(ai; didn't read)
Writing is one of the few things that still forces real thinking. You start with a vague feeling or half-formed idea. You wrestle with it. You throw words on the page, hate most of them, rearrange, delete, add, and eventually something clear and honest emerges. That process isn’t optional. It’s how the thinking happens.
For a long time there was this unwritten promise: if someone published (or even just shared) a piece of writing with you, you could reasonably assume the author had actually done that work. They had lived inside the idea long enough to make it coherent. If you got lost, you could ask “what did you mean here?” and they could rephrase it differently, because the idea originally belonged to them.
With LLMs, that promise is broken.
It’s now trivial to go from “I have an idea” to a 2,000-word document that looks thoughtful and professionally formatted in under ten minutes. The genuine struggle, the editing, the rearranging. This gets skipped entirely. LLMs just fill any existing gaps with plausible nonsense.
What used to be effort upfront (the writing) has been pushed downstream. Readers and reviewers now spend far more time trying to reverse-engineer intent: “What is the actual point being made here?”, “Why are these three paragraphs saying almost the same thing?”, “Is this claim even true or did the model just hallucinate?”.
The really painful cases are when the document is just good enough to avoid immediate rejection. It looks polished, so people assume it’s been thought through. They forward it without review. The AI slop lands on someones desk...
The economics are brutal once you zoom out. One hour saved for the author can easily burn 10–50x as much attention from readers, approvers, and everyone downstream. The larger the audience, the more violently that asymmetry scales.
All of this creates real exhaustion on the reader’s end. People are getting tired of having to decode vague or hallucinated content just because it looks polished. And now there's a shorthand for that: ai;dr. Artificial Intelligence; Didn’t Read. It’s basically tl;dr, but instead of “too long” its “no human thought here, didn’t read”.
If you won’t put in the work to think clearly and write it yourself... why should others put in the work to read and fix it?
Don't outsource thinking...