Generative AI for learning: Investigating the potential of synthetic learning videos
Project Overview
This document explores the application of generative AI in education, specifically focusing on AI-generated synthetic videos for online learning. A study comparing these videos to traditional instructor videos revealed no significant difference in learning outcomes or learner perceptions, suggesting AI-generated videos are a cost-effective alternative for educational content creation. While the study found equal acceptance between the two video types, it acknowledges the potential for positive learning outcomes associated with pedagogical agents and avatars, which are a form of generative AI. A key challenge in using generative AI is the risk of misinformation; however, this study mitigated this concern by utilizing expert-vetted input data. The research emphasizes the potential of AI to revolutionize content creation in education, offering a viable and efficient method for producing engaging learning materials while mitigating the risk of inaccurate information.
Key Applications
AI-generated synthetic videos with virtual instructors
Context: Online professional learning, introductory lesson on energy sources and vectors within the EBA Academy (European Battery Alliance)
Implementation: A mixed-method approach was used, randomly assigning adult learners (n=83) to either a control condition (traditional instructor video) or an experimental condition (synthetic video with an AI-generated character). Pre- and post-learning assessments were collected, and participants were surveyed on their learning experience.
Outcomes: Learners in both conditions showed significant improvement from pre- to post-learning, with no significant differences in gains between the two conditions. No differences were observed in how learners perceived the traditional and synthetic videos.
Challenges: Potential for false or inaccurate information in AI-generated content. The study addressed this by using subject matter experts in the creation of the traditionally produced video, ensuring that the information presented in the AI-generated synthetic video was accurate and reliable. The study also notes that the use of AI-generated synthetic videos alone may not be sufficient to support effective learning. In order to fully leverage the potential of such videos in an educational setting, it is imperative to integrate them within a larger curriculum that is grounded in sound learning science and instructional design principles.
Implementation Barriers
Accuracy of Information
Fear of false or inaccurate information being presented to learners in AI-generated content.
Proposed Solutions: Use input data that is unchanged and comes directly from material produced in collaboration with subject matter experts. Integrate AI-generated videos within a larger curriculum that is grounded in sound learning science and instructional design principles.
Project Team
Daniel Leiker
Researcher
Ashley Ricker Gyllen
Researcher
Ismail Eldesouky
Researcher
Mutlu Cukurova
Researcher
Contact Information
For information about the paper, please contact the authors.
Authors: Daniel Leiker, Ashley Ricker Gyllen, Ismail Eldesouky, Mutlu Cukurova
Source Publication: View Original PaperLink opens in a new window
Project Contact: Dr. Jianhua Yang
LLM Model Version: gemini-2.0-flash-lite