@eniko @thomasfuchs Would you argue the same about AlphaGo or AG0? In those models, the "hard ceiling" seems above the best humans can achieve.