Comment by MrToadMan

Didn't they say that they launched an effort to evaluate their model on all open Millennium Prize problems, and then narrowed down to Navier-Stokes as the most promising after seeing results on simplified versions of the problems?