The Anxiety Paradox of AI Training in High-Stakes Operations and Supply Chain Management: An Integrated Perspective from Control-Value Theory and Organizational Support Theory

In the time-critical, tightly coupled workflows of operations management (OM) and supply chain management (SCM), pervasive employee anxiety regarding job displacement, loss of control, and professional obsolescence constitutes a critical barrier to successful AI adoption. Prior research often conflates this anxiety with generic technostress or treats it merely as an antecedent to performance outcomes, failing to conceptualize it as a distinct, multi-dimensional psychological phenomenon. To address this gap, we conducted a naturalistic field experiment in a Chinese smart-automobile firm, implementing a three-week structured AI training program that integrated technical upskilling with psychological support components. Grounded in an integrative framework of Control-Value Theory (CVT) and Organizational Support Theory (OST), we examined the direct effect of this intervention on AI-induced anxiety and the moderating roles of Technology Self-Efficacy (TSE, as a proxy for perceived control) and Perceived Organizational Support (POS, as a regulator of value appraisal). Our findings reveal a counterintuitive "anxiety paradox": despite its benevolent intent, the training’s short-term net effect was a significant increase in average employee anxiety, effectively functioning as a new "job demand." Further analysis shows that while POS effectively buffered anxiety during the initial training phase, its effect attenuated during the application phase. Conversely, the moderating effect of TSE was statistically non-significant, likely due to the acute cognitive load and algorithmic opacity inherent in high-pressure OM/SCM contexts, which masked TSE-based differences. These findings establish AI-induced anxiety as a core element in understanding human-AI collaboration and demonstrate how standardized AI training can inadvertently exacerbate anxiety by creating a "high-value, low-control" imbalance. Practically, our study underscores the necessity for organizations to adopt staged,精细化 (refined) intervention strategies that prioritize restoring employees’ perceived control, rather than implementing one-size-fits-all training rollouts.