Chess: why man can never defeat AI — history, epochs and the dead end for the grandmaster

The question of who is stronger in chess — man or computer — has troubled minds for half a century. Today the answer is clear: artificial intelligence surpasses any grandmaster to such an extent that an equal match between man and a top neural network has lost its meaning since the mid-2010s. However, the path to total dominance was long, dramatic, and full of legendary battles. We analyze the fall of humanity and reflect on what remains for living chess players.

The race began long before Deep Blue: programs of the 1950s–1990s

The first chess programs appeared in the 1950s with the advent of computers. Scientists regarded chess as an ideal polygon for testing artificial intelligence — strict rules, a finite number of moves, a clear goal. In 1951, Alan Turing wrote the first chess program in history on paper, performing calculations with a pencil. In 1957, Alex Bernstein created the first full-fledged program for a mainframe IBM, which took 3–6 minutes per move. These pioneers played openly weakly — the level of an amateur beginner, but the pace of progress was impressive.

In the 1980s, commercial chess computers appeared: Chessmaster (1986) and products from the Novag company. In 1988, the program Deep Thought (predecessor of Deep Blue) won a tournament game against grandmaster Bent Larsen for the first time. The breath of the computer could be felt in the back of the neck.

1997: the match of the century — Kasparov vs. Deep Blue

In 1996, Garry Kasparov, the reigning world champion and the best player in history, met with the IBM supercomputer Deep Blue. Kasparov won the match 4–2, but lost the first game — a historic moment when a computer won a world champion in a classical control. However, Deep Blue 1996 still made gross positional mistakes, and Kasparov took the upper hand.

Exactly a year later, in May 1997, IBM brought a new version — Deep Blue II ("Deeper Blue"). Increased computational power (about 200 million positions per second) and an improved evaluation function made the monster incredibly dangerous. The match of six games ended in the computer's victory: 3.5–2.5. In the decisive game, Kasparov faltered psychologically, making a blunder, and Deep Blue entered history as the first AI to defeat a world champion in a match. After this, IBM disbanded the team, and Deep Blue never played again.

The era of humans still lingered: 2000–2010s

Deep Blue was a monstrous brute-force computer that played due to blind search. However, the next generation of programs, such as Rybka, Fritz, Houdini, and Stockfish, used more sophisticated heuristics and became accessible on ordinary PCs. Their rating exceeded 3000 Elo points, while no human had ever reached above 2850. Since the early 2000s, professionals have already recognized that the best computer programs play stronger than any human. But the final point was set in the mid-2010s with the appearance of neural network engines.

AlphaZero: the neural network that reinvented chess

In 2017, the company DeepMind presented AlphaZero — an algorithm that learned not from human games but from playing against itself from scratch. The method is called "reinforcement learning". After a few hours of self-training, AlphaZero played hundreds of millions of games, inventing its own, unaccustomed to humans, strategy. It sacrificed material for activity of pieces, built fantastic attacks, and often played a style that professional commentators called "extraterrestrial".

In the match against the best classical engine Stockfish (version 2017), AlphaZero achieved a crushing victory: 25 wins, 25 draws, and no losses with equal time on the move. Stockfish analyzed 70 million positions per second, while AlphaZero — only 80 thousand, but the quality of the decisions was incomparably higher. Computer chess has reached a new level — now it is not hardware that dominates, but the idea.

Power ratio in 2026: a gap of 500 rating points

Today's strongest chess neural networks — Leela Chess Zero, AlphaZero (in its later implementations), the latest versions of Stockfish (with hybrid NNUE architecture) — have ratings of about 3600–3700 Elo. Current world champion Magnus Carlsen (peak rating 2882, current ~2830) is behind the computer by at least 700–800 points. This is roughly the same difference as between Carlsen and an amateur with a rating of 2000.

A modern engine on a good laptop beats the world champion without a chance. The World Chess Championship for humans still exists, but it never puts people against AI in an equal fight — it would be a farce.

Why man cannot catch up with AI — fundamental reasons

The reason is not that "the computer is smarter" and not the amount of memory. Chess AI surpasses man in three key aspects, each of which is insurmountable.

Tactical infallibility. The computer never blunders a piece, misses a checkmate in 2 moves, and does not make mistakes in calculating options due to fatigue. Even the best grandmasters make 1–3 tactical blunders in each game. The computer does not know the word "tired".

Depth of calculation. Man can calculate a move on 8–10 moves in a tense position. The computer — on 30 moves, with dozens of branches, without losing concentration. This is unattainable biologically.

Objective evaluation function. AI is not subject to fear, authority of the opponent, emotions from a lost position, or desire to win beautifully. It always chooses the best move in its own view, without caring about the audience and not getting nervous in time pressure.

What remains for man: chess survived, but has changed

In contrast to the gloomy predictions of the 1990s, chess did not die. On the contrary, thanks to AI, it has become more popular. Top grandmasters use neural networks for analysis and preparation, identifying fundamentally new ideas in openings and endgames. In online tournaments, millions of viewers watch live broadcasts with instant engine evaluation. The ratio of man against man is still captivating, but the computer has become a coach, not an opponent.

Magnus Carlsen himself has said many times: "There is no point in playing against a computer, it dehumanizes the process. Chess is interesting because people make mistakes in it". AI did not defeat chess — it killed the competition between man and machine, but left the beauty of the game between people untouched. And in who is stronger — man or AI — there is no longer any doubt. The answer has been given finally and irrevocably.