Zero pre-AI texts flagged by top detectors across internal, third-party, bounty tests
Afinetheorem · x · 2026-11-17
Follow-up in the Afinetheorem–Timnit Gebru detector debate: beyond internal tests, third-party evaluations and open internet prize challenges all found zero examples of pre-AI writing (from authors of any background) tagged as AI by top detectors — which he presents as decisive evidence against blanket claims that detectors are unreliable.
More from Fun
- "She died of cancer": one-line reply explains where the $400K went — justalexoki · 2026-11-17
- Journalist Shares "AI Agents Hiring AI Agents" Cold PR Email — haydenfield · 2026-11-17
- Wishing for a debug mode for nerve pain: the programmer's-body joke — moultano · 2026-11-17
- 'Feels like March 2020': the AI-impact anxiety meme everyone is sharing — fkasummer · 2026-11-17
- Reading Liu Cixin's 'teacher' passage in the age of LLMs hits differently — sandylikesfrogs · 2026-11-17
- AI-generated series: Schrödinger explains modern dating in 'The Encounter' Ep 1 — FuzzByteStudios · 2026-11-17