Mechanisms underlying university students' perceived AI threat: the mediating roles of perceived controllability and perceived effort–reward imbalance

Introduction As artificial intelligence (AI) becomes increasingly integrated into educational assessment and employment practices, university students are becoming more concerned about their future career opportunities, the value of their skills, and employment competition. However, the mechanisms underlying perceived AI threat remain insufficiently understood among undergraduate and postgraduate students in the humanities and social sciences in China. This study develops an integrated model in which perceived controllability and perceived effort-reward imbalance serve as parallel mediators and examines how human–AI relationship experience, AI anxiety, AI self-efficacy, growth mindset, and role overload influence perceived AI threat. Methods This cross-sectional study analyzed 548 valid questionnaires. SPSS 26.0 and SmartPLS 3.0 were used to conduct descriptive analyses, assess the measurement model, estimate the structural model, and perform bootstrapping tests of the mediation effects. Results Perceived controllability significantly and negatively predicted perceived AI threat, whereas perceived effort–reward imbalance significantly and positively predicted perceived AI threat. Perceived controllability significantly mediated the relationships between each of the five antecedent variables and perceived AI threat. Perceived effort–reward imbalance significantly mediated the effects of AI self-efficacy, growth mindset, and role overload on perceived AI threat, whereas its mediating effects in the relationships involving AI anxiety and human–AI relationship experience were not statistically significant. Role overload produced the largest total effect on perceived AI threat. Growth mindset had no significant direct effect but produced significant indirect effects through both mediators. Discussion University students' threat judgments regarding AI are associated both with whether they believe they can cope with technological change and with whether they believe their current investments can generate commensurate developmental returns. These findings clarify the dual appraisal mechanisms underlying perceived AI threat and provide empirical evidence supporting the implementation of AI literacy education, psychological adaptation support, and career development guidance in universities.