Creative Traits In Ai-Generated Imagery, A Computational Framework To Assess The Semantic Values of Image Properties In Design

This paper proposes a novel computational framework which facilitates an automated and empirically based assessment of creative characteristics within AI-generated imagery. The framework enables a nuanced and ubiquitous evaluation of the Generative artificial intelligence (GenAI) outputs and their use in various design fields. Considering the growing use of GenAI in professional practice and education, the effect of GenAI on the creative design process is yet unknown. The proposed framework assesses design creativity by calculating Class and Image semantic scores. The results from a five-day workshop were analysed by semantic and visual analyses to determine fluctuations in creativity. By providing such an evaluation, the framework can support the filtering or excluding a vast number of AI outcomes, potentially supporting professional designers and students throughout the design process to reflect, correct and make changes while meeting higher creativity values.