Artificial intelligence-generated short stories may be viewed as more engaging and of higher quality than those authored by humans, according to a recent study. Researchers from Villanova University in the United States conducted experiments involving more than 2,500 adult participants to assess perceptions of AI-written fiction compared to human-written works.
In the primary study, 1,682 adults were randomly assigned to read one of six short stories, with three penned by human authors and three generated by the AI model ChatGPT. Each AI story shared a similar theme with a corresponding human-authored story. Participants were informed whether the story was written by a human or AI, although this information was occasionally incorrect. After reading, participants rated the stories based on how absorbing, engaging, and high-quality they found the narratives.
Results showed that stories created by AI received higher ratings for engagement and quality than human-written ones. However, participants gave higher scores to stories they believed were authored by humans, suggesting that knowledge of a story’s origin influenced their evaluation. Furthermore, individuals with more positive attitudes toward AI tended to give higher ratings to stories labeled as AI-generated compared to those who were more skeptical of the technology.
Two subsequent experiments involving 905 participants tested whether readers could reliably distinguish between AI- and human-authored stories when presented with pairs of narratives on the same theme. In one, only about 40% correctly identified the human-written story, a result below random chance, while in the other, 52% were correct. The researchers concluded there was no strong evidence that participants could consistently tell AI-generated writing apart from human work. Familiarity with AI systems corresponded with somewhat better recognition accuracy, although self-reported expertise in fiction reading did not.
Dr. Deena Skolnick Weisberg, senior author of the study, noted that AI can already produce short stories that match or exceed human efforts in perceived quality but emphasized that this does not diminish the value of human creativity. She suggested that AI-generated novels and collaborative human–AI writing might become a growing presence in literature, but the unique human motivation to write—as a form of self-expression, challenge, or making sense of experience—remains significant.
Not all responses to the research have been positive. Luke Kennard, a professor of film and creative writing at the University of Birmingham who was not involved in the study, warned of potential harms associated with AI-generated text. He highlighted that AI models function as predictive code built upon vast datasets, often sourced without explicit consent, and supported by resource-intensive data centers with environmental and social repercussions. Kennard cautioned against viewing AI simply as a benign tool, describing it instead as an existential threat to creative industries and questioning whether the trade-offs are justified.
The study's findings illustrate the evolving landscape of creative writing in the age of artificial intelligence, underscoring both opportunities and challenges as AI-generated content becomes increasingly sophisticated and pervasive.