A Human–AI Orchestration Framework for Mathematics Education

Source abstract

Artificial Intelligence (AI) is increasingly influencing personalized and adaptive education. Existing research has examined Human–AI hybrid adaptivity and coordination between human participants and AI systems, but clearer conceptualization is needed of the pedagogical coordination of functionally different forms of instructional support, particularly in highly abstract domains such as mathematics. This study proposes a Human–AI Orchestration Framework integrating Intelligent Tutoring Systems (ITSs), Generative AI, and direct teacher support while retaining pedagogical orchestration and responsibility with the teacher. Drawing on Cognitive Load Theory, the Framework proposes Cognitive Load Redistribution as a theoretical proposition and introduces coordination mechanisms, including Handover Points and a Meta-Prompt. Its practical implementation was explored through an exploratory design pilot with 32 Grade 8 students in an authentic classroom, using observations and feedback to iteratively refine its configuration. The pilot informed several design refinements, including refinement of the Meta-Prompt, introduction of “Call the Teacher,” and adaptation of interaction formats. These formative observations provide preliminary indications of the practical feasibility of coordinating functionally different and potentially complementary forms of instructional support, while the Framework’s educational effectiveness and proposed conceptual mechanisms remain to be systematically evaluated in future research.