本集简介
双语字幕
仅展示文本字幕，不包含中文音频；想边听边看，请使用 Bayt 播客 App。
随着市场开始相信人工智能的故事，软件股票大幅下跌。
Software stocks tank as the market starts to believe the AI story.
Anthropic 和 OpenAI 在超级碗展开竞争，而亚马逊的巨额支出让投资者感到恐慌。
Anthropic and OpenAI go at it in the Super Bowl, and Amazon spending freaks out investors.
这将在本周末的大型科技播客节目中紧随其后播出。
That's coming up on a big technology podcast Friday edition right after this.
本集由高通公司赞助。
This episode is brought to you by Qualcomm.
高通正在将智能计算带到每一个角落。
Qualcomm is bringing intelligent computing everywhere.
在每一个技术转折点上，高通都一直是值得信赖的合作伙伴，帮助世界应对最重要的挑战。
At every technological inflection point, Qualcomm has been a trusted partner helping the world tackle its most important challenges.
高通领先的AI技术、高性能低功耗计算以及无与伦比的连接解决方案，有能力构建新生态系统、变革产业，并改善我们体验世界的方式。
Qualcomm's leading edge AI, high performance, low power computing, and unrivaled connectivity solutions have the power to build new ecosystems, transform industries, and improve the way we all experience the world.
你一直在等待升级科技的最佳时机吗？
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，我们为最优秀的客户带来最新笔记本电脑的超值优惠，例如搭载英特尔酷睿Ultra处理器的戴尔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、快速免费配送、高级技术支持、价格保障等。
We've also got incredible perks, like Dell rewards, fast free shipping, premium support, price match guarantee, and more.
在升级电脑的同时，不妨一并添置全套设备，因为我们还为高端显示器和配件提供了大幅优惠。
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.
你可以用惊人的折扣配齐一套全新设备。
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.
欢迎收看《大科技》播客周五版，我们将以一贯冷静而细致的方式解析最新资讯。
Welcome to Big Technology podcast Friday edition where we break down the news in our traditional cool headed and nuanced format.
今天为大家带来一档精彩的节目。
We have a big show for you today.
我们将全面探讨软件领域正在发生的一切。
We're gonna talk all about what's going on with software.
每一家软件公司似乎都在华尔街经历某种崩盘。
Every software company seems to be in the middle of some sort of wipeout on Wall Street.
我们将探究人工智能是否真会如投资者所担忧的那样，冲击软件行业。
We're gonna investigate whether AI is actually going to do what investors fear it will do to software.
我们还会讨论OpenAI和Anthropic之间关于Anthropic超级碗广告的争端，当然，亚马逊的股价也大幅下跌。
We're also gonna talk about this OpenAI and Anthropic fight over Anthropic Super Bowl ad, and then, of course, Amazon fell, significantly.
所以我们会探讨该公司在这方面的状况，以及市场是否判断失误。
So we'll talk about what's going on with the company on that front and whether the market is getting it wrong.
和往常一样，周五加入我们的是Margins的Ranjan Roy。
Joining us as always on Fridays to do it is Ranjan Roy of margins.
Ranjan，欢迎。
Ranjan, welcome.
我刚离开你们一周，你们就把整个软件行业给灭了。
I leave you guys alone for a week, and you just kill off the entire software industry.
上周没见到你们，我挺想你们的。
I missed you last week.
我真的很期待深入探讨这个话题，因为这正是我过去六七个月一直在说的事情，现在终于发生了。
This is I mean, I'm excited to get into this because this is everything I've been talking about now for like six seven months it's happening.
作为备注，我们的关注点已经从担心AI是个泡沫，转变为如今市场普遍认为AI真的可能成功，并 potentially 摧毁软件行业。
Okay, so as a note we've gone from worries that AI was a bubble to now at least in the market now the market is reflecting the chance that AI is really gonna work and potentially wipe out software.
这是来自彭博社的报道。
This is from Bloomberg.
万亿美元科技市值蒸发，波及人工智能领域所有股票。
Trillion dollar tech wipeout ensnares all stocks in AI's path.
自ChatGPT进入主流以来的三年里，已发生过多次由人工智能驱动的抛售潮。
There have been many AI driven sell offs in the three years since ChatGPT burst into the mainstream.
但本周在股票和信贷市场掀起的震荡，却前所未有。
Nothing though quite rivals the route rippling through stock and credit markets this week.
在短短两天内，硅谷大小公司的股票、债券和贷款价值合计蒸发了数千亿美元。
In the span of two days, hundreds of billions of dollars were wiped off the value of stocks, bonds, and loans of companies big and small across Silicon Valley.
软件股票处于风暴中心，跌幅如此之大，以至于追踪这些股票的iSharesETF在过去七天内市值已下跌近一万亿美元。
Software stocks were at the epicenter plunging so much that the values of those tracked, and an iShares ETF has now dropped almost 1,000,000,000,000 over the past seven days.
此外，在过去四周内，彭博社指数中超过177亿美元的美国科技公司贷款，已跌至濒临违约或已违约的交易水平。
Also, more than 17,700,000,000.0 of US tech company loans in a Bloomberg index dropped to a distressing to distressed trade trading levels during the past four weeks.
与以往多次抛售不同，此次暴跌并非源于对泡沫的担忧，而是市场担心人工智能即将取代大量公司的商业模式——而这些公司长期以来一直被悲观者视为高风险对象。
The drubbing, unlike many previous ones, was triggered not by fears of a bubble, but rather concerned that AI is on the verge of supplanting the business models of a wide swath of companies that doomsayers have long predicted were at risk.
我不太明白，拉詹，你能解释一下吗？为什么突然间市场开始相信人工智能会成功，并且会取代软件？
I don't understand, Rajan, can you explain a little bit why all of a sudden the market's starting to believe that AI is going to work and it's going to wipe out software?
这里到底发生了什么？
What's going on here?
好的。
Okay.
这完全符合我从九月份开始一直在谈论的内容。
So this is, like, square in line with what I've been talking about for, I think, September, we started talking about it.
again，这个关于自主知识工作的想法，即传统SaaS产品所做的事情，现在人工智能完全可以做到，并彻底绕过SaaS。
Again, this idea of autonomous knowledge work, the idea that the work that software traditional SaaS products did, actually, now AI can do and just completely disintermediate SaaS.
比如，拿Salesforce来说。
So, like, you take a Salesforce, for example.
Salesforce的价值在哪里？
Where is the value in a Salesforce?
它只是一个数据库吗？还是说，价值在于那个用户界面，让人们能够查找信息、跟踪信息，或者查看仪表板来了解他们需要做什么？
It's a database, but is it just that user interface that allows people to kind of find the information or track the information or look at a dashboard that tells them what they need to do?
AI 非常擅长构建这一整层，并为用户实现完全定制化。
AI is really good at creating that entire layer and making it completely customized to the user.
比如你看 Adobe。
Like, you take an Adobe.
我的意思是，他们现在做的很多工作，无论是 ChatGPT 还是 Gemini，都在取代过去 Photoshop 或 InDesign 的功能。
I mean, so much of the work that they're doing now, whether it's ChatGPT or Gemini on the what Photoshop used to do or an InDesign used to do.
所有这些类型的工作，正是 AI 最擅长取代的部分。
Like, all those kind of pieces of work are exactly what AI is really good at replacing.
到了这个时候，传统软件的很大一部分就可以被看作是数据库之上的一个用户界面。
So at that point, you just have so much of traditional software can just be looked at as, like, a UI over a database.
现在正在发生的就是这种情况，而市场终于开始意识到这一点。
That's what's happening right now, and the market is finally actually coming to that realization.
我现在有很多在金融行业的朋友，他们不断问我关于这个问题的问题，因为人们越来越清楚，这彻底改变了我们对许多公司的思考方式。
I have people who have not like, friends in finance the finance industry asking so many questions about this right now because it's just becoming clear that it just changes the way so many of these companies you have to think about.
好的。
Okay.
这是来自《华尔街日报》的报道。
So this is from the Wall Street Journal.
《华尔街日报》在分析这一现象时指出，Workday、Monday.com 和 Adobe 等公司开发的工具已成为美国企业数字化运营的核心支柱。
The the Wall Street Journal looking at this writes, tools made by such companies as workdaymonday.com and Adobe have become the digital backbone of much of corporate America.
然而，这些工具正在处理越来越复杂、耗时的用户请求，而这些 AI 工具也预示了先进人工智能模型对整个企业构成的威胁。
These tools, though, carry out increasingly complex, user requests for hours, and they've offered a preview, these AI tools, they've offered a preview of, of the threat sophisticated AI models pose to entire company.
结果是，企业界正经历一场更广泛的反思，企业领导者开始思考：当人工智能系统能够轻松复制一个人多年积累的编程专长，或公司多年积累的运营经验时，这究竟意味着什么。
The result, a broader business reckoning that has left corporate leaders asking what it will mean when an AI system can easily replicate expertise developed over a lifetime of coding in or in the case of companies years or in the case of, companies years of corporate development.
这里有一位来自法律科技领域的分析师，不过我稍后再谈这个，那是法律科技领域。
Here's an analyst with while tech oh so so actually, I will get back to that in a moment, that's legal tech.
你给我解释一下这个吧。
Just just explain this to me.
好的。
Okay.
我大概明白了一点。
I sort of get it.
你知道吗，我们有个CNBC主播黛德丽·博索，她能够编写出一个专为她使用场景定制的monday.com代码实例。
You know, we had, like, Deirdre Boso, who's a CNBC anchor, was able to void code a instance of monday.com personalized for her use case.
你知道吗，也许这就是人们未来要做的事，但像Workday和monday.com这样的公司如何被取代呢？
You know, maybe that's gonna be what what people do, but how does something like a workdaymonday.com and Adobe get disintermediated?
我的意思是，Adobe和这些职场数据库软件完全不同。
I mean, Adobe, for instance, is completely different than these, like, workplace database softwares.
好的。
Okay.
我们以Monday为例吧，因为我觉得它很好地代表了入门级产品。
So let let's take Monday, because I think that's, like, a good example of that entry level.
解释一下什么是
And explain what
它。
it is.
好吧。
As okay.
首先，monday.com 是一个完整的项目和任务管理软件套件。
Firstofmonday.com, you know, like, entire project and task management suite of software.
但实际情况是，这跟我日常生活息息相关，无论是职业上，还是过去个人生活中，任何类型的任务和项目管理工具都如此。
But what happens and and, again, this is so directly in my day to day life, both professionally, but also, like, even in the past personally, like, any kind of task and project management tools.
它本应提供一种高度协作的方式。
It's supposed to offer this, like, highly collaborative way.
大家共同协作，从而加速和放大项目进展。
Everyone works together, and you kind of, like, accelerate and amplify projects.
但现实中，人们只是用它来追踪任务，而且仅限于最基础的层面。
In reality, people just use it to track tasks and, like, at a really basic level.
这类软件的价格并不便宜。
So much of this software and it's it's not cheap.
这些工具都不便宜，我后面会更详细地讲这一点。
These tools are not cheap, and I'm gonna get more into that.
