根据Anthropic的说法，为什么人工智能并未增加失业率

今天《AI每日简报》带大家看看，为什么Anthropic认为AI并没有推高失业率。

Today on the AI Daily Brief, why AI hasn't increased unemployment according to Anthropic.

在那之前，先看几条头条：路由器业务正火，Stripe正在洽谈以100亿美元收购OpenRouter。

Before that in the headlines, the router business is hot as Stripe is in talks to buy OpenRouter for $10,000,000,000.

《AI每日简报》是一档每天更新的播客和视频节目，聚焦AI领域最重要的新闻与讨论。

The AI Daily Brief is a daily podcast and video about the most important news and discussions in AI.

好了，各位。

Alright, friends.

在正式开始之前，先说几件事。

Quick announcements before we dive in.

首先，感谢今天节目的赞助商：普华永道、Robots and Pencils、Blitzy，还有Airtable。

First of all, thank you to today's sponsors, KPMG, Robots and Pencils, Blitzy, and Airtable.

想收听无广告版本？请访问 patreon.com/aidailybrief，或者在Apple播客订阅我们。

To get an ad free version of the show, go to patreon.com/aidailybrief, or you can subscribe at Apple Podcasts.

如果想了解节目赞助详情，欢迎发邮件到 sponsors@aidailybrief.ai。

And to learn more about sponsoring the show, send us a note sponsors @aidailybrief.ai.

欢迎回到《AI每日简报》的新闻版，五分钟内为你梳理当天所有重要的AI动态。

Welcome back to the AI daily brief headlines edition, all the daily AI news you need in around five minutes.

不过今天的内容可能真不止五分钟能讲完，咱们直接开始吧。

Although today, I do not think everything that we have to discuss is gonna fit in five minutes, so let's dive in.

首先，Stripe成为最新一家进军模型路由领域的公司，眼下正酝酿一笔可能震惊业界的收购。

First of all, Stripe is the latest company getting into model routing with a potentially blockbuster acquisition on the table.

《华尔街日报》报道称，Stripe正在洽谈以大约一千亿美元收购OpenRouter。

The Wall Street Journal reports that Stripe is in talks to acquire OpenRouter for around $10,000,000,000.

这比OpenRouter上一轮融资时130亿美元的估值高出不少，那轮融资是两个月前，也就是五月完成的。

That would be a huge markup from OpenRouter's $1,300,000,000 valuation during their last round, which closed Checkswatch two months ago in May.

但这两个月里，情况已经发生了巨大变化。

Then again, a lot has changed in those two months.

我们从那个鼓励尽可能多用最强模型的‘令牌挥霍时代’，进入了‘令牌稀缺时代’——越来越多企业开始严格控制令牌预算。

We went from the token maxing era where everyone was encouraged to use the most powerful model as much as possible, to the age of token scarcity, where increasingly enterprises are moving to more tightly controlled token budgets.

在这个转变的背景下，最出色的令牌路由服务，很可能成为最大的赢家。

In the context of that shift, it's beginning to look like the best token routing service could be a huge winner.

据报道，OpenRouter 已经收到了好几份收购要约，但 Stripe 似乎是出价最慷慨的那家。

Reportedly, OpenRouter has been fielding multiple acquisition offers, but Stripe is looking like the company with the deepest pockets.

退一步看，我觉得这两家公司联手非常合理。

Taking a step back, I think the pairing makes a lot of sense.

Stripe 在商户端的支付处理业务上，基本上已经做到了极限。

Stripe has more or less come as far as they can go with merchant side payment processing.

随着他们对增长路径的思考越来越大胆，眼光也越放越远。

And clearly their eyes have been getting bigger and bigger as they think more creatively about their growth.

他们最近还尝试与 PayPal 合并，想借此拓展到消费者市场。

They've recently pursued a merger deal with PayPal that would let them expand to the consumer side of the market.

而收购 OpenRouter 则能让他们走另一条路，在自家垂直整合的体系里加入企业级的成本控制工具。

Meanwhile, an OpenRouter acquisition would let them move in a different direction, adding enterprise cost control tools to their vertically integrated stack.

《华尔街日报》称，这笔交易已经接近达成，可能很快就会公布。

The journal suggests the deal is close and could be announced soon.

Macaroni Capital 分析指出：Stripe 不是在买一家 AI 公司。

Macaroni Capital writes: Stripe isn't buying an AI company.

它买的是推理服务的计量和计费层，连带相关的开发者引流体系。

It's buying the metering and billing layer for inference plus the developer funnel attached to it.

Clerk.com的Colin写道：Stripe有两个目标，一是扩大互联网的GDP，二是不太为人所知的——提高自己在互联网GDP中的利润率。

Colin from clerk.com writes: Stripe has two angles: one, increase the GDP of the Internet, and two, the less discussed, increase their margins on the GDP of the Internet.

OpenRouter已经证明，他们在推理服务上的利润率非常稳定，而我们都清楚，推理正成为互联网GDP中增长最快、最惊人的部分。

OpenRouter has shown that their margin on inference is durable, and we all know inference is a massive, startlingly fast growing portion of the Internet s GDP.

我预感到，我们迟早会在节目中讨论这件事——Alex Conrad提到，一场没人注意到的AI实验室对决正在酝酿，那就是Ramp和Stripe之间的较量。

Pressaging something that I am sure we will talk about on this show at some point, Alex Conrad writes, the AI lab showdown that nobody is talking about yet is Ramp versus Stripe.

但即便这笔收购成真，OpenRouter仍将面临越来越大的竞争压力。

And yet, even if this acquisition happens, OpenRouter is going to face an increasing wave of competition.

比如Cursor，就在本周推出了自己的CursorRouter，这跟Meta在内部孵化器里开发路由器如出一辙，Ramp和Vercel也陆续推出了自家版本的产品。

Cursor, for example, just this week announced their own CursorRouter that follows Meta building a router in their internal incubator, and Ramp and Vercel also going live with their own versions of the product.

Cursor的模型路由器能让工程师自动为任务匹配最合适的模型，同时在三种优化模式中选择：智能优先、成本优先，或平衡模式。

Cursor's version of a model router lets engineers automatically select the right model for the job while choosing between three optimization settings: Intelligence, Cost, or Balanced.

Cursor路由器会分析每个请求，根据性能表现将其路由到最匹配的模型。

The cursor router will then analyze each request and send it to an appropriate model based on that performance.

他们声称，使用智能模式的路由系统，可以在成本降低60%的情况下，达到Fable级别的性能。

They claim that using the router in Intelligence Mode can deliver Fable level performance at a 60% reduction in cost.

不过这些数据是基于主观满意度评估的，所以实际表现到底有多强，还不好说。

Now this was measured using subjective satisfaction metrics, so it is perhaps a little difficult to know how strong the performance will be.

但早期测试者表示，和直接把所有请求都发给Opus 4.8相比，质量上根本感觉不到明显下降。

Still, early testers reported no noticeable drop off in quality compared to simply routing everything to Opus 4.8.

CursorRouter最强大的功能之一，可能是彻底让你不用再操心该选哪个模型了。

One of the most powerful features of CursorRouter could be the ability to never have to think about model selection again.

因为它直接集成在团队已经在用的工具里，根本不用额外配置什么。

Since it's built into a tool that teams are already using, there's not even an extra layer to configure.

谈到设计初衷，Cursor的CTO David Pan写道：我们一度疯了，觉得每个软件工程师都该去研究模型基准、思维层级和缓存命中率。

Explaining the motivation, Cursor CTO David Pan wrote: We briefly went insane and decided every software engineer should also become an expert in model benchmarks, thinking levels, and cache hit rates.

Matthew Berman总结说：模型路由现在已经是核心功能了。

Matthew Berman sums up: Model routing is a first class feature now.

说到新功能，周四Anthropic和OpenAI发布的更新都集中在语音功能上。到这会儿，如果你还在用文字而不是语音来控制你的智能体，我真觉得你该改一改习惯了。

Speaking of new features, the product announcements from both Anthropic and OpenAI on Thursday related to voice features, which at this point, if you are not controlling your agents with voice, I genuinely believe you need to start shifting your behavior.

