Reproduce “Enhancing Creativity in Large Language Models through Associative Thinking Strategies” 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
Pronita Mehrotra, Aishni Parab, Sumit Gulwani
Researchers from MindAntix, UCLA, and Microsoft explored whether associative thinking strategies, inspired by human cognition, could enhance creativity in Large Language Models like vGPT-4. Their method, involving explicit instruction to connect unrelated concepts, increased originality by 40% in product design and 28% in storytelling, while the impact on usefulness varied by domain.
Key insights
- Human-inspired cognitive strategies, specifically associative thinking, can effectively enhance the creative capabilities of Large Language Models.
- While associative thinking markedly boosts originality in LLM outputs, its impact on usefulness is nuanced and varies across creative domains.
- LLMs exhibit limitations such as 'context dependency' and 'associative rut' when attempting to form remote associations, mirroring challenges in human creative processes.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.