AI models are increasingly being used to solve complex mathematical problems, including long-standing conjectures that have perplexed mathematicians for decades. OpenAI's Astra model, for instance, has produced ten solutions to various problems, some of which are of considerable difficulty. In May 2026, OpenAI published a counterexample to the 'Unit Distance Conjecture,' which had remained unsolved in geometric graph theory since 1946. This development has sparked a range of reactions within the mathematical community, from cautious optimism to concerns about the potential obsolescence of traditional research methods.
Mathematicians are divided on how to respond to AI's growing role in the field. Abhishek Saha, a professor at Queen Mary University of London, described AI models as being at least as capable as a 'solid and indefatigable PhD student.' He noted that using GPT-5.5 Pro for routine tasks that previously took weeks left him feeling like a 'conductor' rather than the 'whole orchestra.' Saha believes the field may split over how to integrate AI, with some adapting quickly to its potential while others remain skeptical.
The impact of AI on mathematics is being debated in academic circles. Trefor Bazett, a mathematician writing for The Conversation, suggested that AI and human ingenuity could usher in a 'golden age' of mathematics, similar to a prediction made by Fields Medal winner Timothy Gowers in 2000. Gowers envisioned computers handling routine checks while mathematicians focused on deeper ideas, but he also warned that such a golden age may not last long. He expressed mixed feelings about AI's growing capabilities, acknowledging its potential while fearing the 'possible destruction of mathematical culture.'
Source: thedecoder
