OpenAI's Astra Solves 10 Decades-Old Math Problems, Stirring Credit and Job Fears
Updated
Updated · The Verge · Aug 11
OpenAI's Astra Solves 10 Decades-Old Math Problems, Stirring Credit and Job Fears
3 articles · Updated · The Verge · Aug 11
Summary
OpenAI said its unreleased Astra model solved 10 long-standing math problems across fields from sphere packing to non-sofic groups, backing the results with more than 250 pages of papers and Lean-verified proofs.
The advances suggest AI can now crack problems serious researchers have failed to resolve for years; one mathematician said solving even 1 such problem could be enough to land an academic job.
A dispute quickly emerged over credit for the non-sofic groups result after OpenAI initially described problems as seeing no main-result progress for at least a decade, then revised the wording to better acknowledge prior human work.
Researchers told The Verge the breakthrough is accelerating excitement and anxiety because access to top proprietary models is limited and OpenAI estimated the 10 solutions cost about $2,000 in tokens at current API prices.
The broader fear is that AI could squeeze graduate training and academic jobs before mathematicians know whether these systems merely finish existing lines of work or can open genuinely new directions.
If an AI solves a century-old math problem using uncredited human theorems, whose breakthrough is it really?
Could proprietary AI models replace graduate students, or will they force human mathematicians to invent entirely new fields of thought?
With flawless AI proofs bypassing human intuition, who verifies if the machine actually solved the right historical math problem?
$2,000 to Solve 10 Legendary Math Problems: OpenAI Astra’s Breakthrough and the Future of Mathematical Research
Overview
In August 2026, OpenAI’s Astra AI made headlines by solving ten historic mathematical problems and publishing Lean 4-verified proofs, which could be instantly checked by anyone. This leap collapsed traditional peer review timelines and sparked both excitement and anxiety in the academic world. While top mathematicians translated Astra’s results for human understanding, the rapid progress fueled concerns about AI bypassing peer review, eroding training opportunities for young researchers, and shifting research priorities. As AI handles more technical details, human mathematicians are moving toward high-level theory and verification, but the community remains divided over the future of mathematical discovery and academic equity.