Gamasutra recently featured an article titled Intelligent Mistakes: How to Incorporate Stupidity Into Your AI Code by Mick West. It's a great read that I highly recommend for anyone, including non-programmers, regarding the topic of Artificial Intelligence.
For the uninitiated, Artificial Intelligence (AI) is the logic used in a game by your computer "opponent". The fun part of playing a video game in the absence of human vs. human interaction is the ability to play against the machine, which simulates the actions of a human opponent. However, knowing that a computer relies on logic, mathematics, and the ability to perform a million computations per second, there is no reason for the computer not to win every single time.
It wouldn't be fun to play against an opponent that never loses, which is why we need to decrease the computer's IQ! But as Mick points out, simply reducing the number of computations the machine is allowed to perform before making a decision would cause it to make abnormal blunders; it would be too easy to defeat and once again, not fun. Instead, a better approach is for the AI to perform its normal amount of computation and earnestly try to win, occasionally making a move that is more advantageous to the human player. If the human player takes advantage of these "thrown a bone" moves frequently enough, they have the opportunity to win. Yet they can still enjoy the challenge of playing against an opponent that is trying hard to win (and will win if the player fails to take advantage of certain opportunities).
My quick summary hardly does it justice. Read the full article now!
No comments:
Post a Comment