Project Grant 2412675

This $599,411 Project Grant award from the National Science Foundation's (NSF) Computer and Information Science and Engineering (CISE) program (CFDA 47.070) supports the development of a "Trustworthy Toolbox for Double-Correct Predictive Modeling in Sciences." The project aims to create advanced artificial intelligence (AI) and machine learning (ML) models that can make accurate predictions while also providing transparent, scientifically-grounded rationales for their outputs. This is intended to enhance the reliability and trustworthiness of AI/ML systems used in critical scientific domains like extreme weather forecasting. The research will focus on three key areas: improving prediction accuracy for unusual data patterns, maximizing the reliability of model reasoning, and enhancing the scalability of the models to handle growing scientific data volumes. The results will be translated into open-source software, reusable workflows, and AI-ready datasets made publicly available to the broader scientific and technology communities. The University of Delaware, the prime awardee, will collaborate with academic and non-academic partners to integrate the research findings into educational initiatives at multiple levels, from undergraduate research to K-12 outreach.