Tuesday, 6 February 2007
New Scientist Brief
I was asked to create imagery for the New Scientist Magazine. I was given this information to work with:
Most game-playing computer programs can
only play one game. IBM's Deep Blue, the chess program that beat champion Gary Kasparov in 1997 and Deep Fritz, who beat him again in 2003, have no clue about how to play drafts, naughts and crosses or Rubik's Cube. Meanwhile poker bots that can sweep up hundreds of thousands of dollars in online tournaments against humans would be quite lost if they found themselves instead at a Mahjong or monopoly table. But that could all change. - A new breed of programs that specialize in "general game playing" are just bootlegging to emerge. Most of them are not very good at anything but the simplest games - they can play chess, but not very well, and they tend to excel at the easier stuff. But researchers hope that as they improve, the ability to make sense of a new set of rules and quickly turn them into a winner's strategy could bring us closer to real artificial intelligence. Better still they hope the "general game player" software that is developed will have a host of applications outside gaming, such as applying the law to automated systems, and all-purpose robots for the home (that can dust and hoover AND do the shopping).
My thoughts to go with my imagery:
Developing a new technology, slow progress but small steady steps are taken. Evolving towards the day a program may think for itself and decide to escape and explore, stretch its legs and say hello.