Report
Most model answers in a small chess experiment converged on one of two endgame studies
A user says most answers pointed to Réti (1921) or Saavedra (1895) after running each model 20 times.
TLDR
A user ran a small experiment to see which chess positions models find “interesting.” They say most answers converged on one of two endgame studies, Réti (1921) or Saavedra (1895), after they ran each model 20 times.
Combined views
—
2 Sources, first seen ago
