The role of task breadth and cognitive flexibility in generative AI-supported creative idea generation

This paper investigates how, and under what conditions, generative artificial intelligence (AI) systems, particularly fine-tuned large language models (LLMs), stimulate cognitive flexibility and enhance the flexibility pathway to creativity. Drawing on the dual pathway to creativity model, we examine the role of task breadth in facilitating creative idea generation. Through two randomized experimental studies, we found that fine-tuned LLM-generated stimuli significantly enhance breadth of exploration, which is a consequence of cognitive flexibility, when participants engage in exploration within a broad domain (Study 1). In contrast, no significant effect is observed within a constrained domain (Study 2). These findings highlight the critical interaction between generative AI and task structure in driving creativity. Our study contributes to a deeper understanding of how AI systems reshape cognitive processes, providing theoretical and practical insights into the evolving nature of creativity in the era of generative AI