Students are rapidly integrating Large language models (LLMs) into their study practices, yet their role in core engineering curricula remains poorly understood. This study examines how undergraduate students utilize LLMs in an introductory electrical circuits course and how these tools influence their learning behaviors. A survey was administered to 184 students across mechanical, electrical, computer, and bioengineering majors.

Results indicate that LLM use is widespread, with over 80% of participants reporting weekly engagement. Students leverage LLMs for conceptual explanations (81%), answer verification (68%), and exam preparation (61%), rather than direct solution generation (30%). 98% of students reported encountering incorrect or misleading LLM responses—most commonly involving misinterpreted circuit diagrams, algebraic errors, and misapplications of circuit laws. To verify LLM accuracy, students cross-reference course notes (74%), utilize circuit simulators (30%), or consult classmates (30%). Despite these limitations, students report substantial perceived benefits, including clearer explanations (71%), improved mathematical comprehension, and reduced academic stress (61%). However, many students express concern that excessive reliance on LLMs may diminish independent problem-solving proficiency.

These results highlight a need for the strategic integration of LLMs in engineering education— one that leverages LLM’s explanatory benefits while implementing pedagogical safeguards to ensure the development of students' independent problem-solving skills.

Authors:
Ricardo de Castro, University of California, Merced, United States

About the Presenter(s)
Dr. de Castro is an Assistant Professor at University of California, Merced. His research focuses on controls and optimization with applications to electric vehicles.

See this presentation on the full schedule – Monday Schedule

Bookmark