UNDERSTANDING HOW CHATBOTS WORK: AN EXPLORATORY STUDY OF MENTAL MODELS IN CUSTOMER SERVICE CHATBOTS

Abstract

Chatbots are changing customer service interactions, enabling higher reliance on self-serving behavior. There seems to be an emphasis on designing these chatbots to be as humanlike as possible. One drawback is that such humanlike design might lead users to apply the same mental models when interacting with chatbots as they do when interacting with human customer service personnel. Arguably, this may cause issues in the chatbot interaction because the technology may not be capable of handling interactions at the same level of sophistication as human personnel. Thus, the mental models that users apply are important for successful system interactions, but little research have been dedicated towards understanding these mental models. To close this research gap, we asked 16 users to interact with two customer service chatbots to explore the mental models guiding their interactions. Based on qualitative interviews and screen-captured videos of the participants’ dialogues, an exploratory analysis indicated that the participants drew on two types of mental models to understand, predict, and interact with chatbots: a human-oriented model and a technology-oriented model. We discuss our findings and their theoretical and practical implications.