Towards incorporating educators’ perspectives into the meta-system design of intelligent tutoring systems

Intelligent tutoring systems (ITSs) are increasingly used to support personalized learning through adaptive instruction, content, and feedback. However, we suspect that the existing meta‑system design of ITSs emphasizes learner‑oriented components, such as the Domain Model, the Learner Model, and Tutor Models, while underrepresenting educators’ perspectives in real‑world learning environments. This study explores how educator perspectives appear in empirical reports of implemented ITSs in real-world learning environments and how these perspectives align with the existing meta‑system-design of ITSs. Following PRISMA 2020 guidelines, we conducted a systematic literature review across seven major academic databases and identified 1,068 publications from 2014 to 2024. Relevant studies for a full‑text analysis were selected ( n = 21) to form a database of text artifacts describing educators’ perspectives on ITSs and educator-ITS teaming situations implemented in real-world learning environments. In this way, 435 text artifacts were extracted using five defined analytical lenses. These text artifacts were further analyzed using an epistemological network analysis and synthesized through a directed content analysis. The findings highlight mismatches between reported educational practices and the existing meta‑system design of ITSs. Drawing on these insights, we propose an extension of the meta‑system design that incorporates educators’ perspectives on ITSs and educator-ITS teaming situations, grounded in the principles of Learning Design and the Teaching Cycle. The illustrated enhancement aims to inform interdisciplinary research and the EdTech sector in designing and developing ITSs from a holistic perspective on smart learning environments, to strengthen explainability, usability, and transparency for educators.