Federal Project Grant Award Summary

The National Science Foundation (NSF) Division of Mathematical Sciences awarded The Leland Stanford Junior University a $270,000 Project Grant (CFDA 47.049, Mathematical and Physical Sciences) effective September 1, 2026, through August 31, 2029, to develop novel statistical methods for explainability and reproducibility in artificial intelligence (AI) systems. The project will create rigorous methodologies for testing hypotheses about variable importance in complex predictive models while controlling false discovery rates. The research integrates recent statistical advances including knockoff inference, e-values, conditional randomization tests, and explainable AI techniques to produce interpretable, verifiable, and trustworthy insights from sophisticated AI algorithms. A key innovation involves designing inference procedures robust to multiple data passes that avoid selection bias and circular reasoning while adapting to signal in the data.

The deliverables include the development of a methodological framework focused on context-dependent variable importance—examining how explanatory variables influence outcomes under different conditions and through non-additive interactions—with primary applications to genomic data analysis and inference on gene-gene interactions. The project will contribute to workforce development by supporting training for one graduate student in AI research and data science. By strengthening the scientific rigor and interpretability of AI-driven research, this award addresses critical needs for public trust and scientific validity in sensitive domains such as medicine, education, and public policy.