Analysis of the Erosion Mechanism of Social Bots on Online Trust in Para-social Interaction

Abstract

Against the backdrop of artificial intelligence’s deep integration into the communication ecosystem, the role of social robots has shifted from an intermediary application to a cognitive-action-oriented one. This paradigm shift has intensified quasi-social interaction between users and robots. Based on users’ awareness of the robot’s identity and their level of initiative in the interaction, quasi-social interaction can be categorized into two forms: “collusive” (informed and proactive) and “parasitic” (uninformed and passive), both of which follow the logic of “pseudo-reciprocity” in digital interaction. Building on this, this paper introduces AI illusions as a moderating variable to examine how they shape the relationship between quasi-social interaction and online trust. The study found that, in parasitic interactions, trust is built on misidentification of identity, so factual-logical illusions trigger the immediate collapse of trust; in collusive interactions, emotional dependence leads to the sustained accumulation of trust, but knowledge-fictional illusions covertly erode cognitive trust, while emotional-value illusions reinforce AI flattery and transmit the trust crisis to the interpersonal interaction space through user behavior. The theoretical contribution of this paper lies in revealing the differentiated moderating pathways of these three types of illusions across different interaction types, and in providing a dual approach to the governance of online trust: technical correction and the reconstruction of interaction ethics.