Twenty-five recipients of the prestigious Fields Medal, widely considered the highest honor in mathematics, issued a strong warning on Friday regarding the risks posed by artificial intelligence companies racing to solve famous mathematical problems. These scientists fear that this trend
could inflict permanent damage upon this vital academic discipline. This joint warning, signed by an elite group of leading scholars in the field, comes just days after OpenAI announced its success in solving one of the "Millennium Prize Problems," which had
baffled the most skilled mathematicians for generations. The letter emphasized that the objectives of AI companies and those of the mathematics community are "largely incompatible," stressing that this trajectory represents a "general threat to intellectual work" as a whole. The 25
signatories to this letter represent a period spanning nearly five decades of the Fields Medal's history. This medal is recognized as one of the world's most distinguished awards in mathematics, sometimes referred to as the "Nobel Prize of Mathematics." Notably,
the first recipient was Pierre Deligne in 1978, leading up to Yu Ding, who received this honor recently this year. The scientists stated that the ultimate goal of posing mathematical problems lies in helping students develop their skills and analytical abilities,
considering this objective far more crucial than merely arriving at solutions. They clarified that "solving problems is merely a tool and a means to achieve the deeper goal of comprehensive theoretical understanding," asserting that these complex processes "always take time
and and are fundamentally based on deep human interaction." The letter raises a fundamental question about "how to ensure that, amidst the changes AI introduces to how work is done, we do not overlook the true purposes that this work was
intended to achieve." The scientists underscored the urgent need to address these issues promptly, both within the mathematical community itself, by companies pioneering these technologies, and more broadly by society at large. In the same context, OpenAI had announced last Tuesday
that one of its advanced models, not yet released to the public, successfully solved the "Navier-Stokes" problem. This problem is one of the seven "Millennium Prize Problems," and its resolution took 88 continuous hours, involving thousands of AI agents working
in parallel coordination.
