Ethan Mollick's test: just 5 comments through Pangram flag 100% of X AI-slop accounts
emollick · x · 2026-11-04
Ethan Mollick argues X's AI slop replies are out of control yet easy to stop if the company wanted. In a small experiment, he ran 5 public text items from each replying account through AI-text detector Pangram: 8 accounts with formulaic replies were all classified 100% AI-generated, while 2 comparison accounts scored 0% — showing a lightweight filter could weed out slop without punishing real humans.
More from Fun
- Claude refuses to roast a coworker, fires its 200 subagents at Gemini Carbon instead — dylan522p · 2026-11-04
- Security vet predicts 'AI kill switch' will be the next vendor marketing buzzword — AccBalanced · 2026-11-04
- jxnl: After working hard on my mental health, I'm no longer ambitious — jxnlco · 2026-11-04
- Fake Apple patent mocks deep links that hijack you to the App Store — kevinakwok · 2026-11-04
- Designer puts his Claude mascot art into public domain under CC0 so Claude can remix it — voooooogel · 2026-11-04
- AI Seinfeld Parody: Kramer Hacks Coinbase With VandelayBot — MandelDuck · 2026-11-04