AlphaGo Zero beat the version that defeated Lee Sedol by 100 games to nil

The earlier AlphaGo that beat world champion Lee Sedol in 2016 had been trained in part on a large database of human games. AlphaGo Zero, announced by DeepMind in October 2017, discarded that crutch and learned “simply by playing games against itself, starting from completely random play,” with no human game data. Within three days it defeated the published version that had beaten Lee Sedol by 100 games to nil, having, as DeepMind put it, compressed “thousands of years of human knowledge” into a few days of self-play.