Probing the Creativity of Large Language Models: Can models produce divergent semantic association?

TL;DR

This study evaluates the creative potential of large language models using a semantic association task, revealing that models like GPT-4 and GPT-3.5-turbo can surpass human creativity levels under certain decoding strategies.

Contribution

It introduces the use of the divergent association task to measure creativity in large language models and compares different models and decoding methods.

Findings

GPT-4 exceeds 96% of humans in semantic association tasks.

GPT-3.5-turbo surpasses average human performance.

Stochastic sampling enhances creativity but reduces stability.

Abstract

Large language models possess remarkable capacity for processing language, but it remains unclear whether these models can further generate creative content. The present study aims to investigate the creative thinking of large language models through a cognitive perspective. We utilize the divergent association task (DAT), an objective measurement of creativity that asks models to generate unrelated words and calculates the semantic distance between them. We compare the results across different models and decoding strategies. Our findings indicate that: (1) When using the greedy search strategy, GPT-4 outperforms 96% of humans, while GPT-3.5-turbo exceeds the average human level. (2) Stochastic sampling and temperature scaling are effective to obtain higher DAT scores for models except GPT-4, but face a trade-off between creativity and stability. These results imply that advanced large language models have divergent semantic associations, which is a fundamental process underlying creativity.