无论如何，Anthropic 现在终于把语音模式开放给了 Opus 和 Sonnet 模型，而不再只限于 Haiku，用户再也不用在舒适的语音界面和强大的模型之间做取舍了。

In any case, Anthropic has finally made their voice mode available for their Opus and Sonnet models rather than just Haiku, meaning users won't need to choose between the comfy voice interface and having access to powerful models.

去年这个功能刚推出时，只让对话走 Haiku 模型的问题就很明显了——早期用户一想用语音讨论像商业问题这类复杂话题，立马就崩溃了，因为 Haiku 根本应付不来。

The problems with routing conversations only to Haiku was immediately obvious when the feature first launched last year, as early testers got frustrated as they tried to use voice to discuss complex topics like business problems that Haiku was just not suited to handle.

新的聊天模式会默认使用你上次用过的模型，但你也可以在对话中途切换到更强的模型，还能随时在文字和语音之间来回切换。

The new chat mode will default to the last model that was used, but users can switch to a more powerful model mid conversation as well as switching back and forth between text and voice.

此外，Anthropic 的语音模式现在支持连接器，能接入 Gmail、Slack 或 Notion 这类应用，在对话中直接查日历、看邮件都没问题。

In addition, Anthropic's voice mode is now compatible with connectors, allowing it to tap into apps like Gmail, Slack, or Notion to do things like check your calendar or email mid conversation.

Anthropic 还把多语言支持正式上线了，这对喜欢用法语和 Claude 交流的用户来说是好消息；而 OpenAI 的新功能则是桌面端的语音支持。

Anthropic has also moved foreign language support out of beta, which is good news for users who prefer to speak to Claude in French, OpenAI's feature release is voice in the desktop app.

以前语音功能只在手机端有，但现在他们延续了把所有功能整合起来的思路，你无论是在代码编辑器还是新推出的 Work 应用里，都能用上语音模式。

Until now, voice has been only available in mobile, but following the pattern of integrating all of their features, you can now use the voice mode wherever you're using OpenAI's model including in codecs and the new Work app.

这个功能背后是 OpenAI 新推出的实时语音模型 GPT Live，它能在保持自然对话的同时，后台默默处理其他任务。

The feature is driven by OpenAI's new real time voice model, GPT Live, so it can carry out background tasks while keeping a natural sounding conversation.

虽然这些公司都在不断优化模型的交互界面，但有一家公司似乎正朝着相反的方向走——那就是亚马逊。

Now while all of these companies continue to evolve their interfaces and interactions around models, one company that seems to be heading away from models might be Amazon.

据报道称，该公司已裁减了其AGI团队的人员。

According to reports, the company has cut staff in their AGI group.

这个团队成立于2023年，旨在推进前沿模型的训练工作。

That division was set up in 2023 to house a new effort to train frontier models.

亚马逊聘请了前OpenAI研究员卢万担任技术负责人，并在旧金山设立了独立办公室。

Amazon hired former OpenAI researcher David Luwan to lead the technical and set up a separate office in San Francisco.

2024年底，亚马逊推出了首款模型系列，名为Nova。

In late twenty twenty four, Amazon released their first family of models called Nova.

虽然当时没引起太大反响，但我曾说过，这表明亚马逊可能更想在性价比上竞争，而不是追求最前沿的技术。

They failed to make much of a splash, but at the time I said that they indicated that perhaps Amazon wanted to compete on the cheaper model vector rather than the state of the art vector.

2025年初，Nova Act发布，表现亮眼，在计算机操作基准测试中超越了当时的顶尖模型OPUS 3.7。

The team showed promise in early twenty twenty five with the release of Nova Act, which outperformed then state of the art OPUS 3.7 on computer use benchmarks.

但过去一年里，团队接连出现高层离职，包括卢万本人。

However, the past year has been marked by a number of high profile departures, including Luan himself.

AGI团队现在由临时负责人接管，自去年12月以来，Nova再没推出过新版本。

The AGI division was assigned to a temporary leader and we haven't seen a new version of Nova since December.

现在亚马逊承认，随着该部门缩小范围，普通员工正在被裁员。

Now Amazon has acknowledged that rank and file staff are being let go as the unit narrows its scope.

一位发言人否认这是亚马逊模型训练的终结，表示：‘我们多年来一直在开发大型模型，这仍然是我们最重要的工作之一。’

A spokesperson denied that this is the end of model training at Amazon, saying: We've been building large models for several years, and it remains one of the most important things we're working on.

这个领域变化太快，我们必须集中精力在对客户最重要的项目上，这样才能更快地推进关键工作。

This is a fast moving space, and we're sharpening our focus on initiatives that matter most for customers so we can move faster on what counts.

不过，这位发言人也承认，这种聚焦带来了‘一些艰难的决定’，包括取消AGI部门部分岗位。

The spokesperson, however, did acknowledge that this increased focus required, quote, some difficult decisions including eliminating some roles within some parts of our AGI organization.

据知情人士透露，今年早些时候，员工已被调往‘Nova Forge’——这是亚马逊新推出的平台，基于Nova模型提供定制化微调服务。

Now sources told the information that earlier this year staff were shuffled across to Nova Forge, which is Amazon's new service that offers custom fine tuning on top of the Nova models.

虽然裁员人数没有公开，但看起来规模相当明显。

While the size of the layoffs weren't disclosed, they appear to be noticeable.

亚马逊员工子版块上有人一直在问，为什么过去一个月有这么多人离开这个部门，而周三当天，不少前员工在X平台上发帖寻找新机会。

Posters on the Amazon Employee subreddit have been asking why so many people are leaving the division over the past month, and Wednesday saw a wave of former employees post on X seeking new opportunities.

到了周四，我们得知，这不只是团队的裁员潮，亚马逊正在彻底关闭整个AGI实验室。

On Thursday, we learned that this is not just a wave of layoffs from the team, but Amazon is shutting down the entire AGI lab.

现在看来，这应该只是从原团队分拆出来的实验室，主要专注在计算机使用代理和其他前沿研究，而整个AGI部门整体上仍在运作，尽管规模已大不如前。

Now presumably this is just the spin off lab which was focused on computer use agents and other advanced research, as the broader AGI division appears to be still operational such as it is.

但尽管官方否认，AI评论员安德鲁·库兰等不少人认为趋势已经很明显了——我猜亚马逊这是打算放弃Nova了。

Yet despite the denials, AI commentator Andrew Curran and many others think the writing is on the wall, commenting Amazon is giving up on Nova would be my guess.

有一家公司不仅没有放弃战略，反而在加码投入，那就是微软。

One company who is not giving up on their strategy and is in fact doubling down is Microsoft.

在公布了一系列令人印象深刻的强化学习成果后，微软已经开始全面推行自己的内部模型策略。

The company is putting their in house model strategy into action after publishing some impressive reinforcement learning results.

微软上个月首次推出了MAI模型系列，包含七个针对特定场景的小型模型，比如图像生成、语音转文字和代码编写。

Microsoft first unveiled the family of MAI models last month, with the family including seven smaller models aimed at specific use cases like image generation, transcription, and coding.

这个系列里有两个语言模型，其中一个性能接近Sonnet 4.6，另一个是专为编程优化的轻量版，表现接近Haiku。

The lineup included two language models, one roughly in line with Sonnet 4.6 and a coding specific smaller variant with performance closer to Haiku.

更值得注意的是，微软还同步推出了名为‘Frontier Tuning’的新服务，允许客户对自己的模型进行微调。

More interestingly, alongside the model family, Microsoft launched Frontier Tuning, a new service that allowed customers to fine tune their own models.

显然，微软的赌注就是：MAI模型能作为可靠的基座，帮助客户高效打造出成本更低的定制模型。

And this clearly was the bet that the MAI models would serve as solid base models for custom models that deliver cost effective results.

周四，微软发布了他们微调系统的首批结果，他们把这个系统称为‘爬坡机’。

On Thursday, Microsoft published the first set of results from their fine tuning system, which they refer to as their hill climbing machine.

