Was Move 37 really out-of-distribution? A case for long-horizon self-play
weballergy · x · 2026-11-22
weballergy argues AlphaGo's famous Move 37 may well have been within the self-play distribution — we judged its novelty only relative to our own knowledge. If long-running agents with long-horizon incentives shape behavior via self-play, there's no reason AI couldn't produce similar breakthroughs; asking "can AI do it" is the wrong question.
Related event: Debating AlphaGo Moments and Why Long-Horizon Agents Skip Theory-Building(2 posts)→
More from AGI Musings
- Economist Tabarrok: AI now writes better than 99.9% of all humans in every language — jbarbier · 2026-11-22
- Can AI really write? Quote-tweet fight over AI fiction winning contests — TuhinChakr · 2026-11-22
- Gary Marcus: without a Lean for biology, AI can't just 'solve biology' — GaryMarcus · 2026-11-22
- Long-Running Agents With Long-Horizon Incentives Could Build Theories Too — weballergy · 2026-11-22
- Frontier AI outgrows the convex hull view: harnesses extend inference limits — gerardsans · 2026-11-22
- Can human minds exit their sensory distribution? Rethinking the baseline for AI creativity — weballergy · 2026-11-22