ChatGPT as a self-learning assistant: Perceptions among university students

Abstract

The rapid and multifaceted integration of generative AI tools into educational settings requires an understanding of students' perceptions of these tools. This study investigated university students' perceptions of a customized version of ChatGPT as a self-learning assistant from the lens of self-regulated learning. The study followed a descriptive research design in which a customized version of ChatGPT served as a self-learning assistant for students in a programming class for one semester. Then, fifty-five students from two sections of the class completed a questionnaire, and one instructor of one section of the course was interviewed. The results showed that students expressed positive perceptions of the customized version of ChatGPT as a self-learning assistant. Motivation and engagement emerged as the strongest areas, while self-evaluation and regulation were relatively weaker. The instructor observed improved participation, confidence, and independence among students. However, the instructor pointed out that students had variations in their skills in using the tool, and some had faced difficulties in crafting effective prompts. Based on the findings, a customized version of ChatGPT might represent a valuable tool that can be officially integrated into introductory programming courses to act as effective, 24/7 self-learning support. The integration of such a tool and its technical design needs to be aligned with pedagogy to enhance all aspects of self-regulated learning. The integration of a customized version of ChatGPT to support students' learning should be accompanied by AI/prompt training.