A Comparative Analysis of Variability in AI-Generated and Human-Written Text

This study investigates whether AI-generated text exhibits lower linguistic variability than human-written text under identical prompt conditions. Using quantitative linguistic metrics — word count, sentence structure, and lexical diversity — the research compares 172 human responses with 150 AI-generated responses to an open-ended prompt. Findings indicate that AI-generated responses demonstrate substantially lower variance and greater structural consistency, suggesting that statistical regularisation may be an important characteristic of large language model output.