Computers have been able to beat humans at lots of games for years. Computers can beat almost everyone at chess, for example. But there is one game that humans can still reliably beat computers at, a very old East Asian board game called Go. Its rules are much simpler than chess, but it has many more possible moves at every step.
To play chess, computers play several moves ahead and look at every possible outcome before choosing the ‘best’ option. When a chess piece has been captured it is no longer in use so things get simpler. With Go pieces can be used by the side that captures them. The Go board is also larger, made up of 361 squares rather than 64. That makes the number of possible moves huge—too many for a computer to check, or evaluate, quickly enough to play a sensible game of Go.
Humans play Go well because they are good at looking for patterns they’ve seen before and imagining in their head how the game will turn out. Computer scientists have been working on creating computers that are better at thinking like humans. So far, even the best Go programs can only do as well as amateur dan (advanced amateur) human players. But things are always improving, and in a few years, computers might even be able to beat Go champions.
Clever critterness… but you could probably still beat it at Go
More AI stuff →
A bit of cs4fn 