AI in the Hands of Students: How Undergraduate Use and Perceive Generative Tools

Abstract

As generative artificial intelligence (AI) tools such as ChatGPT, Copilot, and BearcatGPT become more common in higher education, undergraduate students are increasingly using them to support learning, often without formal guidance from instructors. This mixed-methods study explores how students use, perceive, and navigate AI in their coursework, focusing on both learning benefits and ethical challenges. With Institutional Review Board (IRB) approval, data were collected from 240 anonymous survey responses and 40 semi-structured interviews in General Chemistry, Biology, and other STEM courses. Results show that nearly three-fourths of students have used AI for study support, most commonly to review concepts, generate practice problems, and clarify feedback. ChatGPT and Grammarly were the most frequently used tools, followed by Copilot and BearcatGPT, an institutionally managed generative AI platform developed for educational use. While students valued AI for its accessibility and immediacy, nearly half expressed uncertainty about what constitutes ethical use, citing confusion about academic integrity policies. Qualitative analysis revealed themes of curiosity, self-directed experimentation, and a strong desire for clearer faculty guidance. Findings suggest that students are using AI as a supplemental learning aid rather than a shortcut, but the lack of consistent instruction can lead to inequitable outcomes. The study concludes with recommendations for embedding AI literacy and responsible-use frameworks into course design to promote effective and ethical engagement with emerging technologies.