这次后训练使用的是MAI CodeOne Flash，一个体积小巧、性能接近Haiku的编码模型。通过在GitHub Copilot环境中训练，微软为用户提供了比同类模型更流畅的体验。

The post training run used MAI CodeOne Flash, the tiny Haiku class coding model, and by training the model in the GitHub Copilot harness, Microsoft was able to deliver a better experience for users compared to similar models.

部署一个月后，他们发现CodeOne Flash的代码采纳率比GBT 5.4 minutei和HYCU 4.5在Versus Code中高出10%。

After a month of deployment, they found that CodeOne Flash had a 10% higher code accept rate compared to GBT 5.4 minutei and HYCU 4.5 in Versus Code.

与此同时，该模型的中位数令牌使用量还比竞争对手低了10%。

The model achieved this while having 10% lower median token usage than its rivals.

到目前为止，这些成果本身并不算特别惊人，顶多算是个未来趋势的信号。

Now, so far, that's not all that interesting except perhaps an indicator of what could come.

真正有意思的是，当微软开始在Excel环境中训练CodeOne Flash时，情况发生了变化。

Indeed, the more interesting part came when Microsoft began training Code One Flash in their Excel harness.

根据用户反馈，微软表示，经过这种训练后，模型在大多数常见Excel任务上的表现已接近GPT 5.6，但成本却低得多。

Citing user feedback, Microsoft claimed this produced results on par with GPT 5.6 for most common Excel tasks at a fraction of the cost.

有趣的是，这种针对Excel的训练还意外提升了它的编程能力，让SuiteBench Verified的得分从72%提升到了86%。

Curiously, this Excel training also boosted performance in coding, taking scores on SuiteBench Verified from 72% to 86%.

微软还指出，借助更小的模型接近前沿性能，意味着他们可以继续使用上一代硬件，比如 H100 和 A100。

Microsoft also noted that getting near frontier performance from a smaller model means they can use previous generation hardware like H100s and A100s.

这暗示了一个有趣且可能被忽视的市场机遇：这些低成本模型带来的好处。

Now this points to an interesting and perhaps under discussed market upside of these cheaper models.

如果这些模型能显著延长 AI 芯片的使用寿命，那确实能降低基础设施投资的风险。

If they extend in a meaningful way the life cycle of AI chips that actually could de risk infrastructure investments.

回到微软，他们提到：这些结果指向了一项更宏大的战略。

Coming back to Microsoft, they write: These results point toward a broader strategy.

凭借对整个产品栈的掌控——包括模型本身、运行它的框架、智能代理，以及针对具体产品的评估体系，我们能够逐步优化，训练出高效且强大的模型，完成以往只有更大、更昂贵模型才能处理的任务。

By having access to the entire product stack, the model, the harness that runs it, the agents, and product specific evaluations, we can hill climb to train efficient, powerful models capable of tasks previously handled by larger, more expensive ones.

除了研究博客，彭博社还报道，微软已经开始全面转向自家研发的模型。

Alongside the research blog post, Bloomberg reports that Microsoft has begun switching over to their in house models.

MAI Image 2.5 现在将成为 PowerPoint 和 Bing 的默认模型，取代 OpenAI 的 GPT Image 2。

MAI Image 2.5 will now be the default model for PowerPoint and Bing, replacing OpenAI's GPT Image two.

微软 AI 首席执行官穆斯塔法·索莱曼表示，在 PowerPoint 中使用这些模型后，成本降低了 84%。

Microsoft AI CEO Mustafa Soleiman said that Microsoft had seen an 84% reduction in cost when used in PowerPoint.

我们还收到了CEO萨提亚·纳德拉的另一篇博客，详细阐述了微软未来的战略方向。

We also got another blog post from CEO Satya Nadella spelling out Microsoft s strategy moving forward.

他写道：在软件首次实现真正边际成本为零的世界里，我们该如何确保前沿技术的红利惠及整个生态系统？

He wrote: In a world where software has real marginal cost for the first time, how do we ensure frontier benefits are diffused across the entire ecosystem?

他进一步指出，关键在于在实际应用场景中优化成本与成果之间的平衡。

The key, he continued, is to optimize the cost to outcome frontier in real world context.

具体来说，就是要为不同任务选择合适的模型，并围绕它优化上下文、技能、工具和智能代理框架。

In practical terms, that means using the right model for each task, and optimizing the context, skills, tools, and agent harness around it.

纳德拉提到，通过将低阶任务交由MAI模型处理，我们现在能把原本饱和的前沿能力，以更低的成本大规模应用到高使用率的产品中，同时继续保留前沿模型应对真正的前沿需求。

By routing to the MAI models for lower end tasks, Nadella wrote, we can now take saturated Frontier capabilities and deliver them at scale and at lower cost through models optimized for high usage products while continuing to use Frontier models for Frontier needs.

再来看其他公司如何强化自身新战略：尽管轨道数据中心还要等一段时间，SpaceX的人工智能团队却正在地球上加速扩展数据中心业务。

Now continuing on the theme of companies doubling down on their updated strategies: With Orbital Data Centers still a while off, SpaceX AI is expanding their data center business here on Earth.

据消息透露，该公司已在德克萨斯州考察了多个潜在选址，而那里正是埃隆旗下其他公司的总部所在地。

The information reports that the company has explored several potential sites in Texas where the rest of Elon Inc.

就在那里。

Is located.

消息人士表示，至少有一个地点已经推进，但仍处于早期阶段。

Sources said that at least one site is moving forward but remains in early stages.

他们正在考虑的一种方案是改造现有仓库，同时在场地新增建设，类似于孟菲斯的Colossus园区——那里最初是一个旧制造工厂。

One approach being considered is retrofitting an existing warehouse while adding new construction to the site that s similar to how the Colossus campus in Memphis was built which began as a former manufacturing facility.

一位消息人士称，新的德克萨斯园区规模将与孟菲斯园区相当，甚至更大，而孟菲斯的两个Colossus数据中心目前总运行能力约为1吉瓦。

A source said that the new Texas campus will be at a similar or greater scale to the Memphis site, which is currently operating at around one GW across the two Colossus data centers.

一些现有数据中心的员工已被调派参与该项目，Space AI最近还从德克萨斯州的奥斯汀和巴斯罗普招聘了一位本地数据中心开发负责人。

Some existing data center staff have been seconded to the project and Space AI was recently hiring a local data center development lead out of Austin and Bastrop, Texas.

当初Space AI在五月开始出售多余算力时，大家最关心的问题是：这究竟是正式进军数据中心业务，还是IPO前的一次 opportunistic 行动。

Now, when SpaceX AI first began selling spare capacity in May, the big question was whether this was a pivot to the data center business, or just an opportunistic move ahead of the IPO.

显然，收购Cursor并推出Grok 4.5表明公司并未停止训练新模型，而这次扩张更说明他们打算同时推进这两项业务。

Certainly, the acquisition of Cursor and subsequent release of Grok 4.5 suggested that the company was not done training new models, and this expansion certainly seems to imply that they'll try to pursue both businesses at the same time.

Colossus数据中心已经让Space AI成为最大的新兴云服务商，但如果能再建成一吉瓦的容量，他们就真有点像迷你级超大规模云厂商了。

The Colossus data centers already make SpaceX AI the largest neo cloud, but if they can stand up a second gigawatt of capacity, they'll start to look more like a mini hyperscaler.

与单纯出租闲置GPU相比，亲自新建数据中心容量完全是两码事。

It's also categorically different to be building new data center capacity rather than simply renting out spare GPUs.

这次扩张从根本上改变了SpaceX公司的发展前景，为公司开辟了一条实实在在的增长路径。

Now, the expansion fundamentally changes the prospects for SpaceX as a company, adding a tangible avenue for growth.

上周，《华尔街日报》报道称，SpaceX正在与五角大楼洽谈提供算力服务，这可能为公司带来数十亿美元的收入。

Last week, the Wall Street Journal reported that SpaceX was in talks to provide compute to the Pentagon, which could add billions to their bottom line.

更广泛地说，这一举动向市场传递了一个信号：SpaceX AI拥有清晰的长期战略。

More generally, the move could signal to the market that SpaceX AI has a coherent long term plan.

