The researchers presented the AlphaZero program, a computer program that "not only achieves superhuman control in some of the most complex boards, known as chess, chess and go, but also teaching to play them without previous knowledge, with the exception of the rules of each games. "
The system was created by experts from DeepMind, a British company owned by Alphabet, the parent company of Google.
The results, according to researchers, represent an important step towards the development of a AI player who can quickly learn and master any game.
Chess, Shogi and Go
Because the chess program IBM Deep Blue defeated Gary Kasparov's human champion in 1997, AI casinos became more advanced and able to defeat man in increasingly complex games.
Also in other abstract strategic games, such as shogi and go, which are so much more complex than chess, they were also dominated by machines.
The novelty of AlphaZero, developed by David Silver and his colleagues at DeepMind, is to avoid the need for information derived from human beings, and instead learn chess, chess, and go, so that they play with each other repeatedly They do not master each of the games
The system succeeded in overcoming the most advanced AI programs specializing in these three games, after only a few hours of self-study.
As a result, researchers say that they now have to turn to a new generation of games, such as multiplayer video games, to cope with the next challenge for AI systems.
<! – Download the attached newsletter ->