I’m becoming AI-blind

我正在变得“AI盲”

Recently I’ve been catching myself having these little moments at work, when I’m trying to read a document someone has sent me and my brain somehow refuses to analyze it. It feels like I’m reading it, but I’m unable to focus on its content. I end up getting dragged into an endless back and forth with the sender, asking questions about things that have been covered in what they’ve already sent me.

最近，我在工作中经常出现一些奇怪的时刻：当我试图阅读别人发给我的文档时，我的大脑似乎拒绝进行分析。我感觉自己确实在阅读，却无法专注于内容。最终，我不得不与发送者进行无休止的往复沟通，询问那些文档中已经涵盖的内容。

It’s rather concerning, because I’ve spent the last year trying to re-learn how to focus and these situations show the exact opposite. I sat down to analyze these situations and realized they all have a common denominator: the documents all show a strong trace to AI.

这相当令人担忧，因为过去一年我一直在努力重新学习如何专注，而这些情况却恰恰相反。我坐下来分析了这些场景，意识到它们都有一个共同点：这些文档都带有强烈的 AI 痕迹。

For example: A design document that looks like a copy-paste from Claude. While it does cover the design of the specific feature in question, it also carries a lot of Claude-specific analysis and lingo. “This cuts just through it”, “The first gate is real”.

例如：一份看起来像是从 Claude 直接复制粘贴的设计文档。虽然它确实涵盖了特定功能的设计，但也夹杂了许多 Claude 特有的分析和术语，比如“This cuts just through it”（这直击要害）或“The first gate is real”（第一道关卡是真实的）。

Or a 20-page marketing concept deck that mixes up (a rather reasonable) marketing strategy with some nonsense product technical architecture gibberish. How does it pitch the idea? “It’s not selling X, it’s selling Y”. “The Redis backbone redefines the product”.

又或者是一份 20 页的营销概念演示文稿，将（相当合理的）营销策略与一些毫无意义的产品技术架构胡言乱语混在一起。它是如何推销这个想法的？“它卖的不是 X，而是 Y”、“Redis 主干重新定义了产品”。

Or a technical requirements document that describes a rather simple concept in a very verbose way. The thing is, a lot of this document reads like someone’s “internal” reasoning that’s not fully sure about certain decisions. Sounds like an LLM to me.

再或者是一份技术需求文档，用极其冗长的方式描述了一个相当简单的概念。问题在于，这份文档读起来就像是某人对自己某些决定并不完全确定的“内心”推理。在我看来，这听起来就像是大语言模型（LLM）写的。

There’s an ongoing discussion of whether humans are good at recognizing AI-generated text. While most research claims that humans don’t really do a good job there, I disagree. It’s not that difficult, at least when we’re talking about the low-effort results. Florian Roth wrote a pretty good summary of the common patterns in the context of social media. I see a similar thing happening for work-related texts.

关于人类是否擅长识别 AI 生成的文本，目前一直存在争议。虽然大多数研究声称人类在这方面表现不佳，但我并不认同。这并不难，至少在谈论那些低质量的产出时是这样。Florian Roth 在社交媒体语境下总结了常见的 AI 模式，写得非常好。我发现工作相关的文本中也出现了类似的情况。

Besides the obvious choice of words, the general flow of sentences and the attempt to pitch every small detail as a breakthrough quickly give it away. If your document describes the checkboxes in an RBAC configuration view for an enterprise application, don’t sell it like you’ve just invented fire.

除了明显的措辞选择外，句子的整体流畅度以及试图将每一个微小的细节都吹捧为“突破”的企图，很快就会暴露真相。如果你的文档只是在描述企业应用中 RBAC 配置视图里的复选框，就别把它吹得好像你刚刚发明了火一样。

I feel like I’ve been “pre-trained” on all the AI-generated LinkedIn posts, emails and websites that are full of text but empty on meaning. My brain learned to quickly spot signs of AI-generated content, at least the content generated with low effort, and it now ignores it and moves on without thinking much about it.

我觉得自己已经被那些充斥着 AI 生成的 LinkedIn 帖子、电子邮件和网站“预训练”过了——这些内容文字堆砌，却毫无意义。我的大脑学会了快速识别 AI 生成内容的迹象（至少是那些低质量的内容），现在它会自动忽略这些内容，甚至懒得去思考。

I’ve heard some people comparing it to “banner blindness”. It’s not surprising. With the amount of content being pushed at us, filtering it out is how we need to stay sane. What’s fascinating to me, is that the same AI that was supposed to make me more productive, is what’s now slowing me down in an unexpected way.

我听说有人将其比作“横幅盲区”（banner blindness）。这并不奇怪。面对海量涌向我们的内容，过滤掉它们是我们保持理智的必要手段。令我着迷的是，原本应该让我提高生产力的 AI，现在却以一种意想不到的方式拖慢了我的节奏。

I don’t usually go on vacation, but this year I really needed a break. One evening I was really hungry, walking past some restaurants on the Baltic coast. There was a single one I immediately ignored, but a minute later something in my head asked “Hey, did they really put up a photo of quiche with mold?” I walked back just to see this.

我通常不休假，但今年我确实需要休息一下。一天晚上，我饿极了，走在波罗的海沿岸的餐馆旁。我立刻忽略了其中一家，但一分钟后，脑子里有个声音问：“嘿，他们真的放了一张长了霉的乳蛋饼照片吗？”我走回去一看，果然如此。