Chess Grandmaster and former World Champion Magnus Carlsen delivered a controversial take on Elon Musk-owned xAI’s latest model, Grok 4. The comment came after its performance in a recent chess match against OpenAI’s o3 model. Carlsen, who was providing commentary for the tournament, reacted with a mix of laughter and disbelief as Grok 4 made a series of blunders, ultimately losing 4-0 to its rival.
Carlsen’s most pointed remarks came during the final, where Grok 4 made questionable knight and bishop sacrifices and repeatedly blundered its queen. The grandmaster said that the spectacle was similar to “watching kids’ games.” He also followed it up with, “Hope everyone feels better about their games after watching this,” especially as the AI chatbot struggled to make coherent moves.
OpenAI o3 model plays ‘like a chess player’
After the one-sided match ended, Carlsen provided a blunt assessment of the AI models’ chess strength. He estimated Grok 4’s skill at a lowly “800,” while rating o3 at “1200.” He described OpenAI’s o3 as looking “like a chess player” that was “fairly ruthless in conversions,” indicating that the AI model had a basic but effective understanding of the game.
In contrast, Grok 4 seemed to have only learned a few opening moves and the rules, but little else.
Carlsen described Grok’s moves as “chess-related moves” but noted they came at “the wrong time and in weird sequences.” During a game where Grok gave away its queen, he said that it was “like that one guy in a club tournament who has learnt theory and literally knows nothing else.”
The World Champion did not spare other large language models from his criticism too, stating that “Both Gemini and Mini were not very good” at chess, while also expressing his disappointment with Claude.
The tournament highlighted a stark performance gap between the AI models, with Carlsen’s commentary providing a humorous yet critical look at the current state of AI in the world of chess.
Will GPT-5 outdo its predecessor?
GPT-5 is yet to hit the chessboard for a clash against its rivals. OpenAI has already claimed that the new AI model is far superior in terms of overall performance and processing complex tasks. Hence, it should come as no surprise if Carlsen finds GPT-5 to be a better chess player than the ‘already impressive’ GPT-o3 model.
