디자인 교육에서 생성형 AI 도구의 반복성과 표현 다양성의 한계 :그래픽 디자인 수업 사례를 중심으로

As generative AI tools become increasingly integrated into design education, issues regarding output repetitiveness and limited expressive diversity have emerged as significant pedagogical challenges. This study empirically examines these limitations through a poster design workshop conducted with 96 graduate students at a four-year university in southern China. Utilizing a mixed-methods approach, this research analyzed 96 AI-generated poster works and conducted structured written interviews. The findings revealed pronounced homogenization patterns, characterized by a predominant use of blue color schemes and an overwhelming prevalence of centralized layouts. Shannon diversity index analysis demonstrated that layout diversity reached only 78.3%, indicating severely constrained variation, while the average inter-work similarity coefficient was measured at 0.693, providing empirical evidence of high repetitiveness. Participants consistently reported several challenges, including limitations in prompt articulation, frustration with repetitive outputs, and inadequate fine-tuned control capabilities. Drawing from these findings, this study underscores the necessity for educational strategies that foster culturally diverse applications of AI tools and proposes a comprehensive framework entitled "AI-Cooperative Design Education for Diversity Enhancement." This research provides foundational insights for developing new pedagogical paradigms in AI-integrated design education.