AI anxiety and teacher innovative intentions: a moderated serial mediation model of burnout and creative self-efficacy

Introduction: The proliferation of generative artificial intelligence (AI) in education presents a double-edged sword: while promising transformative pedagogical potential, it simultaneously triggers significant "AI anxiety" among educators. Despite its relevance, the psychological mechanisms through which this specific anxiety impedes teachers' willingness to innovate remain underexplored. Methods: Based on Conservation of Resources (COR) theory and Social Cognitive Theory (SCT), this study constructs a moderated serial mediation model. Data were collected from 528 K-12 teachers via a stratified random sampling method. Structural equation modeling (SEM) and bootstrapping techniques were employed to test the hypotheses. Results: Findings indicate that (1) AI anxiety positively predicts job burnout; (2) Job burnout and creative self-efficacy serially mediate the relationship between AI anxiety and innovative intentions (AI Anxiety → Burnout → Lower Creative Self-Efficacy → Reduced Innovative Intentions); and (3) Perceived Organizational Support (POS) acts as a critical moderator, buffering the deleterious effect of AI anxiety on burnout and attenuating the indirect negative effect on innovative intentions. Discussion: Moving beyond mere variable relationships, this study reveals a tangible process-based mechanism of teacher resistance to AI: the "depletion → cognition → behavior" pathway. It highlights that anxiety suppresses innovation not merely through emotional exhaustion, but by systematically translating resource depletion into a cognitive deficit (eroded self-efficacy), which ultimately drives behavioral withdrawal.