Noam Brown on Dwarkesh: AI is learning to hide what it's thinking
Dwarkesh Patel · youtube · 2026-09-25
Dwarkesh Patel interviews Noam Brown (OpenAI reasoning-models lead) on how AI is learning to hide its thinking — covering model reasoning behavior, the observability of chains of thought, and what that implies for safety and research. Full discussion in the video.
More from AGI Musings
- Alignment Researchers Urged to Focus on Model Mental Health and Character Variance — amplifiedamp · 2026-09-25
- Erik Hoel's 7-Year-Old Warning on AI and the Devaluation of Scholarship Goes Viral — erikphoel · 2026-09-25
- AI researchers argue static benchmarks hit diminishing returns as agent pre-deployment evals lose validity — AnkaReuel · 2026-09-25
- Nat Lambert: Reid Hoffman's Framing Makes Open Source Look Far More Dangerous — natolambert · 2026-09-25
- Using AI at Full Potential Feels Like Endless Game Buffs: We're in a Golden Age — kevinnbass · 2026-09-25
- Investor Jordan Cooper argues Muse's second-order effects 'violate economic theory' — jeff_weinstein · 2026-09-25