Stimulating Brainstorming Activities with Generative AI in Higher Education

Developing teamwork skills and creativity is becoming increasingly important in higher education. Brainstorming activities can support the acquisition of such transversal skills by encouraging open dialogue in collaborative design processes. In a successful brainstorming, participants propose diverse ideas, combine them, and elaborate new propositions by taking inspiration in the propositions of their teammates. However, the cognitive effect of idea fixation may hinder this process. Idea fixations happens when participants focus on a limited set of ideas and are unable to think about unrelated alternatives. This effect lowers the creativity of the group and prevents the participants to consider diverse aspects of the problem at hand, which is an important mechanism in design-based learning. Leveraging the recent development in Large Language Models (LLM), we propose to integrate a Generative-AI agent to introduce new ideas into a brainstorming process supported by a digital application. Such agent can be designed to bring new topics to be considered that the students may not have brought up themselves. Our study aims to examine how such AI-generated ideas may influence the ideas proposed anonymously by the participants, and ultimately, the outcome of the activity. We use a quasi-experimental design with two conditions: one in which only the students generate ideas (control), another where they receive AI-generated ideas along with their own. We conduct this experiment as a part of a design thinking project carried out by computer science students. Using a qualitative coding combined with Epistemic Network Analysis, we investigate the dissemination of the topics brought by the Generative-AI agent within the ideas of the participants. We also present and discuss practical aspects of the intervention and its implementation by educators in their teaching. Our findings are contributing to the understanding of humanAI interactions in the context of collaborative ideation in educational settings. They can inform the design of new interaction mechanisms with AI agents to support creativity. Furthermore, by introducing previously unconsidered topics to these activities, these AI-based agents can support groups of students in considering a wider range of aspects of the matter at hand, supporting the development of critical thinking. These mechanisms may be useful to educators and facilitators looking for solutions to overcome idea fixation and increase creativity in brainstorming.