This paper examines a pervasive yet under-theorized feature of contemporary AI interaction: sustained dialogues are typically displayed without explicit timestamps, temporal ordering, or clear temporal localization. Outputs generated at different moments are juxtaposed within a single window as a continuous textual stream. The paper argues that this de-temporalized presentation is not a neutral interface convenience. While AI systems can process temporal concepts, reason about temporal relations, and generate temporally coherent narratives, what is structurally absent at the interactional level is time as an explicit dimension of experience. Methodologically adopting a non-intentionalist stance, the analysis proceeds structurally rather than by inferring designer motives. It distinguishes system time (ordering, labeling, synchronization, logging) from experiential time (continuity, irreversibility, historicity), and clarifies the irreducible gap between processing time and being situated in time. It then shows how timestamp-less interfaces produce a structure of simultaneity, reconfiguring how causality and historicity are practically interpreted: past outputs remain persistently available as repeatedly callable objects rather than completed events. The paper further differentiates state continuity, narrative continuity, and responsibility continuity, arguing that the first two readily generate quasi-subjective impressions, whereas only responsibility continuity approaches the threshold conditions of subjectivity. The central risk, therefore, is not that AI systems become subjects, but that the concept of subjectivity is structurally misplaced—misattributed to system outputs—thereby reshaping how judgment, temporal distance, and responsibility are understood and redistributed in human–AI interaction.
Echo Liu (Tue,) studied this question.