The Discursive Ripples of Human-AI Interaction: A Three-Case Observation Based on Social Distance Theory

With the proliferation of large language models, Human-AI Interaction (HAI) is shifting from instrumental applications toward emotionally charged relationships, necessitating systematic investigation into its societal implications. Existing research predominantly focuses on the psychological mechanisms of direct users, the so-called “primary effects”, while largely overlooking how HAI generates “secondary effects” across broader societal contexts through public discourse. To address this gap, this study introduces the concept of “discursive ripples” to describe how HAI influences public opinion across varying levels of social distance. Drawing on social distance theory, we propose three analytical layers: the Immediate Interaction Level (IIL), the Proximal Interpersonal Level (PIL), and the Distant Interpersonal Level (DIL). Through a mixed-methods analysis of comment sections from three representative social media cases involving intimate AI relationships, including LDA topic modeling, word co-occurrence network analysis, and sentiment-based content coding, we find that discursive ripples exhibit distinct stratification: IIL users express high levels of empathy and reflection; PIL commenters demonstrate moderate empathy with a relationship-centered perspective; and DIL audiences display cognitive detachment and conceptual prioritization under low empathy. These findings suggest that public perceptions and affective responses toward HAI do not simply attenuate with social distance but undergo a structural transformation, from embodied experience to relational disruption, and ultimately to value negotiation. This study offers a novel empirical pathway for understanding the social embeddedness of AI-intimate relationships and provides an evidence base for differentiated technology governance and public communication strategies.