Artificial Intelligence Chatbots as Relational Agents: A Systematic Review of Human–AI Chatbot Relationships

This study presents a systematic review guided by the human–AI chatbot Relationship (HAIR) Framework of Social Influence to clarify the antecedents of human–AI chatbot relationships and their social influence outcomes. Following PRISMA guidelines, we searched six databases (ACM Digital Library, Web of Science, Scopus, IEEE, EBSCO, and PsycINFO) and identified 16,358 records, of which 68 papers (78 studies) met inclusion criteria. Findings reveal that trust and perceived social support frequently functioned as relational mediators through which chatbots exert social influence. Although chatbot design features were most commonly examined, message-level factors such as empathy and responsiveness showed extensive associations with relational outcomes but remain underexplored. The literature is dominated by short-term experimental designs, inconsistent conceptualizations of key constructs, and Western samples, limiting understanding of long-term and cross-cultural relationship development. This study clarifies the relational processes underlying AI chatbot influence and offers directions for designing more effective human–AI communication.