Educator and AI agent communication in AI-mediated higher education classrooms: a mini review along the complementarity-substitution axis

Abstract

The rapid integration of generative artificial intelligence into higher education classrooms has reignited the debate over the role of communication between educators and students. This mini-review synthesizes the role of instructional communication constructs (immediacy, rapport, clarity, and credibility) and the relational functions they deliver in AI-mediated classrooms, based on 32 peer-reviewed studies published after 2023, identified through a large-scale literature search (using Scopus and Web of Science-based indexes, OpenAlex, and ERIC). The findings fall into three main categories. First, educator communication behaviors continue to predict student outcomes; a recent meta-analysis reports a moderate-to-large association (r = 0.443), although the evidence base is concentrated in Chinese and EFL samples. Second, students evaluate AI agents using human communication categories, but a systematic gap emerges in the dimension of goodwill, which is partially bridged by cues of immediacy and politeness. Third, when AI replaces human interaction, negative outcomes related to loneliness and a sense of belonging are observed. These three strands are integrated around the complementarity-substitution axis defined at the level of relational functions; tests of measurement invariance, sample diversity, and longitudinal designs are recommended for future research.