Google’s AI program proves advancing Capabilities by beating the best Go player in the world
A huge milestone has been achieved in the artificial intelligence field by Google’s AI program (AlphaGo) defeating the best Go player in the world – Lee Se-dol in the first of the five historical matches being held in Seoul, South Korea. The breakthrough is more significant than IBM defeating world chess champion in 1997 since Go takes a lifetime to master, unlike chess. While playing Go, a computer cannot access all possible moves like in chess but has to rely on something akin to intuition. The game itself involves two player placing black and white markers on a 19-by-19 grid. It is said that the game has more possible playing permutation that the total number of atoms in the universe.
AlphaGo made headline earlier this year when it was reveled it had beaten Fan Hui who is the European Go Champion. For a player to be considered a professional and a champion like Hui it takes years of practice to be in a position of recognizing promising moves and in most case, they will always struggle to explain why a particular position seems promising. Many experts had predicted it would take years if not decades for a computer to be in a position of competing with the best humans.
Lee, who is from South Korea, has been a professional Go player since he was 12 years old.He has already won 18 international titles. During the match with AlphaGo, he resigned with about three and a half hours remaining on his clock. Before the game, he had expressed confidence and noted that it would be a computer’s victory if it even managed to win one game. Lee still has the chance of facing AlphaGo on Thursday, Saturday, Sunday and Tuesday and revenging for the loss. Despite what happens in these remaining games, Wednesday’s victory will remain a colossal moment for AI. Commentators of the game called it a superb game that would be studied for the coming years.
Google acquired a British AI company back in 2014 that established a subsidiary called Google DeepMind. DeepMind created AlpaGo by a combination of several simulated neutral network with other AI techniques with the aim of enabling the program to learn by studying lots of previous game and practice against itself. It has already analyzed data from 100,000 professional human game and has played with itself 30 million times. DeepMind founder Demis Hassabis noted that board games provide a good testing ground for AI algorithms for the purpose of eventually developing the real-world problem-solving applications.
After the match, Lee admitted that he was surprised since he didn’t expect to lose. He noted that he never expected AlphaGo to play in such a perfect manner. Hassabis expressed immense respect for Lee Se-dol and his fascinating skills and referred to the game as hugely exciting and very tense. Team lead David Silver noted that the game was incredible and pushed AlphaGo to its limits.
Featured image credit:UK.BusinessInsider