Dwarkesh podcast: Schulman, Millidge and O'Neill debate recursive self-improvement timelines

Dwarkesh Podcast · rss · 2026-09-12

Dwarkesh Patel's new episode brings together John Schulman, Beren Millidge, and Charlie O'Neill to debate how close we are to recursive self-improvement, what's driving Chinese labs' progress, whether long-horizon RL elicits AGI, why RL is working so well, and rapid-fire timeline predictions.

Original post →

More from AGI Musings

AGI Musings channel →