Impacts of GenAI Technostress on University Students’ Learning Performance: The Mediating Role of Self-Efficacy

Generative artificial intelligence (GenAI) is rapidly entering higher education and reshaping how students complete learning tasks. However, little is known about how GenAI-related technostress affects students’ learning performance. Drawing on the technostressor framework and social cognitive theory, this study examines the relationships between five GenAI technostressors (overload, complexity, insecurity, uncertainty, and privacy invasion), GenAI self-efficacy, and learning performance. Survey data were collected from university students in mainland China (N = 649) and analyzed using PLS-SEM, including tests of specific indirect effects. Results show that GenAI technostressors exhibit differentiated effects: complexity and insecurity are associated with lower learning performance, whereas overload and uncertainty show significant effects in the opposite direction of classic technostress predictions; privacy invasion is not significant. GenAI self-efficacy positively predicts learning performance and mediates the effects of complexity, insecurity, and uncertainty (with evidence of both complementary and competitive mediation), but not overload or privacy invasion. The study extends technostress research to GenAI-enabled learning and highlights self-efficacy as a key lever for educational interventions and GenAI tool design.