If(), else if() alternative in c++(Is this AI?) - c++

Question

First off, I am a noob. I am also a Janitor that has never made a dime writing code. This is just something that I love doing. It is for fun:) That being said, I wrote this console based tic-tak-toe game that has enough ai to not lose every game. (I guess ai is what it should be called.) It has something like 70 if/else if statements for the computers turn. I used 3 int arrays like so:
int L[2], M[2], R[2];
0 = blank;
1 = X;
2 = O;
The board then 'Looks' likeL[0] | M[0] | R[0]L[1] | M[1] | R[1]L[2] | M[2] | R[2]
So I basically wrote out every possible scenario I could think something like:
if(M[0]==1 & M[1]==1 & M[2]==0){M[2] = 2;}//here the computer prevents a win
else if(L[0] ==2&M[1]==2&R[2]==0){R[2]=2;}//here the computer wins
//and so on....68 more times!
I guess my question(s) is(are): Is there a better way? Is there a way to achieve the same result with less lines of code?Is this considered Artificial Intelligence?

Answer 1

The standard algorithm for this is called Minimax. It basically builds a tree, where the beginning of the game is the root, and then the children represent every possible move X can make on the first turn, then the children of each of those nodes are all the moves O can make in response, etc. Once the entire tree is filled (which is possible for Tic-Tac-Toe, but for games like Chess computers still don't have enough memory), you work your way back up, assuming both players are smart enough to make the best move, and arrive at the optimal move. Here is another explanation of Minimax specifically using Tic Tac Toe as an example.

Answer 2

The Wikipedia page on Tic-Tac-Toe has a very good algorithm outline for winning (or tying) every game:
http://en.wikipedia.org/wiki/Tic-tac-toe which is what I used to make a Tic-Tac-Toe game several years ago.
After you understand the algorithm, one of the cleverest ways to implement a Tic-Tac-Toe computer player is with a magic square. The method is discussed here. As far as size goes, I've seen this implemented in about 50 lines of code, I'll post the code if I find it :)
This isn't technically artificial intelligence, as AI usually refers to artificial neurons, neuron layers, gradient descent, support vector machines, solving complex polynomials, and the like. Solving Tic-Tac-Toe

Answer 3

Yes there are better ways.
The most obvious would be to consider how different mirror views of the board would simplify the number of cases.
Also, consider pre-storing "interesting" patterns in arrays and then comparing the game state against the data. For example, one series of pattern would be all the ways a player can win in the next move.
Also, note that with the declaration int L[2], there are only two entries in array L, namely L[0] and L[1]. The references you have to L[2], M[2], etc. are errors that should have been caught by the compiler. Consider turning up the warning level. How this is done depends on the compiler. For gcc it is -Wall.
This counts as a form of artificial intelligence. The series of if statements are accumulated knowledge: how to recognize a situation and the appropriate best reaction to it.

Answer 4

The closest thing to real AI to solve such a game would be to code an artificial network and train it with all combinations of the tictactoe game.
In that case the code would not do so much if then else to solve the problem but would solve the problem by taking the most reasonable choice that solves the problem from a pattern trained in it.
But coding a neural network is not a trivial thing :)

Answer 5

When in need to code a rule-based system (like the AI you are building), you can use a rule engine, like for example CLIPS (which is a tool developed at NASA for creating Expert Systems written in C).
http://en.wikipedia.org/wiki/CLIPS
Perhaps it's an overkill for playing Tic Tac Toe, but if you are in the mood for learning cool AI stuff, Expert Systems is a very interesting area, but different (and perhaps less trickier) than Neural Networks.
Have fun!