Perceived generative AI teaching intelligence and pre-service teachers' professional identity: double-edged pathways through professional agency and AI replacement threat

Abstract

Generative artificial intelligence (GenAI) is increasingly used in teacher education for lesson planning, feedback, and assessment support. However, how perceived GenAI teaching intelligence relates to pre-service teachers' professional identity remains unclear. This study examined a double-edged model linking perceived intelligence to professional identity through professional agency and AI replacement threat, with teacher–AI complementarity beliefs moderating these pathways. In a task-based between-subjects experiment, 598 pre-service teachers were randomly assigned to high- or low-intelligence GenAI teaching-support conditions. Complementarity beliefs and baseline professional identity were measured before exposure; perceived intelligence, professional agency, replacement threat, and post-task identity were assessed afterward. Confirmatory factor analysis and conditional process analyses with 5,000 bootstrap resamples tested the model. Participants in the high-intelligence condition reported greater agency and replacement threat, but no statistically significant between-condition difference in post-task identity was detected. Perceived intelligence was positively associated with both agency and replacement threat, which were, respectively, positively and negatively associated with professional identity. Stronger complementarity beliefs were associated with a stronger agency pathway and a weaker replacement-threat pathway. These findings are consistent with two opposing identity-relevant processes following exposure to capable GenAI. Teacher education should support professional judgment and clarify human–AI role boundaries when integrating GenAI into teaching practice.