Abstract
This article examines the pressing issue of the discrepancy between modern requirements for teacher professional competencies and the results achieved by traditional teacher training models. The study is based on the premise that existing feedback practices do not fully support the development of activity-based competencies, while the haphazard integration of artificial intelligence (AI) often merely automates outdated approaches. As a conceptual solution, an agent-based dialogue model (ADM) is proposed and theoretically substantiated—a pedagogical system that organizes feedback through interactions between the student, the teacher, and AI. The model is based on the principles of student agency, dialogicity, and reflexivity, shifting the focus from control to development.
The central thesis is that ADM can be used as a meta-pedagogical simulator, transforming the student's passive role into an active subjective position. The theoretical analysis demonstrates how the meaningful inclusion of AI as a primary analyzer enables differentiated feedback, freeing the teacher to perform facilitative and mentoring functions, while the student, by initiating the process and developing an action plan, masters self-regulation skills in practice. The model demonstrates how participation in such a hybrid dialogue creates the conditions for the development of an integrated set of professional competencies in future teachers: facilitation, design, reflective, and digital. Thus, the article presents ADM as a conceptual framework for designing developmental educational practices that utilize the potential of AI to shape the professional agency of future teachers.