Abstract
Repair describes the process through which participants in conversation address problems in speaking, understanding, and hearing. In interactions with AI-driven chatbots, user repair addresses chatbots' lack of understanding or misunderstanding of the user's intent.
This paper represents a user-centred description of user repair strategies in interactions with a task-oriented chatbot. It is based on the analysis of simulated user interactions with a chatbot facilitating health appointment bookings. The analysis shows that the repair strategies which users draw on most frequently (e.g., rephrasing) are not necessarily the ones which prompt the bot to correctly recognise intent and provide relevant responses, whereas the less frequently used self-repair strategies (e.g. restating the intent) are more successful in achieving intent recognition. This suggest that the rules of interaction with conversational AI need to be made explicit to users as they lack familiarity with the context, limitations and patterning of interactions facilitated through AI.
•When bots misunderstand or do not understand user intent, users perform repair.•Repair strategies used include e.g., rephrasing, repetition, accommodation.•Users transfer repair strategies from human interaction into bot interactions.•Users deploy the more successful repair strategies less frequently and vice versa.•The rules of communication with conversational AI need to be made more explicit.