A collaborative creation framework for experimental animation pedagogy integrating generative artificial intelligence and vibe coding

Summary

A single-cohort design-based research study of a five-week experimental animation course in the Department of Digital Media Art at Soochow University, in which 40 third-year students used generative AI (Midjourney, ChatGPT and DeepSeek are named) for concept and image work and the Cursor coding environment for vibe coding, writing their own motion mechanics in collaboration with an AI assistant. The paper proposes a generation, feedback and evolution framework across four pedagogical stages, with AI's role moving from tool to collaborator. Its most useful result is a failure: dynamic prompt descriptors such as loop, diffuse and breathe were flattened by image and video generators into rudimentary scaling, drifting or morphing, which is why coding was brought in. Self-reported extraneous cognitive load fell from 3.91 to 2.90 on a five-point scale and students' perception of AI as a creative partner rose from 3.30 to 4.09. There was no control group, the scales are self-report, and the author states that all p-values are unadjusted and exploratory. Every student closed with a collaboration statement on concept, process and authorship.

Related items

• Integrating Creative Problem Solving and Generative AI in Animation Education: Advancing Sustainability-Related Competencies in Higher Education
• Generative AI adoption in 3D animation production: a pipeline-wide empirical benchmark of early-career practitioner perceptions and the HGAI hybrid pipeline framework
• Analysis of Generative AI for Innovative Thinking in Animation Teaching
• Path of practical teaching of AI technology in animation majors

Source

Source: Frontiers in Psychology ↗ (Research)