Refining Ideas with Generative AI: How Textual or Visual Boundary Objects Influence Idea Refinement
(A2025-126027)

AUTHORS

Philipp Gordetzki, University of St.Gallen; Ivo Blohm, University of St.Gallen; Melanie Clegg, University of Lausanne; Felix Schakols, University of St. Gallen; Reto Hofstetter, University of Lucerne

ABSTRACT

Collaboration is crucial to creativity, particularly in refining raw ideas into innovative solutions. With generative AI (genAI), ideators collaborate with both humans and AI systems like large language models (LLMs) and image-generation tools. GenAI aids idea generation and refinement, but its success depends on boundary objects (BOs)—shared materials that contextualize AI outputs. This research examines the impact of textual and visual BOs on creativity during idea refinement. Drawing on BO and mental outcome simulation theories, we hypothesize that textual BOs outperform visual BOs by fostering mental outcome simulation. Across three studies, including surveys and experiments, we find textual BOs enhance creativity more effectively than visual BOs. These findings contribute to ideation, human-AI collaboration, and BO theory, offering guidance for designing genAI tools to enhance collaborative creativity.