Idea Designing a AI that chooses strategic mistakes for Chess

I have playing with this idea for a version of chess ( hypothetical ) , using an AI that either chooses strategic choices that are the best against a player , or chooses strategic mistakes that makes the computer player vulnerable to the human player, allowing a possible defeating move against the computer. I came up with the idea to have the AI mimic human thinking. If the AI is going to mimic human thinking, its going to have to make mistakes. Humans make mistakes, and computers don't. But this idea for an AI works against the player. How do you know the computer is not setting up a trap by weakening its defenses, to lure your most powerful pieces to be captured? You don't. The AI waffles from its computer sided thinking to human thinking to invent mistakes randomly, but not often too much in the game. That's how I came up with this idea. So my argument is if your going to design a AI that will always out think a human player, it makes the game frustrating to the player to loose too many times, but if you design a AI that mimics human thinking and chooses to make a strategic mistake, it makes the game more appealing for the player. Now I figure if you can design an AI, using my idea, for a chess game, how about other games that could be adapted to use an AI for strategic thinking ( for instance sport games such as basketball , tennis, soccer, etc.) ?

What do you think ?
 

Cameron

Member
This is the art of good AI. It's easy to build an AI that can beat humans handily, the real skill is making an AI seem human. This is basically the idea behind the turing test. This is what I aim for when building an AI and absolutely agree you should be as well. The unpredictable nature of humans is what makes us so much more fun to play against. That and our fallability leading to vulnerability. Winning is fun, just as long as it's not too easy. The key is in finding and maintaining that careful balance.
 
Last edited:
Top