Readers rate AI-generated short stories higher than human ones until they learn a machine wrote them
A study involving over 2,500 participants found readers could not distinguish ChatGPT-generated stories from human-written ones. AI texts received higher ratings initially, but scores fell once participants knew the source was artificial.
Researchers conducted an experiment to test human ability to identify AI-authored text. Participants evaluated short stories without knowing the origin, comparing outputs from ChatGPT against human writers.
The data indicates that participants performed no better than random chance when attempting to identify the AI-generated pieces. Interestingly, the machine-written stories were initially rated higher on quality metrics than the human-authored counterparts.
This highlights the growing sophistication of large language models in creative writing tasks. It suggests that AI-generated content may become indistinguishable from human work in certain contexts, raising questions about attribution and value.
However, the study also revealed a bias effect. Once participants were informed that a machine produced the text, their ratings dropped significantly. This underscores the psychological impact of knowing the source behind creative work.
This page provides an editorial summary based on publicly available information. It is not a republished article. Use the source link below for the original report.