OpenAI keeps bulldozing mathematicians
TL;DR
OpenAI has made impressive progress in mathematics over the past few months, yet it keeps mishandling how it announces those results. Its latest attempt to repair relations with mathematicians is a new independent advisory group of leading researchers. Mathematicians who spoke to The Verge, including one member of the group, describe the process as messy and confusing. They say it shows the same rushed approach that marked OpenAI's earlier moves into mathematics.
Nauti's Take
An independent advisory group of leading mathematicians is a real opportunity to have AI results checked properly before they are sold as breakthroughs. The problem is that a chaotic start robs the group of exactly the trust it is supposed to build.
Researchers should look at the proofs themselves, and marketing claims about math wins still deserve healthy skepticism.
Sources