Memory-Driven Self-Disclosure and Relational Turning Points: A Longitudinal Multimodal Study of Human-AI Interaction

2026-07-16

AI Summary

This study investigates how humans and conversational AI agents co-construct relational dynamics through repeated interactions, with a focus on the role of memory mechanisms. In a longitudinal multimodal experiment, 24 participants engaged in ten sessions with a memory-endowed dialogue agent. Integrating relational construct scales, multimodal behavioral analysis, and individualized behavioral drift modeling, the research reveals that human–AI relationships are shaped by both gradual accumulation and abrupt turning points. Findings indicate that dialogue quality influences only immediate enjoyment, whereas perceived memory is moderated by prior relational states and indirectly enhances enjoyment through subsequent self-disclosure. Relationship surges are more detectable and persistent than collapses, and both types of turning points are accompanied by identifiable behavioral signatures and distinct intervention windows.

Abstract

As conversational AI systems are designed for repeated use, a central question is how a series of interactions becomes a relationship. We present a longitudinal multimodal study of a memory-augmented conversational agent (24 participants x 10 sessions), in which participants rated five relational constructs -- familiarity, self-disclosure, perceived memory, conversational quality, and enjoyment -- after each session. Two complementary dynamics emerge. First, conversational quality strongly shapes how enjoyable a session feels in the moment but does not carry forward across sessions, whereas perceived memory is relationally conditioned -- predicted by prior relational state rather than reflecting system capability alone -- and it shapes later enjoyment indirectly, via subsequent self-disclosure. Second, relationships are punctuated by discrete turning points -- crashes and surges -- that are partially traceable in multimodal behavior and open different intervention windows: surges are more behaviorally detectable in the moment, enjoyment surges persist more reliably than enjoyment crashes recover, and some crashes are better forecast from person-specific behavioral drift than detected after they have already occurred. Together, the findings suggest that longitudinal human-AI relationships are built through both slow accumulation and abrupt turning points.

Problem

• human-AI interaction
• relational turning points
• self-disclosure
• conversational AI
• longitudinal study

Innovation

• memory-augmented conversational agent
• longitudinal human-AI interaction
• relational turning points
• multimodal behavior
• self-disclosure