Artificial intelligence models are now helping mathematicians discover and prove new results at an accelerating pace, according to a report by Quanta Magazine.
The shift gained momentum in July 2025 when several AI models successfully solved five out of six problems at the International Mathematical Olympiad. While these competition problems are puzzles with known answers rather than open research questions, the performance stunned the mathematical community.
Mathematicians who previously dismissed AI as too error-prone have begun using the technology to break new ground. Some researchers are now accomplishing in a single day what previously required weeks or months of manual labor, the outlet reported.
In certain instances, algorithms can now formulate a conjecture, prove it, and verify the result with minimal human intervention. Other researchers use extensive dialogue with large language models like ChatGPT, Claude, or Gemini to develop novel proof strategies.
A shift in research methodology
Terence Tao, a prominent mathematician at the University of California, Los Angeles, noted that 2025 marked the year AI became truly useful for diverse mathematical tasks.
“This guy’s got a shovel. This guy’s got a pickax. Together we can bore a tunnel,” Tao said, describing the synergy between human researchers and AI tools.
Tao also predicted that the field will undergo significant structural changes. He suggested that instead of studying one problem at a time, mathematicians will use these tools to solve thousands of problems simultaneously and conduct large-scale statistical studies.
While the technology is changing the workflow, Tao noted that no one expects AI to replace human mathematicians, though significant institutional and cultural shifts are inevitable.
Daniel Litt of the University of Toronto also confirmed that AI is fundamentally changing how mathematics is practiced, according to the Quanta Magazine report.
