Artificial intelligence is proving to be a formidable storyteller, with a new study revealing that readers rated AI-generated short stories as more engaging and higher in quality than those written by humans.
The study, published in the journal Judgment and Decision Making, involved 1,682 adults who each read one of six short stories. Three were written by human authors, while the other three were generated by ChatGPT using similar themes.
Participants were told whether the story they received was written by a human or AI, although the information provided was not always accurate. They were then asked to rate how absorbing, engaging and well-written they found the story.
Researchers found that stories generated by AI received higher overall ratings for quality and engagement compared to those written by humans. However, stories believed to be human-written also received more favourable ratings, suggesting readers still value human authorship.
Dr Deena Skolnick Weisberg, a senior researcher from Villanova University in Pennsylvania, said the findings show that AI is already capable of producing short fiction that many readers consider equal to or even better than human writing.
Despite the results, Weisberg stressed that AI should not replace human creativity.
She said people write not only to produce stories but also as a form of self-expression, to challenge themselves and to make sense of their experiences. While AI-generated novels may become more common, she believes there will always be room for human authors and collaborative AI-human writing.
The researchers also found that participants with more positive attitudes towards AI tended to give higher ratings to stories they believed were AI-generated.
In two additional experiments involving 905 participants, readers were asked to identify which of two similar stories had been written by a human and which by AI.
The results showed that participants performed little better than random guessing. In one experiment, only about 40 per cent correctly identified the AI-written story, while in another, accuracy reached just 52 per cent.
The researchers concluded there was no consistent evidence that readers could reliably distinguish between stories written by humans and those generated by AI.
While self-reported knowledge of fiction had little impact on participants’ accuracy, those with greater familiarity with AI technology were better at recognising AI-generated writing.
Commenting on the study, Professor Luke Kennard of the University of Birmingham, who was not involved in the research, warned against overlooking the broader concerns surrounding AI.
He argued that AI systems rely on vast datasets of existing human writing and questioned whether the technology’s environmental impact and use of copyrighted material justify its growing role in creative industries.
Kennard acknowledged that AI can produce coherent and enjoyable writing, but maintained that its development raises significant ethical and creative concerns that society must continue to address.
