The Promise and the Peril of Equipping Service Chatbots with Emotions and Choices

Abstract

The advance in AI-powered chatbot technologies and their rapid deployment in the service industry have attracted enormous interest from both researchers and practitioners in recent years. However, we still know little about the implications of equipping service chatbots with some important features such as emotional capabilities and choice provision. My dissertation not only investigates how customers respond to chatbots with emotional capabilities and choice-equipped chatbots during a service interaction, but also explores why and when these critical features would benefit or hurt customers. In my first essay, I examine the impact of positive emotion expressed by a chatbot on service evaluations. I show that chatbot-expressed positive emotion does not enhance service evaluations as in human-to-human interactions. Drawing on the emotional contagion and the expectation-disconfirmation theory, I further uncover the dual, affective and cognitive pathways underlying this non-effect and a boundary condition for the cognitive pathway. In my second essay, I explore the impact of empathy expressed by a service chatbot. Extending the social perception literature, I predict and reveal that chatbot-expressed empathy can either help or hurt service evaluations depending on the source of customers’ negative emotions. In the final essay, I draw on the fluency literature and investigate when and why the implementation of choices during a chatbot-driven service interaction enhances or impairs customers’ service experience. Through a series of experimental studies, this dissertation uncovers the promise and the peril of equipping chatbots with emotional and choice-provision capabilities in customer service. It also provides valuable insights for practitioners on the design and implementation of AI-powered chatbots in customer service and beyond.