From tool to companion: an Apparatgeist reanalysis of conversational AI use among Chinese college students

Conversational AI systems are increasingly used not only for information seeking and learning but also for companionship, emotional support, and quasi-social interaction. Existing research has begun to link AI-companion use with depression and loneliness, but this work often treats conversational AI companionship as a unitary behavioral outcome. Drawing on Apparatgeist theory, this article presents an exploratory secondary reanalysis and theoretical reinterpretation of an open survey dataset on Chinese college students' use of conversational AI. We argue that AI companionship is better understood as a set of communicative orientations toward the apparatus. The analysis distinguishes perceived AI agency from perceived AI experience, instrumental from companionate use, and use from acceptance. Results suggest that perceived agency aligns more closely with instrumental and learning-oriented use, whereas perceived experience is more closely associated with emotional-companionate use. Depression is associated with companionate use but not companionate acceptance, which we interpret, tentatively, as consistent with the possibility that AI companionship may function as a reluctant or compensatory substitute rather than an uncomplicated preference (the data do not directly measure reluctance or substitution). Exploratory profiles further suggest multiple Apparatgeist orientations toward conversational AI, including tool, tutor, companion, quasi-person, and reluctant substitute. Because the study is an exploratory reanalysis of cross-sectional secondary data, all associations are correlational and the profile solution is a heuristic rather than a typology. The article contributes to human-machine communication theory by showing that the social meaning of conversational AI depends not only on what users do with it, but on the “spirit” they perceive behind the interface.