Addendum: as portions of my text above have been quoted out of context, I would like to also draw attention to the various caveats listed at https://github.com/teorth/erdosproblems/wiki/AI-contributions-to-Erd%C5%91s-problems regarding the extent to which one can draw broader conclusions about AI mathematics capabilities from the progress in solving Erdos problems.

EDIT (Mar 5, 2026): It should also be noted that in the days after the above posts were made, it was discovered that the methods in these AI results were very similar to that of a 2014 paper of Pomerance, and in fact Pomerance has now released a short note showing how the methods of that paper also provide a solution to problem #728. https://math.dartmouth.edu/~carlp/binom.pdf Nevertheless, the AI-generated solution was the first to explicitly address this problem (though, as noted previously, this was in part due to the fact that the original formulation of the problem was not formulated well and contained some unwanted "trivial" solutions). A full summary of the situation can be found at https://arxiv.org/abs/2601.07421