Human Champions Win Math Olympiad as AI Models Achieve First Gold Scores
Human contestants outperform AI at the International Mathematical Olympiad, but advanced models from Google and OpenAI earn gold-level scores for the first time. The rapid improvement of these artificial intelligence systems highlights significant progress in machine reasoning capabilities.
Human contestants defeat advanced AI models at the prestigious International Mathematical Olympiad, but the technology shows alarming progress. Five young students earn perfect scores, while AI systems from Google and OpenAI fall short of full marks. Despite the human victory, the artificial intelligence programs achieve a major milestone by reaching gold-medal level performance for the first time in the competition's history.
Google's Gemini chatbot and OpenAI's experimental reasoning model each score 35 out of a possible 42 points, securing a gold-level result. This achievement marks a significant leap from last year, when Google only achieves a silver-medal score and takes days to compute the answers. This year, the Gemini model solves five of the six problems within the official 4.5-hour time limit, matching the pace of human participants.
IMO president Gregor Dolinar praises the clarity and precision of the AI-generated solutions, though he notes that organizers cannot verify the exact computing power used or the level of human involvement. Approximately ten percent of the 641 human contestants from 112 countries win gold medals this year. The rapid advancement of these AI models sparks deeper reflection about the future relationship between human intelligence and machine capabilities in complex problem-solving.