Well, the opening books have "learn" functions but they are simply governed by basic stats to raise win % and decrease loss %, so they are not artificial intelligence.
Other than that, I think Hiarcs is the only engine to claim to have a learning function that makes it stronger as it plays, but I'm a little skeptical about it. You can google Hiarcs + learn function for more info.
For a project at school about artificial intelligence, I was wondering if I could write a part about a chess computer. Artificial intelligence is defined as a program that can learn from its own experience and mistakes.
I would like to know if any of the major chess computers like Deep Blue, Rybka, etc. can be seen as artificial intelligence. I of course don't know the insides of their codes, but I am interested in knowing if the computers are just programmed to calculate in some specific way, or if they actually learn from experience, games in the database, their own mistakes, and if they compose their own opening books and such.
Help would be appreciated :)