How users solve problems with presence and affordance in interactions with conversational agents: A wizard of Oz study

Abstract

In this paper, we present a case study that investigates how users cope with the interactional problems with a multimodal agent in a real-life situation. We conducted a wizard-of-oz study on humanavatar interaction at a science fair and analysed the screen-recorded data of 127 user interactions with the help of conversation analysis. The major problems encountered by users were those of the presence and affordance of the conversational agent. Our results show that wizard-of-oz studies in the wild shed light on these more general fundamental conditions on the interaction, most prominently in the opening section.