Tic-Tac-Toe distils combinatorial game theory into a 3×3 grid: place three of your symbols in a horizontal, vertical, or diagonal line before your opponent does. Simple on the surface, but analysing the position after each move reveals a game tree of 362,880 possible sequences (though only 26,830 are distinct due to symmetry). At Easy difficulty the AI makes deliberate mistakes, letting newcomers learn winning patterns; at Hard it plays the textbook minimax algorithm and cannot be beaten — only drawn.
Each of the nine squares carries strategic weight. The centre cell participates in four winning lines (two diagonals, one row, one column), making it the single highest-value opening move. Corner squares participate in three winning lines each; edge squares participate in only two. This hierarchy explains every optimal opening and counter-move: whoever takes the centre against an opponent who replies with a non-corner square can always force a win.
The AI on Hard uses full minimax search with alpha-beta pruning across the entire game tree, meaning it evaluates all possible futures from the current position and picks the move that minimises your best possible outcome. Against a perfect opponent the game always draws from mutual best play — but take the centre or a corner on your first move and the AI will never find a hole in your game if you follow optimal responses. Trying to "trick" the AI into creating a fork is how most players end up defeating the Medium difficulty.
The centre is the most powerful position — it participates in 4 of 8 possible winning lines. If the centre is taken by your opponent, your best response is any corner.
A fork means you threaten two winning lines simultaneously, forcing your opponent to block one while you complete the other. Fork positions require two non-adjacent corners with the centre claimed.
Easy AI won't block a two-way threat. If you hold two separate pairs heading toward wins, the AI can only block one — giving you an uncontested three-in-a-row.