How Chess Computers Work

How do bots like Stockfish and AlphaZero work, and why some chess bots are better than others.


Source: Eugene Chystiakov

A chess set laid out in the streets of Kyiv

By: Salar Nasimov, Journalist

Chess bots/computers are computer programs that are built to play chess. These bots use artificial intelligence, machine learning, and game theory to make their moves and improve how they play over time. At the heart of every chess bot is a set of algorithms that can analyze the state of the board, identify legal moves and evaluate the possible outcomes of each move. In order to do this, the robot must know lots of chess strategy and tactics, which it achieves by training on large data sets of past games and strategies used by human players.

If a chess bot is not as powerful, it means that it has not been played with enough or it has not trained on those data sets. After the robot has analyzed the state of the board, it uses its own algorithms to find the best possible move. This involves examining many possible moves and evaluating each outcome based on its expected effect on the state of the game. After that, the computer chooses the move that it considers to be the most promising based on his evaluations. To keep improving its game, the bot constantly changes its strategies and tactics based on previous games. This may include adjusting its algorithms, analyzing the moves of other top players, and adjusting functions to predict move outcomes better. As a result, the best chess robots are incredibly sophisticated and can compete at a very high level with the best human players in the world.