Slashdot Mirror
Chess Championship: Humans vs. Computer
An anonymous reader writes "X-bit labs has posted very interesting editorial called "Chess Championship: Humans vs. Computer".
During the last 10 years computers penetrated into various spheres of human life. In this article guys try to find out how well computers can
play chess and if it would be correct to say that artificial intelligence is superior to human mind. Interesting read."
I've been busy with computer chess for a few years and I can tell you it's no AI, at least from academic point of view. It's yet a very interesting problem and lots of academic works are done round the world related to this matter (even PhD thesis...). Basically you do a tree/graph search (chess search is something in between as you have a tree with some memory from past levels of the search on the same tree). Also there are some very well-definded knowledge from the opening (and possibly endings) and that's all! The greatest effort is to write a good static positional evaluation function which is rather tricky and there are lots of research on adjusting the coefficients of such an evaluation function using Genetic Algorithms or other (more) advanced optimization methods. All in all you don't need to be an AI expert to write a good chess program and it's not comparabale to other applications like NLP. Tree/Graph search is used in many applications from databases to CAD tools and we don't call them AI applications, so why should a chess program be an AI application? Just because human being does chess too can not signify that as human being deos simple math too but nobody calls a calculator an intelligent machine! At least GO could be a better example as the programs do not just run a simple search (from theoritical standpoint) but should take into account elements of pattern recognition.
Godel's theorem says nothing about the ability to understand Godel's theorem. It simply states that in any formal system (FS) sufficiently powerful to support Peano Arithmetic, there are theorems about FS which can be expressed in FS, but cannot be shown to be true or false (are undecidable) within the rules of FS. I.e. unanswerable questions may be asked of any sufficiently interesting system.
Quite what bearing this has on the ability to "understand" such a system is beyond me. Prove God exists. Prove God does not exist. Can't do either? Oh well, you're obviously not intelligent then.
Well, swap "electrochemical" for "chemical" and I think it may well cover the neural paths you're talking about. And yeah, those neural paths have to be established, but there's nothing mystical about it, it's just very diffucult to understand.
SO YOU'RE GOING TO DIE: The Comic for Dealing with Death
I believe that there is a parameter called the 'contempt factor' that takes this into account.
It is used to increase the computer's score and decrease the opponent's to make the computer take aggressive chances with a novice that it wouldn't with a pro.
It's Christmas everyday with BitTorrent.