Formal verification with AI reveals hole in a Nethack fix: writing specs is the hard part
davidbau · x · 2026-09-28
Researcher davidbau describes using AI assistance to learn formal methods: after finding a bug in Nethack, formal verification exposed a hole in his first fix. His takeaway: spec-writing is itself highly AI-driven — once you can truly say what you want, the proof becomes a triviality on top. He admits it's easy to miss issues while checking assumptions and asks the community for advice.
Related event: MIT Researcher Uses AI Formal Verification to Find NetHack Bugs(2 posts)→
More from coding & agent
- Perplexity CEO ran 50-100 agent workflows on Opus 5.5 vs Fable 5.1, found minimal differences — AravSrinivas · 2026-09-28
- AI agent Muse automates expense reports, emails hotel to recover 3 missing receipts — armand_ruiz · 2026-09-28
- Auditing 1,228 Human Interventions: 91% Weren't Decisions, So They Stopped Trusting Agent "Done" — kraboo_team · 2026-09-28
- Cursor turns a Mac Mini into a remote agent worker with one command, plus Computer Use — mattyp · 2026-09-28
- Indie Dev Gets WhatsApp AI Dental Clinic Receptionist Working End-to-End in n8n — vxdant23 · 2026-09-28
- doodlestein says he now ships 2,000+ commits per day with agents — tokenbender · 2026-09-28