Title: Conversational AI for Education: Design, Evaluation, and Continual Improvement

Date: Wednesday, April 16th 2025
Time: 9:00 AM - 11:00 AM EST
Location (Hybrid):
•    CODA Conference Room 114
•    Zoom Meeting (Meeting ID: 450 568 9105, Passcode: 944041)

Karan Taneja
Ph.D. Student in Computer Science
School of Interactive Computing 
Georgia Institute of Technology 
https://krntneja.github.io/

 
Committee: 
•    Dr. Ashok K. Goel (Advisor) - School of Interactive Computing, Georgia Institute of Technology
•    Dr. Christopher J. MacLellan - School of Interactive Computing, Georgia Institute of Technology
•    Dr. Larry P. Heck – School of ECE and Interactive Computing, Georgia Institute of Technology
•    Dr. Christopher Dede - Graduate School of Education, Harvard University

Abstract: 
The increasing adoption of AI systems in education has enabled a wide range of applications including automated grading, personalized tutoring, content generation, and intelligent feedback mechanisms. These systems aim to enhance learning experience by offering personalized support and streamlining instructional tasks. One emerging area within AI in education is conversational AI, which provides a way for students to interact with course content through natural language. In my thesis proposal, I explore the design and development of LLM-based conversational AI systems for educational settings, addressing critical challenges such as response grounding, content safety, multimodal learning, and trust. By ensuring that AI-generated answers are grounded in instructor-approved documents and integrating visuals alongside textual content, the proposed systems provide accurate responses and an engaging learning experience. My research investigates the impact of these AI agents on learning outcomes and student experience through user studies to shed light on the dynamics of human-AI interaction in multimedia learning through conversational AI. Additionally, I propose a human-in-the-loop framework that leverages human expertise to refine and update LLM-based systems, improving their domain-specific performance with limited feedback and data collected from user interactions. Through these contributions, my work aims to understand conversational AI tools in education through the lens of technology design and multimedia learning.