Recognising Ourselves in AI: Socially Intelligent AI Companions, Congruence, and Mental Well-Being
(A2026-133735)

AUTHORS

Ana Javornik, University of Bristol; Amani Alabed, ‪University of Doha for Science and Technology; Rebecca Casey, Newcastle University; Diana Gregory-Smith, Newcastle University

ABSTRACT

The use of conversational AI agents such as Replika for discussing personal issues and therapeutic services has grown rapidly, as many consumers turn to this technology for emotional support and mental well-being advice. The aim of this paper is to provide an explanation for the popularity of this type of service among consumers. Through three experimental studies we show that social intelligence, as a specific trait of anthropomorphized AI, leads to consumers experiencing congruence with the AI, as consumers feel the AI can genuinely understand them and thus perceive it as similar. In turn, this sense of similarity leads consumers to viewing the AI agent as highly meaningful for them, integrating it into their self-concept (self-AI integration). Finally, we show that AI social intelligence indirectly leads to perceived emotional support and willingness to accept AI recommendations through this psychological mechanism.