Reproduce “A Framework for Collaborating a Large Language Model Tool in Brainstorming for Triggering Creative Thoughts” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Hung-Fu Chang, Tong Li
Researchers from the University of Indianapolis and Ball State University developed the GPS (Goals, Prompts, Strategies) framework, which guides the use of large language models to stimulate creative thought during brainstorming sessions. The framework led to a 385.71% increase in idea fluency, 165.71% in flexibility, 192.86% in originality, and 46.43% in elaboration when compared to unstructured prompting.
Key insights
- The structured GPS framework significantly enhances the overall creativity of LLM outputs across multiple dimensions, moving beyond mere accuracy.
- Applying the GPS template, which involves structuring prompts with clear roles, contexts, and objectives, provides a substantial initial boost in idea fluency, flexibility, originality, and elaboration.
- Specific strategies, such as assigning an 'Unconventional Role,' can effectively trigger unexpected and highly original ideas, sometimes trading idea quantity for increased quality and unique perspectives.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.