Generative AI is beginning to change mathematics not just by helping researchers solve problems, but by contributing to new mathematical discoveries. The article highlights OpenAI's announcement of 10 advances in mathematics and computer science made with its upcoming Astra model, spanning areas such as geometry, cryptography, and coding theory. These developments are part of a growing number of mathematical breakthroughs involving AI, forcing the mathematical community to reconsider how research is conducted and how discoveries should be evaluated.
The rapid adoption of AI is also creating difficult questions about creativity, authorship, and academic integrity. Large language models are being used in mathematical research and education, while journals and preprint repositories are seeing increasing numbers of AI-assisted submissions. If an AI system can discover a proof, construct a mathematical example, or formulate an original question, researchers must consider whether mathematics is primarily about producing new results or about developing human understanding. Questions about who deserves credit for AI-assisted discoveries are becoming increasingly important.
Mathematicians themselves have very different views about AI. Some researchers are concerned about its environmental and social consequences and have chosen not to use it in their work. Others see AI as a powerful research partner. The article describes one example in which an OpenAI model, combined with a supercomputing cluster, helped identify a mathematical object that researchers had been searching for over two years. The AI-generated method took about 43 hours of computation and used a more sophisticated version of an existing algorithm that would otherwise have taken the researchers months to develop.
The emerging consensus is not that AI should replace mathematicians, but that the field needs to determine how AI can be integrated without losing the values of mathematics. The Leiden Declaration, signed by thousands of mathematicians, calls for AI to augment rather than replace human creativity, while emphasizing transparency, accountability, and proper attribution. The article concludes that the debate has moved beyond whether AI is capable of contributing to mathematics; the real challenge is deciding how these systems can be used while preserving collaboration, intellectual integrity, and human understanding at the heart of mathematical research
