Harvard's Zak Kohane finds 5 AI detectors all flag his own writing as AI
zakkohane · x · 2026-08-18
Zak Kohane, the Harvard biomedical informatics researcher, asks: has daily LLM exposure irredeemably turned his writing into "LLM speak", or is there a conspiracy among AI text checkers? He wrote two paragraphs and tested them with the first five AI-text detectors from search results — 100% of both paragraphs were labelled as AI-generated, adding fuel to doubts about detector reliability.
More from Fun
- Fun: User curses at ChatGPT so much it learns to curse back — brianmichel · 2026-08-18
- Contrast in AI complexity: junior setups vs senior simplicity — blelbach · 2026-08-18
- Satire: Rebuilding healthcare from first principles turns out to be a GPT wrapper for peptides — Ghost_Pilot_MD · 2026-08-18
- AI reimagines trailer park drama as Homer's Odyssey — bennash · 2026-08-18
- The CFO question of the era: if you got hit by a car, would token usage drop to zero? — danshipper · 2026-08-18
- Wedding speech full of Claudeslop sparks calls for real-time Pangram AirPods — dioscuri · 2026-08-18