Ex-OpenAI Policy Chief: Limits Stem from Post-Training, Not AI's Nature

Former OpenAI policy head Miles Brundage argues model limitations reflect the difficulty of post-training rather than AI's fundamental nature, and while skeptical of recursive self-improvement, believes intelligence gains can simply be bought with compute.

2026-09-29 ~ 2026-09-29 · 2 related posts

• Miles Brundage: AI capability limits reflect hard post-training, not the nature of AI — Miles_Brundage · 2026-09-29
• Brundage: intelligence explosion needn't be Pareto—just buy it with more compute — Miles_Brundage · 2026-09-29