U.S. Adults Prefer AI-Generated Stories Over Human Writing

A study published in the journal Judgment and Decision Making indicates that adults frequently prefer AI-generated short stories over human-authored ones. Researchers from Villanova University tested 1,682 representative U.S. adults using narratives produced by ChatGPT 4.0 and authors from a literary journal. Participants generally rated the AI stories as higher quality and more absorbing, though they gave the highest ratings to AI stories they were falsely told were written by humans.

Participants struggled to distinguish the sources, with correct identification rates between 39.9% and 52%, a performance described as no better than chance. Dr. Deena Skolnick Weisberg noted that the preference for AI text likely stems from its clarity and predictability, whereas human writing tends to be more complex and subtle.

Professor Luke Kennard criticized the findings, arguing that AI success relies on the exploitation of stolen human databases and poses an existential threat to creativity. He described AI as predictive code physically hosted in destructive datacenters with catastrophic local repercussions.