New Taxonomy Models Misunderstanding in Human and AI Communication

The increasing reliance on AI-mediated communication channels has exacerbated the problem of misunderstanding, as these channels often lack the real-time, in-person cues essential for detection and repair. This research addresses this critical issue by developing a comprehensive, cross-disciplinary taxonomy and formal model of misunderstanding. The paper synthesizes insights from nine distinct fields, including pragmatics and AI agent research, to analyze misunderstanding as a layered process. It meticulously identifies eleven specific failure modes, demonstrating that each operates at a precise point within eight analytical layers of the communicative process. These mechanisms are categorized by function: primarily generating a divergence, amplifying an existing one, or governing its detection and repair. The formal model extends existing information and communication theory to encompass the reconstruction of meaning, providing a robust framework for understanding and potentially mitigating misunderstandings in both human and AI interactions.