Updated
Updated · The Verge · Sep 28
OpenAI’s 9-Member Math Advisory Rollout Sows Confusion as 100-Plus Results Loom
Updated
Updated · The Verge · Sep 28

OpenAI’s 9-Member Math Advisory Rollout Sows Confusion as 100-Plus Results Loom

2 articles · Updated · The Verge · Sep 28

Summary

  • Nine elite mathematicians in the new AGMAI group say OpenAI’s announcement blurred the body’s independence, leaving many researchers unsure whether it was an OpenAI-appointed board.
  • More than 100 unreleased math results are driving the urgency: OpenAI says its internal model has resolved long-standing open problems, and AGMAI was assembled partly to help review and communicate those findings.
  • Martin Hairer, an AGMAI member, said the group gets no funding or technical support from OpenAI and is bound only by standard confidentiality needed for early access to research.
  • Mathematicians told The Verge OpenAI’s earlier releases already damaged trust through poorly written manuscripts, weak engagement with prior literature, and unannounced document revisions that felt "sneakily" altered after criticism.
  • That mistrust is now spilling into the wider field, with researchers fearing sudden announcements could make years of work irrelevant and deepen a sense that AI companies are treating mathematics as a publicity exercise.

Insights

If AI solves a century-old math puzzle without human intuition, did it actually advance science or just break the game?
When AI agents generate complex proofs, who gets the credit if the machine secretly borrowed from unpublished human minds?
Could brute-force AI automation destroy human discovery by robbing future researchers of the struggle required for true genius?