A Cognitive-Interactive Prompt Engineering Framework for AIGC Art and Design

Abstract

Generative AI has been widely applied to art and design across concept generation, visual specification and cross-media delivery. Nevertheless, current prompt research mainly focuses on general language tasks or single-image modification, failing to clarify prompt structure’s role in the full design workflow. Based on a self-built practical material repository, this study adopts qualitative framework construction and directed content analysis to summarize 12 prompt strategies covering generation initiation, task control and system delivery. It verifies that constraint reinforcement, feedback loops and stage inheritance have dual impacts on design output. The paper finally establishes a cognitive-interaction framework, offering methodological guidance for AIGC prompt engineering.