HL-EAI: A Multimodal Framework Enabling Emotional Reciprocity in Human-AI Strategic Decision-Making

摘要

This paper presents HL-EAI, a multimodal framework for studying emotion-driven cooperation in human-LLM, human-human, and LLM-LLM interactions via dynamic game-theoretic tasks. HL-EAI integrates emotional prompting, emotion recognition, and expressive AI avatars to enable bidirectional emotion transfer. By modeling affective influence on trust and alignment, it provides a testbed for developing emotionally aware agents. Our demo shows how multimodal emotional cues shape cooperation, advancing socially intelligent, human-compatible AI for interactive multimedia systems.