When Does an AI Outperform a Human in Service Recovery? A Comparison Between Two Compensation Strategies
Tang Yao, Hanyue Xue, Weiqun Xiao, Qi Qiu · International Journal of Human-Computer Interaction · 2025
Artificial intelligence (AI) has been widely adopted across various service industries. However, consumers often feel that current AI lacks the capability to understand emotions, which can impede its effectiveness in service recovery following service failures. To explore this issue, we examine customer reactions to two compensation strategies—symbolic and utilitarian—provided by different recovery agents (employee vs. AI). A survey and three scenario-based experiments revealed that customers prefer symbolic compensation when it is provided by an employee rather than an AI. Interestingly, this preference gets reversed when utilitarian compensation is provided. Moreover, our research verified that perceived empathy and perceived fairness serve as mediators for this effect, while failure attribution acts as a moderator. These findings enhance our understanding of service recovery and offer important implications for the integration of AI in service industries.