It’s over 2500 years old and the rules are deceptively simple, but Go is an incredibly complex game. The rules are simpler than those of chess, but a player typically has a much more vast choice of moves. The game is played with white and black stones and the aim is to capture territory as well as your opponent’s pieces. Stones can be captured by surrounding them but the control of the board is a much more complicated matter which is why a true competitive AI has been rather difficult to create. In a landmark victory against the European champion in October it was proven that the efforts of the DeepMind team and subsequently Google’s efforts on the AlphaGo project had found success in what had been thought impossible since the early years of computing.
By first studying the patterns repeated in various games, AlphaGo will play different versions of itself millions of times during a single match and learn incrementally. Learning from its mistakes is what makes the AI truly remarkable and what led its developers to take their months of improvements since beating the European champion, to the next level. As of today, AlphaGo has secured its second win against world champion Go master Lee Sedol in a 5-game match worth 1 million US dollars. The games are being played with a 2 hour set time limit for each player followed by three 60 second overtime periods as well as the standard Chinese ruleset.
The magnitude of this achievement opens many doors to advanced functionality of artificial intelligence. The emerging truth is machines can learn real world tasks in much the same way they learn to make moves in a game. This opens new paths of scientific research while augmenting existing industries where machines can learn to identify promising areas of research and help humans in the right direction. Whether it’s fields with a sheer overload of data needing to be sifted through such as biology or security, or more subjective fields such as arts and language. As stated many times in AlphaGo’s 7 matches thus far, regardless of which player wins, it is a victory for humanity.
Game 3 between Lee Sedol and AlphaGo is set to take place tonight, March 11th, at 8:00 PST Watch Here
Source: Deep MindContinue Reading