Social interaction in mixed human-AI groups
本研究テーマでは、社会的学習、社会的承認、社会的促進、社会的抑制などの社会心理学理論を再検討し、人間とAIが混在する集団において、個人の認知や行動がどのように変化するのかを明らかにすることに焦点を当てています。
また、複数エージェントが存在する状況下で、個人あるいは小集団がどのように相互作用するのかを、多様なコミュニケーション文脈(例:対立の発生、困難な意思決定、脆弱な感情の開示など)において検討します。
本研究テーマの目的は二つあります。第一に、従来の対人相互作用に基づく社会心理学理論が、人間と社会的AIとの相互作用にも一般化できるのかを再検討・再考すること。第二に、望ましい社会的行動を形成・促進するためのマルチエージェントシステムを設計することです。
In this line of research, we focus on revisiting social psychology theories (e.g., social learning, social validation, social facilitation, social inhibition, etc) to examine how individuals’ perceptions and behavior change in mixed human-AI groups.
We study how an individual or a small group of people interact with one another in the presence of multi-agents in various communication contexts (e.g., having conflict, making hard decisions, disclosing vulnerable moments). The goal of this project is twofold: (1) to revisit and reflect on the generalizability of social psychology theories from social interaction to human-social AI interaction; (2) to design multi-agent systems for shaping positive social behavior.
Example projects:
Understanding the impact of AI-generated danmaku on presenters’ and audience’ engagement in online presentations (ongoing research with Shigeo Yoshida from OMRON SINIC X Corporation, Japan)
Designing multi-agents for supporting conflict management in cross-cultural communication (ongoing research)
Designing multi-agents for supporting female users to cope with PMS (ongoing research led by Shixian Geng from IISLab at the University of Tokyo; CHI'25)
Understanding and supporting equal gender participation in video conferencing with multi-agents (ongoing research with Wen Duan from Clemson University)
Understanding attachment in human-AI relationships and designing detachable AI for healthy human-AI relationships (Ongoing research)
Keywords: Social AI, mixed human-AI groups, social psychology, multi-agents