Federal Grant Award Summary

Washington State University received a $326,994 CAREER (Faculty Early Career Development) Project Grant from the National Science Foundation's Division of Computing and Communication Foundations under the Computer and Information Science and Engineering program (CFDA 47.070), effective June 1, 2026 through May 31, 2031. The award supports research and development of novel methodologies for improving the reliability and adaptability of foundation models—large artificial intelligence systems used across science and technology sectors. The primary deliverables include a new information flow representation framework that reshapes how large AI models transmit information to enable more reliable assessment of model outputs, and a unified computational framework for systematically integrating diverse data sources with established scientific knowledge. These technical advances are designed to address critical gaps in identifying when foundation model outputs are unreliable, particularly in safety-critical applications such as health monitoring and smart home systems.

The research approach treats internal model stages as information flow snapshots, enabling structured uncertainty tracking in lower-dimensional spaces to provide scalable reliability estimates applicable to large-scale models. Key products include methods supporting adaptation to new data modalities from emerging sensors and mechanisms for incorporating scientific laws and domain knowledge into model operations. Beyond technical research outcomes, the grant supports workforce development activities focused on training practitioners in leveraging foundation model technologies responsibly and implementing trustworthy decision-support systems. The five-year project timeline reflects the CAREER program's emphasis on integrating research innovation with educational components for early-career faculty development.