Grant Award Summary

The University of Chicago received a $200,000 Project Grant from the National Science Foundation (NSF) Division of Mathematical Sciences under the Mathematical and Physical Sciences program (CFDA 47.049), with an award date of June 1, 2026, and completion date of May 31, 2029. The project develops a novel statistical framework for creating interpretable and reliable generative artificial intelligence (AI) systems through nonparametric causal factor models. The research delivers practical models and algorithms that enable stakeholders to understand, verify, and reproduce the internal mechanisms of complex AI systems, moving beyond black-box approaches to transparent, trustworthy systems whose decision factors can be interpreted and diagnosed.

The project integrates methodologies from causal inference, nonparametric statistics, latent variable modeling, and deep learning to investigate how generative models learn reusable causal structures from high-dimensional data such as images, language, and scientific measurements. Key deliverables include rigorous statistical guarantees, reproducible research findings, and frameworks that address how AI systems learn interpretable causal factors without suffering from the curse of dimensionality. No subawards are planned for this award.