From Craftsmen to Curators: Navigating Cognitive Friction in Generative AI Filmmaking Education

While Generative Artificial Intelligence (GenAI) is rapidly transforming media production, higher education faces a critical pedagogical gap in integrating these volatile technologies into arts curricula. Moving beyond large language model for education, this exploratory multiple-case study investigates the educational effectiveness of a 13-week Project-Based Learning (PBL) course involving 32 Master of Arts students for AI filmmaking. Grounded in Cognitive Load Theory and the Revised Bloom's Taxonomy, we theorize a structural transition in student creative identity: from traditional physical “craftsmen” to “curators of probabilities.” Our findings illustrate: (1) how students navigate “cognitive friction” and operational overhead, including prompt fatigue and hallucination management; (2) the emergence of adaptive collaborative workflows; and (3) an “assessment paradox,” where traditional product-oriented rubrics fail to capture complex human-AI co-creation processes. Findings indicate that students experienced an inverse traversal of the Revised Bloom's Taxonomy when using GenAI for filmmaking, shifting from technical execution to aesthetic curation. We synthesize these empirical insights into an exploratory conceptual framework detailing the constructs, dimensions, and boundary conditions of human-AI collaborative creativity in arts education.