In digital learning environments, high-frequency, short-form video content often promotes passive viewing and surface-level processing, inhibiting deep cognitive engagement and higher-order knowledge transfer. This study investigates whether Generative Learning (GL) and Spaced Learning (SL) can address this challenge by optimizing cognitive load and enhancing learning outcomes for high school students. A within-subject experimental design was employed to minimize individual differences. Participants experienced three conditions: Spaced Viewing (SL, B-2), Generative Verbalization (GL, B-3), and a Stacked Strategy (GL+SL, B-3+B-2). To ensure robust internal validity, learning materials spanned three distinct domains—Law, Virology, and Technology—with cross-topic stability confirmed via Kendall’s W coefficient (W = .82, p < .001). The results revealed that while the Stacked Strategy produced the highest overall performance, a ""Transfer Reversal"" phenomenon was observed: the GL-only condition achieved superior scores in divergent Extended Points (4.2 vs. 4.0). This suggests that immediate re-exposure may induce redundancy effects, fixating cognitive resources on structural accuracy rather than autonomous Semantic Integration. Notably, for low-achieving students, GL led to a remarkable 266.7% increase in Core Points, demonstrating a powerful ""cognitive compensation"" effect that effectively narrows the educational equity gap. Beyond experimental findings, this study proposes a ""Two-Stage Differentiated Instructional Model."" We recommend integrating ""Generative Prompts"" into digital platforms to transform passive consumption into active Semantic Integration. This research provides a data-driven framework for AI-assisted personalized learning, where scaffolding is dynamically adjusted based on learner profiles—providing GL-first foundations for struggling students and Stacked Strategies for advanced mastery. By optimizing digital pedagogy, this study directly contributes to United Nations SDG 4 (Quality Education), ensuring inclusive and effective learning in the age of fragmented media.