Most people prefer stories written by artificial intelligence over those written by humans. That is the result of a study published this month and reported by New Scientist, in which researchers tested more than 1,600 adults across the United States.
Sydney Sears and Deena Weisberg at Villanova University in Pennsylvania recruited 1,682 adults selected to be representative of the US population by gender, age and race. Each participant read one story. Half read a story written by a human, sourced from a literary journal. The other half read a story generated by ChatGPT 4.0. Within each group, half were told the story was written by a human and half were told it was written by AI, regardless of the true origin.
The results showed a consistent preference for the AI-generated work. Participants who read an AI-generated story rated its quality at an average of 1.54 on a scale from negative 3 to 3. Those who read a human-written story gave it an average score of just 0.97. On a similar scale measuring how absorbing the story felt, AI-generated stories scored 1.42 compared to 1.0 for human-written ones.
In both cases, scores went slightly higher when participants were told the story was written by a human, regardless of whether it actually was.
The researchers also ran two additional studies in which participants were given one AI story and one human story and asked to identify which was which. In the first test, only 39 percent made a correct identification. In the second test, conducted several months later, 52 percent were correct. The researchers suggested the gap could reflect society slowly adapting to AI-generated text, but offered no definitive explanation.
Claire Hardaker at Lancaster University said the findings force an uncomfortable reckoning. "I think there's a huge sensitivity [around AI generating art]," she said. "When you have AI that is continually improving at the speed that it's improving at, there is going to come a point where every single model can produce Shakespeare-like output. What does that mean for humans?"
Hardaker noted that in her own research on identifying AI-generated speech, text and music, it is AI-generated music that produces the strongest emotional reaction when people discover they were tricked.
The study arrives as AI-generated creative work draws increasing backlash. Earlier this year, a short story awarded the Commonwealth prize and published in Granta was accused by some of being AI-generated, a charge the competition denied. The Villanova findings suggest that negative reputation for AI writing may not match what readers actually experience when they encounter the work without knowing its origin.
