Caltech Mathathon Shifts Focus to AI-Augmented Mathematical Understanding

Caltech mathematicians and community leaders have announced a strategic redesign of the Mathathon, pivoting the event focus from competitive problem-solving to human-centric mathematical exposition. Scheduled for November 13 to 15 at Caltech, the revamped initiative, now themed Old Problems, New Proofs, aims to explore how artificial intelligence can be leveraged to enhance, rather than replace, human comprehension of complex mathematics. Under the new format, participants will select historically solved but notoriously difficult problems, such as the four color theorem, the ABC conjecture, or the Navier-Stokes equations. Over an initial 40-hour intensive period, teams will analyze existing solutions and identify conceptual gaps. They will then spend two months developing alternative proofs or pedagogical explanations, culminating in a public explainer and a corresponding GitHub repository. These submissions will document the entire development process, including computational scripts, visualization code, and large language model interaction logs, ensuring full transparency in methodology. The organizers have deliberately decoupled Mathathon from proprietary artificial intelligence sponsorships to maintain academic independence. Instead, participants will receive discretionary cash grants to fund their preferred computational or educational resources, whether that involves large language model credits, high-performance computing access, or traditional analytical methods. While proprietary systems remain permissible, the event strongly advocates for open-source alternatives. The Foundation for Science and AI Research will partner with Caltech to provide computational infrastructure, host technical workshops, distribute travel stipends, and award prizes to teams prioritizing open models. Algorithmic trading firm XTX Markets serves as the lead financial backer, reflecting its stated commitment to open-source infrastructure and academic research. This structural overhaul responds to growing concerns within the academic community regarding the overemphasis on algorithmic problem-solving at the expense of mathematical intuition and pedagogical clarity. By centering the competition on exposition and conceptual synthesis, the organizers aim to validate the often-overlooked work of mathematical communication and community mentorship. The advisory board and organizing committee deliberately include a wide spectrum of perspectives, ranging from undergraduate students to established faculty, and encompassing both advocates and skeptics of artificial intelligence in research. Organizers emphasize that Mathathon is an exploratory framework rather than a definitive verdict on artificial intelligence in mathematics. The event mandates strict disclosure of AI usage, requiring participants to justify tool selection and evaluate whether smaller-scale models might suffice for their tasks. Ultimately, the initiative seeks to establish a reproducible model for future academic gatherings that integrate computational tools while prioritizing peer education, conceptual depth, and collective scholarly advancement. As the mathematical community navigates unprecedented technological shifts, Mathathon positions itself as a foundational experiment in balancing innovation with human-driven discovery.