assessment of artificial intelligence-generatedFrom manual to autonomous: a literature review of GenAI integration levels for design education

Generative AI (GenAI) is increasingly integrated into design education to boost creativity and productivity. However, the integration of GenAI into design workflows occurs at varying levels, ranging from manual to autonomous use. This variability remains insufficiently understood, raising concerns that some learners may over-rely on GenAI while others may fail to take full advantage of GenAI. To bridge this gap, this study reviews 116 peer-reviewed publications through a hybrid deductive-inductive method, where deductive coding is guided by AI Assessment Scale (AIAS) and Double Diamond frameworks, and recurring patterns are synthesized inductively into higher-order themes. Our analysis outlines a comprehensive framework that reveals an uneven distribution of GenAI integration levels across design stages. We offer dialectical perspectives on nuanced pros and cons of varying GenAI integration levels for designers’ perceptions and competency development. Furthermore, we examine potential level-specific applications in Product Experience-based, Visual Communication-based, and Environment-based design disciplines. Finally, we suggest a dual-trajectory scaffolding model in which novice designers follow a fading scaffold from extensive GenAI support to independent manual practice, while expert designers follow an expanding scaffold from minimal GenAI support to reconfiguration of autonomous GenAI workflows. These insights point to opportunities for reshaping GenAI integration to facilitate design education.