Martin Fowler urges authors not to draft in LLM voice
Original title: LLM Voice
He says edited model drafts still read uncanny, and that handing off prose also hands off the thinking writing forces.
Martin Fowler argues that LLM-written prose triggers an uncanny-valley reaction he calls LLM voice, and that detectors miss the point because the aversion is already human. He advises authors not to let a model write a first draft or accept rephrasing, though outlines may be acceptable if the author writes the prose. He uses models only as a search tool and verifies sources, and his site asks contributors to declare AI help. He allows exceptions for disability, translation, and writing aimed at a process rather than a reader.
- He says one fully model-written draft lost the voice only after a professional editor.
- He warns that practicing writing less will also weaken thinking and comprehension.
- He cites Jim Highsmith’s use of an LLM to write despite Parkinson’s disease.
- He prefers a person’s uneven English to smooth machine translation when humans are the audience.
From the original article
5 min read at martinfowler.com
Sometimes when I read a post, I get the growing feeling that I'm reading the output of an LLM. I can't necessarily define why, although there are the common tells (it's not the em-dashes, it's the load-bearing tapestries). As I
By ibobev · martinfowler.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.