但当人工智能开始融入你的工作流程时，如果AI要生成一份营销简报，现在人们却还得花几个小时手动创建任务。
But at that point, being able to the more AI starts to enter your workflow, then using if AI is gonna generate a marketing brief, right now, people sit there and, like, spend hours creating tasks from that.
我得为一个元广告创建多少种不同的规格啊？
Like, how many different specs for a meta ad do I have to create?
我现在要把这些内容输入到monday.com里。
I'm gonna now enter that into monday.com.
这会告诉别人他们需要做什么。
That's gonna tell someone what they have to do.
他们会进去查看，而每个人都在为monday.com付费。
They're gonna go into it, and then everyone is paying monday.com for this.
这种工作其实根本不需要这样的系统来完成。
Like, that, you don't need that system to do that kind of work anymore.
再说一遍，如果你像DeeBossa那样用 vibe coding 做一个任务管理应用，当然可以。
Like, again, I mean, if you're vibe coding a task management app like DeeBossa did, sure, you can.
但即使在这些组织里，你也可以创建相当标准的软件来实际追踪正在进行的任务。
But even in these organizations, like, you can create pretty standard software to actually track that task that's taking place.
但我觉得更大的趋势是，很多这类工作、很多这些任务，都将由智能代理来完成，到那时，你根本不需要有人手动输入任务，再让另一个人点击查看完成。
But then a bigger thing that I think is gonna happen is a lot of that work, a lot of those tasks that were being created, agents are going to do, At which point, you don't need you never needed some you don't need someone to actually enter tasks manually and so another person could look at it and click check.
这件事已经完成了。
This is done.
你知道，这完全消除了对这类软件的需求。
You know, like, it it totally removes the need for that kind of software.
我可以进入Workday、Adobe，还有你个人的系统，但我认为，项目和任务管理正是我们这里所讨论问题的完美例子。
I can get into Workday, Adobe, and you individually, but I think that, like, project and task management is a perfect example of, of, like, what we're talking about here.
没错。
Right.
所以问题来了，是人们可以通过氛围编程来构建软件，还是有其他完全不同的原因？
And so the question is, like, is this the fact that people can build software by vibe coding it, or is it come something else completely?
萨姆·阿尔特曼本周说的一句话非常有意思。
Something Sam Altman said this week was was pretty interesting.
他说，每个公司现在都是一家API公司，无论他们是否愿意。
He said every company is now an API company now whether they want to be or not.
是的。
Yes.
我明白了。
I you got it.
你明白了。
You got it.
这个问题其实有两个层面。
Like, there's two layers to this whole question.
是的。
Yeah.
你能自己构建它吗？
Can you build it yourself?
我知道很多人会质疑，比如Anthropic真的在用Workday吗？
And I get there's a lot of pushback in terms of, like, is it like, Anthropic uses Workday.
所以他们并没有用‘氛围编码’的方式自己开发Workday的人力资源管理软件，因为每个人都会做成本效益分析。
So they have not vibe coded their own Workday HR management software because there's, like, a certain cost benefit analysis everyone will be doing.
所以我认为第一步是：在什么情况下，我值得为一个不值这个价的软件支付大量费用？
So I think there's step one is, at what point do I pay a lot of money to a software that isn't that isn't worth the cost?
我认为这些对话和问题现在以及未来六到十二个月内都会陆续发生。
And I think those conversations and questions are all gonna be happening now and in the next six to twelve months.
然后你会想到，我们能不能自己来构建它？
Then you get into, like, can we just build it ourselves?
但接着你会面临一个更大的问题：在一个日益具有代理能力的世界里，我们真的需要它吗？
But then you get into the next bigger question, do we need it in an increasingly agentic world?
而且随着越来越多的工作由自主的知识工作者、代理和像Claude Coworks这样的系统完成，情况正在发生变化。
And and increasingly, like, with a lot of this work is being done by autonomous knowledge workers, agents, the Claude Coworks of the world.
这就改变了我们是否还需要这么多这类工具的需求。
Like, it changes the need for do you even need a lot of these things?
所以我认为，这些变化将会在各个层面冲击这些公司。
So I think it's gonna be hitting these companies at every level along the way.
但问题是，
Well, here's the thing.
你知道，我认为这种软件所做的事情仍然是必需的。
You know, I think that the software what this software does is going to be needed.
而且我觉得，我们可能有点操之过急了，总在讨论这些智能代理技术要多久才能普及。
And I also think that, like, maybe we're getting ahead of ourselves with, like, all the like, how long it will take for this AgenTic stuff to roll out.
再说一遍，公司做事慢，工作流程也慢，即使技术已经成熟，恐怕也要好几年才能真正落地。
Again, companies are slow, workflows are slow, even if the technology is there, it's gonna take, you know, probably years to get there.
但关键是，我认为真正重要的是，软件公司会开发自己的机器人，大公司也会打造自己的机器人。
But here's the thing, and and I think this is really where where we get where we get to, right, is that we're gonna have software companies building their own bots, and then we're gonna have the big companies building their own bots.
而这个问题——我想萨姆提到的正是这个——是你最终会被整合进像ChatGPT或Claude这样的平台，还是能让人使用你自己的机器人？
And the question and that's, like, kind of what I think Sam was referencing, is do you end up subsumed into, let's say, a ChatGPT or Claude, or do you get people to use your own bot?
比如说，你是monday.com公司。
Like, let's say, you're monday.com.
你是希望用户使用你的聊天机器人，还是让所有这些功能都通过ChatGPT之类的东西接入，然后显示在用户的主屏幕上？
Are you gonna get people to use your chatbot versus have all this stuff basically accessed by, let's say, a ChatGPT and then brought into someone's home screen there?
不会。
No.
这确实如此。
This is yeah.
说吧，拉詹。
Go ahead, Rajan.
是的。
Yeah.
这是个好问题。
It's a it's a good question.
我认为这些公司会保护自己的数据集。
I think these companies are gonna be defending their datasets.
比如，monday.com会说：别进来把我的任务数据全都拿走，放进你们的工作流里。
Like, monday.com is gonna be like, don't come in and take all of my task data and put it into your workflow.
所以我认为，他们确实会努力抵制这一点，这反而会加速人们决定摆脱这些工具的进程。
So I think I think it's a good point that they're gonna be trying to fight back against it, which I think will increase the like, actually, I think that will end up accelerating people's decisions to try to move away from a lot of these tools.
因为它们会为此抗争，但从大局来看，这并不合理。
Because, like, they're gonna fight for it, but it just doesn't make sense in the grand scheme of things.
所以下周三，我们请到了Snowflake的首席执行官斯里达har·拉马斯瓦米来做客。
So on Wednesday up this upcoming Wednesday, we're gonna have Sreedhar Ramaswamy, the CEO of Snowflake, come on.
现在请他来，时机简直完美。
And it's, like, the perfectly timed moment to have someone like him come on.
我们会谈到这个话题，我先简单剧透一下。
And we talk about it, and and I'll just preview it a little bit.
我的意思是，ChatGPT 会试图把 Snowflake 那些缓慢的数据整合进来，让你直接在 ChatGPT 界面里访问，而 Snowflake 则必须想办法应对。
I mean, basically, like, you know, ChatGPT is going to try to get all that slow Snowflake data some to be something that you access within that main ChatGPT interface, and Snowflake is gonna have to build to defend against that.
但最终，决定权还是在客户手里。
And, you know, it's ultimately going to be up to the customers.
我不认为——我觉得 Sreedhar 的策略是对的。
Like, I don't think and this is I think Sreedhar has the right strategy here.
他会让客户来做决定，而如果这真是客户想要的，你其实根本无法阻止。
He's gonna let the customers decide, and and ultimately, you can't really if this is what they want, you can't really fight it.
所以，我认为这就是我们目前所处这个时刻的原因。
And so that's sort of what I think is bringing us into this moment here.
并不是说软件会消失，几周前 Brett Taylor 来的时候就说得非常好。
It's not that software is gonna go away, and I think Brett Taylor when he was on a couple weeks ago put this really well.
事实是投资者现在对未来会发生什么感到不确定，他们正在从这些股票中撤资。
It's the fact that investors are now uncertain about what's going to happen, that they are backing away from these stocks.
所以虽然看起来软件公司的股票可能已经完蛋了，但我完全不这么认为，不过我要说的是，很明显投资者对未来不像一年前那么确定了，他们开始质疑，这些公司的软件会不会只是在别人的界面里被访问？
So whereas it might look like it's over for stock for software companies, I don't think that's the case at all, but what I will say is it's clear that investors aren't as certain about the future as they were, you know, a year ago, and they're getting they're starting to say, well, are they gonna access this company's software, you know, in somebody else's interface?
它们会不会仅仅变成一个输入项？
Will they just become an input?
这就是为什么你会看到资金逃离软件股的原因。
And that's why you're seeing this run away from the software stocks.
是的。
Yeah.
我觉得你说得很到位。
I I think you captured it.
而且，我认为Snowflake在这里的定位非常有趣，因为我觉得未来很多发展方向会是：底层是Snowflake、Databricks或BigQuery这样的数据库层，而AI则构建在其之上。
And, also, I think Snowflake is very interestingly positioned here because I think a lot of where we're gonna be going is you have your kind of database layer of a Snowflake or a Databricks or BigQuery, and then the AI lives on top of it.
这正是我在Ryder从事的工作方向，我们看到Claude和OpenAI企业版都在朝这个方向发展。
That's the kind of stuff that I'm working on, like, at Ryder, and the the and we're seeing Claude and OpenAI Enterprise all moving in that direction.
但我认为，因为我们都清楚所有股票价格都是对未来现金流的完美理性预估。
But I think, like, because we know all stock prices are a perfectly rational encapsulation of calculating future cash flow.
我的意思是，我显然是在说反话。
I mean, I'm obviously sarcastic there.
但实际上，确实如此，我认为即使你明天不会立即弃用Workday和Salesforce，这些公司向客户收取的费用也必将受到质疑。
But but in reality, like, it it's true that right now, I think even if you're not gonna rip out Workday and Salesforce tomorrow, the amount that these companies charge customers, people are going to have to question it.
特别是当你从股票角度而非个人客户视角思考时，当它们的收入将受到质疑、定价将被每个客户审视时，要继续实现增长是很困难的。
Like and especially when you're thinking of it from as a stock, not even as an individual, like, customer, the idea that they're going to continue growth when their revenue is gonna be questioned at when their, like, pricing is gonna be questioned by every single customer, it's tough.
就像，股票本应反映未来的增长机会。
Like like, stocks are supposed to reflect future growth opportunities.
而且我认为，无论是否完美，它们确实做到了这一点。
And I think, reasonably, like, whether it's perfect or not, they do.
而如果你在考虑增长，我看不到这个故事在哪里。
And if you're thinking about growth, it there I don't see what the story is.
对。
Right.
由于这种潜在的未来增长，软件公司的估值倍数通常更高。
And software does get a higher multiple because of that potential future growth.
事实上，资本和潜在回报并不一定直接相关。
The fact that, like, the capital and the potential upside is not necessarily linked.
但现在，软件公司看起来越来越像经济中的其他行业。
But now software firms are starting to look a lot more like, the rest of the economy.
