earticle

영어

The affective-interactive functions of AI companionship have become an increasingly central component of platform-based content provision. This study examines how such continuous and stable forms of content output are produced, organized, and sustained at the industrial level. Drawing on emotional labor theory, it analyzes two Chinese AI models: Doubao and Tongyi Qianwen to explore how emotions are structured as labor through linguistic construction, functional design, and disclaimer mechanisms. The findings show that both models create relatively stable emotional spaces via narrative strategies and continuous interaction, enabling users’ emotions to be persistently engaged and managed. Simultaneously, disclaimer rules institutionally decouple emotional responses from their potential psychological or social consequences, externalizing responsibility and risk. As a result, AI-mediated emotional interaction is organized as a form of labor space embedded within platform governance and the logic of the content industry.