Single player

Play Tic Tac Toe against the AI

Three computer opponents, from one that makes human mistakes to one that plays the mathematically perfect game.

The three difficulty levels, honestly described

Easy

Easy plays mostly at random, with an occasional sharp move so it doesn't feel completely asleep. It misses open wins and fails to block obvious threats fairly often. This is the right level for young children, for teaching the rules, or for warming up. A player who knows the basics will beat it almost every game.

Medium

Medium follows the two golden rules perfectly: it always takes a winning move when one exists, and it always blocks yours. What it does not do is plan ahead — it cannot see a fork coming. That gap is deliberate. Medium punishes carelessness but rewards real strategy, which makes it the level where most people actually improve. Expect to win roughly one game in five once you have learned to build double threats.

Hard

Hard runs a full search of every remaining position (the minimax algorithm) and plays the objectively best move almost every time. Because perfect Tic Tac Toe from both sides always ends in a draw, the honest goal against Hard is not losing. It slips very rarely — and spotting the one game in many where it does is genuinely satisfying. Beating Hard unlocks an achievement, and it's earned.

What playing the AI teaches you

• Threat scanning. Medium forces you to check, every turn, whether either side can complete a line — the habit that decides most human games too.
• Fork construction. Since Medium can't see forks, it's the ideal opponent for practising double threats until they're automatic.
• Defensive precision. Hard punishes any inaccuracy instantly, which teaches you the small set of replies that never lose.

When you can hold Hard to a draw consistently, you have effectively solved the game — read the strategy guide to get there faster. Then test the theory against an actual human.