Abstract

Artificial intelligence (AI) is rapidly transforming professional work by automating, augmenting, and sometimes replacing tasks historically associated with human expertise. While information systems research has extensively examined cognitive drivers of AI adoption such as perceived usefulness, trust, and performance expectancy, less attention has been devoted to the emotional and identity-based mechanisms that shape employee resistance to AI. Emerging work suggests that AI can threaten professional identity by undermining perceptions of competence, autonomy, and occupational. Building on this foundation, we develop a dual-path framework explaining how professional identity threat produces self-conscious emotions that lead to both AI avoidance and covert AI use behaviors. We distinguish between individual identity threat and group identity threat. Individual identity threat occurs when employees perceive AI as undermining their personal expertise, competence, or professional relevance. This threat is expected to evoke shame, a self-focused emotion characterized by global negative self-evaluation and feelings of inadequacy. Shame motivates defensive self-protection, leading to two forms of AI resistance: (1) AI avoidance, where employees disengage from or reject AI systems, and (2) AI use hiding, where employees secretly use AI tools while concealing this behavior to protect their professional image and avoid social judgment. Group identity threat occurs when employees perceive AI as undermining the legitimacy, status, or future viability of their professional community or occupation. Grounded in social identity theory, we argue this threat evokes guilt, an other-focused emotion linked to perceived responsibility for failing to uphold collective norms and protect group identity. Rather than producing uniform resistance, guilt initiates a repair-oriented motivational state aimed at restoring professional legitimacy. However, we propose a key boundary condition: perceived repair feasibility. When employees believe that meaningful repair of professional identity is possible (e.g., through reskilling, ethical AI governance, or role redesign), guilt leads to constructive engagement with AI. In contrast, when repair is perceived as infeasible due to irreversible technological disruption or institutional constraints, guilt shifts toward defensive coping, resulting in AI aversion and, in some cases, concealment of AI engagement to avoid perceived norm violations. Together, these mechanisms produce a nuanced model of AI resistance that integrates emotion, identity threat, and behavioral outcomes. The framework explains not only why employees resist AI, but also why resistance can take both overt (avoidance) and covert (use hiding) forms. It further highlights that AI aversion is not a uniform response but depends on whether identity threats are experienced at the individual or collective level and whether employees perceive viable pathways for restoring professional identity. This research contributes to the growing human-AI interaction literature by extending AI adoption research beyond cognitive explanations toward emotional and identity-based mechanisms. It differentiates individual and group identity threats and introduces shame and guilt as distinct emotional pathways shaping AI resistance. Practically, it suggests organizations must address professional identity concerns through participatory AI design, transparent communication, and reskilling strategies that enhance perceived repair feasibility.

Download

Share

COinS