User Told AI "Make No Mistakes" — It Made Exactly the Mistakes He Banned
StewartalsopIII · x · 2026-08-18
A user explicitly instructed a model to "make no mistakes," only for it to commit precisely the errors he had forbidden. He quipped: "funny how they say we have AGI."
More from Fun
- Multi-model coding: Enabling Codex and Claude to communicate via text file — OfirPress · 2026-08-18
- AI-generated biopic comedy: — ziv_ravid · 2026-08-18
- Asked an AI to Read My Calendar and Emails, Rendered a Movie of My Week Locally — ziv_ravid · 2026-08-18
- Mimicking Speech with AI and Pink Trombone Simulator — andrew_n_carr · 2026-08-18
- GitHub is still down 🌝 — alexmacgregor__ · 2026-08-18
- Trend Shift: Users Move from Claude Back to GPT — thederbiedone · 2026-08-18