How AI voice affects customer tolerance: the mediating role of warmth perception and the moderating effect of intelligence level

Purpose Voice artificial intelligence (AI) agents are increasingly replacing human agents in customer service roles. However, service failures caused by technological malfunctions are inevitable. To address this, it is essential to investigate effective voice design strategies that can enhance customer tolerance toward such failures. Design/methodology/approach Three scenario-based experiments were conducted to examine the effects of AI agents’ voice types (male-voiced vs. female-voiced) on customer tolerance of service failures, with a focus on the mediating role of warmth and competence perceptions and the moderating effect of intelligence level. Findings The results revealed that customers interacting with female-voiced AI agents exhibited greater tolerance toward service failures compared to those interacting with male-voiced agents. This effect was mediated by customers’ warmth perceptions rather than competence perceptions. Furthermore, the intelligence level moderated the effect of voice type on warmth perceptions. As the intelligence level increases, the effect of voice type on warmth perceptions diminishes. Originality/value This study adds to the limited body of research on AI-gendered voice type in service failure contexts. It highlights warmth perceptions as the key mediating mechanism and identifies intelligence level as a critical boundary condition. The findings contribute to the literature on AI agent voice design and provide valuable practical insights for improving customer experience.