Cross-domain mapping boosts human creativity, LLMs show mixed results

A new research paper explores the impact of cross-domain mapping on creativity in both humans and large language models (LLMs). The study found that humans consistently benefit from drawing inspiration from unrelated source domains, while LLMs, on average, produce more original ideas and do not show a significant effect from this intervention. However, both humans and LLMs showed increased creativity when the inspiration source was semantically distant, with the most capable LLMs benefiting from cross-domain mapping above a certain distance threshold. The research also highlighted differences in how humans and LLMs utilize source concepts, with humans transferring surface features and LLMs transferring structural and functional properties. AI

IMPACT Investigates how LLMs approach creativity and how interventions affect their output, offering insights into AI's creative potential.

RANK_REASON Academic paper on AI capabilities. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• arXiv
• Hugging Face
• large language models
• Qiawen Liu

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →