AI breakthrough sparks debate over future of mathematics
Rapid advances in artificial intelligence are prompting mathematicians to question how their profession could change after an OpenAI system produced a solution to a problem that ha
By The Guardian
A major mathematical breakthrough produced using artificial intelligence has intensified debate over how rapidly advancing AI could transform one of the world’s oldest academic disciplines.
OpenAI announced in early September that an internal AI system had produced a solution to the Navier–Stokes existence and smoothness problem, one of the seven Millennium Prize Problems.
The problem concerns equations used to describe the movement of fluids and has remained unresolved for around 90 years.
OpenAI published a written proof alongside a formalised version produced using Lean, a computer-based proof system that can be used to check mathematical arguments.
The result is particularly significant because the Millennium Prize Problems are among the most difficult challenges in mathematics.
Six remain unresolved, while the Poincaré conjecture was solved by Russian mathematician Grigori Perelman in work published in the early 2000s.
OpenAI said its Navier–Stokes work involved a system of coordinating AI agents powered by an internal model considerably more capable than its publicly available systems.
The group that produced the result involved around 10,000 concurrent agents, according to the company.
Rather than an individual mathematician spending years pursuing a particular line of reasoning, large numbers of AI agents were therefore able to investigate the problem simultaneously.
That difference in scale has become part of a wider debate within mathematics about what increasingly capable artificial intelligence could mean for research.
Some mathematicians have expressed concern about the speed at which AI companies can direct enormous computing resources towards unresolved academic problems.
Questions have also emerged about attribution, access to research and how mathematicians should share work in progress when AI systems are capable of rapidly examining large amounts of existing material.
Others see considerable potential for AI to become a powerful research tool, allowing mathematicians to investigate ideas more quickly, test possible approaches and tackle problems that might otherwise remain beyond reach.
OpenAI itself has acknowledged that the emergence of systems capable of contributing to mathematical research raises questions that cannot be answered by technology companies alone.
The company has said attribution should accurately reflect whether mathematical work was produced by people or AI and that researchers should remain involved in determining how the technology is used within their field.
OpenAI has also said it does not intend to claim the $1 million Millennium Prize associated with the Navier–Stokes problem.
The significance of the submitted solution will ultimately depend on detailed examination by the mathematical community.
But the development has already demonstrated how quickly the capabilities of AI systems are moving beyond solving established mathematical exercises towards attempting original research.
That change could alter the role of mathematicians rather than simply replacing conventional mathematical work.
Researchers may increasingly work alongside AI systems to generate ideas, explore possible proofs and formally verify results, while human expertise remains important in deciding which questions matter, interpreting discoveries and explaining their significance.
The debate is likely to become more pressing as increasingly capable AI systems are applied to other longstanding problems.
For mathematics, the question is therefore no longer simply whether artificial intelligence can contribute to original research, but how the profession should adapt as those contributions become increasingly substantial.