From Individual Chatbot Interaction to the Virtual Tutor with Physical Presence: Comparing Educational Interaction Models with Conversational AI

Conversational artificial intelligence can provide individual study support, but the educational significance of its responses depends on how they are examined and used. We develop a conceptual framework comparing Individual Chatbot Interaction and Standard Classroom Interaction, then specify a Virtual Tutor with Physical Presence as a situated implementation of shared AI inquiry. The framework draws on selected scholarship concerning mediated learning, educational dialogue, self-regulation, cognitive engagement, pedagogical agents, and AI literacy. We define the unit of analysis as an interaction episode and distinguish private consultation, shared instruction, and hybrid activities through observable criteria. The tutor is specified as an animated avatar projected by video-mapping onto a static volumetric surface in the classroom; perceived presence and learning are treated as outcomes requiring measurement. The proposed contribution is a design specification linking public AI responses, teacher orchestration, evidence-based scrutiny, and individual reflection. An explanatory pathway connects these conditions to possible cognitive and metacognitive processes, with four propositions and matched comparisons for future evaluation. A projected avatar is neither necessary nor sufficient for collective verification, and its incremental value must be tested against simpler shared interfaces. No empirical effectiveness is claimed. The framework provides a basis for investigating when a situated conversational tutor supports disciplinary understanding and critical AI use.