Motivation
As AI-generated lecture videos become more common, this project examined whether human-presented and AI avatar-presented videos are equally effective for learners with different levels of autistic traits. It focused on how autistic traits may influence learning experience, attention allocation, and perceived cognitive load during video-based instruction.
Research
The project used a randomized, counterbalanced within-participant design comparing matched human and AI micro-lectures across four topics. It combined AQ-10, knowledge tests, post-video ratings, cognitive load measures, and webcam-based eye-tracking to examine whether autistic traits shaped differences in learning, attention, and format preference.
Impact
The findings showed that human-presented videos led to a slightly better learning experience and quiz performance overall, while autistic traits had limited influence on most human–AI differences. The project provides evidence-based guidance for designing more inclusive and trustworthy AI-supported educational videos.
