Orateur
Simon Frieder
(Oxford University)
Description
Math Copilots are the envisioned successors to the current generation of LLMs that assist mathematicians in various mathematical activities - from devising proofs to finding counterexamples. In this talk, I will survey and categorize the existing landscape of AI-driven mathematical software tools and benchmarks (including the AIMO series of competitions) and show which benchmarks are missing towards pointing LLMs to hill-climb in the direction of Math Copilots.
I will conclude with a call to participate in a new type of benchmark, the BeyondProof benchmark.
