What you should know
Two artificial intelligence (AI) models have won gold medals in the recently concluded International Mathematical Olympiad (IMO). Google and OpenAI models have announced that they have won gold in a competition between high school students.
Both companies' AI models correctly solved 5 out of 6 math questions. They used general-purpose 'reasoning' models to solve mathematical concepts in natural language, according to Reuters. Although AI models have participated in Mathematics Olympiad before, this time they have presented their ability to compete with human intelligence by winning gold. In the
competition, Google used its Gemini Deep Think model. A new experimental model based on OpenAI's 'Test-Time Compute' participated in the competition. With this achievement, some mathematicians and other scientists have now expressed the expectation that they can use AI models more effectively in their studies.
A total of 630 students participated in the 66th International Mathematics Olympiad in Queensland, Australia. 67 of them i.e. about 11 percent of the students won gold medals. In this year's competition, the AI models performed as well as the students and won the gold medal, which is considered an important achievement.
Last year, a math-specific AI model from Google's DeepMind AI unit won a silver medal. According to Google's blog post, earlier models were based on formal languages and Lambethian calculations. However, this year, Google has completely developed a natural language AI model and said that it solved the mathematical problem within the set time limit of 4.5 hours.
OpenAI also prepared an experimental version of its special 'Reasoning Model' for this competition. Researcher Alexander Wei, posting on social network X, said that OpenAI does not plan to release a product with such a high mathematical ability immediately.
