Google computer notches first win versus Go world champ

2016 March 09

In another win for artificial intelligence, Google's DeepMind computer with it's AlphaGo program has beaten South Korean Lee Se-Dol, the world's premier Go player in the first game of a 5 game match set to take place March 8-16th. While other board games have been long solved by computer, the complexity of Go prevented computers from competing with anything but amateur players until recently. Computer scientists had predicted it would be perhaps another decade or more until artificial intelligence caught up with top players and,  while they might be right as there are still four games to be played, the victory is a sign that computers are making inroads into areas of thinking and prediction formerly considered the realm of human intelligence.

DeepMind Go Challenge

The game Go, which originated in ancient China over 2500 years ago, has tantalizingly simple rules which result in very complex and intuitive game play. The board consists of 19x19 lines where black and white stones are placed alternately at lined intersection to try and control territory and cut off enemy controlled spaces. The number of possible game position is an incredible 600 orders of magnitude greater than chess. While the basic rules can be learned in an afternoon, it often takes years of study and practice to gain even an amateur grasp of the subtle strategies involved. 

 

Google acquired then UK based DeepMind in 2014 for an undisclosed sum, reportedly over $500 million, to try and bolster it's artificial intelligence (AI) portfolio amongst intense industry competition. Scientists believe the lessons learned developing AI to compete in intricate board games can also be expanded to solve society's greatest challenges

No Responses leave one →

Leave a Reply

Leave this field empty: