Creative potential of image-generative AI models for conceptual engineering design tasks

Abstract

Creativity is paramount in design engineering, driving design innovation with new ideas. Visual representations are crucial in communicating and refining design ideas. The emergence of image-generative AI presents new opportunities, but a comprehensive understanding of AI creative capabilities, particularly in comparison to human designers, remains unexplored. This explorative study investigates the creative potential of image-generative AI systems by comparing their outputs to those of human designers. Utilizing the metrics of Quantity, Variety, Novelty, and Quality, we evaluated the performance of 6 AI-based image-generation tools representing several diffusion-model families and 20 design engineering students on an assigned design task. Human participants were subject to time constraints, while AI tools were constrained by predefined output parameters. Our findings highlight the complementary nature of human and AI creativity. AI tools excel at generating a large volume of solutions rapidly, but their outputs lack the diversity, novelty, and nuanced quality characteristic of human designs. While AI tools demonstrated promising results in Novelty, achieving high Quality consistently proved challenging across all tools. These results underscore the importance of integrating human oversight and interpretation to guide AI systems towards more meaningful and innovative outcomes, with the aim of optimizing the process of generating creative ideas of designers, using AI systems for image generation.