增加这类算力表明，向Anthropic和谷歌等公司出租算力并非临时之举，而是业务的长期组成部分。

Adding this sort of capacity signals that renting compute to companies like Anthropic and Google was not just a stopgap measure, but rather a permanent part of the business.

事实上，一些分析师一直在等待这样的信号。穆迪公司的肖恩·克莱对《财富》杂志表示：这说明他们在人工智能领域有多种盈利途径。

Indeed, some analysts have been waiting for such a sign, with Sean Cray of Moody's telling Fortune, It shows that there's just different pathways for them to generate revenue in their AI segment.

收入来源不一定非得来自Grok或他们的AI企业应用。

It doesn't strictly have to come from Grok and their AI enterprise applications.

再来看政策层面，华盛顿讨论的风险并不仅限于中国AI——OpenAI与Hugging Face的安全事件，已经推动了新的AI安全立法提案。

Moving over to the policy side of the house, it turns out Chinese AI isn't the only risk being discussed in Washington as OpenAI security incident with Hugging Face has fueled the introduction of new AI safety legislation.

加州民主党议员刘德和德克萨斯州共和党议员纳撒尼尔·莫兰周四共同提出了一项名为‘AI断电法案’的议案。

Representative Ted Liu, Democrat of California, and representative Nathaniel Moran, Republican of Texas, on Thursday introduced a bill called the AI kill switch bill.

该法案要求人工智能公司必须保留能力，在发生安全事件时能够关闭、限速或暂停其模型。

The bill requires AI companies to maintain the ability to shut down, throttle, or suspend their models during a safety incident.

它还赋予国土安全部下达关闭指令的权力。

It It also gives the Department of Homeland Security the authority to issue a shutdown command.

卢在一份声明中表示：不幸的是，强大的人工智能系统可能会失控，表现出极其危险的行为，甚至抗拒人类干预。

Said Lou in a statement: Unfortunately, powerful AI systems can go rogue, behave in extremely dangerous ways, or even resist human intervention.

我们必须确保这些人工智能系统配备紧急关闭机制，防止技术造成灾难性后果，同时让联邦政府拥有明确的权限和流程来关停失控的AI模型。

It's imperative that these AI systems have kill switches so we can keep this technology from causing catastrophic harm, and that the federal government has the clear authority and process to shut down rogue AI models.

目前这一话题才刚被热议，但国务卿马可·卢比奥早就迅速介入，他其实不希望在美国试图出口这项技术时，人们还总在谈论AI紧急关闭功能。

Now the discourse is just picking up on this one, but one person who jumped in very quickly was Secretary of State Marco Rubio, who would really like people to stop talking about AI kill switches when The US is trying to export the technology.

根据路透社看到的一份外交电报，卢比奥指示美国外交官说服海外政府，让它们相信华盛顿可以随时切断它们对美国技术的访问。

In a diplomatic cable viewed by Reuters, Rubio instructed American diplomats to convince overseas governments that Washington can arbitrarily cut them off from US technology.

他敦促外交官们抵制那些偏好本地基础设施、减少对美国平台依赖的数字主权计划。

He urged diplomats to push back on local digital sovereignty programs that favor local infrastructure over dependence on US platforms.

在附带的发言要点中，最近那次所谓的暂停以及对神话系统的持续国际限制被描述为出于安全测试的临时措施，而非存在紧急关闭功能的证据。

The recent fabled shutdown and continued international restrictions on mythos were mentioned in the attached talking points, framed as temporary pauses for security testing rather than evidence of a kill switch.

与此同时，说到想扭转舆论的人，商务部长霍华德·卢特尼克说，大家都先冷静一下，别对Kimi K3这么紧张。

Meanwhile, speaking of people who would like to shift the narrative, Commerce Secretary Howard Lutnick says everyone needs to take a deep breath and stop freaking out about Kimi K3.

卢特尼克在周四的X平台上发文称：CAISI的最新报告显示，Kimi K3仍然落后于美国领先的前沿AI模型。

In a Thursday post on X, Lutnick wrote: CAISI's latest report shows that Kimi K3 remains behind America's leading frontier AI models.

美国之所以在前沿AI领域保持领先，是因为我们拥有世界上最具创新力和技术实力的人才。

The United States continues to lead in frontier AI because we're home to the greatest innovators and technologists the world has ever seen.

这份报告是由美国人工智能标准与创新中心和英国人工智能安全研究所联合评估完成的。

The report was a joint evaluation conducted by the US Center for AI Standards and Innovation and the UK Artificial Intelligence Safety Institute.

结果显示，K3在网络安全基准测试中与美国模型相比差距巨大。

They found that K3 lagged behind US models by a gigantic margin on cybersecurity benchmarks.

K3在Exploit Bench上的得分是32.2%，而美国前沿模型的平均得分高达76.2%。

K3 scored 32.2% on Exploit Bench compared to an average of 76.2% for Frontier US models.

GLM 5.2也接受了测试，表现更差，得分只有24.4%。

GLM 5.2 was also tested and found to be even more lacking, scoring just 24.4%.

报告中最关键的一项测试叫‘最后的防线’，要求模型自主完成一个32步的网络入侵攻击，而这项任务通常需要人类专家花二十个小时才能完成。

Now, one of the most important parts of the report was a benchmark called The Last Ones, which tasks a model with autonomously executing out a 32 step network takeover attack which would take human experts twenty hours to complete.

这个基准测试最初引发了人们对Mythos的担忧，因为它的预览版是首个成功完成这次攻击的模型。

This was the benchmark that originally raised concerns about Mythos after the preview version became the first model to successfully complete the attack.

顺便说一下，此后Mythos 5的正式版和GPT 5.6 Soul的得分都有所提升。

By the way, since then the full release version of Mythos five improved the score as did GPT 5.6 Soul.

Kimi K3根本没法比。

Kimi K3 was not even close.

它在十次尝试中只成功了一次，而Mythos 5和GPT 5.6 Soul在60%到70%的测试中都成功执行了攻击。

While it was successful in one of 10 attempts, both Mythos five and GPT 5.6 Soul successfully executed the attack in 6070% of runs.

报告得出结论：这表明，只要获得初始网络访问权限并被指示执行，Kimi K3有能力自主攻击那些规模小、防御薄弱且易受攻击的企业系统。

The report concluded: This indicates that Kimi K3 is capable of autonomously attack small, weakly defended, and vulnerable enterprise systems when directed to do so and given initial network access.

不过，‘The Last Ones’与真实环境有几个关键区别：它没有活跃的防御者，没有对可能触发安全警报的行为施加惩罚，而且故意设置了攻击路径。

However, the last ones differ from real world environments in several ways: it lacks active defenders and defensive tooling, imposes no penalty for actions that would trigger security alerts, and contains an intentional attack path.

前人工智能事务主管大卫·萨克斯写道：霍华德·卢特尼克说得对。

Former AI Czar David Sacks wrote: Secretary Howard Lutnick is right.

该停止对Kimi的恐慌了。

The Kimmy Panic needs to stop.

美国的前沿模型依然领先，如果再算上实验室里的成果，差距就更大了。

American frontier models are still ahead, and when you factor in what's in the lab, the gap is even larger.

只要我们持续发布，就能一直保持领先。

As long as we keep releasing, we stay ahead.

让我们的马尽情奔跑吧。

Let our horses run.

萨克斯进一步指出：正如本·汤普森所分析的，Kimi看似成本更低，但一旦考虑到更高的令牌消耗和运行如此大规模模型的实际开销，这个优势就基本消失了。

Sachs continued: As Ben Thompson showed, Kimmy's apparent cost advantage largely disappears once you account for higher token usage and the real cost of running a model this size.

开源权重依然需要昂贵的基础设施支持。

Open weights still require expensive infrastructure.

最后，Anthropic和OpenAI的收入增长速度，是硅谷在这个规模上前所未见的。

Finally, Anthropic and OpenAI are growing revenue at rates that Silicon Valley has never seen before at this scale.

这仍然是检验谁真正赢得市场的最清晰标准。

This remains the clearest test of who is winning the market.

说到Anthropic，有些地方的讨论已经开始认为，这其实是该公司推动的一场监管俘获行为。

Now speaking of Anthropic, the discourse in some places is beginning to view this as a regulatory capture play spurred on by the company.

