- Google DeepMind and OpenAI have announced that their artificial intelligence models achieved gold scores in the International Mathematical Olympiad for high school students.
- Both models solved five of the six problems, using general-purpose systems capable of reasoning in natural language.
- The IMO organization officially certified Google's results, while OpenAI was externally evaluated by former medalists.
- The milestone marks a significant advance in the use of AI for complex mathematical reasoning tasks, with implications for scientific research.
The landscape of artificial intelligence in academia has just undergone a significant change following the announcement that Google DeepMind and OpenAI AI models have achieved the score required for a gold medal in the prestigious International Mathematical Olympiad. (IMO). This achievement represents a significant advance in the ability of machines to tackle mathematical problems at the level of the world's best high school students.
Both companies, considered leaders in the development of advanced AI systems, They have announced that their models passed five of the six tests in the competition., obtaining a score of 35 points out of 42, which usually guarantees "gold" recognition in the competition. This result was achieved using general-purpose models that process statements using natural language, thus moving away from previous methods based on translation into formal languages or systematic calculations.
Participation and verification in the IMO
The IMO edition in which these models participated was held in Sunshine Coast, Queensland (Australia), with more than 630 students from 113 delegationsA total of 67 human competitors won the gold medal, a figure that places the AI's feat in a context of high demand and competition.
The validation of the results was carried out by Google's close collaboration with the organizing committee, which officially certified the performance of its AI and established procedures for publishing the data after an independent review. For its part, OpenAI conducted a self-assessment, using three former IMO medalists to qualify their results under the same rules as the competitors.
New models and methods
The main novelty of this edition lies in the use of informal reasoning models, like Google's Gemini Deep Think, that process and solve problems directly in natural language. This ability allows AI to analyze different solution paths in parallel, selecting the most appropriate strategy in a time similar to that of the students: 4,5 hours per exam without external support such as calculators or the internet.
Google highlighted that its model was able to execute the reasoning in a clear and detailed manner, while OpenAI highlighted the possibility of carrying out multiple lines of thought simultaneously through intensive computing, although they acknowledged that these versions will not be available to the public in the short term.
Debate and skepticism about the results
While the progress is undeniable, the The way OpenAI communicated its results has generated controversyThe IMO organization asked participating companies not to disclose their data before the review and awards ceremony, a request that Google respected, but OpenAI did not strictly adhere to.
In addition, some experts and committee members have warned about the lack of transparency in relation to the computational resources used and the possibility of human intervention in some processesHowever, the community generally accepts the mathematical validity of the solutions presented by AI.
Implications for the future of mathematical research
The progress achieved by the Google and OpenAI models reinforces the idea that Artificial intelligence can become a key tool to address unsolved problems in mathematics, as well as in other fields such as physics. Researchers like Junehyuk Jung of Brown University believe we are close to a closer collaboration between mathematicians and AI systems that will overcome significant scientific barriers, driving innovations across multiple disciplines.
In addition to the main competition, the IMO presented the AI Mathematical Olympiad Prize this year., valued at $10 million, aimed at encouraging the creation of open AI models for mathematics. This award went to a team from Nvidia, demonstrating the technology community's strong interest in these types of challenges and the growing professionalization of the sector.
Limitations and future of AI in mathematics
One of the highlights after the competition is the big difference between research models and those accessible to the publicAlthough laboratory AI has achieved impressive results, its open source versions have yet to even achieve a bronze medal in IMO tests. This raises questions about the scalability and democratization of these technologies, as well as the degree of maturity of current business models compared to their experimental counterparts.
The pace of progress in AI in mathematics has multiplied expectations about its ability to solve historical challenges. However, Both the mathematical and technological communities recommend maintaining a cautious stance., ensuring a rigorous and transparent evaluation before accepting final results.
The most recent edition of the The International Mathematical Olympiad has marked a turning point for artificial intelligence., consolidating Google and OpenAI as leaders in the application of advanced reasoning models. Although questions remain about the methodology and public availability of these solutions, the breakthrough indicates a promising path toward ever-closer collaboration between humans and machines in solving complex problems in mathematics and other scientific disciplines.
I am a technology enthusiast who has turned his "geek" interests into a profession. I have spent more than 10 years of my life using cutting-edge technology and tinkering with all kinds of programs out of pure curiosity. Now I have specialized in computer technology and video games. This is because for more than 5 years I have been writing for various websites on technology and video games, creating articles that seek to give you the information you need in a language that is understandable to everyone.
If you have any questions, my knowledge ranges from everything related to the Windows operating system as well as Android for mobile phones. And my commitment is to you, I am always willing to spend a few minutes and help you resolve any questions you may have in this internet world.