“Zen plays well in the opening and in open areas that are difficult even for top pros to judge,” says Michael Redmond 9P in his commentary on Game 3 in the recent Cho-DeepZen match, in which Cho bested the AI program. “It makes local mistakes, a problem that seems to be a characteristic weakness of the neural network system. It is playing at a pro level, and is approaching the strength of AlphaGo that we saw in March 2016. In Game 2 I saw some patterns that Zen has been playing for years, so it has not turned into a copy of AlphaGo, but has kept it’s original ‘style’.”
Michael’s game 3 commentary:
