AlphaGo Zero Learns The Game 'Go' From Scratch And Defeated The Original AlphaGo


With Google making progress in the field of machine learning, using AutoML that has the ability to self-replicate, DeepMind has created AlphaGo Zero.

AlphaGo that has 48 AI processors. With data built on thousands of Go matches. AlphaGo, under the direction of humans, began to learn the game. And eventually, the AI has become so advanced that it was able to defeat the world’s top human Go player and establish AI’s supremacy in the game.

This newer version of AlphaGo has never been taught by humans like the original.

With AutoML that has the ability to self-replicate, Google's DeepMind is able to teach AI to be better than the humans who created it can. This is where AlphaGo Zero was born.

Evolving from the original AlphaGo, this newer version only has four AI processors. And the only data it was given was the rules of the game. After 40 days of learning, it was able to defeat the original AlphaGo at its own game.

According to Google on its blog:

"This technique is more powerful than previous versions of AlphaGo because it is no longer constrained by the limits of human knowledge. Instead, it is able to learn tabula rasa from the strongest player in the world: AlphaGo itself."

The first AlphaGo was initially trained to play Go by thousands of human amateur and professional games. AlphaGo Zero skips this step and learns to play by just playing the game against itself, starting from completely random play. In doing so, it quickly surpassed human level of play and defeated the previously published champion-defeating version of AlphaGo by 100 games to 0.

AlphaGo Zero was able to defeat the original AlphaGo because DeepMind have taken "self learning" to the next level by not making it restricted to human capabilities.

This way, it was able to overcome limits that are unique to humans.