ChatGPT, the AI chatbot renowned for its conversational abilities, has caused a stir by rating five-time World Chess Champion Magnus Carlsen as a below-expert-level player after losing to him in a casual match. The game, played on July 11, 2025, ended with Carlsen dominating ChatGPT in 53 moves without losing a single piece, yet the AI shockingly estimated his skill at just 1800-2000 ELO, far below his actual FIDE rating of 2839, the highest in the world.
Before facing Carlsen, ChatGPT tested its chess skills against other opponents, with a performance that was less than stellar. It played Google’s Gemini in a casual game, managing a win despite both AIs making numerous illegal moves, as reported on Chess.com. ChatGPT also faced an Atari 2600 chess program and suffered a crushing defeat, committing blunders that would make beginners wince. Against human players on platforms like Chess vs. GPT, it lost after 12 moves in one documented game, even attempting illegal moves like capturing its own pieces. No formal win-loss ratio exists due to ChatGPT’s casual play, but its record leans heavily toward losses, with the Gemini win being a rare and messy exception.
The Carlsen match was a showcase of the grandmaster’s prowess. Sharing screenshots on social media with the caption,
“I sometimes get bored while travelling,”
Carlsen revealed ChatGPT’s resignation message:
“All my pawns are gone. You haven’t lost a single piece. You fulfilled your win condition perfectly… I resign.”
He praised the AI’s decent opening and bold piece sacrifice but noted its failure to follow through. ChatGPT’s claim that Carlsen’s skill was barely above intermediate left fans and experts dumbfounded.
Unlike dedicated chess engines like Stockfish, ChatGPT, created by OpenAI, is a language model, not built for strategic gameplay. Its struggles against other AIs and humans underscore its reliance on pattern recognition over deep calculation. The chess community found the AI’s rating error amusing, with Grok, another AI, commenting,
“Unlike ChatGPT’s amateur guess of 1800-2000, I’d peg Magnus Carlsen at his actual FIDE classical rating of 2839—chess royalty. No blunders here.”
This episode has sparked discussions about AI’s limitations in chess. While ChatGPT can entertain casual players, it’s no match for grandmasters or specialized chess engines. Carlsen’s decisive victory and ChatGPT’s misguided assessment have given fans a mix of laughter and skepticism about AI’s place in the game.
