The Effect of Artificial Intelligence-Supported Education on Students’ Academic Resilience: The Mediating Role of Psychological Capital

Abstract

This study examined the effect of artificial intelligence (AI)-supported education on students’ academic resilience, with particular emphasis on the mediating role of psychological capital. A quantitative, cross-sectional survey design was employed, and data were collected from 323 undergraduate and postgraduate students enrolled in public and private universities across Pakistan. Participants were recruited using convenience sampling, and data were collected through a structured questionnaire measuring AI-supported education, psychological capital, and academic resilience. The data were analyzed using IBM SPSS Statistics, employing descriptive statistics, reliability analysis, Pearson correlation, linear and multiple regression, and PROCESS Macro Model 4 with 5,000 bootstrap samples to assess the indirect effect. The illustrative results indicated a positive association between AI-supported education and academic resilience (r = .58, p < .001), with AI-supported education explaining 34% of the variance in academic resilience (R² = .34). Furthermore, AI-supported education and psychological capital jointly explained 48% of the variance in academic resilience (R² = .48). The mediation analysis suggested a positive indirect effect of AI-supported education on academic resilience through psychological capital (B = .24, 95% bootstrap CI [.17, .33]), indicating a potentially significant mediating relationship. These findings suggest that AI-supported learning practices and students’ psychological resources may contribute jointly to academic resilience. The study highlights the importance of integrating AI-supported educational practices with strategies that foster students’ psychological capital to help them cope with academic challenges. However, the reported statistical values are provisional illustrative estimates and must be verified against the actual dataset before the abstract is submitted as an empirical research report. The cross-sectional design and convenience sampling also limit causal interpretation and generalizability.