LLM Voice
- ID
- 41054
- Status
- summarized
- Published
- 29 Oct 2026, 10:57 PM
- Fetched
- 30 Oct 2026, 1:06 PM
- Provider
- Hacker News
- Category
- dev-community
- Original URL
- https://martinfowler.com/bliki/LLM-Voice.html
- Source URL
- https://hnrss.org/best
Summary
- Score
- 6.5
- Created
- 30 Oct 2026, 1:07 PM
- Tags
- Audience
- developersvibe_coderssaas_founders
What happened
Martin Fowler's 29 October 2026 post argues that LLM-generated prose produces an 'uncanny valley' reaction he calls LLM Voice, and that authors should not let an LLM write drafts or even suggest rephrasing because the voice is hard to edit out. He allows LLM help only for outline/structure, says his site requires disclosure of AI assistance, and notes he uses AI for writing only as a search tool with source verification. The Hacker News thread drew 385 points and 205 comments.
Why it matters
If your team uses LLMs to draft docs, changelogs, blog posts, or launch copy, decide now whether to restrict LLM use to outlines/search and whether to disclose AI assistance. Fowler's one clean exception needed a professional editor at a major publication, so light editing is not a reliable fix for LLM Voice.
Discussion angle
What disclosure or usage policy should teams adopt for LLM-assisted docs, changelogs, and marketing copy if even edited LLM drafts can still read as LLM Voice?