AI-detection drama: researcher claims prize committee's own investigation report was AI-written
Afinetheorem · x · 2026-11-17
Afinetheorem lays out three claims in an AI-detection drama around a literary prize: the winner's post-2024 work differs completely in form, the writing shows obvious AI tics, and — most damning — the prize committee's published investigation report itself appears obviously AI-written, with no supporting evidence ever shown. He also explains how classifiers work: thousands of features 10% more common in AI text, most of which humans pick up unknowingly and couldn't replicate deliberately.
Related event: Caine Prize Confirms 2026 Winner Amid AI Ghostwriting Allegations(2 posts)→
More from Fun
- Gamer AI plays through Doom straight from the pixels — minchoi · 2026-11-17
- Founder jokes his cloud startup will succeed because his CTO showed him a katana — tekbog · 2026-11-17
- Local model Poodle's default prompt yields a dog giving a talk to cats — SuB8u · 2026-11-17
- Why were early image models so naturally psychedelic? X users revisit the eerie resemblance — ZeroStateReflex · 2026-11-17
- GPTDisguise Chrome extension disguises ChatGPT as Google Docs, hits 10K users — yuljg · 2026-11-17
- 'Why would a gifted writer offload an 8-page personal story to AI?' — prize drama continues — Afinetheorem · 2026-11-17