Many humans never learned to play chess in their lifetime but an artificial intelligent (AI) machine nicknamed Giraffe learned to play the game in just three days.

The machine did not just teach itself how to play chess, it has also excelled in the game it is now better than most humans. It even managed to get up to the International Master level in just a span of 72 hours.

Computer programs that can compete and beat chess champions is not new but what makes the new program different is that it uses an artificial "neural network" that mimics the human brain and adapts over time.

Matthew Lai, a student at the Imperial College London, who developed the new chess AI, said that he developed the program by combining human efficiency and processing power.

Instead of studying millions of possible moves simultaneously just like IBM's Deep Blue, which beat chess champions Garry Kasparov in 1997, the Giraffe AI "thinks" intuitively using a strategy that involves automatic feature extraction and recognition of patterns.

Giraffe assesses a chess move in three stages. It initially checks whose turn it is to make the move and what pieces are currently available.

It then evaluates the state of the board figuring out the locations of each of the pieces. For the final step, the AI considers the moves that its pieces can make relying on millions of programmed scenarios that it has tried and tested itself earlier.

The machine, in effect, learns which of the moves work and which do not as it plays along just as human players learn the game.

Giraffe, however, proved to be a fast learner. After spending only three days playing against itself, it reached a rating equivalent to that of an international chess grandmaster, or the top 2.2 percent of the chess players.

"Unlike most chess engines in existence today, Giraffe derives its playing strength not from being able to see very far ahead, but from being able to evaluate tricky positions accurately, and understanding complicated positional concepts that are intuitive to humans, but have been elusive to chess engines for a long time," Lai wrote in his report on the program.

Lai said acknowledged that his program is not as good as the most sophisticated chess-playing software in the world but he said that his program can be applied to other board games and that it could learn well enough to beat human opponents.

Photo: Liz West | Flickr