A Systematic Review and Empirical Framework for Human-AI Co-Creation in the Conceptual Design Process

Generative Artificial Intelligence (GenAI) is rapidly transforming creative practice, particularly within the conceptual design process, by augmenting human creativity and enhancing design productivity. Despite widespread adoption, fundamental questions persist about how GenAI influences design cognition, collaboration, and, critically, human agency. This paper synthesizes findings from a systematic literature review of GenAI's role in conceptual design with original empirical evidence drawn from two qualitative studies (Study 1, n = 6; Study 2, n = 7) exploring how individuals experience creative agency during co-creative tasks involving AI-generated images and text. Using think-aloud protocols and semi-structured post-task interviews, we identify four central dimensions of creative agency in human-AI collaboration: creative self-efficacy, control over creative action, autonomy in the creative process, and ownership of the creative product. We further identify self-regulatory and metacognitive mechanisms—including progressive refinement, selective appropriation, and counter-inspiration—that users employ to sustain agency when navigating unpredictable AI outputs. Building on these findings and anchored in the Co-Creative Framework for Interaction Design, we propose the Generative AI Enhanced Conceptual Design (GAECD) Framework, which delineates roles, responsibilities, and interaction modes for effective human-AI co-creation. This paper contributes both a critical evaluation of the current state of GenAI-human collaboration and a practical roadmap for designers, developers, and researchers seeking to harness GenAI's full potential while preserving meaningful human creative agency.