Understanding How AI-Assisted Evaluation Influences Innovative Processes, Collaboration, and Creative Confidence in Design Education

Abstract

Generative AI is increasingly embedded in design practice, yet its effects on convergent stages such as concept screening and selection remain underexplored in team settings. This study examines how a transparent AI co evaluator shapes screening practices, collaboration, and creative confidence in design education. We conducted a mixed-methods workshop where design student teams ranked ideas, then reconsidered choices after AI feedback. ProSight is a co-evaluator that returns rubric scores with brief evidence-linked rationales from each concept’s image and description. Teams reviewed the feedback, discussed agreement and conflict, and finalized selections. Data sources included ranking trajectories, pre post surveys on trust in AI, teamwork collaboration, and creative confidence, as well as post task discussions and interaction logs. Across groups, AI input rarely replaced initial preferences, but it shifted how options were compared and justified by making evaluative criteria more explicit and surfacing considerations that were less salient in early deliberation. Participants treated the AI as a credible reference for reflection and justification, but verified suggestions against intent and constraints, and kept authority over trade-offs and final decisions. The findings clarify how explanation rich AI can support reflective screening while reshaping collaborative dynamics in time limited group decision making.

Keywords: Human–AI Collaboration, Transparent AI, Team Decision-making, Design Education, Creativity Support Tools

DOI: 10.54941/ahfe1007380