A Study on Consumer Responses to AI Voice Assistant Service Failures: Focusing on the Effects of Emotional Interaction, Problem-Solving Ability, and compensation proposal appropriateness

This study investigates how emotional interaction, problem-solving ability, and compensation proposal appropriateness influence customer forgiveness and repurchase intentions in the context of service failures involving AI voice assistants. A unique methodological contribution of this research lies in identifying key service failure attributes using qualitative interviews combined with Term Frequency-Inverse Document Frequency (TF-IDF) analysis, ensuring empirical grounding in actual user experiences. Based on this founda tion, a survey was conducted with 350 users who had prior experience with AI voice assistants such as Apple Siri, Amazon Alexa, and Naver Clova. Structural equation modeling revealed that all three service recovery attributes had significant positive effects on customer forgiveness, which in turn strongly predicted repurchase intentions. Moreover, frequency of use and brand loyalty were found to moderate the relationships between service recovery attributes and forgiveness. These findings contribute to the literature by extending the service recovery framework to AI-based interactions and offer practical insights for enhancing AI service strategies through emotional and personalized engagement.