实头盔、穿着虚拟化身参加工作会议的那种形式——我其实还挺想体验一次的，但一直没机会。

Metaverse is alive, just not not legless avatars while wearing a virtual reality headset and sitting in a meeting because I still wish I did that once but never Yeah.

抽空去做了。

Got around to

不幸的是，我觉得当时我所在的公司还没到那个阶段。

I unfortunately think I was just not working in a company for when the time that came around.

哦，还有一件事我们该提一下，我认为关于元宇宙的这场讨论，如果不谈它所带来的结果，那就是不完整的——也就是，如果我们当初讨论过的消费级AI还有希望，那很可能通过某种设备实现，而Meta在这方面确实领先一步。

Oh, one last thing that we should say, I I think that this conversation about the Metaverse, you know, is incomplete without talking about what it led to, which is like, if there is a hope for consumer AI, which we debated in the beginning, it's probably through some form of device, and Meta certainly has a head start on that.

我的意思是，Ray-Ban Meta智能眼镜，你我都非常喜欢，这些新项目正是源于这场VR布局的直接成果。

I mean, the Ray Ban Metas, which which you and I both really like, and these newer projects are direct results from this VR move.

VR头显依然存在，只是元宇宙的概念已经淡了。

And the VR headsets live on, just the Metaverse know.

抓住这次胜利吧。

Take the win.

我认为，如果他们公开说，这些技术让我们在可穿戴设备和整个AI新领域中占据了先机，没人会质疑这一点。

I think, like, if they just said openly, like, these technologies gave us a head start in wearables in the entire new world of AI.

没人会对此有异议。

No one would question it.

这没错。

It's it's right.

所以你不必直接关闭Horizon世界。

So you don't have to just shut down Horizon's world.

没关系。

It's okay.

你不必假装继续维持它开放。

You don't have to pretend to keep it open.

你不必这么做。

You don't have to.

你们

You guys

不。

No.

我只是在心里对自己说，我觉得他们基本上是在说，你刚才提到的那种版本，就像是，你知道的，Fire Phone 带来了 Echo。

I I was I was just saying I to myself, I think they are basically saying that that version of what you just said, which is like this was it's almost like, you know, like the, fire phone led to the echo

或者

or

什么的。

something

类似这样。

like that.

没错。

Exactly.

没错。

Exactly.

这里有一个英雄故事。

There's a there's a hero story here.

这确实是一个英雄故事。

There's a definite hero story.

既然他们不必再花那么多钱支持它，现在就可以真正投入精力来开发人工智能。

And now that they don't have to spend all that money supporting it, they can actually work to build build AI.

好吧。

So alright.

说到亚马逊，有一个非常有趣的故事，我觉得我们不该忽略，那就是据《华尔街日报》报道，杰夫·贝佐斯正在洽谈筹集1000亿美元用于一个AI制造基金。

Speaking of Amazon, there's this really interesting story that I don't think we should leave without talking about, which is that, Jeff Bezos, according to the Wall Street Journal, is in talks to raise 100,000,000,000 for an AI manufacturing fund.

贝佐斯正在早期洽谈筹集1000亿美元，用于设立一项新基金，收购制造企业，并试图利用人工智能技术加速它们的自动化进程。

Bezos is in early talks to raise 100,000,000,000 for a new fund that would buy up manufacturing companies and seek to use AI technology to accelerate their path to automation.

这位亚马逊创始人正在与全球一些最大的资产管理公司会面，为该项目筹集资金。

The Amazon founder is meeting with some of the world's largest asset managers to raise funding for the project.

几个月前，他曾前往中东，与该地区的主权财富基金代表讨论这项新基金。

A few months ago, he traveled to The Middle East to discuss the new fund with sovereign wealth representatives in the region.

最近，他去了新加坡。

More recently, he went to Singapore.

根据投资者文件中的描述，该基金被称为制造业转型工具，旨在收购芯片制造、国防和航空航天等主要工业领域的公司。

The fund described in investor documents as a manufacturing transformation vehicle as aiming to buy companies in major industrial sectors such as chip making, defense, and aerospace.

它将远远超过世界上一些最大的收购基金的规模，并与软银1000亿美元的科技导向愿景基金相抗衡。

It would dwarf the size of some of the world's, dwarf the size of some of the world's largest buyout funds and rival SoftBank's 100,000,000,000 tech focused Vision Fund.

关于这个，容我分享几点想法。

Couple of thoughts on this, if I may.

第一，贝佐斯为什么要筹集这笔钱？

Number one, why is Bezos raising this money?

他不是已经有钱了吗？

Doesn't he have it?

我不知道，也许你并不总是想用自己钱，但如果你真的如此相信这个想法，为什么不直接用自己的资金投入呢？

I don't know, maybe you don't always want to use your own money, but if you believe that much in the idea, why don't you just put your own in towards it?

