I Don't Want to Read What You Didn't Write
Colin Breck
Opinion — commentary, not a factual news event.
A TLDR Dev essay says AI is making writing worse to read, not better. It’s great as a helper, but the human voice is what people actually trust.
Based on reporting by Colin Breck — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
People are using AI to churn out design docs, business plans, tickets, PR summaries, blog posts, and meeting notes, and the result, this essay argues, is a flood of text that feels less like communication than an obstacle course. The complaint isn’t about AI helping with writing. It’s about AI replacing the writer’s judgment, voice, and context with something that can be detailed and still say almost nothing.
The sharpest example is the retrospective document: AI is used to build something, then again to explain it after the fact. That turns a design proposal into a machine-made summary stuffed with facts but stripped of the slow, messy work that actually builds consensus. The same problem shows up in pull request summaries and meeting notes. They may list every change, but they often fail the basic test: why should anyone care, and what, exactly, are they supposed to do with this?
The author draws a hard line between reading and prompting. When someone uses AI on their own work, they already have the context, so the output can be useful. A stranger reading the same text gets none of that. They have to inspect every line to figure out what matters, which is exactly the kind of labor people are increasingly refusing to do. That, the essay says, is why readers stop, tune out, or simply avoid the writer next time.
But the essay is not anti-AI. It says AI was genuinely helpful while drafting an academic paper: checking citations, spotting grammar problems, simplifying clumsy sentences, drawing diagrams, and catching one subtle technical mistake missed by four human reviewers. The key is that the human still wrote the paper. AI helped with verification and cleanup. It was also best at the most mechanical part of the job: the abstract.
The broader argument is that writing is not always about compression. In technical work, in operations, in incident response, even in performance reviews, the narrative matters because the uncertainty matters. The essay points to tools like ASD-STE100 for clear technical instruction and Pangram for detecting AI-written text, then ends on a simple preference: people should keep writing in their own voice, imperfections included, instead of handing readers a polished summary with nobody in it.
My take — AI-written commentary, not fact-checked reporting
This is the right fight. AI should catch typos, verify claims, and clean up the sludge; it should not turn every human into a committee of one. The real tell is simple: if a message sounds like it was optimized to offend nobody, it usually says nothing worth reading.
Read more about this at: Colin Breck
