AI Computer Beats a Go Champion to Prove Machines Are Winning!
We have always known the potential of, and been excited for the future of Artificial Intelligence. However, the research in that field has not yet reached the limits that Hollywood flicks have been warning us of for a long time.
What we have achieved, nonetheless, is an AI computer system that excels at playing the game Go. Not only that, the program is so good at it that it has recently beaten the best that humans could produce in the board game.
For those who do not know, Go is an ancient game that been around for more than 2500 years, thanks to the Chinese. The game is so complicated due to the potential moves one can take at any given point that it has always proven a tough nut to crack for the developers working to master it with their artificially intelligent programs.
Point worth noting here: this game is certainly not as easy to replicate in a computer as chess, backgammon and draughts.
Almost all the attempts made at an AI computer aimed to be the best at Go have been handicapped one way or another but DeepMind’s new AlfaGo program is better than that.
AlfaGo took on European Go champion Fan Hui sometime ago and managed to beat him! Although the results have been shared only now, they have reiterated what the developers at DeepMind have boasted about the Value and Policy Neural Networks that allow the AI Computer to pick up the best choices from among the tons of possible moves in the game. Watch the video below for a demo: