Is Chess Solvable? Tech Titans Elon Musk and Demis Hassabis Think So

Elon Musk and Demis Hassabis Believe Chess Can Be “Solved”

Chess purists might balk at the idea, but some of the biggest names in technology seem to believe that their beloved game could be “solved”.

Elon Musk and Google DeepMind’s Demis Hassabis have independently said that chess can be “solved”. Solving chess refers to the goal of determining — with absolute certainty — the optimal first move and complete game strategy for chess, such that a perfect game would always end in a predetermined outcome (either a win for white, a win for black, or a guaranteed draw). Solving chess is currently impossible for computers because of the high number of variations any single game can take, but if technical advancements could create sufficiently powerful computers, they could eventually solve chess. This could mean that playing chess could end up having a simple strategy that can be simply be read out from a (massive) database.

Google DeepMind’s Demis Hassabis, who also won the Nobel Prize this year, had congratulated India’s Gukesh Dommaraju on X after he won the World Chess Championship. “Huge congratulations @DGukesh my friend!! The new and youngest ever world chess champion at 18 – absolutely incredible !” he had posted. Elon Musk too had personally congratulated Gukesh on X with a post saying “Congratulations!”. But he replied to Hassabis’ post as well. “This is cool, but I am quite certain that Chess can be fully solved, like Checker,” Musk said.

Demis Hassabis, who was an extremely good chess player before he became involved in AI, seemed to agree. “I agree that chess is likely solvable, but that doesn’t mean it won’t still be very fun for humans to play for a long time yet. It’s also great training for the mind and for kids to learn it at school,” Hassabis replied.

Now solving chess isn’t straighforward. Chess’s combination of different pieces with different moving styles means that an extremely large number of unique games can take place. A standard chess game has approximately 10^120 possible unique game variations, a number so large that it far exceeds the number of atoms in the observable universe. Chess also has an estimated 10^43 board positions.

But technology has made rapid strides in chess. In 1997, IBM’s Deep Blue had beaten the then-best player in the world, Garry Kasparov, in a match that had been dubbed as a match between man and machine. Over the next two decades, computers became steadily better at chess, and the current best chess engines are significantly better than the best human chess players. The highest rated human player, Magnus Carlsen, current has an ELO rating of around 2800, but the best chess engines can work at 3400 levels, which are far beyond what any humans are capable of.

Solving chess, though, would be a different challenge. The strongest chess engines today can evaluate positions and suggest optimal moves, but cannot conclusively state whether chess inherently favors one side if played with perfect strategy. Mathematically “solving” chess remains an open challenge in computational theory and game strategy, representing one of the most complex computational problems imaginable. But with advancements in AI — and more recently in quantum computing — two of the biggest tech minds in Elon Musk and Demis Hassabis seem to believe that it will eventually be surmounted.