Why University Students Choose Generative AI for Academic Tasks: An Exploratory Mixed-Methods Factor Analysis

Abstract

This exploratory sequential mixed-methods study examines how university students evaluate generative artificial intelligence when choosing tools for academic tasks. Qualitative interview coding informed a 24-item questionnaire, followed by exploratory factor analysis of 70 undergraduate and master’s students. Seventy-two responses were collected and 70 were included in the factor analysis; the reason for excluding two responses was not recorded. Sampling adequacy was modest (KMO = .560), Bartlett’s test was significant, and Cronbach’s alpha was .668 (.695 standardized). Seven components with eigenvalues above one explained 65.097% of observed variance: reliability, multifunctionality, exploratory capacity, suitability, prompt understanding, information clarity, and awareness. The findings show that student choice combines appraisal of the tool, fit with the academic task, and self-appraisal of the user’s ability to prompt, interpret, verify, and use generated material responsibly. The study therefore reframes adoption as evaluative AI literacy rather than simple convenience. Because the sample is small and diagnostic values are modest, the factor structure should be replicated with larger samples and confirmatory analysis.