This paper defines Relay-Discontinuity as a structural condition in human–AI interaction, where continuous human problem-solving processes encounter session-based discontinuity in AI systems. We show that this condition gives rise to a Cognitive Relay Loop, in which thought is iteratively transferred and reconstructed between human and AI. Rather than treating this as a limitation, we argue that Relay-Discontinuity enables cognitive restructuring through repeated interaction, positioning AI as a cognitive amplification interface rather than a mere tool. Finally, we outline interface design implications for stabilizing and leveraging this structure.
Byung Gil Moon (Mon,) studied this question.