Alex Gerko, founder of quantitative trading firm XTX Markets, is offering a $10 million prize through his company’s new AI-MO Prize competition. The bounty will go to the first team to develop an AI model capable of winning a gold medal in the prestigious International Mathematical Olympiad (IMO) high school maths competition.
You may also be interested in this $25,000 AI voice detection prize: FTC Offers $25,000 Prize for Detecting AI-Enabled Voice Cloning
The IMO features incredibly difficult maths problems that require creative problem solving skills beyond textbook knowledge. As such, current AI such as ChatGPT cannot reliably solve IMO problems that require original idea generation. However, Gerko believes that advancing AI to conquer such complex reasoning is both achievable within 5 years and will prove hugely valuable to mathematics and technology.
The new AI-MO Prize aims to incentivise machine learning researchers to make breakthroughs in mathematical reasoning AI. The details of the competition are still taking shape, but the $10 million reward would go to the creators of the first publicly shared model that matches a gold medal IMO performance. Additional money will also be offered for achieving incremental milestones.
Gerko has personal and professional motivations for sponsoring this AI maths challenge. A gifted Russian mathematics student, he narrowly missed out on a chance to compete in the IMO years ago, but remains passionate about mathematics competitions and educational programmes. His quantitative trading firm is also looking to hire the quantitative talent that often excels in maths competitions like IMO.
As AI advances to higher cognitive levels, uniquely human skills will become even more important, according to experts such as Carnegie Mellon professor Po-Shen Loh. Loh believes that an AI that wins IMO gold medals could arrive sooner than expected, thanks to the pace of progress in machine learning. The journey of developing such an AI through incremental breakthroughs could prove as rewarding as reaching the ultimate goal.
Ultimately, the AI-MO Prize aims to accelerate AI’s mastery of mathematical reasoning by offering financial incentives tied to concrete benchmarks. Similar past contests have driven innovation in AI for recommendation systems, self-driving cars, and other applications. Even fallout progress in mathematical AI could found valuable commercial applications. The $10 million carrot simply makes the pursuit of this difficult but potentially groundbreaking research more enticing.
Read more about the prize at the orignial article on the WSJ.