Irving: Alignment Research Needs TCS Barrier Culture

Geoffrey Irving recommends a paper proving an impossibility theorem on why monitoring cannot scale to superintelligence, and argues that alignment research would benefit from theoretical computer science's 'barrier culture' of formal negative results, which machine learning lacks.

2026-08-25 ~ 2026-08-25 · 3 related posts

• Geoffrey Irving cites paper proving monitoring can't scale to superintelligence — geoffreyirving · 2026-08-25
• Geoffrey Irving: ML lacks obstacle culture compared to theoretical CS — geoffreyirving · 2026-08-25
• Irving: TCS's 'obstacle culture' offers hope for alignment research — geoffreyirving · 2026-08-25