Re: TIB: Artificial Intelligence

To: ti-basic@lists.ticalc.org
Subject: Re: TIB: Artificial Intelligence
From: KeysDezes@aol.com
Date: Tue, 1 Feb 2000 15:56:25 EST
Reply-To: ti-basic@lists.ticalc.org

In a message dated 1/31/00 11:56:00 PM Eastern Standard Time, 
Jared173@aol.com writes:

> I made a tic tac toe game for the PC not too long ago like that.  I use an 
>  array with nine numbers that all start at 0.  The computer moves into the 
> box 
>  that has the highest value.  When I move, the boxes around where I go are 
>  increased in value by one point so that the computer is more likely to 
move 
>  there next turn.  If it senses that I have 2 next to each other, the 3rd 
box 
> 
>  gets jacked up in value so it will block a winning move from me.  You can 
>  make it so that it is more offensive or defensive, like making the values 
>  around itself higher in value is offensive and making the values around my 
>  moves higher is defensive.
>  As for learning from its mistakes, I set it up to play 1,000 games against 
>  itself..which it did, in about 3 seconds.  Now I can't beat it any more...
>  
>  -Jared C.
>  tivault.com
>  isignupnow.com

Well, the problem with that is it isn't learning how to play, it's learning 
how you play.  If you play enough games, it would work though.  The way I do 
it has to do with backtracking to determine why it lost.  Quite neat.  The 
other advantage to the way I do it is that you can port it to all sorts of 
different games in just a few minutes.  The underlying logic to the computer 
player remains the same.