关键词: AI工作身份威胁, AI特征, 工作退却, 工作重塑, 员工创新行为

Abstract: With the rapid advancement and widespread application of artificial intelligence (AI), AI technologies are increasingly encroaching upon domains once considered uniquely human, and employees increasingly experience threats to their professional identity as a result. Although AI work identity threat has emerged as a salient organizational phenomenon, its connotation, antecedents, and behavioral consequences remain theoretically underdeveloped: existing research largely conflates AI-induced identity threat with other threat types (e.g., stigma-related threat), overlooking AI's unique characteristics as a threat source. Grounded in identity threat process theory, this research develops an integrative framework—organized around “novel dimensions, triggering mechanisms, and consequences”—through three interrelated studies.
Study 1 attempts to explore what AI work identity threat actually entails. Using the symbolic-instrumental framework as a heuristic lens, and jointly considering the uniqueness of the work-identity domain and of the AI threat source, this study attempts to identify the possible new dimensions of AI work identity threat. In the AI context, threat may arise not only at the symbolic level of “who am I / how valuable am I,” but also at the instrumental level concerning whether an individual can retain an irreplaceable position as labor/human capital. Accordingly, this study divides AI work identity threat into two interrelated yet distinguishable categories: symbolic threat, referring to challenges to employees' self-concept, such as perceived threats to identity value, identity meaning, and identity enactment; and instrumental threat, referring to challenges to one's capability/function as labor, such as perceived threats to identity uniqueness and identity continuity. This reconceptualization broadens the understanding of identity threat and extends the explanatory scope of identity threat theory.
Study 2 focuses on two core AI characteristics—anthropomorphism and explainability—to examine the triggering mechanisms of AI work identity threat, proposing two parallel pathways. On the self-referenced pathway, AI anthropomorphism makes AI appear more human-like and capable, lowering employees' perceived employability (internal and external), which heightens AI work identity threat. On the leader-referenced pathway, AI explainability increases the perceived credibility of AI outputs and, correspondingly, employees' perception of their leader's trust in AI; this perceived leader AI trust likewise heightens AI work identity threat, as employees infer a greater risk of being judged replaceable. AI application pattern (automation vs. augmentation) is proposed as a boundary condition: automation, which takes over tasks entirely, strengthens both pathways, whereas augmentation, which mainly extends human capabilities, weakens them. This dual-pathway, boundary-conditioned model offers a more fine-grained, characteristic-level explanation of AI work identity threat than prior research's coarse focus on the degree and frequency of technological change.
Study 3 uncovers a “double-edged sword” effect of AI work identity threat on employee innovative behavior—an outcome central to organizational competitiveness yet previously unexamined in this context. Rather than assuming a uniformly negative consequence, this study proposes that the behavioral translation of AI work identity threat hinges on employees' trait regulatory focus. Employees high in trait prevention focus are theorized to respond via work withdrawal—a defensive, identity-protective strategy—which suppresses innovative behavior. Employees high in trait promotion focus are theorized to respond via job crafting—an identity-expansive strategy encompassing cognitive, relational, and task reshaping—which promotes innovative behavior. By specifying this individual-difference boundary condition and the underlying mediating mechanisms, the dual-pathway model reconciles the largely negative findings of quantitative identity threat research with the more optimistic, qualitative literature on adaptive identity responses.
In summary, this research contributes to identity threat theory in three ways. First, it theorizes new dimensions of instrumental threat within AI work identity threat, extending the content coverage of existing constructs. Second, it identifies AI anthropomorphism and AI explainability as characteristic-specific antecedents and specifies the mediating mechanisms (employability, perceived leader AI trust) and moderating mechanism (AI application pattern) through which they operate, offering a more fine-grained causal explanation than prior generic-technology perspectives. Third, it uncovers the mechanism by which AI work identity threat produces a double-edged sword effect on employee innovative behavior through competing mediating pathways (work withdrawal vs. job crafting) moderated by trait regulatory focus, extending identity threat theory into a previously unexplored, performance-relevant outcome domain. Practically, the framework offers organizations concrete points of intervention: calibrating AI anthropomorphism and explainability in interface and communication design, deliberately configuring automation versus augmentation modes according to desired identity outcomes, and designing regulatory-focus-informed interventions that guide identity-threatened employees toward job crafting rather than withdrawal. In sum, this research deepens theoretical understanding of how AI reshapes the meaning and boundaries of individual work identity, while offering a systematic, mechanism-based approach to managing AI work identity threat in the contemporary workplace.

Key words: AI work identity threat, AI characteristics, work withdrawal, job crafting, employee innovative behavior