Extant use of Artificial Intelligence in developing Design Variations in Design Studios in Kumasi Technical University

Abstract

This study investigated the extant use of artificial intelligence (AI) for generating design variations in studio-based design education, focusing on the Departments of Construction Technology and Quantity Surveying and Interior Design & Materials Technology at Kumasi Technical University. A convergent parallel mixed-methods design was employed in which 130 students completed counterbalanced ideation tasks under AI-assisted and traditional conditions, and ten lecturers provided expert evaluation and qualitative insight. Outputs were assessed using expert rubrics (novelty, diversity, feasibility, aesthetic quality), automated CLIP-based diversity/novelty metrics, task logs, and interviews. Quantitative analyses showed that AI substantially increased ideation productivity (Mean variations: AI = 10.6 vs Traditional = 5.2; F(1,129) = 118.45, p < .001, η² = .48) and yielded higher novelty and diversity scores (expert means: Novelty 5.8 vs 3.9; Diversity 6.1 vs 4.2; both p < .001), while feasibility was reduced (4.6 vs 5.4; p = .021). Prompt specificity and instructor scaffolding were strong mediators of output quality (prompt specificity r = .61, p < .001). Qualitative findings revealed benefits for divergent exploration alongside risks of design fixation, stylistic homogenization, and ethical concerns (authorship, copyright, and cultural bias). The study concludes that AI functions best as a complementary ideation tool when embedded within scaffolded pedagogy emphasizing prompt literacy, critical evaluation rubrics and ethical guidance. Practical recommendations include curriculum modules on AI and prompt engineering, institutional policies for IP and cultural dataset development, and further longitudinal and industry-linked research to assess long-term impacts on creativity and professional practice.