Robot service failure: interaction effect of robot language style and customers’ sense of humor on service failure recovery

Purpose This study examines the interaction effect of service robot language style (rational vs humorous) and customers’ sense of humor on recovery from artificial intelligence (AI) service failure and explores the recovery effects’ mechanisms and boundary conditions. Design/methodology/approach Three experiments corresponding to three situations (airports, restaurants and hotels) and inter-group control experiments were performed. Participants (N = 627) were exposed to robot service failure scenarios with manipulated language styles and responsibility attributions. Hypotheses were tested using ANOVA, moderated mediation analysis and F-tests for the interaction effect. Findings Rational language enhanced satisfaction among low-humor customers by triggering perceived empathy, whereas humorous language was more effective among high-humor customers. Responsibility attribution moderated these effects: attributing failures to service robots (vs employees) strengthened the interaction between language style and sense of humor. Research limitations/implications This study has three limitations: (1) It focuses solely on robots’ linguistic styles (e.g. humor vs rationality), neglecting multimodal behavioral expressions (e.g. facial cues, gestures) that may enhance recovery effectiveness. (2) This study only explored the recovery effects for a single robot humor language type and the general severity of robot service failures. In the future, it is suggested to explore the recovery of robots by increasing the types of humor and different degrees of service failures. (3) Human–robot–customer triadic collaboration is unexplored. Further investigations should address how employee–robot coordination impacts satisfaction and whether human intervention remains critical in AI-driven service contexts. Practical implications Service companies should use AI technology to evaluate customers’ sense of humor and adopt corresponding language style recovery strategies. Training employees to clarify responsibility boundaries between humans and robots can enhance recovery effectiveness. Originality/value This study highlights the importance of matching service robot language style with customer humor in the recovery from robot service failures and verifies the moderating effect of robot failure attribution, which has rarely been studied.