Workshop on Collective States in Multimodal Interaction

This workshop explores how multimodal sensing and AI techniques can be utilized to detect and interpret the collective states that emerge in group interactions, in conjunction with individual participant states. Although many previous studies have examined human group sensing, the rise of agentic AI that acts as a peer or a facilitator in team environments introduces new challenges and opportunities. In these situations, sensing technologies will play a crucial role in helping AI agents identify effective strategies for engagement, coordination, and adaptation. This will enhance team brainstorming and collaborative tasks within a human–AI hybrid group, where AI can sense, model, and guide the social dynamics of the team.

Rather than focusing on traditional dyadic interactions between two individuals, this workshop will investigate multi-party settings in which group dynamics and emergent states, such as collective engagement, shared attention, and group affect, are central. The goal is to understand and track how group-level behaviors emerge from multimodal interpersonal signals and how these patterns influence individual participants within the group. By bringing together expertise in artificial intelligence, multimodal interaction, and organization science, this half-day workshop will establish a research agenda for AI-enhanced group interactions.

---