Groups of humans or crowds can be remarkable when coming up with ideas. However, not everyone has a group of humans at their disposal to brainstorm. With recent advances in AI, however, generative large language models (LLM) might be capable of contributing ideas in a brainstorming session, turning individual work of a human into joint work of human and AI. It is, however, unclear, how group effects known from human brainstorming groups transfer to such a human-AI setting. In our mixed-method study (qualitative emphasis) with 24 participants, we investigate how a human brainstorms together with the generative LLM ‘GPT-3’, and how they perceived their experience. Our results highlight known effects like cognitive stimulation but also a risk of free riding. We thereby contribute to the understanding of how generative AI, which is becoming broadly available, can be used to address the challenge of human-AI collaboration for solving open-ended problems.
Memmert, Lucas and Tavanapour, Navid, "TOWARDS HUMAN-AI-COLLABORATION IN BRAINSTORMING: EMPIRICAL INSIGHTS INTO THE PERCEPTION OF WORKING WITH A GENERATIVE AI" (2023). ECIS 2023 Research Papers. 429.