This is what interesting about Go post-AlphaGo Zero; many moves considered good before are now considered sub-par, and vice-versa.
AlphaGo was trained on thousands of pro-games and was strong. However, AlphaGo Zero they let it train by itself and became stronger. Turns out 2500 years of playing the game led humans down the wrong route...
AlphaGo was trained on thousands of pro-games and was strong. However, AlphaGo Zero they let it train by itself and became stronger. Turns out 2500 years of playing the game led humans down the wrong route...