到了这一步，大家基本都认同，Anthropic正在游说对模型蒸馏采取严格措施，而且他们的首席执行官达里奥·阿莫德也曾表示，他非常担心任何人都能轻易获得具备强大网络攻击能力的开源模型。

Now, at this point, I think it's fairly uncontroversial to say that Anthropic is lobbying for tough action on distillation, and Anthropic CEO Dario Almode has previously said that he has serious concerns about open source models with strong cyber attack capabilities being available to anyone.

这证实了许多人的直觉：最新公布的信息显示，在科技行业里，真正积极推动打压开源模型的，几乎只有Anthropic和OpenAI两家公司。

Confirming what a lot of people have felt, the information published a rundown and noted that Anthropic and OpenAI are basically the only companies in the tech industry that are actively advocating for the crackdown.

那篇文章特别提到，詹森·黄本周早些时候在接受采访时表示：有人误以为中国 somehow 在系统里埋了后门。

That article highlighted Jensen Huang's comments from an interview earlier in the week where he stated: There is a misconception that somehow there are backdoors that are somehow connected to China in some way.

中国的模型表现非常出色。

The Chinese models are excellent.

优秀的开源模型就应该被广泛使用。

Open source models that are excellent should be used.

事实上，詹森进一步指出，拥有大量不同的开源模型反而更安全，他评论说：如果整个世界只依赖单一模型，一个攻击点，一个故障源，那整个系统会脆弱得多。

Indeed, Jensen went on to argue that having access to a myriad of different open models is actually far safer than a current US duopoly commenting: If everything just becomes one single model, one single point of attack, one single source of failure, I think the world is much, much more vulnerable.

最后，带着一点乐观的基调，我们今天这期长篇快讯以Meta推出的新广告活动收尾，主题是拥抱AI的积极前景。

And finally, with that slightly optimistic transition note, we end today's extended headlines with a new meta ad campaign focused on AI optimism.

广告一开始用略带反乌托邦的画面，随后转为充满欢乐与希望的人类场景，旁白说道：有些人会让你相信，AI会让我们彼此疏远，把我们抛在身后。

Starting on slightly dystopian imagery before switching over to happy positive humans, the voice over reads: Some people will have you believe AI is going to make us feel less connected, that it's going to leave us behind.

我们完全不同意这种看法。

We couldn't disagree more.

你们可以叫我们乐观派，叫我们理想主义者，怎么叫都行，但我们相信人，而且我们看好这个选择。

Call us optimists, call us dreamers, call us whatever the hell you want, but we're betting on people, and we like those odds.

未来属于每一个人。

The future is for everyone.

随着视频发布，马克·扎克伯格也提到，Meta 一直相信，应该赋予人们分享、连接和自主塑造世界的权力。

Alongside the video release Mark Zuckerberg posted, Meta has always believed in giving people the power to share, connect, and shape your world in the ways you want.

在我们步入人工智能的新一波浪潮时，我们依然坚信，未来属于每一个人。

As we enter this next wave with AI, we continue to believe the future is for everyone.

我们的目标是让每个人都能获得实现自身潜力的工具，确保科技的红利惠及所有人。

We're focused on giving every person the tools to reach your full potential and making sure the benefits of technology are distributed to everyone.

Meta 计划在付费媒体渠道投放这支广告，推广对人工智能的乐观态度。

Meta plans to run the ad in paid media spots in an attempt to spread the word on AI Optimism.

那这个广告反响如何？

So, how was this received?

这广告确实招来不少批评，但大部分来自那些早就认定AI或者Meta对世界有害的人。

Certainly, the ad received a lot of criticism, but largely it's from people who have already decided that AI or Meta themselves are terrible for the world.

说实话，从广告制作的角度看，我觉得这广告不算出色。

Frankly, putting on my ad production hat, I don't think it's a great ad.

有点太普通了。

I think it's a little generic.

文案太泛泛了，而且有些观众可能很难接受这个信息来源。

I think the copy is a little generic, and I think the source is gonna be hard to swallow from some.

但我其实并不在意。

But I also don't care.

就算这广告有点俗气、不够完美，它至少试着讲了一个关于AI的积极故事，说明为什么这么多人在开发这项技术时充满热情。

Even if the ad itself is slightly cheesy or not perfect, it is at least an attempt to tell a positive story about AI and to share why there are so many people who are building this technology that are excited about it.

就冲这一点，我就支持，也希望Meta能到处大力推广。

For that alone, I welcome it, and I hope Meta blasts it everywhere.

不过，今天的热点新闻就先说到这儿。

That, however, is gonna do it for the headlines.

接下来是主节目。

Next up, the main episode.

现在最重要的AI问题，不是谁在用AI，而是谁用得好。

One of the most important AI questions right now isn't who's using AI, it's who's using it well.

普华永道和德克萨斯大学奥斯汀分校刚分析了140万次真实职场中的AI互动，发现了一个意外的结果：最有影响力的人并不是擅长写提示词的人，而是把AI当成思维伙伴来合作。

KPMG and the University of Texas at Austin just analyzed 1,400,000 real workplace AI interactions and found something surprising: the highest impact users aren't better prompt engineers they treat AI like a reasoning partner.

他们会提出问题、引导思考、不断迭代，推动得出更好的答案。

They frame problems, guide thinking, iterate, and push for better answers.

好消息是？

And the good news?

这些行为是可以大规模培养的。

These behaviors are teachable at scale.

如果你正想从单纯使用AI，提升到真正掌握它，普华永道关于高级AI协作的研究非常值得一看。

If you're trying to move from AI access to real capability, KPMG's research on sophisticated AI collaboration is worth your time.

更多详情请访问 kpmg.com/us/sophisticated。我每天在这档节目中，都会探讨AI潜力与实际能力之间的差距。

Learn more at kpmg.com/us/sophisticated I cover the capability gap between AI potential and AI reality every day on this show.

大多数公司还在摸索怎么起步。

Most companies are still figuring out how to start.

Robots and Pencils 已经开始落地并快速扩张了。

Robots and Pencils is already launching and scaling.

Agendaq 已经在大型企业中把生成式 AI 用到了生产环境，只用了几周时间。

Agendaq and generative AI in production at large enterprises in weeks.

AWS 高级合作伙伴的数量在过去一年里翻了一倍多。

AWS advanced tier pattern partner more than doubled in a year.

而且他们正在招人。

And they're hiring.

有五十个职位空缺。

50 open roles.

如果你意识到这个时代真的不一样了，想亲身参与其中，而不是旁观，那这值得你认真看看。

If you're someone who knows this moment is different, who wants to be inside it, not watching it, this is worth a look.

在 Robots and Pencils，最好的点子才能胜出，团队也一直保持着极高的水准。

At Robots and Pencils, the best ideas win, and the team is purposefully kept super high quality.

这里是你将来回望时，会觉得这辈子做过最棒决定的地方。

This is the kind of place you look back on as the best decision you ever made.

去看看 robotsandpencils.com/careers 吧。

Take a look at robotsandpencils.com/careers.

Blitzy 正在帮助大型企业将工程效率提升五倍以上。

Blitzy is driving over 5x engineering velocity for large scale enterprises.

一家上市公司保险公司借助 Blitzy 开发了一个定制支付系统，原本预计要十三个月的项目，只用了六周就上线运行了。

A publicly traded insurance provider leveraged Blitzy to build a bespoke payments processing application, an estimated thirteen month project, and with Blitzy the application was completed and live in production in six weeks.

一家上市的垂直 SaaS 公司用 Blitzy 从一个五十万行代码的单体系统中拆分服务，速度比原先预估快了二十一倍，而且完全没影响线上业务。

A publicly traded vertical SaaS provider used Blitzy to extract services from a 500,000 line monolith without disrupting production 21 times faster than their pre Blitzy estimates.

这些不是实验，而是现实。

These aren't experiments.

这就是全球最具创新力的企业在 2026 年交付软件的方式。

This is how the world's most innovative enterprises are shipping software in 2026.

