Psychosocial Determinants of Artificial Intelligence Adoption among Public University Students: A PLS-SEM Approach

Abstract

This study aimed to analyze the influence of psychosocial factors on the responsible adoption/appropriation of generative AI chatbots among students from public universities in northern Peru using an extended UTAUT2 framework. A quantitative, non- experimental, cross-sectional design was applied to a sample of 430 students from five departments (Piura, Tumbes, Lambayeque, La Libertad, and Cajamarca), gathered through quota and convenience procedures. Constructs were assessed with adapted Likert-type reflective indicators and tested through a structural equation model, showing adequate psychometric properties (loadings 0.74–0.91; AVE 0.62–0.80; α 0.84–0.92) and acceptable model fit (SRMR = 0.073; NFI = 0.915; χ2/df = 2.18). All hypotheses were supported: AI learning self-efficacy emerged as the strongest predictor (β = 0.34; p < 0.001), followed by social influence (β = 0.28; p < 0.001), while performance expectancy, AI readiness/anxiety, perceived enjoyment, and ethical awareness had significant but smaller effects. The findings indicated that appropriation was driven more by perceived agency and social legitimation than by instrumental usefulness alone. Recommendations emphasized prioritizing training to strengthen self-efficacy and critical AI literacy, complemented by institutional and teaching guidelines to foster ethical use and reduce anxiety.