How do we measure creativity? A systematic review and taxonomy of evaluative methods in human-AI co-creativity

As artificial intelligence (AI) rapidly integrates as a cognitive partner in day-to-day creative tasks, to truly understand the value and effects of this integration, it is critical to know how co-creativity is defined, measured, and understood. Without a coherent understanding of the evaluative practices in the field, it is challenging for researchers to determine which metrics are appropriate, for practitioners to adopt effective evaluative methods, and for the field to build cumulative knowledge. To address this pressing need, we conduct a systematic review of Human-AI (HAI) co-creativity measurement metrics and evaluative methods. Through a critical analysis of how creativity is conceptualized and measured, we synthesize existing evaluation practices into a taxonomy for interpreting HAI co-creativity research. This taxonomy provides a multidimensional framework for interpreting how creativity is operationalized across the literature. By mapping these measurement patterns and gaps, this study ultimately offers researchers and practitioners a deeper understanding of how creativity can be meaningfully assessed in Human-AI co-creative tasks and provides a starting point for the development of more rigorous and context-sensitive measurement strategies for HAI co-creativity and cognitive support.