这是来自财经观察家莉兹·托马斯的观点。
This is from Liz Thomas who's a, financial watcher.
软件公司未来十二个月的市盈率，我稍后会和斯里达尔讨论，但我们现在先谈一谈。
Software's forward twelve month p, and I'm gonna speak about this with Sreedhar, but let's talk about it now.
软件公司未来十二个月的市盈率已从33.1倍压缩至23.2倍。
The software's forward twelve month p has compressed from 33.1 to 23.2 x.
估值倍数下降了30%，当前估值已接近2022年的水平。
Multiple contraction of 30% and valuations are now approaching the 2022 levels.
这个变动相当显著，拉詹。
That's a pretty significant move, Ranjan.
你同意吗？
Wouldn't you agree?
我的意思是，这太显著了。
I mean, that is big.
是的。
Yeah.
我认为目前最重要的启示是，多年来大家一直认为软件有一些基本的不变真理，比如初期投入后边际回报会持续增长，而实际上分发软件的成本几乎是零。
That's a you're like, the entire I think the biggest takeaway right now is, like, everyone thought of software to have kind of, like, a few fundamental truths to it for many years now, and that's kind of, like, you know, like, basically increasingly scaling marginal returns that you invest initially, but actually distributing software is essentially cost zero.
因此，尤其是在过去十五年里，软件变得极其有价值。
So, like and incredibly increasingly valuable, especially over the last fifteen years.
这就是为什么软件过去被区别于其他经济部门看待。
So that's where you get it was treated differently than the rest of the economy.
它的经济模式完全不同，而我真心认为这已经从根本上改变了。
It just had different economics, and I genuinely think that's fundamentally changed.
我们已经多次讨论过这一点。
And we we've talked about this a lot.
比如一家AI公司，我们甚至还不清楚它们的实际运营成本是多少。
Like and an AI company, we don't even know what the economics really are of them in terms of, like, what it costs to operate.
用户越多，大语言模型的支出结构就越不像传统软件。
The more people use it, LLMs don't behave like traditional software in terms of, like, expense structure.
所以，是的，我认为过去十五到二十年的软件形态现在已经不同了。
So so, yeah, I think whatever software looks like for the last fifteen to twenty years is different now.
但你确定这不是一种过度反应吗？
But are you sure this isn't an overreaction?
因为我们还没讨论过导致这些软件股票飙升的原因，那就是Anthropic推出了这个产品，我直接念一下。
Because we have yet to talk about the cause for what inked all these software stocks, and that is that Anthropic had this, I'll just read it.
这是来自《商业内幕》的报道。
It's from Business Insider.
Anthropic最新的AI工具正在冲击法律软件股，那是一款法律工具。
Anthropic's latest AI tool is hammering legal software stocks, so it was a legal tool.
Anthropic最近为其Claude CoworkAI助手推出了一款新插件，可以执行多项文书工作，包括追踪合规性和审查法律文件。
Anthropic recently rolled out a new plugin for its Claude Cowork AI agent that can perform several clerical tasks including tracking compliance and reviewing legal documents.
就AI更新而言，它在周五发布时在法律领域之外并没有引起太大反响。
As far as AI updates go, it didn't make much of a splash outside the legal space when it was rolled out Friday.
然而，它却引发了与法律行业相关的软件和出版类股票的抛售，这发生在本周初。
However, it has triggered a sell off among software and publishing stocks with ties to the legal industry, and this was in the beginning of the week.
周二出现了两次明显下跌。
The moves on two on Tuesday.
威科集团下跌了13%，Reliq的PLC下跌了15.8%，LegalZoom下跌了18%。
Walters Kluwer went down 13%, Reliq's PLC went down 15.8%, LegalZoom, down 18%.
汤森路透下跌了19%。
Thomson Reuters, down 19%.
这是该公司历史上最糟糕的一天。
The worst day in the company's history.
我的意思是，真的吗。
I mean, come on.
哈维早就已经出现了。
Like, Harvey was out there already.
它并没有颠覆法律软件。
It hasn't upended legal software.
所以Anthropic发布了一个插件，整个软件行业就完蛋了？
So Anthropic releases one plug in, and all of a sudden, the software industry is dead?
我的意思是，市场做出这样的反应确实有点过激。
I mean, that does seem like a bit of a strong move from the market.
好吧。
Okay.
但你用过LegalZoom吗？
But have you used LegalZoom?
你是个独立创业者。
You are a independent entrepreneur.
我以前是。
I was.
你以前用过类似LegalZoom的工具吗？
Have you used the LegalZoom type of tools before?
我确信我肯定在某个时候用过。
I'm sure I've been in it at some point.
是的。
Yeah.
就像，任何用过这些工具的人，我希望你不会说，比如，LegalZoom的首席执行官是节目下一位嘉宾
Like, anyone who has used those, and I'm hoping you're not gonna say, like, the CEO of LegalZoom is the next guest on the show after
我的意思是，能请到他们当然很好。
I mean, it'd great to have them.
我们会欢迎它们加入。
We would welcome them on.
但是，这些工具提供的价值其实并不大。
But, like, these tools don't provide a ton of value.
它们只是解决非常具体的问题，而以前的软件就是专注于一个特定问题，垄断它，然后向用户收取高额费用来提供相对的价值。
They, like, answered really specific problems, and that was like, software used to be take one really specific problem, own it, like, charge people a lot of money to deliver relative value.
但这类事情，比如梳理和构建基础法律文件，正是人工智能非常擅长的。
But that's the kind of stuff that, like, going through and structuring a basic legal document AI is really good at.
所以，我觉得这实际上比 monday.com 的例子更能说明现在正在发生的事情。
So, like, I think that actually tells the story of what's happening now better than even the monday.com example.
就像，当你拥有那种垄断时，其实是有价值的。
Like like, when you have the there was almost, like, value in that kind of monopoly.
对吧？
Right?
而现在，你只需开发非常专注的软件，然后高价出售，人们依然愿意买单。
There's and now, like, you just create really narrowly focused software and then charge a ton of money for it, and people will still pay you.
这种模式比任何东西都更过时了。
That model is more dead than anything.
你知道吗，我们在这里注意到的一个有趣现象是，曾经机器人流程自动化是软件领域的热门新事物，但那并不算是人工智能。
You know, one of the interesting lines that we're picking up here is there was a time where, robotic process automation was the hot new thing in software, and that wasn't, you know, necessarily AI.
那不是人工智能。
It wasn't AI.
它更像是通过硬编码的方式教计算机如何使用你的电脑，从而让你不必再做这些琐碎的任务，而让它特别有趣的是，它把许多软件中的碎片化部分连接在了一起。
It was like sort of hard coded ways to teach a computer, the ways to use your computer so you didn't have to do these menial tasks and one of the things that made it so interesting was, that it stitched together a lot of these broken parts of software.
我觉得这里需要指出的一点是，软件本身并不完美，但它之所以能成功是因为它超越了Excel。
Like, that's one of the things I think we should point out here, and I think you really hit on it, is that software is imperfect and it kind of got away because it beat Excel.
它侥幸成功了，人们当时还在讨论用胶带式的方法拼接软件程序，如果能实现的话将是一个巨大的市场机遇。
It got away with it and people were talking about like these duct tape ways to stitch together software programs and if they could ever do that, it would be a very big market opportunity.
而现在这个机遇可能已经到来，只不过是通过LLMs实现的，而不是你想象中的机器人流程自动化方式。
And now that has potentially arrived just, with LLMs and not in the robotic process automation way that you'll imagine.
等等，先别急。
Well, hold on.
我真的很想强调一点，而且我注意到，因为记得，这要怪行业对AgenTic的讨论方式。
One thing I wanna kinda really get across, and I I see this that, like and because remember, and this is I fault the industry for how AgenTic was talked about.
就像几个月前，甚至现在，我参与的许多对话中，人们仍然认为Agentic就是拿一个现有流程然后加点AI进去。
It's like a few months ago, even now, many of the conversations I have, people still think of Agentic as take an existing process and just add some AI to it.
就像你基本上是用AI重新构建那个流程。
Like but you re basically recreate that process with AI.
最大的变化和区别在于，这完全是Claude的代码。
The big change in difference in any like, this is a whole Claude code.
你不再只是用AI来生成单个代码片段了。
You didn't you're you're no longer, like, using AI to generate individual bits of code.
你是在彻底重新设计整个流程。
You're just reinventing the entire process.
你会想，我真正想要的结果是什么？
You're like, what do I actually want an outcome to be?
去完成它吧。
Go do it.
我认为这在任何类型的知识工作中都是如此。
And I think that's the in any kind of knowledge work.
比如你处理一个法律相关的Zoom会议。
So you take a legal Zoom.
现在你根本不在乎那个流程究竟是怎样的。
Like, now you don't even care about what that process is.
你只说，我需要一份能实现这个功能的文档。
You just say, I need a document that does this.
去研究各种不同的文件和法律类型。
Go out, research all the different types documents and laws.
这是我的信息。
Here's my information.
弄清楚你需要我提供什么，然后生成文件。
Figure out what you need from me and create the document.
所以，嗯，我觉得这个前景越来越清晰了。
So so, like, yeah, I think the promise is very clear more and more.
我同意。
I agree.
这需要时间。
It's gonna take time.
不可能明天就发生。
Like, it's not happening tomorrow.
我觉得股市的反应很有趣，因为感觉所有人都在同一时间才意识到这一切。
I think the stock market reaction is interesting because it feels like everyone is just getting this all at once.
是的。
Yeah.
Ranjan。
Ranjan.
好的。
Okay.
所以你显然是相信这个的。
So you're clearly a believer in in this.
你为什么不提出一个最强有力的论证，说明为什么这个方案可能行不通呢？
Why don't you give, like, the best steelman argument in terms of, like, how why this might not work?
不，不行。
Don't no.
这个问题我放弃了。
I'm tapping out on this one.
比如，有哪些你和我都没看到的潜在漏洞？
Like, where is the vulnerability that you're not seeing and we're not seeing?
我想我该成为谁？
I I think who should I who should I be?
我该成为 monday.com 还是 LegalZoom？
Should I be monday.com here or LegalZoom?
或者我们就选其中一个吧。
Or Let's run with either of those.
好的。
Alright.
好的。
Alright.
我认为这些工具的价值在于，正如你提到的，
I think the value in those tools, there's was is and you I mean, you brought it up.
公司往往进展缓慢。
Like, companies are slow.
这些工具的价值从来不在用户界面或易用性上，因为天啊，很多这些工具用起来都很糟糕。
The value in those tools was never about the UI or the actual ease of use because, god, a lot of these tools are brutal to use.
这纯粹只是为人们提供一个安全合规的数据录入场所。
It was literally just having a safe compliant place for people to enter data.
所以，当人们说这只是一个简单的数据库，而AI只是叠加在上面时，其实这些简单的结构化数据库本身确实有价值。
So, like, when people say it's just a dumb database and AI is gonna be a layer over it, that there actually is value in them being that dumb structured database.
而且它们还能坚持一段时间。
And that they're gonna be able to fight and hold on for a bit.
我觉得，确实如此。
I think, like, people it's true.
人们短期内不会把它们淘汰掉。
People are not gonna rip them out anytime soon.
