OT: I made a Tic-Tac-Toe learning AI.

[narf poit chez BOOM]

Well, I've never really done an AI before, much less a learning one.

It's on my webpage.

Didn't take long, once I figured out what I was doing.

[Arkcon]

Great. Now expand it to be a checkers learning program, just add a few more states and rules. After that, a chess program is just a few more steps away.

[narf poit chez BOOM]

Chess AI? Chess AI?! *Blink Blink* *Faint*

It's not a true learning program, all it does is record the part of the board that is three X's or O's in a row when it wins, then next game it checks that list to see if anything on the board looks close to one of those patterns. If it does, it tries to fill in the missing parts. In v2 each AI should also check their enemy's tokens to see if they are close to any winning solution it knows, then if they are, place a peice in that pattern to interfere. Very simple. Just learns patterns and runs them through if-then statements. Have to re-write it for checkers...And I still have a chat program to finish.

[Makinus]

try to make a bigger grid (like 4x4 or 5x5) still using 3 tokens to win so it can learn more complicated strategies...

[douglas]

That won't work. With a larger grid and still only 3 in a row needed to win, the first player to move can easily win every time.

[Kamog]

If it's 4 in a row needed to win, it should work, right?

[narf poit chez BOOM]

I could increase the grid and row size, but there really wouldn't be any point. This was just made so I could so if I could do a very basic AI. I did. It's not likely to be expanded; it's more likely I'll make a different AI test program.