Leveraging “Khanmigo” Generative AI-Powered Tool for Personalized Tutoring to Learn Scientific Concepts
Nedim Slijepcevic, Ali Yaylali · Journal of Teaching and Learning · 2025
This mixed-methods study investigated the effectiveness of Generative AI (GenAI) powered intelligent tutoring systems (ITS) in undergraduate physics education, specifically comparing learning outcomes between students using Khanmigo (Khan Academy's AI tutor) and Google search engine. The study involved 69 undergraduate students divided into two groups (Khanmigo and Google search engine), with a third Paper-only group emerging during the experiment. Participants completed pre and posttests using the Lunar Phases Concept Inventory (LPCI) and participated in structured interviews about their learning experiences. Quantitative analysis revealed significant learning gains across all conditions but found no statistically significant differences between groups in terms of learning outcomes. Qualitative findings indicated that students perceived Khanmigo positively, appreciated its step-by-step guidance, practice problems, and personalized interactions. However, students viewed it as a supplementary tool rather than a replacement for traditional instruction. The study's findings suggest that while GenAI-powered tutoring systems can effectively support learning, their immediate impact on learning outcomes may be comparable to traditional methods. However, the short duration of exposure to the AI tutor and the quality of the printed materials may have affected these results.