Enhancing Contextual Understanding in AI-Powered Tutoring: Evaluating the Oliver System for Effective Learning Support
Haoran Zhu, Michael Cooper-Stachowsky, Zille Huma Kamal · 2025
In recent years, advancements in conversational AI have led to the development of intelligent tutoring systems to enhance learning experiences through interactive conversation. This paper presents Oliver, an innovative virtual teaching assistant and course management system that leverages contextual memory and response strategies designed to promote active learning and critical thinking. Unlike traditional models that frequently offer direct answers, Oliver encourages exploration and comprehension. We also evaluated Oliver against ChatGPT-4o mini in a controlled environment with over 100 real class interactions by using Bloom's Taxonomy as a framework. Results indicate that Oliver retains lecture-related context and promotes learning more effectively, with 90% of responses fostering higher-order thinking and critical engagement, compared to 60% from ChatGPT-4o mini. These findings underscore Oliver's potential to serve as a powerful tool in education, supporting learners in developing deeper cognitive skills rather than relying on rote memorization.