Abstract

Limited funding and organizational constraints hinder educators’ ability to provide personalized sup-port. Intelligent Tutoring Systems (ITSs) present a scalable solution; however, they receive insufficient attention in information systems research and lack a robust pedagogical foundation. This paper em-ploys a design science research (DSR) approach to develop an ITS and a Learning Analytics Infor-mation System (LAIS), utilizing large language models (LLMs) for distance learning in higher educa-tion. The ITS design integrates constructivist learning principles, such as connecting new information to existing knowledge, providing adaptive scaffolding, fostering learning motivation, and enhancing metacognitive skills. Through two design cycles, the ITS was developed and evaluated with a focus on implementing these constructivist principles. In a third cycle, the LAIS was introduced to capture and analyze qualitative interaction data, enabling targeted, data-driven interventions by instructors. Evalua-tion results provide preliminary evidence that the ITS can support active learning, while LAIS-informed instructor interventions were followed by reduced topic help-seeking. Based on these find-ings, we formulate transferable design principles for integrating pedagogical theory into LLM-based ITSs and guiding the collaboration between ITS and LAIS in higher education. This paper makes three key contributions: it presents a constructivist ITS design that leverages LLMs, proposes an integrated LAIS for instructor interventions based on qualitative data, and outlines how these principles can fa-cilitate ITS-LAIS integrations in digitally supported learning environments.

DOI

10.17705/1jais.01010

Download

Share

COinS