第二，这有点令人不安——贝佐斯想要自动化他能接触到的一切，虽然亚马逊在引入机器人后确实增加了仓储中心的员工，但贝佐斯似乎正准备通过这一举措自动化真正的蓝领工作，这确实令人不安。

Number two, it is somewhat horrifying, that Jeff Bezos is going to after well, I guess Amazon did increase its employees in the fulfillment centers after it brought the robots in, but it is somewhat horrifying that Bezos, who wants to automate everything he can lay his hands on, seems like he's ready to automate, you know, real, you know, blue collar jobs with this push.

第三，根据我对贝索斯的了解——当然，这种了解来自于阅读关于他的报道以及与了解他的人交谈——他总是对这些事情有准确的判断，我真的相信他在这里发现了关键：人工智能驱动的制造业变革已经启动，即将迎来重大飞跃，其中蕴含着真正的机遇。

And third of all, you know, knowing knowing Bezos the way I do and that is, of course, reading about him and speaking to people who know him, he just tends to be right about these things all the time, and I I really do believe that he's on to something here, that, manufacturing transformation with AI is is already underway, but is about to make a major major leap and there's real opportunity there.

所以我认为贝索斯的判断非常准确，我也有类似的感受。

So I think Bezos is on the money, and I have a lot of these feelings.

你觉得呢，拉詹？

What do you think, Ranjan?

如果你继续阅读接下来的两段内容，贝索斯最近被任命为新创公司‘普罗米修斯计划’的联合首席执行官，这家公司正在开发能够理解和模拟物理世界的AI模型。

Well, if you continue reading the next two paragraphs on this, Bezos was recently appointed co CEO of Project Prometheus, a new startup that is building artificial intelligence models that can understand and simulate the physical world.

尽管人工智能革命的大部分焦点都集中在大型语言模型上，但数十亿美元的资金已经开始流向那些致力于将空间感知型AI系统应用于机器人和制造业等领域的公司。

While much of the AI revolution has been focused on large language models, billions of dollars have begun to flow to companies that are seeking to apply spatially focused AI systems towards industries including robotics and manufacturing.

所以，是的，我之前在我们的准备材料里根本没看到这一点。

So, yeah, I I had not even seen that in the the our our prep doc here before.

世界模型，这将是下一个大趋势。

World models, that's that's gonna be the next big thing.

而且，再说一遍，我很认同这个观点。

And and, again, I like the point.

贝索斯不仅经常在这些事情上是对的，就像亚马逊对整个仓储领域所做的那样——引入更多人力，但正是自动化水平让我们所有人都沉迷于两天送达甚至一天即时送达。

Not only is Bezos someone who is often right about this kind of thing, again, like what Amazon did to the entire warehouse space, bringing in more people and but still the level of automation is what led to us all getting addicted to two day delivery and one day instant delivery.

他们能够推动的这种技术创新，他真的懂。

Like like the the technological innovation that they were able to push, like he gets it.

他一次又一次地证明了这一点。

He he's shown it time and time again.

所以我认为这绝对是值得我们密切关注的事情。

So I think he will definitely this is this is something we will watch very very closely.

不。

No.

完全同意。

Totally.

我同意。

I agree.

我的意思是，贝索斯知道，有些公司会采用这项技术，有些则不会，我认为他正在下一笔相当大的赌注——当然，用的是别人的钱——但他会大胆押注，并且很可能是对的。

I mean, the thing with Bezos is he knows that there is like there are gonna be companies that will implement this and companies that don't, and I think he's making a pretty sizable bet, of course, with others money, but he's gonna make this big bet and and probably be right.

好了，拉詹，在我们离开之前，我有个问题想问你。

Alright, Ranjan, before we leave, I have a question to ask you.

这有点敏感，但我真的需要问一下。

It's somewhat sensitive, but I just need to ask.

有。

Do

你有干聊过吗？

you dry chat?

我干聊过。

I have dry chatted.

我干聊过。

I have dry chatted.

所以，《华尔街日报》记者梅根·巴布罗夫斯基在推特上说，这似乎是她收到的一个推销方案。

So Wall Street, Journal reporter, Megan Babrowski tweets, what the and this is seemingly a pitch that she got.

她说，四分之一的人承认他们确实这么做了。

She goes, one in four admit they that's she did.

这封邮件来自曼迪。

The email goes from Mandy.

可怜的曼迪，当着大家的面被公开批评。

Poor Mandy getting put on blast in front of everybody.

四分之一的人承认，他们在进行情感上困难的任务前会先闲聊。

One in four admit they try chat before emotionally difficult, tasks.

什么是闲聊？

What is try chatting?

这是推销内容。

Here's the pitch.

嗨，梅根。

Hi, Megan.

在一场艰难的对话前感到紧张。

Jittery before a tough conversation.

你试过Jai聊天吗？

Have you tried jai chatting?

干聊？

Dry chatting?

显然，这是一种趋势，因为超过一半的成年人承认他们难以表达自己的情绪。

Apparently, it's a trend, as over half of adults admit they find it hard to articulate their emotions.

在紧张的对话中，许多人正转向人工智能进行练习。

During tense conversations, many are turning to AI to rehearse.

这就是干聊。

Enter dry chatting.

在现实生活中进行情感挑战性的对话之前，先与人工智能进行练习。

Rehearsing emotionally challenging conversations with AI before having them in real life.

