+- MKLab (https://mklab.gr)
+-- Forum: [INDEX] (https://mklab.gr/forumdisplay.php?fid=1)
+--- Forum: ARTFICIAL INTELLIGENCE (AI) (https://mklab.gr/forumdisplay.php?fid=5)
+---- Forum: ARTICLES (https://mklab.gr/forumdisplay.php?fid=33)
+---- Thread: Math Can’t Go On Like This (/showthread.php?tid=2030)
Math Can’t Go On Like This - mklabgr - 09-16-2026
The Atlantic argues that mathematics is entering a fundamentally different era because AI systems are becoming capable of attacking problems that once represented the limits of human mathematical research. The immediate catalyst is OpenAI's reported solution to the Navier–Stokes Millennium Prize problem, using a swarm of 10,000 AI agents in less than four days.
The development follows other recent AI mathematical breakthroughs and Terence Tao's warning that AI is creating a “crisis in our mathematical values and practices.” The article emphasizes that the controversy isn't simply whether AI can solve difficult problems, but what mathematics is supposed to accomplish: a result can be correct while the reasoning behind it remains opaque to humans.
The deeper concern is that AI could change the incentives and culture of mathematics. AI companies can spend enormous computational resources pursuing spectacular benchmark problems, while academic mathematicians may lack the resources to compete. The article also discusses Tristan Buckmaster's allegation that OpenAI may have benefited from his AI-assisted work; OpenAI says its systems did not access or use his inputs and says its proof was independently developed.
Twenty-five Fields Medalists have meanwhile criticized the industry's emphasis on solving mathematical problems as benchmarks. The central issue is therefore understanding versus merely obtaining answers: mathematicians may increasingly have proofs that are formally correct but too complicated or unfamiliar for humans to understand. The field may consequently need to reward explanation, conceptual insight, and the development of reusable mathematical ideas—not just spectacular solutions.
Key takeaways
- AI is moving from assisting mathematicians to potentially doing frontier-level mathematics.
- OpenAI says 10,000 AI agents solved the Navier–Stokes problem in under four days.
- The breakthrough depended heavily on decades of accumulated human mathematics, rather than emerging from nowhere.
- Correctness is not the same as understanding: an AI can potentially produce a valid proof that humans struggle to comprehend.
- The 166-page Navier–Stokes proof is reportedly difficult for mathematicians to digest, and its broader mathematical value remains to be determined.
- AI companies have vastly greater computational resources than most university mathematics departments, potentially changing who gets to pursue major problems.
- The Buckmaster episode raises important questions about research privacy, attribution, and AI-assisted mathematical work; OpenAI disputes that its system used his inputs.
- Twenty-five Fields Medalists have argued that using mathematical problems primarily as AI benchmarks can conflict with the goals of mathematical science.
- Historically, the process of solving difficult mathematical problems can generate valuable techniques even when the original problem has little immediate practical use.
- Mathematics may need new ways to recognize explanation, conceptual understanding, and reusable insights, rather than measuring achievement mainly by whether a problem has been solved.
- The article's biggest question is not “Can AI do mathematics?” but “What happens to mathematics when machines can solve problems faster than humans can understand their solutions?”
ARTICLE