Recent advances in conversational AI increasingly challenge the traditional instrumental framing, where users in interaction not only utilize AI to accomplish tasks but also begin to place it in a subject position. However, existing theoretical constructs primarily focus on the psychological attributes users ascribe to AI, rather than the normative practices of how AI is concretely treated in relationships. To bridge this conceptual gap, this article proposes the concept of AI Subjectification, defining it as the process by which interactants place AI in a subject position and, on that basis, extend specific treating rules and status expectations to it. Drawing on objectification theory and recent findings in human–computer interaction, we conceptualize AI subjectification as an observable and multidimensional mode of treating in interaction, and construct a Layered Framework of AI subjectification. This framework organizes AI subjectification into the Value Layer (End-regarding orientation), the Relational Interaction Layer (Treat as Autonomous, Treat as Agentic, Treat as Experiencing, Treat as Unique, and Treat as Dignified), and the Institutional Layer (Independence Conferral and Power Conferral), which leads to a series of testable propositions. This article further clarifies the boundaries between AI subjectification and adjacent concepts (anthropomorphism, mind perception, and the Computers as Social Actors paradigm). AI subjectification provides a novel and operationalizable vocabulary for the fields of psychology and human–computer interaction, and offers important implications for the design of AI ethical guidelines, user governance, and the management of human–AI responsibility chains.
Xingyi Liu (Sat,) studied this question.