Generative AI-Assisted Preschool Instruction and Children's Creative Learning Engagement: The Mediating Role of Autonomous Motivation and the Buffering Role of Perceived Teacher-Child Interaction Quality.

Drawing on Self-Determination Theory (SDT) and the Classroom Assessment Scoring System (CLASS) framework, this study tested a moderated mediation model in which two dimensions of GenAI-assisted teaching predict children's creative learning engagement through autonomous motivation. Cross-sectional survey data collected from 486 preschool teachers across 52 kindergartens in three Chinese provinces were analysed with structural equation modelling and bias-corrected bootstrap procedures. AI-enhanced interactive pedagogy was positively associated with autonomous motivation, whereas AI-dependent passive instruction was negatively associated with it; only passive instruction exhibited a direct negative link with creative learning engagement. Autonomous motivation fully mediated the interactive pedagogy pathway and partially mediated the passive instruction pathway. Perceived teacher-child interaction quality significantly moderated the passive instruction to motivation path, attenuating the indirect association to non-significance at high levels of perceived interaction quality. Because all constructs were assessed via teacher self-report at a single time point, the observed associations reflect teachers' perceptions rather than objectively measured behaviours, and causal inference is not warranted. The study offers theoretical and practical insights for designing developmentally appropriate, teacher-scaffolded GenAI integration in preschool education.