How Do Design Students Use ChatGPT When Ideating?

Abstract

This paper explores how design students use Generative AI (GAI) tools, in particular ChatGPT, during the ideation phase of a design process. Although design students use GAI tools to support ideation, little is known about how they apply them and how this relates to their creativity. This explorative study describes how design students interact with ChatGPT during ideation and considers what this might mean for teaching creativity in design education. The study took place within an existing project course and analysed the individual concept generation of 44 User Experience Design students. The dataset includes ChatGPT chatlogs, three resulting concepts per student, and short self-reported descriptions of tool use. Two analyses were conducted: a qualitative content analysis of the chatlogs and a comparative analysis of the concepts developed from the generated output.

These findings show that ChatGPT-supported ideation is often initiated through broad, task-driven prompts with limited output steering. Iterative interaction occurs, but is not the default mode. At the concept level, generated content often informs concepts directly, with purpose, form, and meaning more frequently retained, and interaction more often adjusted. The most common configuration consists of concepts in which all four dimensions remain the same, indicating direct transfer of generated content. Alignment analysis further shows that concepts often remain more closely aligned with generated output than students’ descriptions suggest.

Taken together, these findings indicate that AI-supported ideation unfolds through differing configurations of interaction and concept development. The prevalence of limited steering and direct adoption suggests a lack of consistent creative engagement. At the same time, instances of more iterative use and loosely inspired concepts show that alternative forms of engagement do occur. This highlights the need for more deliberate and reflective engagement with generative tools and a crucial role for design education in teaching and supporting this type of engagement.