Student Opinion Scale
Category: Student Learning
Abstract: Human tutoring is among the most effective educational interventions, yet high costs and intensive resource demands make it difficult to provide broadly and at scale. We conceptualize this tension as an “iron triangle” of tradeoffs between effectiveness, affordability, and scalable reach. This essay examines how an emerging genre of tutoring, hybrid human-AI tutoring, reconfigures these tradeoffs by redesigning how the work of tutoring is divided between people and technology. We first synthesize developments in human- and AI-only tutoring models, with a focus on introducing concepts across the fields of learning analytics and AI in education (AIED) for a broader education research audience. We identify four emerging hybrid architectures, compare how tutoring responsibilities are assigned to humans versus AI and coordinated within each architecture, and examine patterns of human involvement. We conclude by highlighting unresolved questions and outlining a research agenda focused on which combinations of human and AI tutoring are most promising for different learners and settings, and what conditions are needed to implement hybrid tutoring effectively, equitably, and sustainably at scale.
Connect Education's weekly roundup draws on four EdWorkingPapers. It leads with Chicago research by Elaine Allensworth and colleagues finding that schools matter more than neighborhoods for post-pandemic attendance, and also covers papers on hybrid human-AI tutoring and Texas's four-day school weeks.