Fields Medalist says ChatGPT 5.5 Pro delivered "PhD-level" math research in under two hours with zero human help

In a groundbreaking demonstration, Fields Medalist Timothy Gowers utilized ChatGPT 5.5 Pro to tackle complex open problems in number theory, achieving results that he claims equate to “PhD-level” research in under two hours without any human intervention. This development raises significant questions about the evolving role of AI in advanced mathematical research and its implications for the field.

Gowers focused on improving an exponential bound to a polynomial one, a task that typically requires extensive human expertise and time. The model’s ability to achieve this in less than an hour has drawn attention, particularly from an MIT researcher who described the key idea behind the solution as “completely original.” Gowers emphasized that the new benchmark for mathematical contributions now hinges on proving concepts that large language models (LLMs) cannot yet accomplish, suggesting a shift in how mathematical research may be conducted in the future.

This advancement could have profound implications for researchers and educators, potentially democratizing access to high-level mathematical insights and accelerating the pace of discovery. As AI tools become increasingly capable, they may redefine the landscape of academic research, prompting both opportunities and challenges for human mathematicians.

Looking ahead, the AI community will be watching closely to see how these developments influence the collaboration between human researchers and AI, as well as the broader implications for academic integrity and innovation in mathematics.