Latent Profiles of University Students’ Artificial Intelligence Learning Characteristics: Epistemic Capacities, Psychological Attitudes, Facilitating Conditions, and Behavioral Engagement

Understanding university students’ differences in their AI learning characteristics is increasingly important for supporting their engagement with artificial intelligence (AI) in higher education. This study examined AI learning characteristics across four domains—epistemic capacity, psychological attitudes, facilitating conditions, and behavioral engagement—among 733 university students. A revised measurement model was first evaluated using confirmatory factor analysis and measurement invariance testing across gender, educational level, and disciplinary background. Then, latent profile analysis (LPA) was used to examine person-centered configurations across 11 indicators, followed by R3STEP analyses to examine associations between demographic characteristics and profile membership while accounting for classification uncertainty. Four profiles were identified by differences in their overall levels across the indicators, and named as broadly low-scoring, moderately high-scoring, moderately low-scoring, and broadly high-scoring. After Holm adjustment for multiple pairwise comparisons, gender showed the most consistent association with profile membership, remaining significant across four profile contrasts. Disciplinary background was associated only with the broadly high-scoring versus moderately low-scoring contrast, whereas educational level was not significantly associated with any profile contrast. These findings demonstrate the value of a person-centered approach for examining how multiple AI learning characteristics co-occur within university students while also indicating that the identified profiles should be interpreted as empirical configurations under the present operationalization rather than as fixed learner types. The findings may inform differentiated AI learning opportunities in higher education, although such applications require further investigation.