AI Models in Interpreting Basic Design Elements & Principles

This study addresses the challenge of evaluating how effectively artificial intelligence (AI) models can interpret and visualize basic design principles and elements—a skill essential in creative disciplines such as art and design education. The core aim is to examine whether AI-generated visual outputs can reflect a deliberate understanding of design concepts, particularly within abstract symmetrical and asymmetrical compositions. To explore this, the study analyzes the performance of 18 widely used text-to-image AI models, based on five design principles—rhythm, movement, contrast, emphasis, and balance—and five design elements—dot, line, shape, color, and texture. Each model was prompted to generate compositions aligned with these criteria, and the outputs were evaluated based on three primary criteria: understanding the design elements, understanding the design principles, and understanding the balance. Scores ranged from 1 (least accurate) to 5 (most accurate), and averages were calculated for comparative analysis. The results revealed that while models like Microsoft Designer and DALL-E occasionally produced strong outputs, they lacked consistency across different balance types. In contrast, models such as Craiyon and Flux.ai consistently underperformed. Notably, high scores in asymmetrical balance often inflated overall averages, masking deficiencies in design comprehension. The study concludes that symmetrical balance serves as a more reliable indicator of AI proficiency and emphasizes the need for holistic evaluation frameworks. These findings offer valuable insights for integrating AI tools into design education and assessing their pedagogical potential.