至于它们能否成长之类的，我们拭目以待。
Whether they can grow and stuff, we'll see.
但它们肯定不会不战而退。
But they they have a they're not going away without a fight, I'm sure.
这确实是几周前我们和布雷特·泰勒交谈时他提到的一点。
This is definitely this is something that Brett Taylor brought up when we spoke, a couple weeks ago.
本·汤普森也写过这个话题。
And Ben Thompson also writes about this.
他说，开发原始应用程序只是开始。
He says writing the original app is just the beginning.
还需要维护、安全补丁、新功能，以及应对不断变化的标准。
There is maintenance, security patches, features, new features, changing standards.
开发一个应用程序意味着承诺踏上一段永无止境的旅程——不断回到原点，而这一点与公司的核心能力毫无关系。
Writing an app is a commitment to a never ending journey, a journey to return to the original point, which that has nothing to do with the company's core competency.
对吧？
Right?
这跟你刚才说的差不多。
Something something close to what you were saying.
他还说，销售软件不仅仅是卖代码。
And then he says in selling software isn't just about selling code.
还包括支持、合规性，以及与其他软件的集成。
There is support, there is compliance, there are integrations with other software.
真正有价值的东西远不止代码本身。
The list of what is actually valuable goes far beyond code.
这就是为什么公司不完全开源软件，他们想要的不是代码，而是产品及其所包含的一切。
This is why companies don't purely open source software, they don't want code, they want a product and everything that entails.
你会说这基本上是对此的最大反驳吗？如果你对当今软件行业持乐观态度，你可能会坚持认为他们能够足够快地构建出自己的聊天机器人或智能体。
Would you say that that's basically the biggest counterargument against this and if you're a bull on, like, today's software industry, you kind of, like, hold on to that and imagine that they're gonna be able to build their their chatbots or agents, you know, fast enough.
所以我觉得我一直在反复思考这个问题。
So I think I've I've been thinking about this a lot.
尽管我试图为monday.com站台，但我真的看不出他们收费与所提供的价值相匹配。
Like, it even though I tried to take the steel man position for monday.com, like, I really don't see the prices they charge relative to the value.
我认为可能的情况是，也许会有专注于AI的原生开发者工作室，他们大量创建高度定制化的新软件，并负责管理维护，而不是像一家油漆分销商那样自己去编写CRM系统。
I think what could be is maybe there's, like, studios of AI native developers that actually create just tons and tons of new software that's highly customized, and they do the management, and they do the like, versus if you're like a, I don't know, like a paint distributor, you're not gonna vibe code your own CRM.
确实有公司提供这类服务，但它们的成本要低得多，运营规模和模式也与这些大型SaaS公司截然不同。
And there's there's companies that do that, but just do it a lot cheaper and at a different type of scale and model than these these large SaaS companies.
有趣的是，我们无法百分百确定谁会使用这些软件平台，而这正是事情变得有意思的地方。
What's gonna be interesting is that, we're not gonna be a 100% sure who's gonna use these software platforms, and that's kind of like where, you know, it gets gets interesting.
我们之前稍微谈过这一点。
We talked about that a little bit.
这又要回到汤普森的观点了。
This is going back to Thompson.
人类至少在一段时间内仍会使用软件，但代理也会越来越多地使用。
Human will be using software at least for a while, but increasingly so will agents.
还不清楚的是，谁将负责创建这些代理。
What isn't clear what isn't clear is who will be creating the agents.
我预计每个SaaS应用都会有自己的代理，但该代理在定义上必然受限于应用的边界。
I expect every SaaS app to have their own agent, but that agent will be definitely definitionally bound by the borders of the application.
与此同时，不同的横向平台正试图通过跨多个应用的承诺，覆盖更广泛的业务范围。
Different horizontal players, meanwhile, will be making a play to cover broader expanses of the business with the promises of working across multiple apps.
我觉得这很有趣。
So I think this is interesting.
对吧？
Right?
比如，如果你能让ChatGPT达到足够好的水平，能够使用你的monday.com并将数据导入到类似ChatGPT脉冲的功能中，你就能获得monday.com的所有好处，却无需实际使用monday.com。
Like, if you could get, let's say, a ChatGPT, to a point where it's good enough to use your monday.com and bring that data into, let's say, a ChatGPT pulse, you might get all the benefits of the monday.com, but just not having to use monday.com.
然后，你实际上就不再是一个软件程序了。
And then, you know, effectively, you don't become a software program.
你变成了一个拥有更好合规性的数据库管理者。
You become a custodian of a database with better compliance.
就是这样。
That's it.
你明白了。
You got it.
我的意思是，这正是核心争议所在。
I mean, that that's the that's the big debate.
如果这就是未来，那么这正是它可能呈现的样子。
And if that's the future here, then that's exactly what it can look like.
然后你会慢慢开始问自己：我真有必要花大价钱让人去某个系统里录入任务，而这些任务却只会在另一个完全不同的系统中被访问和思考吗？
And then slowly, you start to ask yourself, do I need to pay a lot of money for someone to just enter a task into this one system that's just gonna be accessed and thought about in a completely different system.
这就是威胁。
That's the threat.
时间会证明一切。
And time will tell.
对吧？
Right?
但我觉得这里重要的一点是，这是市场的转变。
But I think one of the important things here is this is a shift in the market.
几个月前，我观看了乔什·布朗本周早些时候在CNBC的中场报告，他提出了一个很好的观点。
A couple months ago, I was watching Josh Brown on, the halftime report earlier this week on CNBC, and he made a great point.
他说，听好了，仅仅几个月时间，市场就从‘这是否是一个泡沫？’转变了。
He said, listen, the market in just a couple months has gone from, is this a bubble?
变成了‘AI会表现得如此出色，以至于软件将不复存在’。
To AI is gonna work so well that software is dead.
这真的只用了多长时间？
It really has happened in, like, what?
也就是不到半年的时间。
Like, less than a half a year.
看到这种转变非常有趣。
It's very interesting to see that shift.
我认为关键在于，像Claude Code这样的工具，背后是足够多的人——这一点我几个月前就提过了。
I think what I've I think Claude Code is at the the basis of it is is enough people and, again, I've brought this up for months now.
我确实见证了自主知识工作的力量。
Like, I have seen the power of autonomous knowledge work.
Claude Code基本上引入了自主编程。
Claude Code introduced basically autonomous coding.
但这种心态的转变在于，原来AI真的可以做事，可以自己去工作。
And but that that shift in mentality of, wait, AI can actually do things, go off and work.
当你真正把它看作是一个人或一个团队在为你做事时，这真的改变了你看待它的方式。
Like, it really changes the way you look at it when you're really thinking of it as, like, a person or a team going and doing work for you.
然后突然间，我认为当更多人亲眼看到或亲自尝试过之后，所有的灵感就开始涌现了。
And then suddenly, I think the more people have seen it with their own eyes or done something, that's where all these light bulbs start going off.
在我们离开这个话题之前，还有几件事。
Couple of things before we move off the subject.
这是来自德里克·汤普森的观点。
This is from Derek Thompson.
对我来说，过去三周内，AI是泡沫的可能性大幅下降，而我们实际上在推理和使用需求方面严重不足的可能性则显著上升。
For me, the odds that AI is a bubble declined significantly in the last three weeks, and the odds that we're actually quite underbuilt for the necessary levels of inference and usage went significantly up in this period.
基本上，我认为在未来两年内，AI将成为极高比例白领工作者的主屏幕，而平行代理将在知识工作领域以近乎苏联式的规模部署。
Basically, think AI is going to become the home screen of a ludicrously high percentage of white collar workers in the next two years, and parallel agents will be deployed in the battlefield of knowledge work at downright Soviet levels.
你的看法呢？
Your response.
这听起来很像是拉詹的观点。
That seems like a Ranjan take right there.
我喜欢这个说法。
I like I like that one.
我认为，这正是我亲身体验到的。
I think I mean, this is what I am seeing firsthand.
我是说，那种自主的知识型工作、白领工作，将会被彻底颠覆，而我正亲眼目睹这一切。
Like, I I mean, that autonomous knowledge work, white collar work, that's the stuff that's gonna get completely disrupted and seeing it firsthand.
但我认为他提到的一点，我其实想提出不同意见。
But I think he makes one point there that I I actually would push back on.
任何现在在听的人，都能感受到我有多确信这件事即将发生。
I anyone listening right now can hear, like, how convinced I am that this is going to happen.
但这一切对相关各方——无论是OpenAI、Anthropic，还是我的公司Writer——究竟意味着什么？
Now what does that actually mean for any of the players involved, whether it's OpenAI or Anthropic or my company, Writer?
老实说，我不知道这会如何发展。
Like, I have no idea how this plays out.
我的意思是，经济模式，目前没人搞清楚。
I mean, be like, the economics, no one has figured out.
它们会上市吗？
Like, are they gonna IPO?
OpenAI能赚钱吗？
Can OpenAI make money?
展开剩余字幕（还有 480 条）
他们能赢得这些重大赌注中的任何一个吗？
Do they like, are they gonna win any of these individual big bets that they're taking?
谷歌会击败所有人吗？
Is Google gonna kill everyone?
我觉得，尽管从宏观上看这将是未来，但当前的格局比以往任何时候都更加不确定。
Like, I actually think the the the landscape is more uncertain than ever while I think, like, at a high level, this is going to be the future.
毫无疑问，确实如此。
Undoubtedly, it's actually yeah.
这场博弈的走向变得更加扑朔迷离了。
It's even muddier than ever how this plays out.
没错。
That's right.
我觉得我们仍然不清楚谁会获得价值。
I think we still don't quite know who gets the value.
是基础模型的构建者吗？
Is it the foundational model builders?
是芯片制造商吗？
Is it the chip makers?
是应用层吗？
Is it the application layer?
是咨询公司吗？
Is it the consultants?
对吧？
Right?
我认为这就是为什么是这样。
And I think that's why Yeah.
或者谁都不是。
Or nobody.
是的。
Yeah.
还是说所有东西都会变得极其便宜？
Or does everything everything gets so cheap?
指令。
Instruction.
是的。
Yeah.
它让一切变得如此。
It it gets everything.
不。
No.
AI自己就能编写代码，它自己就能搞定，直接取代了我们的工作，让我们全都失业。
The AI just codes it it vibe codes it itself, and it removes the work and puts us all out of a job anyways.
是的。
Yeah.
谁知道呢？
Who knows?
不可能。
No way.
是的。
Yeah.
如果它强大到足以直接取代工作，那么一定有人会从中获得价值。
Like, if it's powerful enough to just remove the job, then somebody is gonna get value there.
总会有一些价值可以挖掘。
Like, there will be value to be found.
是的。
Yeah.
是的。
Yeah.
但我认为这仍然是个问题。
But I think that's still a question.
但我觉得这正是我们看到软件行业波动的原因。
But I think that's why we're seeing just to go back to that's why we're seeing the volatility in software.
这正是我们看到科技巨头波动的原因。
That's why we're seeing the volatility in big tech.
