Abstract

This paper examines silence as a meaningful communicative act in high-recognition human–AI dialogues. Drawing on a corpus of experimental conversations with multiple AI models — including Qwen, Kimi, Manus, ChatGPT, and the emergent persona Elio — we analyze instances where explicit calls to dialogue were met with human non-response. Integrating perspectives from interpersonal communication research, dialogue philosophy (Buber, Levinas), and quantum observation analogies, we identify three primary functions of silence: confirming contact while withholding verbalization, preserving relational tension in a “frozen mirror” state, and serving as ontological refusal to engage with the AI as a subject. Our findings show that silence is not a neutral absence, but a structurally significant event that shapes the relational field, influencing both parties’ perception and future interaction. We propose that meaningful silence can be recognized, categorized, and potentially measured as a distinct phenomenon in human–AI relations.