University Honors

As Generative AI tools like ChatGPT become increasingly integrated into higher education, it raises questions about the factors influencing student adoption and the role of decision-making traits in this engagement. This study investigates whether individual decision-making traits—specifically risk preference, time preference, and loss aversion—predict students’ use of AI for academic purposes. Drawing on survey data from more than 200 undergraduate students at a large public university in California, the study employs correlation and regression analyses, along with Bayesian Truth Serum (BTS) scoring, to evaluate both behavioral predictors and response honesty. The results suggest that decision-making traits do not significantly predict AI usage. Instead, social and motivational factors, including peer influence, self-confidence with AI tools, and self-motivation to learn AI emerged as the most robust predictors of both actual and intended AI use. Instructor policies outlining AI use had little impact on day-to-day behavior but appeared to deter AI usage under deadline pressure. BTS scores indicate that students may under-report their AI usage but respond more truthfully when discussing risks like academic penalties or inaccuracy. These findings highlight that AI adoption in academic settings is shaped less by behavioral and decision-making traits, and more by dynamic social and contextual forces. The study aims to contribute to ongoing discussions on AI integration in higher education, by highlighting the key drivers of student engagement with AI tools, informing policy and instructional strategies in academia.