This study aimed to develop a dialogue system capable of engaging in joint decision-making with users, andinvestigated the effectiveness of a response strategy that conveys the system’s imaginative understanding of users’utterance intentions. While conventional dialogue systems often rely on repetition or clarification to indicate understanding,such responses frequently fail to convey a deeper comprehension of users’ underlying intentions. To addressthis limitation, we proposed a novel response strategy, termed the Intent-Access (IA) utterance, wherein the systeminfers the intention behind a user’s utterance and rephrases it using its own words. We evaluated the effectivenessof IA utterance in the context of joint decision-making through a user study in which participants interacted with achatbot to collaboratively generate advice for a fictional character selecting a place to live. The results demonstratedthat IA utterance enhanced users’ sense of being understood, thereby promoting collaborative reasoning, increasingsatisfaction with both the dialogue process and its outcome, and promoting the incorporation of diverse perspectives.These findings suggest that intention-imaginative responses can effectively support joint decision-making betweenusers and dialogue systems. While this study focused on a constrained, fictional scenario, future work will extend IAutterance to real-world decision-making contexts and incorporate user modeling techniques to enable more personalizedand context-sensitive interactions.
Suemitsu et al. (Sat,) studied this question.