The Co-Creative Design Framework for Hybrid Intelligence

With the rapid advancement of generative AI, co-creation has emerged as a key interaction paradigm, enabling humans and AI to collaborate in creative processes. However, despite decades of research on co-creativity, recent AI developments often lack a structured framework to integrate these insights effectively. To address this gap, we propose the Co-Creative Design Framework (CCDF), which formalizes human-AI co-creation through cognitive and interaction principles. The framework is structured around three core dimensions: agency, which defines the balance of autonomy and control between user and AI; interaction dynamics, which describe the evolving relationship between collaborators and their shared creative product; and communication, which governs information exchange between human and AI. The CCDF provides a systematic approach to modeling co-creative AI and hybrid intelligence systems, defining key dimensions of variance that shape the interaction space of co-creation. In particular, it highlights agency and interaction dynamics, which have been underexplored in recent co-creative AI frameworks. This paper details the iterative development of CCDF, synthesizing insights from co-creativity literature and AI research. We apply the framework in a comparative analysis of Traditional ChatGPT, ChatGPT Canvas Mode, and DALL-E, demonstrating its ability to capture fine-grained differences in system design and user experience.