posted on 2025-11-18, 11:53authored byH Gao, B Zhao, X Hu, C Liu, H Chen, X Jiang, H Zhang, Huiyu ZhouHuiyu Zhou
<p dir="ltr">Introduction: Although group discussion plays a crucial role in collaborative learning, it often falls short of achieving optimal effectiveness. The introduction of conversational agents has the potential to enhance the effectiveness of group discussion; nevertheless, the interaction strategies between conversational agents and human participants remain an issue that requires further investigation. The present study aims to examine how the diverse viewpoints provided by the conversational agent and participants’ attention to them affected discussion effectiveness.</p><p dir="ltr"><br></p><p dir="ltr">Methods: This study involved 129 university students who discussed an open-ended question in an adaptive discussion system. A 2 (adaptive cognitive diversity: homogeneity vs. diversity) × 2 (attention: with vs. without instruction) between-subjects design was employed, with an additional control condition. Participants in the experimental conditions interacted with a conversational agent, while those in the control condition discussed in pairs without it.</p><p dir="ltr"><br></p><p dir="ltr">Results and discussion: The results indicated that discussions in the diversity condition exhibited greater breadth, whereas those in the homogeneity condition demonstrated significantly greater depth, suggesting that diverse perspectives promote broader idea exploration, while similar perspectives facilitate deeper elaboration. Compared with the control condition, the diversity-with-instruction demonstrated greater discussion breadth. Participants under the with-instruction condition perceived the conversational agent’s viewpoints as obstructing their own idea generation; by contrast, those under the without-instruction condition generated a higher proportion of valid ideas and achieved deeper and better understanding of the discussion topic. These results suggest that attention plays both positive and negative roles in the discussion process. The present study examined the roles of adaptive cognitive diversity and attention in group discussion and explored how manipulating these factors within a human-computer interaction system can shape discussion effectiveness.</p>
Funding
Science and Technology Tackling Project of Henan Province "The influence of adaptive cognitive diversity and fine elaboration on discussion effect in intelligent discussion system" (232102320156), Research and Practice Project on Higher Education Teaching Reform jointly supported by Henan Province (2024SJGLX0391) and Xinxiang Medical University (2024-XYJG-12) "Reform of psychiatry teaching based on generative AI in the context of new medical science"
History
Author affiliation
University of Leicester
College of Science & Engineering
Comp' & Math' Sciences
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: figshare: https://figshare.com/s/7db9d1f93d5ca9e47208 and DOI: https://doi.org/10.6084/m9.figshare.28255064.