Microsoft Research said on October 21, 2026 that senior researcher Xinzhi Zhang and co-author Zhao Song have proved the Kannan-Lovász-Simonovits conjecture, a 1995 problem in high-dimensional geometry, using a process in which AI agents explored roughly 100 research trajectories while the researchers chose which ones to pursue.
The KLS conjecture already has proof: the optimal flat cut does not depend on the dimension. Three independent demonstrations endorse it. The Microsoft Research entry solves the KLS conjecture: high-dimensional geometry changes after 31 years aparece primero en Merca2.