TL;DR:
- XTX Markets launched the ‘AI-MO Prize,’ a $10 million challenge fund to stimulate the development of AI models capable of achieving gold medals in the International Mathematical Olympiad (IMO).
- The grand prize is $5 million for the first AI model to match IMO gold medal standards.
- Additional progress prizes totaling $5 million will be awarded for AI models reaching key milestones.
- The initiative aims to go beyond existing AI models and nurture new ones with mathematical reasoning abilities.
- It seeks to compare AI problem-solving capabilities with those of exceptional young mathematicians.
- An AI-MO Advisory Committee will oversee prize design, with competitions starting in early 2024.
Main AI News:
In a bold move to foster innovation in the realm of artificial intelligence, XTX Markets has introduced the ‘Artificial Intelligence Mathematical Olympiad Prize’ or ‘AI-MO Prize,’ a prestigious $10 million initiative. This challenge fund is strategically designed to catalyze the development of a publicly-shared AI model that can achieve the coveted gold medal status in the International Mathematical Olympiad (IMO).
The pinnacle of this endeavor will see a staggering $5 million grand prize awarded to the first publicly-shared AI model that successfully competes in an AI-MO approved competition, demonstrating mathematical prowess equivalent to securing a gold medal at the esteemed IMO. Additionally, a series of progressive prizes, amounting to a total of $5 million, will be earmarked for publicly shared AI models achieving significant milestones en route to the ultimate grand prize.
The IMO, renowned as a prestigious high school mathematics competition, holds a dual distinction as a yardstick for exceptional mathematical performance and a reliable predictor of future accomplishments. In fact, it’s worth noting that half of all Fields Medallists, the highest honor in mathematics, once participated in the IMO during their formative years.
Taking inspiration from these illustrious traditions, the AI-MO Prize aims to push the boundaries beyond the capabilities of large language models such as ChatGPT, embarking on a journey to nurture novel AI models with a profound aptitude for mathematical reasoning. In essence, the AI-MO Prize has the potential to serve as a contemporary Turing Test, gauging the mathematical prowess of AI against human intellect.
Gregor Dolinar, President of the International Mathematical Olympiad, expressed his enthusiasm, stating, “Using challenging mathematical problems as a litmus test to evaluate whether AI can match the pinnacle of human cognitive capabilities is a brilliant concept. Each year, over 600 of the brightest students from more than 110 countries grapple with six exceedingly complex IMO problems, with typically only one or two students successfully tackling all six within the allotted nine hours. I am confident that many will closely follow the AI-MO Prize with bated breath, eagerly awaiting the day when AI can rival the world’s most brilliant young minds.”
Terence Tao, a Fields Medallist from UCLA, weighed in on the intricate nature of using AI to solve advanced mathematical problems. He remarked, “Despite recent advancements, leveraging AI for the resolution, or at the very least, assistance in solving intricate mathematical problems, remains an exceptionally intricate and multifaceted challenge. It is imperative to explore diverse approaches to this objective and assess the performance of each rigorously. The AI-MO Prize pledges to provide a valuable set of benchmarks that will facilitate the comparison of various AI problem-solving strategies on a technical level, making it accessible and engaging for a broader audience.”
The development and formulation of these prizes will be orchestrated by an AI-MO Advisory Committee, to be appointed shortly, comprising distinguished mathematicians, AI and machine learning experts, and experienced Olympiad problem-setters.
Anticipated to commence in early 2024, the inaugural AI-MO approved competitions will offer participants a chance to make their mark in the realm of AI-driven mathematical problem-solving. A presentation showcasing the progress achieved will be held at the 65th IMO, scheduled to take place in Bath, England, in July 2024.
Further insights into competition regulations and entry criteria will be unveiled in due course, promising an intellectually stimulating and rigorous challenge for participants and spectators alike.
Geoff Smith, former President of the International Mathematical Olympiad and member of the AI-MO Prize Advisory Committee, applauded the remarkable capabilities of modern AI systems while highlighting the nuances involved in replicating the creative thinking of mathematicians. He affirmed, “The ability of contemporary AI systems to emulate human communication in written form is truly remarkable, particularly when there is a vast corpus of relevant information available on the internet. However, emulating the creative thought processes of mathematicians engaged in abstract reasoning presents a far more daunting task. The global IMO community is thrilled to be actively supporting this competition, as we endeavor to measure AI systems against the exceptional young mathematical minds of our time.”
Simon Coyle, Head of Philanthropy at XTX Markets, expressed excitement about the AI-MO Prize and its potential to cultivate AI models proficient in mathematical reasoning. He eagerly anticipates collaborating with mathematicians, scientists, and technologists to refine and shape the prizes, all the while observing how AI models stack up against the cream of the crop in high-school mathematics.
Conclusion:
XTX Markets’ introduction of the $10 million AI-MO Prize signifies a pivotal step in advancing the field of artificial intelligence. By fostering the development of AI models capable of achieving mathematical excellence akin to IMO gold medals, this initiative not only fuels innovation but also opens up new possibilities for AI applications in various industries. It provides a platform for benchmarking AI against human cognitive abilities, which has significant implications for the evolving AI market, pushing the boundaries of what AI can achieve in the realm of mathematical problem-solving and beyond.