Graduate Students’ Self-Efficacy in Using ChatGPT for Academic Tasks

Artificial intelligence (AI) continues to transform higher education, particularly through generative tools such as ChatGPT, which support academic writing, research development, and data analysis. This study examined graduate students’ self-efficacy in using ChatGPT for academic tasks at a private higher education institution. A quantitative descriptive–comparative research design was employed, and data were collected from 54 graduate students using a structured questionnaire measuring five domains: academic writing, research skills, data analysis and interpretation, critical thinking and responsible use, and overall academic confidence. The results revealed a high level of self-efficacy across all domains, with an overall mean of 3.13. Among the domains, academic writing self-efficacy obtained the highest mean score, indicating strong confidence in utilizing ChatGPT for writing-related tasks in academic settings. An independent samples t-test showed no significant difference in self-efficacy when grouped by sex. However, a one-way analysis of variance (ANOVA) revealed a significant difference based on the frequency of ChatGPT use, with students who reported always using the tool demonstrating higher self-efficacy than those who rarely used it. Effect size analysis indicated a small practical effect for sex and a moderate effect for frequency of use. These findings suggest that experiential engagement with AI tools plays a more substantial role in shaping academic confidence than demographic characteristics. This study underscores the importance of structured exposure, institutional guidelines, and ethical training to promote responsible and effective AI integration in graduate education.