Google Notebook LM boosts EAP learning: study finds AI materials enhance rather than slop

A new study from researchers at a Hong Kong Community College examines whether AI-generated educational content is pedagogical enhancement or merely 'AI slop'—high volume, low quality material. Using Google Notebook LM, a RAG (retrieval-augmented generation) tool, an instructor transformed course materials into videos, podcasts, infographics, and individualized feedback reports for 106 English as a Foreign Language learners in an English for Academic Purposes (EAP) course. The study employed an explanatory sequential mixed-methods design including surveys, interviews, and correlation analysis with academic scores, framed through the Technology Acceptance Model and Cognitive Load Theory.

The results show that students perceived the AI-generated materials as highly useful and easy to use, with a clear preference for assessment-linked content in visual and multimodal formats—especially videos and infographics. Notably, video preference positively correlated with academic performance, indicating that well-designed AI media can boost learning outcomes. However, higher cognitive load was negatively associated with course grades, emphasizing that material complexity must be carefully calibrated. Interestingly, lower-performing students independently adopted the materials as remedial scaffolds, demonstrating the potential for scalable personalized feedback that traditional methods cannot easily provide. The paper concludes that when teacher-prompted AI generation aligns with student goals and cognitive principles, it meaningfully enhances learning rather than producing AI slop.