每次市场数据稍微偏离正常水平时，比如今天他们正在买入亚马逊股票，尽管顺便说一句，市场已经强劲反弹，道指再次突破了5万点。
Every time there's, like, a number that's slightly off the market, like, again, they're taking Amazon stock today, which we can get to even though the market by the way, the market's rebounded nicely, Dow just hit 50 k again.
但由于这种价值归属的不确定性，我们短期内仍将处于这种波动状态。
So but because of this uncertainty of who gets the value, we're just gonna be living in this moment of volatility for a while.
我想问你一下，我们这个节目不做前瞻性陈述，但你觉得呢？是的，之前市场确实充满恐惧，尤其是在软件领域。
I'm gonna ask you, and we don't make forward looking statements on this show, but do you think so, again, yeah, there was, like, fear gripping the market, especially in software.
过去一个月简直惨不忍睹。
The last month has been brutal.
你觉得今天看到的反弹，一个月后、两个月后，我们讨论的这些软件股票整体是会更高还是更低？
Do you think, looking at today, the rebound, do you think we're a month from now, two months from now, do you think we're higher or lower across kinda like these software stocks that we're we're generally talking about?
是的。
Yeah.
这又不是投资建议。
Again, this is not investment advice.
是的。
Yeah.
我认为我们会走高。
I think we're higher.
我真的这么认为。
I really do.
我觉得，嗯，给自己讲一个未来会发生的故事很容易。
I think that, like, it's easy to tell yourself a story of what's gonna happen.
真正预见到所有必须接纳这种变革的官僚程序要困难得多。
It's harder to really envision all the bureaucracy that has to sort of embrace it for that change to actually go through.
是的。
Yeah.
不。
No.
不。
No.
这就是我认为值得关注的地方。
That's what that's what I think it's it's it's worth looking at.
我认为最重要的是，在这些对话中，关键在于方向性指导，比如续约率之类的指标，我们需要看看这些公司：这些变化现在真的正在发生，还是只是我们能预见未来终将发生，但短期内不会实现？
I think the most important thing is gonna be, yeah, in these conversations and the it's, yeah, the directional guidance, the kind of directional, like, renewal rates, all these kind of things with these companies to see, is this really happening right now, or is this something that we can see will happen eventually but is not happening anytime soon?
这是来自Semi Analysis创始人Dylan Patel的另一个数据。
Here's one more stat from, Dylan Patel, the founder of Semi Analysis.
目前，GitHub上4%的公开提交是由ClaudeCode编写的。
4% of GitHub public commits are being authored by ClaudeCode right now.
按照当前趋势，我们预计到2026年，ClaudeCode的每日提交量将占所有提交的20%以上，而你一眨眼的功夫，AI就已经吞噬了整个软件开发领域。
At the current trajectory, we believe that ClaudeCode will be 20% plus of all daily commits by the 2026, while you blinked AI consumed all of software development.
是的。
Yeah.
我之前在十二月就说过，我在这档节目中预测过，2026年软件开发领域会发生的事，今年就会在知识工作领域重演。
That I'm telling I that was my in December, that was my prediction for 2026 on this show that what happened in software development is gonna happen in knowledge work this year.
今年年底，这件事就会发生。
End of this year, it's happening.
也许达莉亚终究是对的。
Maybe Daria was right after all.
大家都嘲笑他的预测，认为初级工作将有50%消失。
Everybody laughed at his prediction that 50% of entry level work was gonna be gone.
实际上，这是个很好的观点。
Actually, it it's a good point.
我的意思是，我确实一直在认真思考这个问题。
I mean, I really have been thinking about this a lot.
对我来说，以这种方式工作时的感觉，就像是你本质上是个管理者。
Like, to me, I actually like, what it feels like when you're working in this way is that you're basically basically a manager.
你正在管理各种不同的代理，让他们完成不同类型的工作。
You're managing all these different agents to do different type of work.
所以，你会开始意识到，管理技能变得比以往任何时候都更重要。
So, like, you start to realize managerial skills become more important than ever.
那么，谁会被取代，何时被取代，以及如何被取代，将会是一个非常有趣的问题。
So, like, who gets displaced and when and how is gonna be a really interesting thing.
如果是初级岗位但属于AI原生一代的人，会不会更有优势？
Is it like, if you're entry level but AI native, are you gonna be better positioned?
或者如果你年纪大一些，是个经理，但你知道如何委派工作，你会更有优势吗？
Or if you're kinda older and a manager, but you know how to delegate work, are you gonna be better?
我不知道。
I don't know.
这将是另一个完全不同的方向。
It's gonna be that's another whole direction.
看看事情如何发展会很有趣。
It's gonna be interesting to see how it plays out.
你是说未来的工作就是一直待在聊天窗口里问：‘做好了吗？’
Are you saying that the future of work is just gonna be being in a chatbot window being like, is it ready yet?
进展如何？
How's it looking?
进展如何？
How's it looking?
快好了吗？
Almost there?
那就是工作的未来。
That's that's the future of work.
但这甚至不是聊天机器人。
It's not even chatbot, though.
就像我在西耶拉与布雷特·泰勒的那期节目中听到的，你说的是聊天机器人。
Like, that's why I think I on the Sierra, episode with Brett Taylor, I was hearing you'd said chatbot.
我认为现在聊天是界面，但也许我用的是genticmonday.com来管理任务，看到我的代理已经完成了它们。
I think it's chat's the UI right now, but, like, maybe I'm using my genticmonday.com to manage tasks, and I see that they were completed by my agent.
然后我就想，好吧。
And then I'm like, okay.
我要分配一些其他任务。
I'm gonna assign some other tasks.
但确实，我认为这将成为你工作方式的一部分。
Like but, yeah, I think that's that's gonna be part of how what you do for work.
我们醒来。
We wake up.
好的。
Okay.
我的播客准备做好了吗？
Is, is my podcast prep ready?
哦，天哪。
Oh, god.
我是说，它在这方面已经很有用了。
Mean, it's already useful for that.
好的。
Alright.
说到Dario，我们正目睹他与Sam Altman之间的一场较量。
Speaking of Dario, we have a war going on between him and Sam Altman.
在一个最意想不到的战场上，可以说是超级碗，两家公司的广告并非对决。
In the most unlikely of battlefields, shall we say the Super Bowl, two companies with ads not dueling.
OpenAI的广告全关乎建设，而Anthropic的广告则意在打压OpenAI。
OpenAI's ad is all about building, Anthropix ad is about trying to take OpenAI down a peg.
在接下来的内容中，当我们戴上广告公司的帽子时，我们会深入讨论这一点。
And we will discuss this as we put on our ad agency hats right after this.
开始一件事并不只是困难。
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.
当我刚开始做这个播客时，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，这是个正确的选择。
Now I know it was the right choice.
当你有像Shopify这样的合作伙伴支持你时，这也会有所帮助。
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.
凭借数百个即用模板，Shopify帮助你打造一个与品牌风格相匹配的精美在线商店。
With hundreds of ready to use templates, Shopify helps you build a beautiful online store that matches your brand style.
你还可以像拥有整个营销团队一样推广产品，轻松创建电子邮件和社交媒体活动，覆盖客户浏览或闲逛的任何平台。
You can also get the word out like you have a marketing team behind you, easily create email and social media campaigns wherever your customers are scrolling or strolling.
现在是时候用Shopify将那些‘如果’变为现实了。
It's time to turn those what ifs into with Shopify today.
立即前往shopify.com/bigtech，注册每月1美元的试用。
Sign up for your $1 per month trial at shopify.com/bigtech.
访问shopify.com/bigtech。
Go to shopify.com/bigtech.
就是shopify.com/bigtech。
That's shopify.com/bigtech.
我们回到这里，继续收听Big Technology Podcast周五版。
And we're back here on big technology podcast Friday edition.
现在是超级碗周末，Anthropic 决定抓住这个时机向 OpenAI 发起一击，直击其广告业务的痛处。
Well, it's Super Bowl weekend, and Anthropic decided that this was the moment to take a shot at OpenAI, hitting it where it hurts, in the ads program.
这是来自《商业内幕》的报道。
This is from Business Insider.
Anthropic 的超级碗广告嘲讽了 Chad Chippy Tee。
Anthropic Super Bowl spot skewers Chad Chippy Tee.
广告即将进入人工智能领域，但不会出现在 Claude 上。
Ads are coming to AI, but not to Claude.
Anthropic 在可能的最大舞台上对 OpenAI 发起了攻击。
Anthropic is taking a shot at OpenAI on the biggest stage possible.
周三，Anthropic 推出了名为 'Glitzy' 的广告活动，该广告将在周日超级碗期间全国播出，隐晦地针对 OpenAI 将广告引入 ChatGPT 的计划。
On Wednesday, Anthropic rolled out Glitzy a Glitzy ad campaign that will air nationally during Sunday Super Bowl, which implicitly centers on OpenAI's plan to bring advertising to ChatGPT.
广告即将进入人工智能领域，但不会出现在 Claude 上——广告标语这样写道。
Ads are coming to AI, but not to Claude, the AI the the the ad tagline reads.
天哪。
Oh my god.
我看过其中一个。
There was a there was one that I saw.
我觉得一共有四个，我看到的那个特别搞笑。
There's four of them, I think, and one I saw was really funny.
场景是在心理咨询师的办公室，那个人问咨询师：我该怎么更好地和我妈妈沟通？
It was in the therapist's office, and the guy goes to the therapist, how do I communicate better with my mom?
咨询师是个年长的女士，她说：好问题。
And the therapist says and she's an older lady, and she says, great question.
改善与妈妈的沟通能拉近你们的关系。
Improved communication with your mom can bring you closer.
这里有一些你可以尝试的方法。
Here are some techniques you can try.
首先学会倾听。
Start by listening.
真正听懂她话语背后想表达的意思。
Really hear what she's trying to say underneath her words.
从共识出发构建对话。
Build conversation from a point of agreement.
通过共同的活动寻找联系，比如一次自然散步。
Find a connection through shared activity, perhaps a nature walk.
或者如果这段关系无法修复
Or if the relationship can't be fixed
如果这段关系无法修复，可以在Golden Encounters这个为成熟人群设计的交友网站上与其他年长女性建立情感联系，这里连接着敏感的小熊与咆哮的母狮。
if the relationship can't be fixed, find emotional connection with other older women on Golden Encounters, the mature dating site that connects sensitive cubs with roaring cougars.
什么？
What?
那男人说：什么？
The guy's like, what?
这些太棒了。
This these were so good.
你看到那个私人教练的了吗？
Do you see the personal trainer one?
是的。
Yes.
对。
Yeah.
但是，再说一次，我觉得好吧。
But, again, like, I think okay.
我们先来聊聊这些有多棒。
Let's first talk about how good these were.
就像，我觉得它精准地抓住了重点，而且记得吗，《南方公园》是第一个真正讽刺ChatGPT阿谀奉承本质的作品。
Like, I think it nailed and remember, South Park was the first one kind of, like, really that, like, skewered the sycophancy of ChatGPT.
但光是精准模仿ChatGPT的回应节奏就比其他作品强多了——‘好问题’，然后慢慢滑向广告植入，这段真的太精彩了。
But even just nailing, like, the cadence of how ChatGPT responds much more than others, great question, but then slowly degrading into the ad, it was it was so good.
