Divergent_Creativity_in_Humans_and_Large_Language_

2024

Open publication workspace · Sign in to read the full PDF.

AI-generated summary

1) This paper systematically evaluates and benchmarks the divergent creativity of state-of-the-art large language models against a large cohort of 100,000 humans, demonstrating that models like GPT-4 can match or surpass human performance in specific semantic association and creative writing tasks.
2)
- Evaluates various LLMs against 100,000 human participants using the Divergent Association Task (DAT), Divergent Semantic Integration (DSI), and Lempel-Ziv complexity.
- Examines how prompt engineering strategies (such as etymology and meaning opposition) and hyperparameter tuning (temperature) modulate LLM creative outputs.
- Benchmarks performance across diverse creative writing formats including haikus, movie synopses, and flash fiction.
3) Tags: Artificial Intelligence, Creativity, Large Language Models, Semantics, Divergent Thinking

Check the original publication for accuracy and context.