New research reveals people automatically downgrade content they believe is AI-generated, even when humans wrote it.
Why it matters: As AI writing tools become more sophisticated, understanding public perception and bias against AI-created content is crucial. This research exposes a significant credibility gap that could impact AI adoption across industries.
- The study comes as organizations increasingly explore AI for content creation and storytelling.
Key finding: ChatGPT-written stories performed nearly as well as human-written ones, but people rated stories poorly when told they were AI-generated – regardless of actual authorship.
“People don’t like when they think a story is written by AI, whether it was or not.”
Haoran “Chris” Chu, Ph.D., professor of public relations at University of Florida
The process:
- Researchers created two versions of identical stories – one AI-written, one human-written
- Participants rated engagement levels while researchers manipulated story attribution
- Study measured “counterarguing” and “transportation” elements
Keep in mind: AI still lags behind humans in creating truly engaging narratives that transport readers into the story world.
Real-world impact: The findings have significant implications for:
- Public health communications
- Content marketing
- Creative industries
- Public trust in AI-generated content
- The bias against AI authorship suggests organizations may need to carefully consider how they disclose AI use in content creation.
TL;DR
- AI can write coherent and logical stories that nearly match human quality
- People show strong bias against content they believe is AI-generated, regardless of actual authorship
- The findings could impact how organizations implement and disclose AI use in communications
Read the Paper
Can AI tell good stories? Narrative transportation and persuasion with ChatGPT