REU Site: Exploring the Limits of Intelligent Systems

Harvey Mudd College, Claremont CA

Investigators

Abstract

REU Site: Exploring the Limits of Intelligent Systems (#2243941) As machine learning and AI systems become ever more common, encroaching on every aspect of society, the need for understanding the systems and their fundamental limitations becomes even more pressing. This project promotes the progress of science and supports education and diversity by training undergraduate students to research questions on the limits of intelligent systems. It builds a cohort of undergraduate students who can develop research ability, presentation skills, and further interest in research-related computing careers. The project focuses on understanding the boundaries in what intelligent systems can achieve both theoretically and in complex real-world scenarios with non-expert users. By educating undergraduate students to probe and understand the limits of these systems, the project helps build an empowered citizenry capable of grappling with the complex ethical and social issues these systems present. Through this award, students will work on cutting-edge subprojects in computer vision, programming language analysis and synthesis, human-robot interaction, and information-theoretic understanding of machine learning systems. These research topics give students valuable academic and industry skills that extend beyond current AI models and frameworks towards the broader reaches of what computing may achieve in the future. The topics for this project are designed to be modular and hierarchical to enable successful undergraduate research milestones. Students will be housed at Harvey Mudd College, an undergraduate-only institution, and experience the most compelling aspects of a graduate school environment during a ten-week summer program. Students will actively engage with the entire research process, from literature search, to articulating problems of interest, to investigations of specific pieces of these problems, and focusing results for presentation and publication. 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.