AI Rivalry Moves to the Math Olympiad

Artificial Intelligence (AI) is also taking over the world of elite mathematics. Models developed by OpenAI and Google have won gold medals at the prestigious International Mathematical Olympiad (IMO) - a competition dedicated to the world's brightest young mathematicians. These performances highlight recent advances in AI, but also the fierce competition between the two tech giants.

• Two Models, Two Gold Medals

OpenAI kicked off the party on the X platform (formerly Twitter), where it announced that its latest mathematical reasoning model had achieved a score equivalent to a gold medal at the current edition of the IMO. The model was designed to simulate and solve complex math problems, using advanced "experimental reasoning” methods. Just a few hours later, Google issued a similar statement, saying that its advanced AI model had solved five of the six Olympiad problems, a feat that also falls into the gold medal category. This result marks an improvement over last year, when Google's AI only won the silver medal in the same competition.

• Reasoning or simulating reasoning?

While both companies claim that their AI "reasons,” several experts in artificial intelligence and mathematical logic point out that what the current models do is more of a sophisticated form of reasoning simulation, not a proper understanding of mathematical concepts. "Large language models (LLMs) do not think in a human sense. They build coherent logical chains based on learned data, but they do not have consciousness or real mathematical intuition,” Oxford mathematician Professor Marcus du Sautoy told The Guardian.

• Results validation dispute

Google was quick to criticize OpenAI's premature announcement, claiming that the organizers of the Olympiad expressly require that the results be published only after they are verified by independent experts. OpenAI quickly responded, claiming that it was unaware of this rule and that it called on a separate team of independent specialists to validate the model's performance, without directly consulting the IMO committee. This procedural divergence could raise questions about the transparency and comparability of the results, but it does not seem to overshadow the technical achievements themselves.

• What is the IMO, and why does it matter for AI?

The International Mathematical Olympiad (IMO) is the most important mathematics competition for high school students, having been organized annually since 1959. The proposed problems are recognized for their extreme difficulty, requiring not only advanced knowledge, but also creativity and mathematical intuition. By using this framework as a benchmark, AI companies are validating their ability to approach and solve abstract problems, beyond the usual linguistic routine. Such achievements bring AI models closer to human symbolic reasoning, an important step towards artificial general intelligence (AGI).

• A technological duel with global impact

The competition between OpenAI and Google is not new, but in the context of IMO it has become a symbol of the "race” to develop the most intelligent machine. Google has invested heavily in the Gemini model, and OpenAI is constantly developing the GPT family, which is already in its fourth generation (GPT-4) and testing prototypes of advanced mathematical reasoning.

In a landscape marked by ethical dilemmas, fears of losing control over AI, and regulatory challenges, such technical demonstrations have a double value: on the one hand, they are spectacular engineering achievements, on the other hand, they become weapons of influence in the battle for technological supremacy. As AI models begin to outperform humans in logic and math competitions, the scientific community remains divided between admiration and caution. OpenAI and Google have not expressed their intention to publish the details of their methods, but both companies have suggested that the models will continue to be trained on IMO problems as a key test of progress. AI is no longer just an office assistant or a text generator. It is already an Olympic contender. It remains to be seen whether it will one day become a Nobel laureate.