In chess, is man doomed to lose against computers?

La Croix: Is the computer always stronger than the human being at chess?

Lisa Rougetet: In theory, the computer is not stronger. But the machines have two advantages: their computing power to better predict the progress of the game and their phenomenal memory of the games already played. The computer stores a quantity of information that a human being would be unable to remember and knows how to "sort" it with exceptional speed.

The basic idea of ​​chess is not to play piecemeal but to anticipate the opponent's reactions and build a strategy. But the computer anticipates more than the man: a very good chess player can predict up to 7 or 8 exchanges in advance, while the computer can go a long way.

But there can be configurations where man beats the machine, thanks to his intuition or his unpredictability. For example, in 2007, the American champion Hikaru Nakamura managed to spin the Rybka computer in circles by playing a completely random game, without tactics. In total, almost 150 moves were exchanged between the two, which is huge in chess, and Hikaru Nakamura ended up winning this game.

Aren't computers now more "intelligent"?

LR: Computers specialized in chess are supercomputers, which also have a basic form of artificial intelligence. Back in the days of Deep Blue in 1997, it was basically big computers that kept track of games that had already been played to “learn” from their mistakes. Then there was an improvement in the algorithms, thanks to heuristics, that is to say calculation methods that quickly provide a solution, even if it is not optimal.

Henceforth, knowledge of chess, knowledge of games and strategies, has been “inculcated” into computers, which have all possible combinations and can give meaning to positions. A computer can thus choose to sacrifice certain parts or to lose a part, because the experiment, that the human being informed him, teaches him that it is useless to continue.

Can we soon imagine a computer that would improvise a chess strategy?

LR: A computer is not capable of improvising, it has been programmed by man. But sometimes, in the history of chess, machines have unveiled answers to certain positions that were quite innovative and surprising at that moment in the game, only to prove fruitful later in the game.

Due to the great diversity of strategies in memory and the number of calculations it can perform, a computer can thus give the impression of improvising by simply exploiting positions that had never before been done by man. Sometimes computers can therefore destabilize their human adversaries, and even the engineers who designed them! This idea of ​​surprise is very important because it highlights the limits of the machine and that of man.

The machine does not allow itself to be destabilized, it does not tire, it has no feelings at stake during a game of chess. Chess computers are not aware of the identity of their opponent, they have no feelings about the match. The human player is human with all that that implies psychologically speaking. This is both its handicap and its great wealth.