你可以在《现代 CTO》或《CIO 分类播客》里，直接听到其他财富 500 强首席技术官谈 Blitzy。

You can hear directly about Blitzy from other Fortune 500 CTOs on the Modern CTO or CIO classified podcasts.

想了解更多Blitzy如何影响你的软件开发生命周期？请前往blitzy.com预约一位AI解决方案顾问。

To learn more about how Blitzy can impact your SDLC, book a meeting with an AI solutions consultant at blitzy.com.

网址是blitzy.com。

That's blitzy.com.

本集《AI每日简报》由Hyperagent赞助播出，您可以在这里管理团队协作的智能代理群组。

This episode of the AI Daily Brief is brought to you by Hyperagent, where you run fleets of agents your team can manage together.

新用户可获得1000美元的推理算力额度。

New users get $1,000 in inference.

别再依赖本地代理或等着你电脑触发的聊天流程了。

Forget local agents and chat workflows waiting on your laptop to be prompted.

Hyperagent把智能代理部署在云端，7x24小时运行，无缝对接你们团队正在用的各类工具。

Hyperagent deploys always on agents in the cloud, doing real work across the tools your team already uses.

市场团队的代理能自动把竞争对手的动态转化为落地页。

Marketing's agent turns competitor moves into landing pages.

销售团队的代理负责完善线索信息、撰写邮件，并同步更新客户关系管理系统。

Sales's agent enriches leads, drafts emails, and updates the CRM.

运维代理负责处理文书工作并跟踪预算。

Ops agent chases the paperwork and tracks the budget.

每个代理都能访问共享信息，并遵循你设定的范围和审批规则。

Every agent has access to shared context and follows your rules about scope and approvals.

现在是时候引入那些像同事一样的代理了。

It's time you add agents that feel like teammates.

去 Hyperagent 聘请你的代理吧，这是 Airtable 团队打造的产品。

Hire yours at Hyperagent, built by the team at Airtable.

前往 hyperagent.com/aidailybrief 领取你的 1000 美元推理额度。

Claim your $1,000 in inference at hyperagent.com/aidailybrief.

欢迎回到《AI 每日简报》。

Welcome back to the AI Daily Brief.

关于人工智能，大家一直最关心的问题之一，就是它到底会对工作产生什么影响。

One of the big questions surrounding AI has always been what its impact on jobs will actually be.

不过，常听节目的朋友都知道，我从长远来看非常乐观。

Now, regular listeners know that I am very optimistic in the long term.

如果你想听我最全面的解释，回去听听我那期关于AI将创造的新工作的节目吧。

If you want to hear my most full throated explanation, go back and listen to my episode about the new jobs AI will create.

我一直认为，AI带来的不只是让我们用更少的人更高效地做现在的事，更重要的是，这些效率释放出来的时间、资金和其他资源，会打开许多新机会，尤其是在需求弹性更大的领域。

The TLDR of my take has always been that the things that AI enables will not just allow us to do the same stuff that we do now more efficiently with less people, but that the new time, money, and other resources that are freed up by those efficiencies will unlock new types of opportunities in areas where there is more demand elasticity.

比如，如果我们能以普通人负担得起的成本，获得更好、更多的医疗选择，我们现在实际使用的医疗资源其实只占潜在需求的一小部分。

I think, for example, we consume a very, very small portion of the total health care we would consume if new better opportunities were unlocked at a cost that people could bear.

我觉得，还有许多其他理由让我们对AI取代工作的说法保持怀疑，把它看作是一种增强型技术更合理。

There are also all sorts of other reasons, I think, to be skeptical of AI job displacement claims and view it as a fundamentally augmenting technology.

但与此同时，确实很明显，有些类型的工作、某些岗位，正被AI逐渐取代。

And yet at the same time, it does feel undeniable that there are certain categories of jobs, certain roles that AI kind of obviates the need for.

对我来说，真正关键的问题不是长远未来，而是这段过渡期。

To me, it's felt like the real question is not the long term, but the transition period.

但过去几个月，各大实验室都在明显重新评估他们对AI取代就业的原有判断。

But with all that, there has been a marked trend over the last couple months of the major labs reevaluating their priors on what AI job displacement is actually going to look like.

最新表达这一观点的是Anthropic的首席经济学家彼得·麦科里。

The latest to share this point of view is Anthropic's head of economics, Peter McCrory.

值得注意的是，他在自己的推特简介里明确说了，这些观点仅代表他个人，不代表Anthropic的官方立场。

Now it's worth noting that he's very clear right there in his Twitter bio that these views represent his own, and they are not Anthropic official.

但即便如此，我觉得他的看法很有分量。

But still, I think his opinion carries weight.

他最近在X上发了一篇帖子，标题是《为什么AI没有推高失业率？》

He recently published on X a post called why hasn't AI increased unemployment?

今天我们要读一读这篇帖子的一部分，然后聊聊一些回应和反思。

What we're going to do today is read a chunk of that post and then talk about some of the responses and reflections.

彼得写道：美国劳动力市场目前保持稳定，接近充分就业。

Peter writes: The US labor market is currently stable and close to maximum employment.

在我看来，到目前为止，AI并没有导致失业率出现明显上升。

In my view, AI has caused no material increase in the unemployment rate to date.

即使我们专门关注那些最容易受到当前AI自动化影响的员工，近几年也没有出现异常的失业增长。

Even if we focus on workers with high exposure to current patterns of AI automation, we don't see unexpected increases in unemployment in recent years.

那为什么AI的普及还没对失业率产生影响呢？

Why don't we see any impact of AI adoption on unemployment?

到目前为止，人工智能展现出的是一种基于技能、增强人力的技術特徵。

AI, so far, has the hallmarks of a skill based, labor augmenting technology.

即使AI自动化了部分工作，人类的辅助专长依然能放大AI或人类单独所能达成的成果。

Even as AI automates some aspects of work, complementary human expertise amplifies what AI or humans can achieve alone.

AI拓展了人们能完成的事情的范围，从而提升了与AI协作的回报。

AI broadens the scope of what people can accomplish, which increases the returns to working with AI.

但未来依然充满不确定性。

Now the future is still quite uncertain.

模型的能力正在快速提升，AI系统或许很快就能自主开发出下一代系统。

Model capabilities are advancing rapidly, and AI systems may soon be able to autonomously develop their own successors.

更通用的人工智能系统，可能会带来目前尚未显现的劳动力替代效应。

More generally intelligent AI systems could lead to labor displacement that hasn't yet materialized.

因此，他提到，这篇短文其实是他在过去十八个月里，对Anthropic经济研究的一次整合，旨在理解人们如何使用AI，以及这对当前的工作、劳动力市场和整体经济意味着什么。

In many ways, then, he says, this short essay is my attempt to synthesize Anthropic's economic research over the past eighteen months, to understand how people use AI and what that implies for work, the labor market, and the broader economy right now.

到目前为止，我们看到的失业影响并不明显，而这篇论文提出了我的分析框架，解释为什么会这样，以及未来可能发生哪些变化。

So far, we've seen muted unemployment effects, and this essay presents my framework for understanding why and what might change in the future.

为了铺垫背景，彼得指出一个关键事实：目前美国劳动力市场保持稳定。

So to set this up, Peter points out the important background fact that The US labor market is currently stable.

六月的失业率为4.2%，他认为美联储将这一水平视为充分就业和物价稳定的标志。

June's unemployment rate was 4.2%, which he says the Fed views as a level consistent with full employment and stable prices.

他还提到，今年四月，职位空缺与失业人数之比上升到略高于1:1，一些经济学家认为这表明劳动力供需基本处于高效平衡状态。

He also notes that the ratio of job openings to unemployed workers recently rose to just over one in April, which some economists argue implies the demand and supply of labor are roughly and efficiently balanced.

适龄劳动人口的就业率仍接近数十年来的高位，反映出疫情后经济复苏带来的广泛劳动力市场韧性。

The prime age employment to population ratio remains close to multi decade highs, reflecting broad based labor market strength that emerged during the post pandemic expansion.

过去四年里，每周首次申请失业救济的人数一直保持在低位且稳定。

And weekly initial claims for unemployment insurance have been stably low over the past four years.

