The latest to come out of Google’s AI group known as DeepMind, the AI has been able to learn how to play the ancient Chinese game ‘GO’ all by itself. Further to this, it even went ahead and developed its own strategies for the game – within 3 days.
When made an opponent against it’s predecessor AlphaGo, which made headlines when it beat the South Korean Grandmaster Lee Sedol in 2015, the new AlphaGo Zero managed to win 100 points to 0. Crazy
Previously, AlphaGo learnt by analysing human matches. AlphaGo Zero however, mastered the game by simulating millions of matches against itself, and using this method to develop its own strategies.
“For us, AlphaGo wasn’t just about winning the game of Go,” said DeepMind CEO Demis Hassabis.
“It was also a big step for us towards building these general-purpose algorithms.”
