Human vs Machine - Lee Sedol wins against Google

When AlphaGo - a program driven by Google’s DeepMind AI, won the third game in the best-of-five Google DeepMind challenge, it was hailed as a huge milestone for Artificial Intelligence development. But humans are not the ones to give up easily. Lee Se-dol, the world's top-ranked Go player, finally managed to stop the winning run of this computer program in the fourth game of the challenge on Sunday.

AlphaGo might have taken home $1 million prize money but what Lee Se-dol earned by this one game cannot be put into words. “This one win is so valuable and I will not trade this for anything in the world,” said Lee amidst thunderous applause after winning the match.

Grandmaster’s Key

South Korean Go grandmaster believes to have found a couple of weak points in the artificial intelligence (AI) program based on deep neural networks and machine learning. According to Lee, AlphaGo falls short in reading unexpected moves, indicating the machine lacked the ability to deal with surprises, and appeared to have more problems playing with black stones.

After the loss, Google DeepMind co-founder DemisHassabis tweeted that Lee “was too good for us today and pressured #AlphaGo into a mistake that it couldn’t recover from.” He added that the loss is a valuable learning tool and would help to identify the weaknesses in the program that needed to be worked upon.

Go, Ancient Chinese board game

Go, a game for two has players taking turns to put black or white stones on a 19-by-19 grid trying to gain territory by arranging stones in strategic shapes or patterns. The winner is the player who surrounds more territory than the opponent. Despite its seeming simplicity, the permutations of Go vastly exceed those in chess.

The first move of a game of chess offers 28 possibilities; this figure changes to 361 for the first move in a game of Go. A game of chess lasts around 80 turns while Go games can last 150. These almost endless possibilities make it difficult to craft in advance and follow a particular strategy in Go.

AlphaGo’s win special

Experts had not expected an AI program to beat a human professional in the game of Go for at least a decade until AlphaGo defeated a European champion player last year. This ancient Chinese board game was long considered impossible for computers to play at a world-class level due to the high level of intuition required to master its intricate strategies unlike chess, in which IBM’s Deep Blue famously beat former world champion Garry Kasparov in 1997.

This Human vs Machine challenge has one more face-off to go tomorrow which will once again test Lee’s intuition against AlphaGo’s ingenuity.