How perceived learning difficulties shift in a short-term GenAI-integrated design course

Abstract

This study examines shifts in students’ perceptions of support and learning difficulties within a short-term Generative AI (GenAI)-integrated design course. A pre-post survey design was implemented in a four-week (32 contact-hour) undergraduate design course where GenAI was embedded as a pedagogical component. Matched responses (paired n = 36; post n = 39) assessed perceived efficiency support, creative support, self-efficacy, and future intention, alongside post-course mechanisms (guidance improvement, workflow formation, iteration intensity) and persistent bottlenecks. Pre-post changes were modest: efficiency support remained high but unchanged, creative support showed a small non-significant increase, and self-efficacy improved with marginal significance (p = 0.058). Most students reported an improved ability to guide GenAI and an emerging workflow, while common bottlenecks centred on controllability, consistency, and integration. The findings rely on perception-based measures and should be interpreted as indicators of students’ experienced support and difficulty during the course. The results suggest that GenAI compresses early production barriers but shifts difficulty toward higher-order control and synthesis.