Computerized checkers player can't be beaten

Could an autistic savant beat the computer at it's own game of checkers?

I remember researching this topic a couple of years ago, fascinating really. I looked particularly at Naughts and Crosses with it's 3X3 board for simplicity. Even here, the calculations are much larger than you'd intuitively expect. The point about analysing combinations is the crux of the matter.

I watched a show the other day about one of the few female chess grandmasters and saw her participating in speed chess I think it's called, her manipulation and quick fire calculation of the board and current pieces was astounding. She was beating players left, right and centre with incredible speed. They went on to talk about memory and how she apparently see's chess combinations in her mind (and therefore recognises them) much like we all recognises faces (her brain lights up in the same way). So given this, you can imagine just how fast she can play.

It's completely intuitive, she recognises a chessboard combo and reacts accordingly like we recognises the face of a familiar friend. The savants tap into the kind of incredible computing power and memory capabilities that give computers their edge but if I think given the draw nature of Checkers as noted in the article, the best outcome for a human player will be a draw which would represent a perfect game from both sides. You can imagine the path of a game in contrast with all possible game routes like a tree structure, starting of with the 1st move on a board and trickling down to end games with wins, losses and draws.

Now, when these guys come up with a computer program that can beat human players in a chinese board game called 'Go', that'll be unbelievable. The fascinating thing with Go is that the computations needed are so massive (much bigger than for chess or checkers) that a program using the same techniques as I imagine have been used for this checkers program would take an entire day to make a single move because the computer goes through pretty much all possible combinations whereas human players instinctively dismiss huge lists of possible move combinations based on experience of play and learning which I don't fully understand. What makes Go so difficult for computers and one reason why humans seem to excel at it against computers is that there are in fact so few rules compared to say, Chess. This means the limit to creativity and ingenuity placed on a game of Go is much smaller than for Chess (as far as I can tell) so humans needn't have huge computational powers to the same degree and computers have traditionally struggled to both be creative and to deal with creativity itself. In Go, from the perspective of computers, human think outside the box because the box is too big for computers to deal with.

I think it revolves around brute force search methods that a lot of the chess/checkers programs use involving looking at much larger combinations of plays compared with selective searching which is more akin to human play, using expert knowledge. You get things like machine learning, neural networks and genetic algorithms in artificial intelligence, attempts to mimic the more human process of learning which contrast greatly with brute force methods, I think there is real promise here.