The "Mathematical Foundations of Artificial Intelligence" (MFAI) program, as outlined in NSF 24-569, is a funding opportunity provided by the National Science Foundation (NSF). The program aims to address critical gaps in the mathematical and theoretical foundations of artificial intelligence (AI). Despite AI’s remarkable achievements across various scientific fields, deeper mathematical insights are essential to further advance the technology and address its limitations.

The MFAI program encourages research that explores the capabilities and limitations of AI, develops new mathematical principles, and enhances our understanding of AI systems. Key areas of interest include the mathematical aspects of foundation models, deep learning, federated learning, and other emerging AI paradigms. The program also seeks to promote mathematically grounded design principles that could lead to more explainable, interpretable, and scalable AI technologies.

Up to 15 awards are expected to be granted per competition cycle, with funding ranging between $500,000 and $1.5 million per project. The total anticipated funding is $8.5 million annually, contingent on the availability of funds and the quality of proposals. Each project will typically have a duration of 36 months. Proposals should focus on collaborative interdisciplinary research, drawing expertise from mathematics, computer science, engineering, and social sciences, as relevant.

Eligible applicants include U.S.-based institutions of higher education and non-profit, non-academic organizations associated with research or education. Each proposal must be led by a Principal Investigator (PI) who holds a tenured or tenure-track position, or a full-time paid research or teaching role at an eligible organization. Proposals must be submitted through Research.gov or Grants.gov by the deadline of October 10, 2024, with subsequent deadlines in 2025 and 2026.

The program places emphasis on fostering collaboration across disciplines and institutions, including underrepresented groups and Minority-Serving Institutions (MSIs). Proposals will be evaluated based on intellectual merit, broader societal impacts, and the proposed collaboration plan. The NSF encourages projects that contribute to STEM diversity and build a workforce capable of advancing AI research in the future.

Applicants are advised to carefully review the NSF Proposal and Award Policies and Procedures Guide (PAPPG) for submission guidelines. The program does not require cost-sharing, and successful proposals must demonstrate innovative mathematical contributions and a clear plan for collaboration. All submissions must include budgets that account for participation in an annual PI meeting.

This funding opportunity presents a significant chance for researchers to contribute to foundational AI research while addressing real-world challenges and promoting responsible AI development.