AI-generated stories rated better quality than human-written ones, study finds
Key Points:
- A study published in Judgment and Decision Making found that AI-generated short stories by ChatGPT were rated as more absorbing and higher quality than human-written stories, though participants preferred stories they believed were written by humans.
- Participants struggled to reliably distinguish between human and AI-authored stories, with correct identification rates ranging from 40% to 52%, indicating no strong ability to tell them apart.
- Positive attitudes toward AI correlated with higher ratings for AI-labeled stories, and expertise with AI systems improved participants' ability to recognize AI-generated writing.
- Researchers suggest AI-generated stories may be easier to read, but emphasize that human creativity and self-expression remain valuable and distinct from AI outputs.
- Critics like Professor Luke Kennard highlight ethical concerns, viewing AI writing as a contentious and potentially harmful technology that relies on extensive human-created data and poses existential threats to creative industries.