我想我们找到了。

I I guess we found it.

这就是人工智能消费者的使用场景。

This is the use case AI consumer.

就在这里。

Here it is.

消费者的应用场景就是干聊吗？

Is the consumer use case dry chatting.

我看到这个的时候，说实话，'干聊'这个词对我来说就是……我不知道怎么说。

I kind of when I saw this, obviously the term dry chatting is just like I don't know.

它是一种能给你带来某种帮助的东西。

There's it's it's something that gives you the x.

但当你开始深入了解时，你会想：等等。

But but then when you start to see it, you're like, wait.

我得承认，如果你得写一封紧张的邮件，先让AI帮你看看，问问建议，明确你想要表达什么，这其实是个非常好的做法，我觉得大多数人应该都这么做。

This is so I have admit, like, if you have to write a tense email, running it through an AI and asking for some pointers and saying, like, what you want out of it is a pretty good thing I think most people should do.

我承认，我还没直接用过语音干聊。

I I I haven't gone straight for the the voice dry chat, I'll admit.

我没跟ChatGPT、Gemini、Claude或者其他任何AI工具实际演练过完整对话，但也许这值得一试。

Like, I haven't I haven't talked to chat gbt Gemini, Claude, whoever else, and, tried to rehearse the conversation in full, but maybe it's maybe it's worth it.

也许我很快就会试试干聊了。

Maybe I I might try jet dry chatting soon.

这就像，你知道吗？

It's, it's like it you know what?

如果这能帮助你以更友好的方式真正解决问题，难道我们不都应该进行干聊吗？

If it helps you actually resolve the situation in a in a much more amicable way, shouldn't we all be dry chatting?

首先，我要说，我确实用过语音进行干聊。

First of all, I will say I have used voice to dry chat.

你用语音干聊过？

You've you've voiced dry chatting?

我用过。

I have.

在我的一些面试前，我会先列个提纲，想预判一下面试官可能会说什么，所以有时候我会和机器人角色扮演，假装你是那个人。

Some of my some of my interviews, before I like I have like a rundown and, you know, I wanna anticipate what the interviewee is gonna say, so I role play with the bot sometimes and I'm like, you're this person

那后来怎么样了？

What happened to that?

而我就是我。

And I'm me.

所以我要把干聊和角色扮演区分开来。

So I'm gonna separate dry chatting from like role playing.

哦，好吧。

Oh, okay.

排练。

Rehearsing.

因为我觉得干聊特别适用于社交这类场景。

Because this this is specific I feel dry chattingly applies to, like Social.

比如具有挑战性的对话。

Like challenging conversations.

比如你即将解雇某人，或者你要分手，或者必须传达一些非常糟糕的消息。

It could be if you're about to fire someone, if you're, like, like, you know you have to break up, you have to, like, deliver some really bad news.

对我来说，我会继续使用干聊，确保我们都理解清楚。

To me, I'm gonna I'm gonna keep dry chatting, make sure we all understand.

天哪。

Oh my god.

这纯粹是阿谀奉承和干聊，我觉得这两者根本搭不到一块儿。

This is Just the the sycophancy, and the dry chatting, I don't think it goes well together.

好吧。

Like, alright.

没事的。

It's alright.

Chat GPT，你当我的女朋友，我呢就做我自己，我有些消息要告诉你。

Chat GPT, you're gonna be my girlfriend and I'm gonna be me and I have some news to to share with you.

然后Chat GPT会说，好的，说吧。

And Chat GPT, like, okay, go ahead.

而你会说，宝贝，我们分手吧。

And you'll be like, baby, we need a break up.

然后Chat GPT会说，好主意。

And chat GPT will be like, great idea.

它们总是能想出最聪明的点子。

They're always coming up with the smartest ideas.

那么，干巴巴的聊天反而更好吗？

Well, so but so does dry chatting work better?

评估标准是什么？

What's like the eval?

干聊的基准是什么？

What's the dry chat benchmark?

因为如果四四四很讨好，当你真正要传达坏消息时，你会被撕得粉碎，因为那个阿谀奉承的GPT四四说你什么都对。

Because if four o four o sick and then see would then when you go to deliver the actual bad news, you'll get ripped apart because sycophantic GPT four o told you you're right about everything.

也许这就是我们需要制定的新基准。

Maybe that that's my that's the new benchmark we need to, come up with.

这必须是一种强化学习，你奖励那些不会让人在现实生活中被打的对话。

Well, you have it has to be this, like, sort of reinforcement learning where you reward conversations that don't get the person slapped in real life.

所以有一些可怜的AI工程师得来应付我们。

That's the so there's some poor scale AI guys who have to go through us.

我得走了。

Gotta go.

好吧。

Alright.

听好了。

Listen.

你这周的任务是。

You're you're tasked this week.

我准备好了。

I'm ready for it.

你得和你女朋友分手。

You gotta break up with your girlfriend.

为了科学。

For science.

我们要看看她会不会打你。

We gotta see if she's if you're getting slapped.

我跟你说，老兄。

I'm telling you, man.