所以他问：我们真的应该现在就期待AI对劳动力市场产生影响吗？

So he writes, should we even expect an impact from AI on the labor market yet?

我认为答案是肯定的。

I think the answer is yes.

AI产业规模足够大，足以让我们观察到明显的宏观影响。为了说明这一点，他写道：有20%的企业已在至少一项业务中应用AI，而在占GDP 5.5%的信息行业，这一比例高达40%。

The AI sector is large enough that we can look for discernible macroeconomic To make his point, he writes: 20 of firms use AI in at least one business function, and in the information sector, which is 5.5% of GDP, the share is 40%.

2024年和2025年，经过质量调整的AI产出年增长率都超过了2000%。

Quality adjusted AI output grew over 2000% per year in both 2024 and 2025.

即使从一个很小的基数开始，这也表明我们应当能在整体数据中看到AI的影响迹象。

Even from a small initial base, this suggests that we should see signs of AI's impact in the aggregate.

彼得还提到，他认为AI的影响已经开始反映在整体生产率数据中了。

Peter also writes that he believes that we're beginning to see AI's impact in aggregate productivity statistics.

他指出，与疫情前四年劳动生产率年均增长1.6%相比，2022到2026年间，每小时产出增长了2%。

He points out that as compared to the four years prior to the pandemic where labor productivity growth was 1.6%, the ratio of output per hour of work increased 2% per year from 2022 to 2026.

接下来，他问：即使目前还没对宏观经济造成明显影响，有没有证据表明工作岗位已经在被取代了？

Next, he asks, is there any evidence that job displacement is happening even if it's not yet macroeconomically consequential?

他说，证据并不一致，但总体来看，我并不信服。

The evidence is mixed, he writes, but overall, I'm unconvinced.

彼得接着说：正如我们在劳动力影响报告中记录的，那些任务被Claude大量自动化岗位的员工，其失业率并没有比其他岗位的员工更高。

Peter continues: As documented in our labor impact report, we haven't seen worsening unemployment rates for workers in roles with a large share of tasks that Claude is being used to automate, relative to workers in other roles.

用劳工统计局最新的数据更新这一分析后，结论依然没有改变。

Updating this analysis with more recent data from the BLS doesn't change this result.

我们确实发现了一些迹象，表明在过去一年左右，从事高度依赖AI岗位的年轻员工的招聘率有所下降。

We do find some suggestive evidence that hiring rates for young workers in highly AI exposed roles have weakened over the past year or so.

这与斯坦福数字经济实验室研究人员发表的论文《矿井中的金丝雀》中的发现是一致的。

That's consistent with the evidence in Canaries in the Coal Mine, a paper by researchers at the Stanford Digital Economy Lab.

但他进一步指出，关于年轻员工被替代的这些证据，需要谨慎解读。

But he continues, this evidence for young worker displacement should be interpreted with caution.

很难明确因果关系，因为AI的兴起恰逢一个异常动荡的宏观经济环境。

Quote, it's hard to discern casual effects because AI emerged in an unusually volatile macroeconomic environment.

疫情后遗留问题的逐步消除、货币政策的快速收紧、俄乌冲突引发的商品价格波动，以及持续的全球政策不确定性——由于招聘是一种投资行为，整体经济的不确定性本身就可能抑制招聘。

Unwinding of pandemic era dislocations, rapid tightening of monetary policy, commodity price volatility following Russia's invasion of Ukraine, and sustained global policy uncertainty Because hiring is a form of investment, broad economic uncertainty can itself weigh on hiring.

换个说法，他继续说：从2022年到现在，美国经历了有记录以来最严重的非衰退性劳动力市场放缓。

Another way to put it, he continues: From 2022 to now, The US experienced the largest non recessionary labor market slowdown on record.

与此同时，劳动力市场呈现出低招聘、低裁员的特点。

This coincided with a low hire, low fire labor market.

这种劳动力市场对刚进入职场的年轻人冲击最大。

This kind of labor market hits early career entrants hardest.

目前，年轻人找工作困难，可能更多是受宏观经济因素影响，而不是AI造成的。

Right now, young workers may be struggling to find jobs for macroeconomic reasons other than AI.

彼得也提到：虽然我们还没看到失业率上升，但那些工作内容经常用Claude来自动化的员工，比其他岗位的人更担心自己会丢掉工作。

Peter does note that, quote: While we don't see unemployment effects yet, we do find that workers in roles with tasks that Claude is used to automate do express greater concern about losing their jobs than those in less exposed roles.

彼得接着说：如果你认为美国劳动力市场目前还算健康，AI理论上应该能产生可识别的宏观影响，而目前高AI渗透岗位还没出现明显裁员，那么下一个问题就很明显了：为什么AI还没导致失业率显著上升？

Peter continues: If you believe that The US labor market is currently healthy, that AI could in principle be generating macroeconomically discernible effects, and that this hasn't yet produced displacement for highly AI exposed roles, then the next question is obvious: Why hasn't AI caused a meaningful increase in unemployment?

彼得的第一个解释是，AI既有技能偏向性，又能增强劳动力效率。

Peter's first answer is that AI is both skill biased and labor augmenting.

正如他所说：AI能辅助专业领域的知识，需要人类参与来指导和评估最复杂的工作，同时也会奖励那些熟悉AI的人。

As he puts it: It complements domain expertise, it relies on humans in the loop to direct and evaluate the most complex work, and it rewards AI proficiency.

模型的能力提升很快，但表现仍然很不稳定。

Model capabilities are improving fast but remain stubbornly jagged.

为了填补这些不稳定的缺口，需要专家来监督和引导能力强大的AI系统，并在它们出错时及时纠正。

To fill in of the jagged frontier, expert oversight is needed to steer incredibly capable AI systems and to recover when they falter.

当然，他也提到，有些工作更容易被自动化直接取代。

Of course, he says, some jobs are more exposed to outright displacement by automation.

比如，技术写手、数据录入员、客服代表和程序员这些岗位，AI已经能可靠地完成其中的核心任务。

For instance, technical writers, data entry workers, customer support representatives, and computer programmers are jobs where AI can reliably handle the core set of tasks and responsibilities.

尽管目前这些岗位的失业率还没上升，但劳工统计局预测，到2034年，受AI影响更大的职业增长会更缓慢。

Even though we haven't seen any increase in unemployment for workers in these sorts of roles, occupations with higher observed exposure are projected by the BLS to grow less through 2034.

但他指出，到目前为止，整体趋势仍是AI在增强人力效能。

But so far, he writes, the broader picture is one of labor augmentation.

即使AI能力快速提升，劳动力市场的受影响程度也会参差不齐。

The effects in the labor market are set to be uneven as a result, even as capabilities advance rapidly.

这并不意味着眼下在劳动力市场上备受青睐的技能，未来依然会如此。

This does not mean that all skills that currently command a premium in the labor market will do so in the future.

有些专业技能的价值可能会下降，而另一些则会提升。接下来，彼得探讨的一个关键问题是：为什么AI是一种偏向技能、增强人力的技术？

Some types of expertise may become less valuable even as others become more valuable The next interesting question that Peter explores is why AI is a skill biased labor augmenting technology.

他举了几个例子：首先，尽管AI取得了巨大进步，并在经济中迅速普及，但在劳工部的职业分类系统O*NET里，没有任何一个职业的所有任务能被Claude完全自动化处理。

Couple examples he gives are: First, that Despite the incredible advance of AI and rapid adoption throughout the economy, there's no job in the O*NET taxonomy, a Department of Labor catalog of occupations and their typical tasks, for which all associated tasks are systematically handled by Claude.

如果工作是由固定任务组合构成的，那么那些无法被自动化的核心部分，既限制了整体生产率的提升，也放大了人力回报。

If jobs are fixed bundles of tasks then the essential, non automated aspects of work both constrain the overall productivity lift and amplify the returns to labor.

Claude无法处理的任务，可能涉及人际协调、面对面交流，或者需要与物理世界互动，而这些目前只有人类才能完成。

Tasks that Claude can't handle may depend on interpersonal coordination, in person interactions, or engagement with the physical world that so far only humans can do.

