AI in creative brainstorming: between idea expansion and the risk of herding

Abstract

The incorporation of generative artificial intelligence into creative ideation has reconfigured contemporary brainstorming practices, particularly in design, communication, innovation, and higher education. Although large language models can increase ideational fluency, accelerate the production of alternatives, and reduce initial cognitive barriers, recent literature highlights an ambivalent effect: the same ability to generate fast, plausible and formally coherent ideas may induce premature convergence, cognitive fixation and collective homogenization. This article presents a critical literature review on the relationship between artificial intelligence, divergent creativity and the risk of herding in brainstorming processes. Drawing on classical research on group creativity, idea generation, fixation, divergent thinking and creativity support systems, the review connects the contributions of Osborn, Guilford, Amabile, Paulus, Nijstad, Runco, Boden, Lubart and Shneiderman with recent studies on generative models, AI-assisted creativity and reduced semantic diversity. The review argues that AI should not be understood merely as a tool for creative expansion, but as a techno-cognitive agent that intervenes in attention, semantic exploration and ideational autonomy. An interpretive model is proposed around four tensions: expansion/convergence, assistance/dependence, diversity/homogenization and autonomy/herding. The article concludes that the academic and practical value of AI in brainstorming depends less on its capacity to produce ideas and more on the pedagogical, methodological and critical protocols governing its use.