Making the Case for Audience Design in Conversational AI: Users’ Pragmatic Strategies and Rapport Expectations in Interaction with a Task-Oriented Chatbot

Abstract

Abstract With chatbots becoming more and more prevalent in commercial and service contexts, they need to be designed to provide equitable access to services for all user groups. This paper argues that insights into users’ pragmatic strategies and rapport expectations can inform the audience design of chatbots and ensure that all users can equally benefit from the services they facilitate. The argument is underpinned by the analysis of simulated user interactions with a chatbot facilitating health appointment bookings, users’ introspective comments on their interactions, and users’ qualitative survey comments. The study shows that users’ pragmatic strategies show considerable variation. It also shows the negative impact of user experiences when the chatbot’s language and interaction patterns do not align with users’ rapport expectations. In closing, the paper uses these findings to define audience design for chatbots and discuss how audience design can be realized and supported by research.