Identity Reshaping in Human-AI Collaboration: Why Do High Performers Actively Mentor Colleagues in Using AI?

While generative AI enhances efficiency, it simultaneously undermines the professional authority of high performers, thereby posing an expert identity threat to these high performers. Contrary to traditional perspectives that view identity threats as triggers of passive defensiveness, this paper constructs a moderated mediation model based on the dual-dimensional structure of identity (centrality-solidarity). A two-stage questionnaire analysis of 323 employees reveals that: (1) high performers perceive stronger expert identity threats; (2) these threats conversely prompt them to guide colleagues in using AI; (3) task-AI fit positively moderates this mediation effect. This study breaks through the traditional defense paradigm, confirming that high performers achieve identity compensation and power regeneration through helping others and empowering them, providing insights for organizations to promote the transformation of excellent employees into "human-AI collaboration mentor ".