Developing core probabilistic reasoning for data science using AI-driven modeling

The National Science Foundation Division of Undergraduate Education awarded $400,000 to the University of Delaware on October 1, 2026, for a Project Grant supporting development of core probabilistic reasoning for data science using AI-driven modeling under the STEM Education program (CFDA 47.076).

The University of Delaware will incorporate large language models into probabilistic modeling courses and study how artificial intelligence tools influence student reasoning processes in advanced undergraduate mathematics. The project explores how interactions among AI tools, faculty, and peers collectively shape student probabilistic thinking about real-world models and model development. Data collection includes videos of student group work during in-class modeling activities, student assignments, and surveys; findings will be synthesized into model diagrams capturing students' modeling approaches and reasoning. An expert advisory board will provide formative and summative feedback throughout the project.

Performance occurs in Newark, Delaware, with a period of performance from October 1, 2026, through September 30, 2029. This Level 1 IUSE ESL project serves the national interest by exploring how AI can support student reasoning while maintaining student thinking and understanding as the primary objective, rather than replacing human cognition.

Abstract

This level 1 iuse ESL project from the university of delaware aims to serve the national interest by exploring how AI tools can be used to support student reasoning in advanced undergraduate mathematics courses. Specifically, the project will incorporate large language models into probabilistic modeling courses and study how the use of artificial intelligence tools influence student modeling processes. Interactions with AI tools, faculty, and peers can all contribute to changes in student reasoning, and this project will explore how those three factors work together to improve modeling skills and understanding. Current research highlights the potential risks and rewards of AI use in undergraduate stem courses. Using AI tools to support and guide, instead of replace, student thinking can be a powerful approach to strengthen student learning and understanding. Project research will explore how students? Probabilistic thinking about real-world models develops in settings that incorporate the use of AI. Of particular interest is how students? Reasoning about model development is impacted by interactions with AI tools, peers, and faculty. Data streams include videos of student group work during in-class modeling activities where students will use AI to support model development, student work from in- and out-of-class assignments, and surveys. These data will be used to create model diagrams that capture students? Modeling approach and reasoning. An expert advisory board will provide formative and summative feedback across the project to strengthen implementation and progress towards project goals. The NSF iuse: EDU program supports research and development projects to improve the effectiveness of stem education for all students. Through the engaged student learning track, the program supports the creation, exploration, and implementation of promising practices and tools. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the foundation's intellectual merit and broader impacts review criteria.- subawards are not planned for this award.