Sometimes when I read a post, I get the growing feeling that I'm reading the output of an LLM. The author calls this the uncanny valley of prose—a human reaction that detection tools cannot resolve. By the time you reach for a tool to tell you what you're reading, it's already too late, even if it confidently tells you there's no LLM present.
The author urges writers to be extra-wary of LLM-Voice. Authors should not get an LLM to write a draft, even if edited by a human, because it's very hard to expel the voice. Outline help may be acceptable, but authors should fill the prose themselves. On this site, anyone using AI to help write must declare how they used it, usually in the acknowledgments.
Writing is an almost-essential tool for thinking, and surrendering to an LLM undermines progress. Exceptions include Jim Highsmith, who finds LLMs essential to overcome the bane of Parkinson's disease, and translation assistance for colleagues uncomfortable in English. As more people delegate writing to LLMs, they also delegate thinking and comprehension.
Source: Hacker News · Summarized by HeadlinesBriefing