DeepMind researcher says AI safety is still crude empiricism, far from rigorous standards
Geoffrey Irving argues that AI labs' safety practices remain ad-hoc empiricism, three levels below the rigorous standards of aviation and nuclear engineering, and that safety arguments sit between casual empiricism and cryptographic proofs.
2026-10-23 ~ 2026-10-23 · 4 related posts
- AI safety researcher: labs are at 'slapdash empirics' level, three tiers below aerospace-grade rigor — geoffreyirving · 2026-10-23
- AI Labs' Safety Is 'Slapdash Empirics,' Says Researcher Calling for Systems Safety Engineering — geoffreyirving · 2026-10-23
- DeepMind's Geoffrey Irving: alignment may need cryptography-grade rigor as AI finds 12 OpenSSL zero-days — geoffreyirving · 2026-10-23
- Irving: AI safety sits between slapdash empirics and cryptographic proofs — geoffreyirving · 2026-10-23