Paper Info
Title
Action-a-Bot: Exploring Human-Chatbot Conversations for Actionable Instruction Giving and Following
Abstract
ABSTRACTConversation serves as one critical mechanism for knowledge-sharing and instruction-giving in collaborative work. Conversation allows people to take turns to make contributions, plan joint actions, align shared understanding of work status and resolve action failure. However, when such collaboration involves non-human AI actors like chatbots, there is a lack of understanding of how human participants may respond to the chatbot’s prompts and guidance, and whether the interaction can similarly improve the actionability of instructions given to people. In this study, we prototyped a chatbot system, ActionaBot for providing task instructions to novice workers, and conducted an initial study to explore its effects on procedural instruction giving and following. Our results indicate that, novices although might perceive instructions to be inactionable due to prior experience and how instructions were authored, they were able to follow conversational guidance and willing to adapt to the chatbot through turn-taking to calibrate working states back-and-forth. Besides, users’ awareness of the work status increased with the conversational prompts from the chatbot.
Year
DOI
Venue
2022
10.1145/3500868.3559476
Computer Supported Cooperative Work
DocType
Citations 
PageRank 
Conference
0
0.34
References 
Authors
0
3
Name
Order
Citations
PageRank
Qingxiaoyang Zhu101.01
Yi-Chieh Lee2262.08
Hao-Chuan Wang329645.80