Do you want to see an AI and a human try to beat one another in a game of Go? Well, you can!
Today, Google’s AlphaGo, part of the DeepMind AI group, will be going face to face with a human in a game of Go. His opponent is South Korea’s Kee Se-dol, one of the world’s best Go players.
This competition between man and machine will be played out in a series of five matches that will be held in the Four Seasons hotel in Seoul over the next few days. The winner of the series will win USD 1 million. Should AlphaGo win, Google intends to donate the winnings to charity.
AlphaGo’s showdown with Kee wouldn’t be the first time that the AI faced a human opponent. Back in October., the AI beat three-time European Go champion Fan Hui 5-0 back in October. The key to AlphaGo’s success is a novel mix of deep neural network machine learning and tree search techniques. Simply put, DeepMind taught AlphaGo how to play the game by watching professional human Go players play, and then honed its strategy by pitching instances of AlphaGo against each other, in a process known as reinforcement training.
The first match between AlphaGo and Lee is scheduled to begin at 12pm Malaysian time, so the game should be ongoing by the time you read this. As a game of Go can go on for quite a while, I highly recommend you have some snacks ready.
Source: Ars Technica
