Presentation Information
[3Yin-A-03]Designing AI-Mediated Facilitation for Visual Thinking Strategies:Exploring the Tension Between Neutrality and EmpathyAn Exploratory Study of Neutrality and Empathy in AI-Mediated Dialogue
〇Asuka Saito1,2 (1. SMFL, 2. TOKYO CITY UNIVERSITY)
Keywords:
VTS,Generative AI,Communication,Learning Design,facilitation
Visual Thinking Strategies (VTS) is a dialogic learning approach widely used in educational and professional training contexts, where participants discuss artworks to develop observation, reasoning, and communication skills through facilitated dialogue. However, its adoption is constrained by the availability of skilled human facilitators.This study explores whether generative AI can function as an AI-mediated facilitator for VTS, focusing on how prompt-based interaction control shapes dialogue quality.
We designed and iteratively refined multiple prompt patterns based on core VTS principles—neutrality, empathy, and thought elicitation—examining how variations in questioning depth and response tone affect interaction. These prompt strategies were evaluated through exploratory trials and a small-scale user study with both experienced and novice participants. The results indicate that while strict neutrality helps maintain dialogic structure, it often reduces engagement and perceived interaction safety for novices. In contrast, prompts incorporating limited empathetic responses improved dialogue sustainability without violating VTS principles.These preliminary findings reveal a design tension between neutrality and empathy in AI-mediated learning dialogues.
We designed and iteratively refined multiple prompt patterns based on core VTS principles—neutrality, empathy, and thought elicitation—examining how variations in questioning depth and response tone affect interaction. These prompt strategies were evaluated through exploratory trials and a small-scale user study with both experienced and novice participants. The results indicate that while strict neutrality helps maintain dialogic structure, it often reduces engagement and perceived interaction safety for novices. In contrast, prompts incorporating limited empathetic responses improved dialogue sustainability without violating VTS principles.These preliminary findings reveal a design tension between neutrality and empathy in AI-mediated learning dialogues.