Sydney — In a major international mathematics competition, humans outperformed generative AI models created by Google and OpenAI. However, these AI programs achieved impressive gold-level scores for the first time, showcasing a notable rate of improvement that warrants human reflection. While neither AI model reached full marks, five young individuals at the International Mathematical Olympiad (IMO) – an esteemed annual competition for participants under 20 years old – attained perfect scores. Google announced that an enhanced version of its Gemini chatbot successfully solved five out of six math problems at the IMO, hosted in Queensland, Australia. The U.S. tech giant confirmed that Google DeepMind achieved a significant milestone by earning a gold medal score of 35 out of 42 points, as reported by IMO president Gregor Dolinar. The solutions presented by the AI models were praised for their clarity, precision, and ease of comprehension by IMO graders. Approximately 10% of human contestants received gold-level medals, with five achieving perfect scores of 42 points. OpenAI also celebrated success as its experimental reasoning model secured a gold-level score of 35 points on the test, marking a significant achievement in the realm of AI. Google had previously achieved a silver medal score at last year’s IMO, but this year, the Gemini model demonstrated remarkable progress by solving the problems within the 4.5-hour time limit. The competition organizers noted that tech companies privately tested closed-source AI models on this year’s problems, aligning with the challenges faced by 641 students from 112 countries. IMO president Dolinar expressed excitement over the advancements in the mathematical capabilities of AI models. The competition organizers could not verify the extent of computing power used by the AI models or if there was any human involvement. In an interview earlier this year, a leading AI researcher from Google predicted that computers would attain human-level cognitive abilities within the next five to ten years, a milestone known as “artificial general intelligence.” Google DeepMind CEO Demis Hassabis emphasized the rapid progress of AI technology and its potential to understand complex problems and develop imagination due to increased investment and talent in the field.
