In recent years, the rapid development of artificial intelligence (AI) has facilitated the widespread adoption of service robots within the service industry. However, owing to existing technological constraints, service robots may experience service failures. In the face of such failures, service robots need to adopt certain remedial measures to improve customers' negative attitudes. While existing research has explored the effect of non-verbal interaction features (e.g., anthropomorphism and cute design) on forgiveness, the role of verbal communication styles, particularly humor, remains underexplored. In fact, the positive effects of humor have been confirmed in other contexts, while its mechanisms and remedial effects in service robot failure scenarios still require further investigation. Furthermore, there remains significant potential for discussions on internal psychological mechanisms and boundary effects of personal traits. To clarify the above issues, this study synthesizes the Benign Violation Theory, Social Response Theory, Cognitive-Affective System Theory of Personality, and Regulatory Focus Theory to construct theoretical model and employs a scenario-based experiment method to test hypotheses. The results of three experiments reveal that: In the case of low failure severity, compared with neutral communication style, humorous communication style has a more significant positive effect on customer forgiveness; perceived intelligence and perceived warmth play parallel mediating roles in this process; regulatory focus significantly moderates the mediating pathways. Specifically, promotion-focused customers perceive higher levels of intelligence and warmth in humorous communication scenarios compared to neutral ones, while no significant differences are observed among prevention-focused customers. This study innovatively explores the remedial effect of humorous communication in the failure situations of service robots. Theoretically, it enriches the research results on service robot recovery strategies and extends the application domains of relevant theories. Practically, the findings offer effective insights for enterprises to focus more attention on AI-driven service failure management, optimize service robot deployment strategies, and implement differentiated remedial approaches.

Artificial Intelligence, Service Robot, Service Failure, Communication Style, Humor, Consumer Forgiveness