AlphaGo was not primarily trained to mimic humans, but to win games. This included playing many games against itself and semi-random tree searches for better st

AlphaGo was not primarily trained to mimic humans, but to win games. This included playing many games against itself and semi-random tree searches for better strategies. If it was only mimicking humans it probably would have lost to the world's best.