This page will include links and resources about AI, to lectures about AI reported in various blogposts, and to projects that include AI experimentation.
AI+ Math projects polymath style (April 2026)
AI and Math has become a hot topic (and a source of some worries) among and beyond the mathematical community. Nissan Hajaj proposed in the blogpost Polymath Plus AI to run “AI polymath projects” based on a similar concept of polymath projects but with participation of AI agents. We launched a preliminary version of Nissan’s proposal in this post: Optimal Monotone Families for the Discrete Isoperimetric Inequality. Other proposals for such projects are: Polynomial Hirsch Conjecture; Acyclic Unique Sink Orientations, the The sunflower conjecture, and more. For more details see the page “project“.
Pages and blogposts on AI
Conversation with the AI program GPT3, April 25 2022
ChatGPT: Möbius Randomness – ChatGPT offered 7 (and later 21) conjectures that combine computational complexity and Mobius randomness. See also this post for the context.
ChatGPT Meets Elchanan Mossel’s Dice Problem (In 2023 ChatGPT gave the wrong answer 3; in 2025 it gave the right answer.)
Will AI take over mathematics? Dec. 31, 2024
Ten Recent Questions for ChatGPT (Nov. 2025);
ChatGPT is a Useful but Demanding Mathematical Collaborator (Sept. 25)
Amazing: Erdős’ Unit Distance Problem was Disproved! It was achieved by AI! (May 26)
MathOverflow question:
Examples for the use of AI and especially LLMs in notable mathematical developments
Verification and simplification
Verification of difficult proofs is also a major direction. See this post
ICM 2022. Kevin Buzzard: The Rise of Formalism in Mathematics
Of course, it is very important to find simplifications and alternative paths to difficult proofs. AI can be helpful also for simplifications and for finding good avenues for explaining and teaching difficult proofs.
AI, Riddles, and Psychology. A project with Maya Bar-Hillel
In 2022/3 I had a project with Maya Bar-Hillel regarding certain class of puzzles that tend to cause people to fail. We tried to see if various AI and search tools exhibit similar “way of thinking” as humans. See the post: Riddles (Stumpers), Psychology, and AI. (Maya and I did not agree on the interpretation of our findings.)
First proof and other challenges and Benchmarks for AI
There are various projects aiming to challenge AI and benchmarks its ability.
First proof is a notable such project.
Other projects are the Ramanujan challenge for AI; Daniel Litt’s problems;
Concerns and ethics
Quite a few mathematicians are concerned with AI’s applications to mathematics.
There is a notabe Leiden Declaration on Artificial Intelligence and Mathematics.
Michael Harris’s critical writings on the matter are worth reading.
Amnon Shashua’s lecture and new company
A very nice lecture from May 2023:
Amnon Shashua’s lecture at Reichman University: A Deep Dive into LLMs and their Future Impact.
Apparently a question I asked during the lecture served as a trigger to a new AI company of Amnon!
Combinatorics and more 
