AI Anxiety, Learning Outcomes, and Cognitive Engagement in Adolescents: A Moderated Mediation Analysis

A growing number of adolescents between the ages of 14 and 16 are reporting symptoms of “AI anxiety” as a result of the rapid proliferation of deepfakes and generative AI technology. This study investigated the associations between AI anxiety and three important developmental outcomes: cognitive engagement, learning performance, and self‐confidence. Drawing upon social cognitive theory and the stimulus–organism–response framework, this study examines a moderated mediation model in which digital literacy moderates the relationship between AI anxiety and cognitive engagement, and self‐confidence mediates the association between AI anxiety and learning performance. The results indicated that AI anxiety significantly and negatively predicted learning performance ( β = −0.186, p < 0.01), self‐confidence ( β = −0.360, p < 0.001), and cognitive engagement ( β = −0.336, p < 0.01). Self‐confidence and cognitive engagement were also positively associated with learning performance ( β = 0.260, p < 0.001; β = 0.151, p < 0.05). Digital literacy significantly moderated the AI anxiety–cognitive engagement pathway ( β = 0.219, p < 0.001), such that higher digital literacy attenuated this negative association. Self‐confidence partially mediated the AI anxiety–learning performance link (indirect effect = −0.077, p < 0.01). The conditional indirect effect of AI anxiety on learning performance through cognitive engagement suggests that digital literacy reduces this negative impact. These findings have implications for educators, policymakers, and mental health practitioners regarding the importance of promoting digital literacy and self‐efficacy as protective factors against the adverse associations of AI anxiety.