Presentation at the ICMI 2026 Conference in Naples, Italy
08 Oct 2026
Presentation at the ICMI 2026 Conference in Naples, Italy
The 28th ACM International Conference on Multimodal Interaction (ICMI 2026) took place from October 5 to 9, 2026, in Naples, Italy. Under the theme “Context and Cultural Awareness for Multimodal Interaction,” the conference brought together researchers working on artificial intelligence, multimodal communication, and interactive technologies.
The HCI and PIIS groups at the University of Würzburg contributed a full paper and two Late-Breaking Results posters. A particular highlight was a review of ethical guidelines and stakeholder perspectives on conversational AI in psychotherapy, connecting ethical principles with the perspectives of the people who may use or be affected by these systems.
Paper
Ethical Guidelines and Stakeholder Perspectives on the Use of Conversational AI in Psychotherapy – A Scoping Review
Paula Friedrich, David Obremski, Marc Erich Latoschik, Carolin Wienrich
How can conversational AI in psychotherapy be developed with the needs and concerns of stakeholders in mind? This scoping review brings together ethical considerations and research on stakeholder perspectives to inform human-centered development.
Late-Breaking Results Posters
Alongside the scoping review, the groups presented two contributions in the Late-Breaking Results poster track, sharing ongoing research with the multimodal interaction community.
How Local AI Framing Shapes User Experience and Perceived Data Security with an Embodied AI Psychotherapist in VR
Paula Friedrich, Lukas Polifke, David Obremski, Marc Erich Latoschik, Carolin Wienrich
This study explores how describing an embodied AI psychotherapist as locally hosted or cloud-based affects users’ perceptions. The findings highlight how information about AI deployment can shape the experience of conversational systems in sensitive contexts.
CEMAS - a Configurable Embodied Multi-Agent System
David Obremski, Lukas Polifke, Marc Erich Latoschik, Carolin Wienrich
CEMAS enables a human user to interact with multiple embodied conversational agents in extended reality. The system combines a modular conversational AI pipeline with a moderator that coordinates speaking turns and an embodiment layer that synchronizes speech with non-verbal behavior. It provides a configurable foundation for studying group conversations with virtual agents, with potential applications in education, mental health, and collaborative problem solving.
