<![CDATA[Artificial intelligence-mediated communication introduces new forms of technological participation in interpersonal interaction, raising questions about trust, social presence, and relational connectedness in higher education. This study examined how existing scholarship characterizes these relationships and identified the conditions associated with variation in relational outcomes. A scoping review and framework-based relational mechanism synthesis were conducted using 34 scholarly sources published between 1976 and 2026, identified through bibliographic and publisher-based searches. The evidence base included experimental, cross-sectional, longitudinal, mixed-method, design-oriented, and conceptual investigations. Sources were examined according to communication configuration, degree of AI assistance, disclosure conditions, relational outcomes, and contextual moderators. The synthesis identified three interconnected domains: trust calibration and authenticity, social presence and conversational agency, and interpersonal connectedness in educational communication. Across the reviewed literature, AI assistance was associated with communicative efficiency and responsiveness, while relational evaluations varied according to perceived authenticity, transparency, technological reliability, and human control. Conversational and anthropomorphic cues contributed to perceived social presence without consistently establishing durable interpersonal trust. Evidence involving university students particularly emphasized AI literacy, perceived accuracy, risk perception, and technology adoption, whereas broader relational interpretations drew partly on other populations. The study proposes calibrated trust as an interpretive principle emphasizing proportionate reliance, verification, and human accountability. These findings support differentiated assessments of AI-mediated communication that distinguish technological performance, perceived social presence, and sustained interpersonal relationships.]]>
