KAGE: A Knowledge-Augmented Generative Engine for Cognitively Sustainable Cultural Heritage Design

Generative artificial intelligence (AIGC) has transformed creative practices, yet its promise of “universal creativity” is constrained by high cognitive demands. Users must translate abstract ideas into structured prompts while maintaining conceptual and cultural coherence, a process that often overloads working memory. This challenge becomes critical in cultural heritage creation, where accuracy depends on specialized domain knowledge. To address this issue, this study proposes the Knowledge-Augmented Generative Engine (KAGE), integrating integrates domain-specific knowledge graphs and intelligent agents to support cognitively sustainable creation. Grounded in Cognitive Load Theory and Distributed Cognition Theory, KAGE externalizes complex knowledge retrieval, allowing users to focus on high-level creative decisions. A controlled experiment using NASA-TLX, expert evaluation, and user experience measures, results show that KAGE significantly reduces extraneous cognitive load and improves creative quality. Mediation analysis further demonstrates that cognitive load reduction indirectly enhances perceived experience. These findings establish the Cognitive–Quality–Experience (CQE) Framework, and offer design principles for building cognitively friendly generative systems that enhance, rather than replace, human creativity.