OpenAI Chief Scientist: Models Could Get Much Better at Math, but RSI and Alignment Come First

Dr_Singularity · x · 2026-09-08

OpenAI Chief Scientist Jakub Pachocki says the company could make its models significantly better at mathematical research with additional focus, but is prioritizing RSI (recursive self-improvement) and automated alignment research instead.

Commenting on this, DrSingularity notes OpenAI's models are already delivering breakthroughs at or above the level of the world's top mathematicians — and the company is basically calling that a toy it hasn't started taking seriously yet.

Related event: OpenAI Chief Scientist Prioritizes RSI and Auto-Alignment Over Math(2 posts)→

Original post →

More from AGI Musings

AGI Musings channel →