Empirical Study of Problem-solution Co-evolution in Human-GAI Collaborative Conceptual Design

The problem-solution co-evolution model is a foundational framework for under-standing the emergence of creativity in both individuals and teams. With the ad-vent of Generative Artificial Intelligence (GAI), a new paradigm of co-creation in conceptual design has arisen. However, the dynamics inherent to human-GAI col-laborations remain largely unknown. In our investigation of the co-evolution dy-namics of human-GAI interaction, we employed retrospective protocol analysis to examine the verbal reasoning processes of twenty novice designers co-designing with GAI (text-to-text and text-to-image models). Drawing from the outcomes of our creativity assessments, a key revelation emerged: GAI has the potential to amplify team creativity by fostering human abductive reasoning. In further discourse, we introduce a novel human-GAI co-evolution model, which elucidates the significant role of GAI in aiding human problem-framing explora-tion. Central to our exploration, we spotlight "introspection" and "retrospection" as pivotal constructs in probing human-GAI collaborations.