Psychological Profiles Associated with Trust in Artificial Intelligence Among University Students: A Machine Learning Clustering Approach

This study investigates psychological profiles associated with trust in artificial intelligence (AI) among university students using an interpretable machine learning (ML) approach. Although AI tools are becoming increasingly embedded in higher education, limited research has examined how students’ psychological resilience, stress levels, and trust in AI interact to shape their perceptions of AI systems. This study analyzes cross-sectional survey data from 107 students at a Historically Black College and University (HBCU) in the United States. K-means clustering was applied to identify student profiles based on psychological resilience, perceived stress, and AI trust. Clustering solutions were evaluated using multiple internal validation metrics and stability analysis, followed by ANOVA and chi-square analyses to characterize psychological and demographic differences across the identified profiles. The analysis identified three student profiles: (i) moderately stressed AI-positive students, (ii) high-resilience low-stress AI-adopters, and (iii) psychologically resilient AI skeptics. Significant differences were observed across all psychological variables (p < 0.001), with stress and AI trust demonstrating the greatest differentiation across profiles. Gender was significantly associated with profile membership (χ2 = 10.34, p = 0.006), whereas age group, academic level, employment status, and STEM background were not significantly associated with profile membership. Overall, the findings highlight heterogeneity in students’ psychological characteristics and AI-related perceptions, suggesting that psychological factors may provide additional insight into variations in AI trust beyond demographic characteristics.