Hacker News front pageColin Breck13 min readintermediate
I don't want to read what you didn't write
Summary
The essay laments the unreadability of AI‑generated design docs, PR summaries, and other engineering writing, citing a survey where 78% of readers quit when they suspect AI involvement. The author finds AI useful only when they control the prompt and context—e.g., for grammar checks, citation generation, and diagram creation while writing a paper—but not for producing the core narrative. He highl…
- AI‑generated engineering prose often lacks context, making it hard for reviewers to extract relevance or risk, leading to disengagement.
- A developer survey shows 78% stop reading AI‑scented articles and 71% avoid the author; 98% prefer the author's original voice.
- When the writer is in the loop (prompting, providing source code, logs, metrics), AI can accelerate tasks: grammar fixes, citation formatting, TikZ diagram generation, and abstract drafting.
- AI struggles to produce accurate, context‑rich summaries of complex systems; human‑written narratives remain superior for conveying intent and nuance.
Engineering communication is a bottleneck; unreadable AI‑generated artifacts waste reviewer time and erode trust. Understanding when AI adds value (editing, verification) versus when it harms (uncontextualized summaries) helps teams maintain effective knowledge transfer.
5/10
