The Impact of Generative AI Use on University Students' Self-Directed Learning

Abstract

Conversational generative AI systems such as ChatGPT are becoming increasingly popular, and their use in higher education as a learning tool is spreading. At the same time, there are concerns that the use of generative AI may undermine agency and autonomy when learning. Against this backdrop, this study examined changes in self-directed learning (SDL) resulting from in-class use of generative AI by university students. Generative AI-based learning activities were implemented as part of a course offered by a management-related department at a private university. A survey was conducted before and after the trial, and statistical analyses were performed using matched responses from before and after the trial by 116 students.

Results indicated that SDL scores increased significantly from before to after the trial (t(115)=5.37, p<.001), with a moderate effect size (dz=0.50). The mean difference was 0.33 points. Although this is not a major change on a five-point scale, the findings suggest that self-evaluations of SDL by students can improve after brief in-class practice. Furthermore, analysis of the generative AI usage method (eight categories) showed that mean changes in SDL were positive across all categories, whereas differences among categories were not statistically significant (F(7,224)=0.508, p=.828). These results suggest that, rather than the selection of particular functions, the experience of integrating generative AI into coursework as learning support—and of experiencing progress in learning—may be associated with heightened awareness of SDL. However, because mediating processes such as self-efficacy and strategy use were not directly measured, this interpretation should be regarded as tentative. Overall, this study provides foundational evidence for examining the effects that generative AI use has on learning in higher education.