ChatGPT Stories Better Than Humans — Unless You Know They're Not

A recent study found that stories written by ChatGPT 4.0 were rated higher than those appearing in elite literary journals (1.54 vs. 0.97 for quality; 1.42 vs. 1.00 for immersion). Participants in this experiment (n=2,500) weren’t able to tell the difference between AI-written and human-written stories (they performed at chance levels).

However, once participants learned that some of the stories had been authored by AI, the ratings for those stories dropped precipitously. Thus, it seems that people don’t dislike AI-written stories because they’re bad, they just dislike them because they’re AI-written. And sooner than you might imagine, the boundary between literary fiction and AI-generated stories will crumble.