From institutional support to pedagogical innovation: AI literacy as a mediator and boundary conditions in pre-service teacher education

Educational institutions worldwide are investing heavily in AI training, yet the extent to which this support is associated with pedagogical innovation among pre-service teachers remains unclear. Drawing on Organizational Support Theory, Human Capital Theory, Social Cognitive Theory, and Diffusion of Innovations theory, this study examines whether AI literacy statistically mediates the association between institutional support and pedagogical innovation, and whether school culture and technology self-efficacy condition different stages of this association. A cross-sectional survey was conducted with 350 pre-service teachers from five universities across eastern, central, and western China (all urban settings). Data were analyzed using covariance-based structural equation modeling with robust maximum likelihood estimation and conditional-process analysis with 5000 bootstrap resamples. AI literacy accounted for 53.4% of the total association between institutional support and pedagogical innovation. The association between institutional support and AI literacy was stronger in more supportive school cultures (simple slope: 0.65 vs 0.41), and the association between AI literacy and pedagogical innovation was stronger among pre-service teachers with higher technology self-efficacy (simple slope: 0.67 vs 0.37). The study contributes a conditional-process account that connects institutional support, capability development, and contextual and individual conditions. The practical implications are presented as evidence-informed possibilities rather than causal prescriptions.