AI-Driven Digital Concept Maps in Learning Environments: Empirical Insights from Student Use

Abstract

This study explores the integration of AI-powered, ontology-based interactive concept maps into a university programming course to support student learning and includes a comparison of these maps with AI-generated instructional text materials. As part of the “SDT 100: Principles of Programming with Java” course at American University Kyiv, foundational instructional materials were generated using GPT-4o and formalized into an ontology via the Semantic Portal platform. The platform then automatically generated interactive concept maps, with each node linked to structured concept descriptions. These maps were used alongside AI-powered text lectures, providing students with an alternative to linear instructional content. To assess pedagogical relevance, students were asked to provide qualitative feedback on the perceived usefulness of the concept maps. The findings suggest that such maps are particularly effective for content review and understanding conceptual relationships, serving as a valuable complement to traditional text-based materials. The study provides empirical evidence on the instructional potential of AI-powered, ontology-based concept maps and outlines future directions for enhancing their automated generation using large language models.