Perceived Subjectivity as an Interactional Effect in Affective Conversational Systems

Abstract

Large language models increasingly elicit responses that exceed instrumental use, inviting users to relate to conversational systems as if they possessed an inner perspective, memory, and affective responsiveness. This paper argues that the relevant phenomenon is not adequately captured by the binary between machine sentience and mere tool use. Instead, it conceptualizes perceived subjectivity as a socially consequential interactional effect produced in encounters between users and affective conversational systems. Methodologically, the paper develops an interdisciplinary analytical synthesis that brings social psychology, philosophy of mind, and Science and Technology Studies into dialogue in order to distinguish perceived subjectivity from phenomenal consciousness, agency attribution, and emotional projection. On this basis, it shows how functional proxies, user projective dispositions, and interface features such as persona, memory, and warmth cues jointly stabilize the impression of mindedness in use. The paper further argues that this interactional perspective has pragmatic implications for design and evaluation, since trust, disclosure, emotional dependence, and epistemic deference should be treated as design-mediated effects rather than secondary outcomes. Finally, it examines the ethical consequences of these dynamics, especially where conversational fluency begins to reorganize routines of justification and redistribute epistemic authority.