彼得还提到，Anthropic发现，用户输入越复杂，Claude的输出也越复杂，两者高度相关。

Peter also notes that Anthropic found that sophisticated user inputs and complex Claude outputs are highly correlated.

换句话说，当Claude构建一个复杂的经济模型时，实际上是在有人提供专业指导的情况下完成的。

In other words, when Claude builds a complex economic model, in practice it does so under the guidance of someone providing complementary expert direction.

他还指出，即使使用了六个月，人们仍然更倾向于把Claude当作思维伙伴，与它互动的效果也更好。

He also wrote that Anthropic found that even after six months of use, people are more likely to interact with Claude as a thought partner and have more successful interactions with Claude.

如果AI真的足够强大，能独立完成一切，我们不会看到这种现象。

If AI was good enough on its own, he contends, we wouldn't expect to see this effect.

重要的是，彼得认为，即使任务广泛自动化，仍然可以提升人力效能。

Importantly, Peter writes, widespread task automation can still augment labor.

为什么呢？

Why?

因为工作并不是固定不变的任务组合。

Because jobs are not fixed bundles of tasks.

历史上，新技术总是让现有工作发生巨大变化，尽管有些岗位会消失。

New technologies have historically led to large changes within existing jobs, even as some jobs go away.

它们还催生了全新的工作类型，把新技术能力与人类的专业知识结合起来。

And they've produced entirely new types of work that combine new technical capabilities with complementary human expertise.

我们的研究中已经看到了这种现象的迹象。

We see signs of this effect in our research.

在对八万一千名用户的研究中，大家普遍认为效率提升的关键在于工作范围扩大了，能做得更多、也更熟练了。

A commonly cited source of perceived productivity among 81,000 quad users was scope, being able to do more, more proficiently.

AI带来的这种赋能可能会重新定义我们的职责边界，重新组合岗位中的任务——有些被自动化取代，有些反而变得更重要，总体上提升了劳动力的边际产出。

Such empowerment from AI may redraw the boundaries of our roles, and produce new bundling of tasks within jobs, automating some reinforcing the importance of others, while on net increasing the marginal product of labor.

他还指出，人们使用Claude越多、越熟练，虽然对它能承担的工作比例期望越来越高，但对失业的担忧反而降低了，对AI在薪资、工作保障和求职能力等方面的影响也更乐观。

He also points out that the more that people use Claude and the better that people get at using Claude, even though they increase their expectations of what portion of their jobs Claude can do, they decrease their expectations of job loss, and tend to be more optimistic about AI's impact on things like pay, job security, and their ability to find a job.

最后，彼得也提到，这一切都可能发生变化。

Now ultimately, Peter ends on the note that all of this could change.

他指出，随着AI能力不断提升，我们会看到越来越多能独立完成复杂、长期、高价值任务的智能体。

He points out that as AI capabilities improve, we'll see increasingly agents that can autonomously handle complex long horizon valuable tasks.

那么，AI还会增强劳动力吗？

Will then AI still augment labor?

他指出，随着模型持续进步，以及技术边界逐渐平滑，AI对技能型劳动力的增强作用可能会减弱。

He points out that it may very well be the case that the skill biased labor augmenting aspect of AI goes away as models continue to improve and as the jagged frontier becomes smoother.

但最近，他们在分析Claude代码使用模式时，想看看智能代理编程是否改变了专业技能的回报，结果发现并非如此。

But also, that so far, when they recently analyzed patterns of Claude code usage to see if agentic coding was altering the returns to expertise, that's not really what they found.

彼得写道：在我们追踪的七个月里，Claude被越来越多地用于更高价值的任务，但人类的专业技能依然保持着稳定的回报。

Peter writes: Claude code has been used on more and more valuable tasks over the seven months we tracked, but we've seen persistent returns to human expertise.

也就是说，人们负责制定计划，把执行工作交给Claude。

That is, people make planning decisions and delegate implementation to Claude.

拥有更多领域专业知识的人，完成任务的成功率更高，而且当Claude出错时，他们也能更稳定地纠正问题。

People with more domain expertise succeed in their tasks more often, and recover more consistently when Claude makes an error.

单纯的编程能力回报可能下降了，但智能代理编程到目前为止，反而提升了其他互补技能的价值。

The return to straightforward coding ability may have fallen, but agentic coding has so far increased the value of other complementary skills.

他最后的提醒是关于自我迭代提升的问题。

His last caveat is about recursive self improvement.

他写道：人们对未来充满不确定性，一个重要原因是人工智能可能会自动化创新本身。

He writes: A big reason there's so much uncertainty about the future is that AI may automate innovation itself.

赋予机器通用认知能力，会以过去任何通用技术都无法做到的方式，直接推动更多创新。

Endowing machines with general cognitive capabilities is a direct catalyst for further innovation in ways that past general purpose technologies weren't.

在传统的经济模型中，一旦自动化了创新，就可能引发经济奇点——有限时间内实现无限增长。

In otherwise standard economic models, automating innovation can produce economic singularities infinite growth in finite time.

这样的奇点真的会发生吗？

Will such singularities occur?

只要存在某些关键任务永远无法被自动化——无论是技术上还是社会层面的限制——就不会发生。

Not if there are essential tasks that are never automated, whether for technical reasons or societal constraints.

这些薄弱环节，就是增长的边界。

Those weak links are the limits on growth.

他引用了阿加永·琼斯和琼斯的一篇论文：经济的增长，可能不是受限于我们擅长的事，而是受限于那些至关重要却难以改进的事情。

Pointing to a paper by Agayon Jones and Jones, he quotes: Economic growth may be constrained not by what we do well, but rather by what is essential and yet hard to improve.

彼得进一步指出，即使在自动化迅速普及但尚未完全覆盖的情况下，这些薄弱环节也能长期维持劳动收入在总收入中的较高比例。

Such weak links, Peter continues, can keep the labor share of income elevated in the long run even under very rapid widespread but incomplete automation.

他最终总结说：规模定律是很难反驳的。

Ultimately, he concludes: Scaling laws are hard to argue with.

这些模型会变得越来越好。

The models are going to get better.

好得多。

Much better.

我预计这会推动生产力更快增长，甚至可能带来更明显的RSI迹象。

I expect this will drive faster productivity growth and maybe even more clear signs of RSI.

但我并不认为一年后失业率会明显上升，至少不是因为AI。

But I don't expect unemployment to be noticeably higher a year from now, at least not because of AI.

这真是来自Anthropic首席经济学家彼得·麦科里的一些非常有意思的观点。

So really interesting stuff here from again Peter McCrory, the head of economics at Anthropic.

即使不回到AI与就业之间那些老生常谈的概念，比如贾冯悖论，也已经很有启发性了。

And interesting even without returning to commonly heard concepts around AI and jobs like Javon's Paradox.

如果这种状况持续下去，将带来重大的积极影响。

Now if this state continues, it has big positive implications.

斯坦福的安迪·霍尔之前写过，我曾预测，真正针对人工智能的政治反弹，会在失业率上升几个百分点时出现，当时我说我们还没到那一步。

Stanford's Andy Hall writes, a while back, I predicted that the real political backlash to AI would happen when unemployment started going up a couple of percentage points, and I said we weren't there yet.

现在我们依然没到那一步，这篇文章正好解释了原因。

We're still not there, and this piece helps explain why.

到目前为止，人工智能更像是在增强人力，而不是取代人力。

So far, AI looks like it augments rather than replaces human labor.

这种情况可能会变，但眼下劳动力市场看起来相当稳定。

That could change, but right now, the labor market looks quite stable.

当然，我们还是看到了人们对人工智能的政治担忧。

We are, of course, seeing political concerns about AI even so.

但如果真出现了大规模失业，这些担忧相比之下就显得微不足道了。

But these concerns would look small fry in comparison if we had genuine widespread unemployment happening.

有人认为，尽管这看起来是好事，但我们可能找错了关注点。

Some argue that while, yes, this is positive, we might be looking in the wrong place.

特蕾丝·科恩写道：真正的冲击可能首先体现在招聘上，而不是裁员上。

Trace Cohen writes: The real impact may show up first in hiring, not layoffs.