The Efficacy of AI-Assisted Learning in Mitigating Anxiety Among University Students: A Multimodal Pedagogical Perspective

Abstract

Grounded in a multimodal pedagogical framework, this study investigates a cohort of 100 university students participating in multimodal foreign language classrooms to provide a granular analysis of the primary etiologies underlying foreign language anxiety, while further elucidating the functional role and underlying mechanisms of AI-assisted learning in alleviating such psychological distress through a mixed-methods research design incorporating both questionnaires and interviews. The empirical findings reveal that higher levels of perceived competence, correlate with intensified communication apprehension, whereas elevated levels of perceived autonomy are associated with heightened concerns regarding their inherent capabilities. Notably, the high AI-usage group exhibited significantly higher levels of test anxiety and fear of negative evaluation than the low-usage group, yet simultaneously demonstrated greater perceived autonomy and a higher degree of endorsement regarding AI efficacy, with students of lower perceived competence manifesting a more pronounced affirmation of the assistance provided by AI. Collectively, the results suggest that anxiety within multimodal environments originates from a “performance-pressure paradox,” wherein AI functions as a dual-faceted instrument capable of both exacerbating and mitigating psychological strain; consequently, pedagogical practices should strive to construct an empowering instructional model that leverages AI to manage performance-related pressures while providing individualized affective support.