Let technology serve learning: a flipped instructional model with generative AI support to mitigate educational technology risks and strengthen learner agency

Despite its growing adoption, flipped learning still struggles with limited pre-class engagement, superficial in-class interaction, and insufficient individualized support. Meanwhile, the rapid uptake of generative AI (GAI) in higher education introduces opportunities for learning scaffolding alongside concerns about overreliance and diminished learner agency. To address these tensions, this study proposes a six-phase, human-centered flipped framework that extends the WSQ model to include Peer Practice, Create, and Feedback. Within this structure, GAI functions as a cognitive scaffold that aids comprehension, creativity, and reflection rather than substituting for student thinking. Using a sequential mixed-methods and cross-over quasi-experimental design, quantitative analyses showed improvements in self-efficacy, motivation, higher-order thinking, collaboration, learning experience, satisfaction, and project performance. Thematic analysis further revealed that structured peer interaction and feedback reduced passive reliance, stimulated critical questioning, and supported creative application, while GAI served as a catalyst for idea refinement. Recognizing the risk that GAI may weaken learner agency or higher-order thinking, the findings demonstrate that with appropriate design and positioning, human–AI collaboration can enhance flipped learning while preserving critical thinking, problem-solving, and creativity. Overall, learning gains appear to stem not from AI alone but from the interplay of structured tasks, peer collaboration, and AI-mediated scaffolding.