Kasparov, Garry

former World Chess Champion; pro-democracy activist; author of "Winter Is Coming" and "Deep Thinking"

Garry Kasparov: Chess, Deep Blue, AI, and Putin

Lex Fridman Podcast · Oct 27, 2019

• Garry Kasparov
• chess psychology
• Deep Blue
• closed systems vs open systems
• AlphaZero
• human-machine collaboration
• algorithmic bias
• Vladimir Putin
• Soviet history
• election interference

Core ideas

Kasparov attributes his chess success not primarily to fear of losing or love of winning but to an unshakeable belief that game outcomes were always a direct result of the quality of his own play — a conviction he says paradoxically freed him to make bold, uncertain decisions rather than hesitating, since "fear of a mistake guarantees mistakes." He distinguishes top players from the very best by the ability to commit to a decision under genuine uncertainty (without predictable consequences) through inner confidence rather than exhaustive calculation, and frames his own post-chess career (democracy activism, human-machine relations writing) as driven by the same underlying need to "create something new" and make a difference, even knowing his influence there would be less decisive than at the chessboard. Reflecting on his rivalry with Karpov, he singles out Game 16 of their 1985 World Championship match — where he found a genuinely novel, at-the-time-unrefuted opening idea that was later shown to be theoretically flawed — as illustrating how deep preparation converts into something like real-time intuitive brilliance even when the specific analysis doesn't directly "work," since the discipline itself builds a capacity for finding ideas under pressure.

Asked to rank Magnus Carlsen among the chess greats, Kasparov resists direct cross-era comparisons, arguing each generation inherits accumulated chess knowledge unavailable to its predecessors (a modern amateur with a chess engine already knows more codified opening theory than Kasparov did at his peak), so the fairer comparison is the size and duration of a champion's gap over contemporaries. He describes Carlsen's game as a rare fusion of Bobby Fischer's relentless, resourceful fighting energy with Anatoly Karpov's minimalist efficiency (extracting maximum effect from minimum material advantage), sustained over unusually long games, and notes physical conditioning (chess players burn significant calories per game) as an underrated factor in modern elite play.

On his 1997 loss to IBM's Deep Blue, Kasparov states this was painful specifically because it was his first-ever tournament loss of any kind, and that he harbored real suspicions (which he still partly maintains, while calling it "water under the bridge") that IBM's conduct of the rematch was not entirely fair, given IBM's commercial incentive and his belief he was objectively the stronger player at the time. He reframes the broader cultural narrative around that match as based on a category error: chess was never truly a "pinnacle of human intellect" but a "closed system" — a finite, fully rule-bound domain where a chess-playing machine only needs to reduce its own mistake rate relative to a human, not achieve any deeper understanding, since the game's ~10^46 possible legal move sequences (per Claude Shannon) can never be exhaustively solved by any computer regardless of processing power. He extends this "closed system" framework to Go, Shogi, and video games, arguing machines will always eventually dominate any well-defined closed system through consistency (lack of human-style instability) rather than superior intelligence — evidenced by the enormous modern gap between top chess engines (~3500 rating) and even Magnus Carlsen (~2850), comparable, in his framing, to the gap between a Ferrari and Usain Bolt.

Kasparov identifies AlphaZero as qualitatively different from earlier "brute force" AI (including Deep Blue), because it represents machine-generated knowledge and pattern discovery (learned from ~60 million self-played games) rather than merely faster processing of human-generated data and heuristics — describing his own excitement studying AlphaZero's willingness to sacrifice material for long-term positional advantages that resembled genuine intuition. He notes a key limitation, though: AlphaZero-style systems, once they develop a flaw or blind spot, require enormous numbers of games (hundreds of thousands) to self-correct, whereas a human collaborator could identify and correct such a flaw far more quickly — the basis for his broader argument that the future is not human-versus-machine competition but human-machine collaboration, where humans should identify the narrow "few decimal points" of territory where their judgment genuinely adds value and otherwise avoid interfering with a machine's superior performance (illustrated by his suggestion that an experienced nurse reading radiology scans might outperform a top professor specifically by trusting rather than second-guessing the machine's read). On algorithmic bias, he argues it's incoherent to expect machines to be less biased than the societies that produce their training data — "like looking in a mirror and complaining about what you see" — while noting society holds autonomous systems to a much higher (and, in his view, somewhat inconsistent) safety standard than it holds human drivers, who cause far more fatalities in aggregate.

The episode's second half turns to Kasparov's political views. He argues the 20th century's clear historical lesson is that totalitarian, centrally-planned systems cannot sustain innovation and inevitably fail, and states his belief that communism caused more overall damage than any other 20th-century ideology, criticizing the absence of a "Nuremberg for communism" comparable to the postwar condemnation of fascism — citing his own public efforts (including a speech on the Molotov-Ribbentrop Pact) to argue Stalin's regime bears direct responsibility for enabling the start of World War II. Discussing his own safety as a vocal Putin critic living in exile in New York, Kasparov acknowledges genuine risk to his life, references his family's inherited dissident ethic ("do what you must and so be it"), and states his belief that Putin's regime, like past dictatorships, will end suddenly and unpredictably — expressing confidence he will return to Russia "sooner than many people think." Asked directly about Russian interference in U.S. elections, Kasparov states unequivocally that he believes the Russian government actively worked to elect Donald Trump in 2016 and would do everything possible to keep him in power through 2020, framing Putin's public posture toward Trump as resembling how a KGB officer regards an intelligence asset, and expressing concern less about Trump specifically than about the absence, in his view, of a credible political response to the broader threat he believes Russian interference poses to democratic institutions.

Excerpts