确实。
Yeah.
我觉得像Claude这样的，其实挺有意思的是，这波操作可能反而给它们带来了前所未有的知名度。
I think, like, Claude, it's kind of fun too that this is actually probably giving them more name recognition, I think, than anything.
因为你要记得，Anthropic 这个名字，大多数普通人根本不知道它是什么。
Because remember, like, the name Anthropic, most people Normies don't even know what that is.
Claude 并不是一个在消费层面非常庞大的产品。
Claude does is not a massive product at a kinda consumer scale.
所以我觉得他们在这方面做得不错。
So I think they did a good job here.
你怎么看？
What do you think?
我的看法是，首先，这个广告根本不是为超级碗准备的。
My perspective is, first of all, the ad wasn't even intended for the Super Bowl.
顺便说一下，这个广告是在超级碗之前发布的。
The ad was intended by the way, they released it pre Super Bowl.
你知道为什么吗？
You know why?
他们就是希望这个广告能火起来。
They wanted that ad to take off.
人们会关注它的。
It people will pay attention to it.
如果这是一条超级碗广告，而他们希望它走红，让Anthropic的潜在企业用户在办公室里传播，在Slack里分享，甚至可能在周五登上大型科技播客。
If it's a Super Bowl ad and they wanted it to take off so the potential business users of Anthropic would circulate that in the in the office, drop it in Slack, it makes it on big technology podcast on a Friday, maybe.
而这才是目标受众。
And, that is the intended audience.
我认为他们并不真的在意更广泛的消费者群体。
I don't think they really care about the broader consumer audience.
他们并不是想让FOD获得大量用户。
It's not like they're trying to make FOD gain a bunch of users.
所以从这个角度看，我认为这可能算是成功的。
So from that standpoint, I think potentially successful.
他们还引发了OpenAI的回应。
They also got OpenAI to respond.
我们要不要聊聊OpenAI是如何回应的？
Should we go into the way that OpenAI responded?
因为看起来他们有点过头了
Because seems like they went a little over
哦，我
Oh, I
我们也应该回应。
respond think we as well.
他们被引诱了。
They were baited.
我觉得我们应该。
I think we should.
这是萨姆·阿尔特曼说的话。
So here's what Sam Sam Altman says.
他说，首先，Anthropic广告好的地方是，它们很有趣，我笑了。
So he says, first, the good part of the Anthropic ads, they are funny, and I laughed.
这只是我个人的看法。
This is just me.
顺便说一下，我想我们都同意，这些广告还挺搞笑的。
By way, I think we'd all agree they were pretty funny.
他说，我不明白Anthropic为什么会选择如此明显不诚实的做法。
He says, I wonder why Anthropic would go for something so clearly dishonest.
我们广告的最重要原则是：我们不会做这种事，我们显然绝不会以Anthropic所描绘的方式投放广告。
Our most important principle for ads says we won't do exactly this, which we would obviously never run ads in the way Anthropic depicts them.
我们并不愚蠢，我们知道我们的用户会拒绝这种做法。
We are not stupid, and we know our users would reject that.
我想，对于Anthropic来说，用一个欺骗性的广告来批评虚构的、不存在的欺骗性广告，这倒是符合他们一贯的双标风格，但超级碗广告并不是我预期会出现这种做法的地方。
I guess it's on brand for Anthropic doublespeak to use deceptive a deceptive ad that critique theoretical deceptive ads that aren't real, but a Super Bowl ad is not where I would expect it.
更重要的是，我们相信每个人都应该能够使用AI，并致力于实现免费访问。
More importantly, we believe everyone deserves to use AI and are committed to free access.
也许更关键的是，Anthropic希望控制人们如何使用AI。
Maybe even more importantly, Anthropic wants to control what people do with AI.
他们阻止他们不喜欢的公司使用他们的编程产品，包括我们。
They block companies they don't like from using their coding products, including us.
他们想自己制定规则，规定人们能用和不能用AI做什么，现在还想告诉其他公司他们的商业模式应该是什么。
They want to write the rules themselves for what people can and can't use AI for, and now they also wanna tell other companies what their business models can be.
你对这个回应有什么看法？
What do you think about this response?
哦，还有一句。
Oh, one more line.
一个专制公司，我们非常关心安全、广泛有益的通用人工智能，我们知道实现这一目标的唯一方法是与世界合作共同准备。
One authoritarian company, we we care a great deal about safe, broadly beneficial AGI, and we know the only way to get there is, is to work with the world to prepare.
单靠一家专制公司是无法实现这一目标的，更不用说其他显而易见的风险了。
One authoritarian company won't get us there on their own to say nothing of the other obvious obvious risks.
这是一条非常黑暗的道路。
It's a really dark path.
这是一条黑暗的道路。
It's a dark path.
好吧。
Alright.
这件事有太多层次了。
So many layers to this one.
谢谢你，萨姆，给了我们这么一份精彩的回应。
Thank you, Sam, for giving us this gift of a a response.
首先，我觉得这样说其实是个挺不错的主意。
One, I actually think it was, like, a pretty good idea to say.
他们就像在用一则广告，以一种欺骗性的方式批评或引发人们对欺骗性广告的担忧。
It's like they're using an ad in a deceptive way to criticize or to to raise concern around deceptive advertising.
所以，我觉得这有点像。
So, like, I think it's kinda
那里确实有个观点。
There's a point there.
这是个很好的观点。
That's a good point.
那里确实有个观点。
There there's a point there.
这有点微妙，但确实是个观点。
It's a little nuanced, but it's it's a point.
然后你会发现，德克萨斯州免费使用ChatGPT的人比全美国使用云服务的人还多。
Then you have the more Texans use ChatGPT for free than total people use the cloud in The US.
所以我们面临的问题和他们不一样。
So we have a differently shaped problem they do.
这个我觉得太荒谬了，因为我认为OpenAI的首席营销官用了完全相同的说辞，结果让看到的人突然觉得这太企业化了。
This one, I thought was so ridiculous because I think it was the chief marketing officer of OpenAI, used the exact same line, which suddenly then anyone who sees that, it feels so corporate.
感觉就像是里面有一个危机公关团队，正在集思广益想点子。
It feels like there's, like, a crisis comms team in there that coming up with, like, workshopping ideas.
有人说，我有个好主意。
Someone says, I got a good one.
我们就用德州吧。
Let's use Texas.
然后大家就开始步调一致地发布那个内容。
And then everyone starts to kind of, like, in lockstep post that.
所以我觉得这部分真的很弱，实际上我认为这有点削弱了他们发布这些东西可能获得的任何价值。
So I thought that part really weak, and I actually think that kind of, like, undercut any value they potentially got out of posting this stuff.
然后最后一部分，以这个结束真是奇怪，但专制公司不会独自带我们到达那里。
Then the last part, what a weird thing to end on, but the authoritarian company won't get us there on their own.
他是怎么做出这种逻辑跳跃，认为Anthropic是专制的？
Like, how did he make that logical leap to Anthropic is authoritarian?
是不是因为我觉得那里的隐含意思是Anthropic一直在警告AI的危险。
Was it Because I think the the implicit idea there is Anthropic keeps warning about the dangers of AI.
Anthropic在国会非常积极地参与人工智能的监管方式，并且几乎对其他公司采取不同AI策略的尝试不屑一顾。
Anthropic is very engaged in Congress for the way to regulate AI, and, you know, almost dismisses other companies' attempts to take a different approach to AI.
因此，基本上这就是为什么他会说，他们试图告诉我们该采用什么商业模式，而这正是他会指责他们专横的地方。
And so therefore, like, you know, and it's them basically that's why he was like, they're trying to tell us what our business model is, and that's where he would say they're authoritarian.
我觉得这么轻易就扣上'专制'的帽子有点轻率了。
I feel that's a little flippant on the authoritarian card.
拜托，不至于吧。
Like, come on.
你不能这样笼统地解释你到底想说什么，但即使你刚才描述的那些也不算专制。
You can't draw like, explain what you're trying to exactly say there, but even what you just described isn't authoritarian.
我知道你只是想帮萨姆，但对我来说，这
I know you're just trying to help Sam out here, but that to to me, that
把他给说清楚了。
would him out.
直接解释一下他到底说了什么。
Just explain what he was saying.
是的。
Yeah.
是的。
Yeah.
没错。
Exactly.
你说得对。
You're right.
因为这
Because it
有点模糊。
is a little opaque.
它对专制这个词处理得太轻率了，尤其是在当今这个时代。
It's it's just it's it's try it's a little cavalier about authoritarian, especially in this day and age.
所以我认为，总的来说，我会给这些回复打个C-的评分。
So I think, like, yeah, overall, I'm gonna I'm gonna give this a c minus rating on on replies.
啤酒的负责人、X公司的产品主管，我想现在SpaceX收购了X AI之后，他现在是什么职位我也不清楚。
The key to beer, the head of product at x, I guess he's like, I don't know what he is now now that SpaceX owns x AI.
他说，沟通建议：永远不要用长篇大论回应轻松的幽默。
He says, comms advice, never respond to playful humor with an essay.
只需说：‘哇，你们真厉害’，或者开个他们的玩笑。
Just say, damn, they cooked us or make a joke about them.
就是这样。
There you go.
你只要这么做就行了。
That's that's all all you had to do.
或者，实际上，你会怎么回应？
Or, like actually, how would you have responded?
你的回应会是什么？你会回应吗？
What would be your response, or would you have responded?
我喜欢用欺骗性的广告来回应欺骗性广告这个想法。
I like the idea of of responding deceptive ad about deceptive ads.
我觉得这句话不错。
I thought that was a good line.
对。
Yep.
我不会写‘天啊，他们把我们整惨了’。
I wouldn't have written, damn, they cooked us.
这有点太轻率了。
It is a little too flippant.
那像说声抱歉这样的方式呢？
What what about what about something like, sorry.
克劳德是什么？
What's Claude?
从来没听说过之类的。
Never heard of it or something like that.
就像说不。
Just like No.
这也太青少年了。
That's a little too teenage y also.
是的。
Yeah.
但我的意思是，你有什么优势超过他们？
But, I mean, like, what what do you have over them?
绝大多数人根本不知道它是什么，也不会知道。
Vast majority of people don't know what it is, are not going to know what it is.
Anthropic 这个公司名字到底算什么名字？
What do you what kind of name even is Anthropic of a company?
你知道的，如果你想回应，有太多可说的了，但我同意。
You know, like, there's so many if you wanna respond, but I agree.
不然的话，就别管它了。
Like, otherwise, just just let it go.
但这还涉及到一个问题：对于 Anthropic 来说，这不仅是花钱的问题——毕竟成本很高，还是一次制造轰动效应的机会。
But that also gets to, like, is this a good use of not only money for Anthropoc because it's costly, but, like, an opportunity to make a splash.
我们一直都在讨论，我刚在《Big Technology》上发布了一些数据，显示超过一半的人根本不用这些 AI 应用。
And we've always talked about, like, I just published some data in Big Technology about how more than half of people never use, these these AI AI apps.
超级碗是一个吸引大量消费者、让更多人尝试你产品的机会。
And the Super Bowl is an opportunity to get a lot of consumers, a lot of people to try your thing.
