Karan Taneja presented at the 26th International Conference on Artificial Intelligence in Education (AIED 2025) in Palermo, Italy, on July 24. The paper, Towards a Multimodal Document-Grounded Conversational AI System for Education, was co-authored with Anjali Singh and Ashok K. Goel.
AIED 2025
AIED 2025 brought together over 700 participants from more than 50 countries to celebrate a quarter-century of advancing intelligent human-technology learning ecosystems. The conference theme—“AI as a Catalyst for Inclusive, Personalised, and Ethical Education: Empowering Teachers and Students for an Equitable Future”—emphasized the role of AI in driving equity, personalization, and ethical practices in education.
Introducing MuDoC
The study introduced MuDoC, a Multimodal Document-grounded Conversational AI system designed to enhance multimedia learning. Unlike traditional text-only AI systems, MuDoC generates responses using both text and visuals from documents, producing multimodal, blog-like explanations.
Findings on Engagement and Trust
Researchers compared MuDoC to a text-only system to assess learner engagement, trust, and educational outcomes. Findings indicated that visuals and content verifiability improved learner engagement and fostered trust. However, no significant differences in overall performance were observed.
