Machines Like Us

During the Go Tournament in Paris, staged between 22 and 24 March 2008 by the French Go Federation (FFG), the MoGo artificial intelligence (IA) engine developed by INRIA -- the French National Institute for Research in Computer Science and Control -- running on a Bull NovaScale supercomputer, won a 9x9 game of Go against professional 5th DAN Catalin Taranu. This was the first ever officially sanctioned 'non blitz' victory of a 'machine' over a Go Master.

More Complex Than Chess, More Possible Combinations Than the Number of Particles in the Universe

In 1997, a computer beat the then World Chess Champion, Garry Kasparov, for the first time. But the game of Go remained the exclusive preserve of Man. More complex than chess, with 10 to the power of 600 different possibilities -- more than the number of particles in the known Universe -- the game of Go is a remarkable school for strategy. So this latest officially-endorsed performance by INRIA and Bull represents a real achievement.

The victory was won as part of the 'IA-GO Challenge', in a so-called 9x9 game (played over nine lines and nine columns). Although Catalin Taranu beat the computer in a 19x19 configuration with a nine-stone handicap, the Go Master nevertheless rated the IA system as 'approaching Dan standard' in a performance that promises some formidable battles to come between man and machine.

"The software used in this victory -- the result of a collaboration between INRIA, the CNRS(1), LRI(2) and CMAP(3) -- is based on innovative technologies that can be used in numerous different areas, particularly in the conservation of resources which is such a vital issue when it comes to tackling environmental problems," explained Olivier Teytaud, Researcher at INRIA, MoGo team leader. "We are particularly delighted that, during the tournament, we were able to welcome researchers working in these areas who are interested in this technology, and we would like to thank the FFG and the tournament organizers for having given us the opportunity to learn more about these advances stemming from fundamental research."

"Over and above the intellectual and emotional challenge that is involved in designing a computer capable of facing up to the Grand Masters of Go, our collaboration with the MoGo team has helped us to make further advances in parallelization techniques. The users of our supercomputers will reap the benefits of this work in their own environments," affirmed Eric Monchalin, Director of High-Performance Computing R&D at Bull.

Bull