Back in 1997, the world watched in awe as IBM’s Deep Blue super computer beat Chess grand master, Gary Kasparov. Fast forward to Seoul, 2016 and this time its Machine Vs Man in a game that’s much more advanced than Chess. More precisely… it is Deep Mind’s (Google’s British AI company) AlphaGO AI program Vs Lee Se Dol, 18 times World champion of the Chinese board game of Go.
The game of Go, believed to be more than 3000 years old, is said to be much more complex than Chess because of its incomputable number of move options. The game involves black and white stones placed on a grid and players gain the upper hand by surrounding their opponents pieces with their own. As opposed to 20 possible moves in Chess, Go offers a choice of 200 moves making the number of move options exponentially more complex.
According to Deep Mind’s team “there are more possible positions in Go than there are atoms in the Universe”. Experts suggest that the challenge with Go is that unlike Chess, here the computer must have an almost human like intuitive ability to win. So here was a challenge worthy of testing advancements in AI.
Stephen Evans who covered the event for the BBC reported that Lee Sedol appeared nervous at the start of the game.
As the game picked up speed, Mr. Lee seemed to have an upper hand, which continued until things took a turn in the last 20 minutes where AlphaGO took the lead. Once the AI took the lead it just turned the game around and kept gaining. Eventually Mr. Lee conceded defeat. In an interview later he said that “he was stunned by one unconventional move that the AI program had made that a human never would have played”
One has to sit back and register this comment from the world champion…the fact that the program had deployed a strategy that a human could not have played… or in other words, the intuitive capability that AlphaGo had developed in the course of learning this game.
According to Deep Mind CEO Demis Hassabis, in preparation for the game, the AI program first studied common patterns repeated in past games. “After it’s learned that, it’s got to reasonable standards by looking at professional games. It then played itself, different versions of itself millions and millions of times and each time get incrementally slightly better – it learns from its mistakes”
This was the first of 5 games being played and the opponents will battle for a $1m Prize. Irrespective of whether its Man or Machine that emerges as the final winner, one thing is clear…machines with intuitive capabilities are a not too distant realty.