Can a Chatbot Be Your Friend? Using Topic Modeling to Examine Customer Reaction to Feeling AI
(A2024-119826)

AUTHORS

Salma André, University Jean Moulin Lyon 3, iaelyon School of Management, UR Magellan; Margherita Pagani, Skema Business school

ABSTRACT

Despite the promising role of feeling AI in services, customer reaction to this technology, when employed in a relational strategy, is still underexplored. This research addresses this gap by examining customer reviews about one of the most advanced relational feeling-AI chatbots within the context of mental health services, where emotional aspects are primordial. We employed Bertopic modeling to identify twelve prominent topics in customers’ discussions, followed by a thematic analysis to gain a deeper understanding of the extracted topics. We examine the impact of the extracted topics on customer satisfaction using multiple regression analysis on score ratings. Our findings reveal customer mind attributions to feeling-AI, through perceived agency and feeling abilities. We introduce two distinct human-AI relationship dynamics where AI can be either an extended-self or a competitor. These results carry significant managerial implications and academic contributions to Human-Robot Interaction research, service research, and consumer behavior research.