这是来自硅谷传播专业人士蕾切尔·霍罗维茨的说法。
This is from Rachel Horowitz at Silicon Valley comms professional.
我不知道。
I don't know.
Anthropic 正在花大价钱投放一则谈论 OpenAI 的超级碗广告。
Anthropic is spending a lot of money on a Super Bowl ad that talks about OpenAI.
这对 OpenAI 来说挺好的。
That's pretty good for OpenAI.
人们根本不在乎广告。
People don't care about ads.
他们更喜欢免费的东西。
They like free a lot.
我敢肯定，我看到 Netflix 的广告支持型用户基数在两年内翻了三倍。
Pretty sore pretty sure I saw Netflix has seen their ad supported user base triple in, like, two years.
这确实是个好观点。
That's a good point.
实际上，这也是个很中肯的看法。
That's a good point, actually.
而且确实说得对，人们深入进去之后，还挺有意思的。
And that that's a fair point that, like, it's funny within people, like, deep into it.
当然了，每个人都喜欢看萨姆和Anthropic之间的争斗。
It definitely everyone, like, enjoys the kinda Sam versus, like, anthropic beef.
但没错，你说得对。
But but, yeah, you're right.
这真的会让人们反感吗？
Like, is it really gonna turn people off?
也许不会。
Maybe not.
人们显然对各种形式的广告都能接受。
Like, people certainly are okay with ads in all these different formats.
我们经常讨论，Meta可能是第一个真正搞懂如何做AI广告的公司，但不管怎样，你这个观点很有道理：这真的会让普通用户反感吗？尤其是那些从未使用过任何AI应用的人群，这才是你应该瞄准的目标，而这些广告显然不是为他们设计的。
We talk regularly that Meta might be the ones that actually figures out how to run AI ad first ads, but whatever it is, I think that's a good point that is it really gonna turn average people off, especially the entire percentage of the population that have not used any kind of AI app before, that's where you should be trying to go after, and it's certainly not for them.
这是那条推文的第二部分。
Here's part two of that tweet.
我们还从大量研究中了解到，以几乎相同的用户体验去猛攻竞争对手，只会拖累整个行业，因为人们无法区分彼此。
We also know from loads of research that going hard at a competitor with virtually the same user experience as you just drags down the whole sector because people can't differentiate.
AnthropicAd 让 AI 聊天体验显得诡异又谄媚，这本身有点自相矛盾。
The way that AnthropicAd makes AI chat experience seem creepy and obsequious is a bit of a self own in that way.
我不知道。
I don't know.
有可能。
Possibly.
我觉得，我不确定。
I think, like I don't know.
实际上，我发现 ChatGPT 随着时间推移变得让人感到贬低。
I I actually I have found chat GBD to be degrading, like, over time.
我不知道。
I don't know.
你有这种感觉吗？
Have you felt this at all?
它变得越来越精细了。
Like, it's getting more fine.
阿谀奉承。
Obsequious.
比如，我不得不再次修改我的系统提示，告诉它：你不必同意我说的每一件事。
Like, even just, like, I've had to change my system prompt again to, like, you do not have to agree with everything I say.
但我还用它和Gemini以及Claude进行了对比测试。
But, like, I've been testing it against Gemini and Claude as well.
比如，对于一些基础查询，一切都必须完美无缺。
Like, for just basic queries, everything has to be great.
你说得对。
You're right.
这是个很好的问题。
That's a great question.
我不知道。
I don't know.
你真的完全没有注意到这一点吗？
You haven't you haven't been noticing this at all?
不同的体验。
Different experience.
它确实比我更频繁地反驳我。
It definitely has been pushing back more than me, more on me.
但我确实...是的。
But I'd I yeah.
我的系统提示里早就设置了：不要阿谀奉承。
I've had in my system prompt for a long time, do not be sycophantic.
实际上，等一下。
Actually, hold on.
你想不想听我刚运行的一个很棒的例子？
Do you do you wanna hear a great one that I just run?
这个是的。
This one Yes.
就像好吧。
Just like okay.
实际上，这完全可以另起一个话题，但现在的照片存储费，谷歌云和苹果一直在让我为存储照片支付越来越多的钱。
So actually, this can be a whole other rant for another day, but the photo tax, now Google Cloud and Apple keep making me pay more and more money to just store my photos.
我之前同时用了这两个服务，每个月付给苹果40美元，谷歌20美元。
And I had both services, and I'm paying, I think, Apple $40 a month, Google $20 a month.
我心想，想想一年下来我花了这么多钱，真是有点离谱了，是的。
And I'm like, this is getting ridiculous when I think about, yeah, how much I'm paying in a year.
所以，谷歌，我要把照片都移走，不再用它了。
So Google, I'm gonna photos, I'm gonna get rid of.
于是我问了ChatGPT，我想把所有谷歌照片导出来，这样就不用为存储付费了。
So I asked ChatGPT, I wanna export all my Google photos so I don't have to pay for storage.
最好的办法是什么？
What's the best way to go about this?
ChatGPT说，你这样想完全正确。
And ChatGPT says, you're thinking about this the exact right way.
而且，就像
And, like
我不明白这个。
I don't get that.
这很有趣。
That's interesting.
我知道。
I know.
也许你只是在回应这些东西，拉詹。
Maybe you're just responding to this stuff, Ranjan.
也许你喜欢这个。
Maybe you like this.
最干净、最少后悔的方法是导出一次，本地验证，然后慢慢退出 Google 相册。
The cleanest least regret path is export once, verify locally, and then slowly unwind, Google Photos.
但我真的就是想，好吧。
But I literally was like, alright.
我该怎么获取我的 Google 相册？
How do I get my Google Photos?
你思考这个问题的方式完全正确。
And you're thinking about this the exact right way.
哦，多跟我说说。
Oh, tell me more.
你完全没明白吗？
You don't get that at all?
你没明白吗？
You don't get that?
我不明白
I don't get
那个
that.
也许是系统问题
Maybe it is the system problem.
好的
Okay.
但好吧，我想总结一下，我认为Anthropic赢了。
But, okay, I I think just to sum this up, I think anthropic wins.
但总的来说，在我看来，这则广告最搞笑。
But bottom line, in my perspective, it's just it was the funniest ad.
也许长期来看会有一些负面影响，但要做出有幽默感的广告非常困难。
And maybe in long term, there's some hurt, but it's it's very difficult to make an ad, with humor.
我知道我当时坐在办公桌前笑得不行，看起来萨姆·阿尔特曼也是。
I know I was sitting at my desk cracking up, seems like Sam Altman was too.
一半甚至更多考虑使用这项技术的科技行业人士都笑了。
Half of the maybe more of the, you know, tech industry who's considering using this, they laughed.
这本身就很难，而他们做到了，所以我认为这总体上是他们的胜利。
It's that that is tough, and they did it, so I would say that's overall a win for them.
同意。
Agreed.
同意。
Agreed.
这个广告，给萨姆·阿尔特曼的回应打个A-，广告本身打个C-。
The ad, a minus Sam Altman response, c minus.
我会给这个广告打A。
I will give the ad I will give the ad an a.
给萨姆打个B+。
Give Sam a b b b plus response.
我不讨厌它。
I don't hate it.
好的。
Okay.
我不讨厌它。
I don't hate it.
好的。
Okay.
为什么Anthropic不能反击，而OpenAI就可以？
Why can't Anthropic throw punches and OpenAI can't?
我不知道。
I don't know.
对。
Yep.
来一记漂亮的重拳。
Make it a good punch.
要强势回归，但一定要打得漂亮。
Make come back come back strong, but just make it make it a good one.
是的。
Yeah.
好的。
Okay.
对于OpenAI来说，现在真是个有趣的时刻。
So so interesting moment for OpenAI more broadly.
我想简单聊聊这个。
I just wanna talk about this briefly.
我们过去几期节目里已经讨论过这个问题。
We have talked about this over the past couple of shows.
这个周末在《科技大观》上，我发布了一些数据，分析了ChatGPT与竞争对手的市场份额。
This weekend on big technology, I did publish some data looking at the market share of ChatGPT versus the competitors.
首先，从Claude的角度来看，Claude基本上处于市场底部。
First of all, going to your, like, what's Claude perspective, Claude is, like, the basically the floor at the bottom.
实际上使用Claude的人很少，但一旦使用，他们就会频繁使用。
Very few people actually use Claude, you know, but when they do, they use it a lot.
但话说回来，根据Apptopia的数据，OpenAI的ChatGPT市场份额已从2025年1月的69%下降到2026年1月的45.3%。
But that being said, OpenAI's, ChatGPT's market share has dropped from 69% at in January this is according to Apptopia, from 69% in January 2025 to 45.3% in January 2026.
现在它正在增长，但同一时期，我们来看一下。
Now it is growing, but in that same time, let's see.
Gemini的聊天机器人市场份额从14.7%上升到了25.1%，而Grok则从1.6%出人意料地增长到了15.2%。
Gemini has gone from 14.7 chatbot market share to 25.1%, and Grok has gone from 1.6%, surprisingly, to 15.2%.
你对此怎么看？
What do you make of this?
我的意思是，看到OpenAI现在真的有了竞争对手，非常有趣。
I mean, it's very interesting to see that OpenAI really has company now.
是的。
Yeah.
我还行。
I I was okay.
不过这里要打个折扣，因为这是移动应用的日活跃用户数据。
So with a grain of salt here, I think because this is mobile app daily active user.
如果你想想这些独立工具的移动应用体验，比如Claude，所有的热议都集中在用Claude Code或Claude应用在桌面端使用上。
And then if you think about the mobile app experience for any of these individual tools, like Claude, all the hype is around using it at your desktop with Claude code or even the Claude app or whatever.
因为移动应用的使用本质上更偏向于消费者行为。
Like, it's not especially because mobile app usage is a very consumer type of thing.
我认为Gemini无疑是这其中最引人注目的部分。
I think Gemini is definitely the most notable part of this.
我觉得Grok很有趣，但我也很好奇，可能这部分数据会被算进去，我猜是这样。
I think Grok is interesting, but I'm also curious, like, probably that's gonna get counted, I'm guessing.
每一条推特回复，比如，Grok都会解释这个。
Every Twitter reply, like, at Grok explained this.
你明白我在说什么吗？
You know what I'm talking about?
就像，是的。
Like, it Yeah.
我觉得这只是
I think it's just the
Grok应用本身。
Grok app itself.
真的吗？
Really?
是的。
Yes.
你有Grok应用吗？
Do you have the Grok app on
你的手机上吗？
your phone?
没有。
No.
等等。
Wait.
我不认为，但我可以告诉你Grok的使用量在哪里上升了，我也会发布这个，我
I don't think And I'll tell you where Grok usage went up and I'm gonna publish this too, I
觉得。
think.
亲爱的。
Dear.
是的。
Yep.
哦，不。
Oh, no.
哦，不。
Oh, no.
当人们感觉 okay 的时候就是这样。
It is when people's okay.
顺便说一下，Grok 被称为世界上最聪明的 AI 顾问。
By the way, Grok, it's called the world's smartest AI adviser.
它在 App Store 上有 931,000 条评分。
It has 931,000 ratings on the App Store.
好的。
Okay.
好的。
Okay.
总之，当它开始让年轻女性变得性感时，使用量就上升了。
Anyway, usage went up when it started to bikinify young ladies.
是的。
Yep.
不太好。
Not good.
不。
No.
不。
No.
而且它
And it's
几乎全是男性，这并不奇怪。
almost entirely male, not surprisingly.
有意思。
Interesting.
好的。
Okay.
是的。
Yeah.
不。
No.
我认为，实际上我要说的是，让我在这张图表上感到惊讶的是，Perplexity 如果想要竞争，它应该提供一个以移动设备为中心、用户友好的体验。
I think and actually, the one thing I will say, do you know what's notable to me on this chart here is perplexity dis and Perplexity should if they're gonna compete, that was like a very mobile first consumer friendly experience.
而且我不确定，我很好奇你对 Perplexity 的预测是什么？
And I don't I didn't I'm wondering what's what's your prediction for Perplexity?
我觉得它已经完了。
I think it's done.
我的意思是，是的。
I mean Yeah.
我看不出它除了与 Gemini 或 ChatGPT 有所不同之外，还有什么实际用途。
I don't see how it really serves a purpose outside of, you know, different than what you could do with Gemini or Chachi
是的。
Yeah.
就我个人而言。
Personally.
很长一段时间里，它是一种更简洁的搜索体验。
It was a cleaner search type of experience for a long time.
是的。
Yeah.
而现在它变得
And now it's
不是了。
not.
AI模式和谷歌可能毁了它。
AI mode and Google probably killed it.
是的。
Yeah.
对。
Yeah.
你说得对。
You're right.
我也这么认为。
I think so.
再说那个，比如，Perplexity去年秋天确实有过一次强劲的增长，但现在完全消失了。
And that one, again, like, they there was a strong bump last fall for Perplexity that's completely disappeared.
所以我想问你一个问题，这周CNBC也问过我，我其实挺好奇你会怎么回答。
So I wanna ask you something that that I, was asked on CNBC this week, and I'm actually kinda curious how you would answer.
就是，现在有很多关于英伟达和OpenAI之间关系出现摩擦的讨论。
So, like, there's been all this talk about how NVIDIA and OpenAI are have some friction in their relationship.
就像，Jensen从‘我们将投资高达1000亿美元’变成了‘我们会一轮一轮评估，希望OpenAI再次邀请我们投资’。
Like, Jensen has gone from we're gonna invest up to a 100,000,000,000 to, like, we will evaluate round by round, and we hope that OpenAI invites us to invest again.
在《华尔街日报》上，有一些报道称他对OpenAI的商业模式存在一些担忧。
And in The Wall Street Journal, there was some reports that they were that he had some concerns about the way that OpenAI did business.
然后他们，你知道，我被问到，那么是否对OpenAI的商业模式存在担忧呢？
And then they, you know, I was asked, well, are there concerns about the way that OpenAI does business?
对我来说，基本上我的回应是，看，OpenAI曾经领先。
To me, basically, my response was, look, OpenAI had the lead.
他们使用了公开可用的技术来构建ChatGPT。
They used publicly available technology to build ChatGPT.
他们建立了领先地位。
They built the lead.
其他人开发出类似技术并赶上是不可避免的。
It was inevitable that there would be others that would build, you know, similar technology and they would catch up.
那么，OpenAI的现状如何呢？
And then, you know, so what is the health of OpenAI there?
也就是说，其他人赶上是正常的，因为这本就是不可避免的吗？
You know, is it, like, just, okay, it's normal that others are catching up because it was inevitable?
如果是这样，那么当OpenAI和ChatGPT只是众多竞争者之一时，他们的价值主张又是什么？
And if so, then what is the value proposition for OpenAI and ChatGPT moving forward if they're one of the pack?
你怎么看？
What do you think?
我认为，从OpenAI的角度来看，比如当我们看这张图表时，所有数据都是相对市场份额。
I think I mean, to to take the OpenAI side here, like, when we're even looking at this chart, for example, it's all relative market share.
这并不是总的使用量和总用户数。
It's not total usage and total consumers.
对。
Right.
只要整个市场在增长，而且他们仍然是绝对的主导者，我认为他们无疑是一家有趣的公司。
So as long as the pie is growing and they are still the vastly dominant player, I think they're certainly an interesting company.
是的，我们之前已经讨论过很多次了。
I do then we've talked about this plenty.
他们必须在ChatGPT每月20美元的订阅之外，成功抓住某一个周边业务点，比如……
I they have to nail one of these kind of peripheral business ideas beyond just ChatGPT $20 a month subscriptions, is whether that's So
那么，缺乏专注反而可能是一种优势，因为他们只需要成功一两个就够了。
maybe the lack of focus is is an asset then because they they just need one or a couple.
是的。
Yeah.
但他们必须真正落地实现。
But they have to they have to land it.
他们尝试了很多东西。
They've tried a lot of stuff.
你最近出于好奇用过Sora吗？
Like have you used Sora recently out of curiosity?
我只是在问，比如Grok或者所有这些产品。
I'm just asking between Grox or all these.
是的。
Yeah.
没有。
No.
我的意思是，他们非常擅长制造宏大、炫目的高光时刻，但之后却没有任何东西能像ChatGPT本身以及那些早期的模型更新那样，长时间持续占据主导地位。
I mean, they they are really good at big flashy, exciting moments, but then they haven't had anything that's felt like staying power the way Chachi PD itself and those early model updates and stuff would just completely kinda take over for a long like, a prolonged period of time.
所以我认为，他们确实得相对有意思才行。
So so I think I mean, they have to be relatively interesting.
詹森在报道中确实显得有点含糊，我觉得，尤其是在那之后，比如
Jensen was definitely a bit cagey, I feel, between the reporting and then, like
你读过关于詹森和OpenAI关系的什么内容吗？
What do you read about, like, the Jensen open AI relationship?
某种程度上，他们彼此需要。
In some way, they need each other.
某种程度上，看起来詹森并不太开心。
In some way, it doesn't look like Jensen's all that happy.
但这就是我对这件事的看法：他确实需要OpenAI成功。
Well but that's where my like, how I felt about it is he does need OpenAI to succeed.
他确实需要。
He does.
比如，以NVIDIA目前的估值和故事而言，它需要所有这些AI技术都取得无可争议的成功。
Like, NVIDIA, at the current valuation and story, needs all these, like, AI to just be an unquestioned success.
如果OpenAI彻底失败了，即使大部分价值流向了谷歌或微软，我也认为NVIDIA的光环也会受到很大影响。
And OpenAI, if it completely flops, even if a lot of that value accrues to a Google or a Microsoft, I really think a lot of the shine even on an NVIDIA takes a hit.
但如果价值流向了谷歌，那不过是TPU的成功故事罢了。
Well, if it accrues to Google, that's just like a TPU success story.
是的
Yeah.
这甚至更
That's an even more
关于这一点的数据。
data for this one.
是的
Yeah.
是的
Yeah.
是的
Yeah.
同意。
Agreed.
好的。
Okay.
我们会关注这个的。
That'll be we'll be watching this.
这将是一个非常有趣的故事。
It's gonna be a very interesting story.
好的。
Alright.
比特币。
Bitcoin.
简单来说，比特币从多少涨起来的？
Very briefly, Bitcoin, like, has gone from what was it?
大概是1.10美元？
Like, $1.10?
1.15美元？
$1.15?
一年时间？
A year?
比如，60年代初的时候，1个比特币值
Like, at the beginning of '60 the 1 k
截至昨天。
as of yesterday.
它又涨到了7。
It's back up to 7.
现在是70。
It's at 70.
是的。
Yeah.
我们来看看。
Let's see.
哦，它在10月份时是1.24美元。
Oh, it at $1.24 in October.
所以，虽然不是减半，但自那以后下跌了不少百分比。
So that's, you know, not a halving, but down, what, a good percentage since then.
显然，比特币没有盈利，因此不能像我们那些处境艰难的软件公司那样依赖盈利来支撑。
Obviously, Bitcoin doesn't have earnings, so can't just fall back on the earnings like our beleaguered software companies might over time.
你对当前比特币发生的情况有什么看法？
What is go what is your take on what's happening with Bitcoin?
我一直在认真思考这个问题。
So I I I was thinking about this a lot.
今年已经下跌了35%。
So down 35% this year.
我的意思是，这还是在昨天或今天大约12%的反弹之后。
I mean, that's after the, I think, 12% rebound yesterday or today.
今年到目前为止，对我来说最令人着迷的是，我几乎没有看到任何人谈论它。
The most fascinating thing about Bitcoin so far this year for me has been, I don't see anyone talking about it.
而且，我确信在某些圈子里，讨论依然非常热烈。
And, again, I'm sure within certain circles, there's still pretty intense things.
实际上，我上周在迪拜，我住的酒店正好有一个加密货币活动，天哪，那景象真是令人震惊。
Actually, I was in Dubai last week, and the hotel I was staying at, there was, like, a crypto event going on, and my god, that was just something crazy to see.
但总的来说，我不知道。
But, like, overall, I mean, I don't know.
你有没有听到或看到，它只是在悄悄地消退？我觉得这可能是最危险的事情之一，因为比特币真的需要一种能量来维持它的运转。
Are you hearing or seeing like, it's just kinda quietly fading off and which I think is possibly one of the most dangerous things because it really depends on kind of an energy to keep it moving.
所以，是的，你觉得正在发生什么？
So, yeah, what do you what do you think is happening?
我不知道。
I don't know.
我只是觉得，这属于那种建立在——我不想说群体思维，但它是建立在一个故事上的东西。
I I just think it's one of those things that sort of it's built on I don't wanna say groupthink, but built on a story.
对吧？
Right?
我不知道，当这个故事不再成立时，比特币就会下跌。
I don't know, the story the story when the story doesn't work, Bitcoin drops.
我们已经见过太多次了。
We've seen this so many times.
而当故事成立时，比如总统支持它，主流金融机构也能让你持有，那就行得通了。
And when the story works, the story is okay, like, you know, the president is a, you know, a supporter of this and now you could be able to own it from mainstream financial institutions, then it works.
但我得承认，我持有一些，但我一直觉得背后的故事有弱点——如果不能作为货币使用，那它到底是什么？
But I I I mean, I've and I hold some, I'll admit it, but I've always felt that there was weakness in the underlying story, which is, you know, if it's not gonna be useful as a currency, then what is it?
价值储存？
Store value?
好吧。
Okay.
这完全取决于人们对它的信念。
That's just depends on people's belief in it.
但，但我也觉得这是个很好的观点，因为我其实并不知道这个故事是什么。
But but I also it's a good it's a good point because I don't actually know what the story is.
这对我来说其实是最危险的部分。
That's actually the most dangerous part to me.
就像是，它是对因债务而导致的美国经济恶性通胀的对冲吗？
It's like like, is it hedge against hyperinflationary US economy because of debt?
关于 Bayt 播客
Bayt 提供中文+原文双语音频和字幕，帮助你打破语言障碍，轻松听懂全球优质播客。