AI can write a competent story, but new research shows it consistently removes the ambiguity that makes fiction memorable — and making the models larger does not help.
The Research
Researchers at the University of North Carolina at Chapel Hill developed an automated evaluation framework called CASPER to compare how characters in AI-generated stories stack up against those written by humans. The study, led by graduate student Anneliese Brei in computer science, analyzed thousands of stories across eight distinct axes drawn from literary theory — dimensions like whether a character seems realistic or exaggerated, whether they evolve over time, and whether they remain mysterious or fully explained by the end.
The team found that AI models systematically "play it safe." They wrap up storylines neatly, answer every mystery, and force characters into flat, predictable archetypes. Human authors, by contrast, routinely leave questions unanswered, keep characters morally gray, and allow them to remain unresolved — the kind of structural ambiguity that makes a story stick with a reader.
"We found that AI models tend to 'play it safe' with their characters, in the sense that they wrap up storylines neatly," Brei said. "Human writers, on the other hand, are sometimes more willing to leave questions unanswered and let characters remain mysterious. That difference matters because ambiguity is often what makes a story linger with a reader."
One of the study's most striking findings concerns scale. "The illusion of scale" refers to the revelation that scaling up parameter size does not solve the problem. Massive, state-of-the-art flagship language models generated characters just as flat and archetypal as smaller, less complex models. The deficit is rooted in how models understand storytelling, not in processing power. The results arrive as AI writing tools like Sudowrite and Squibler gain traction in publishing, film, and television.
Why It Matters
This research is not just for novelists. It touches on something fundamental about human cognition: we are drawn to unresolved questions. Ambiguity forces the brain to fill gaps, weigh evidence, and hold contradictory ideas at once — a core feature of flexible thinking. When AI consistently resolves every mystery, it flattens the cognitive workout that fiction provides. Understanding this bias also helps you recognize it: if your own thinking always seeks a tidy resolution, you may be skipping the ambiguity that leads to deeper insight. The AI pattern mirrors a human shortcut we should all notice.
What You Can Do
- Use AI as a brainstorming partner, not a final author — deliberately rewrite tidy endings into open ones.
- When reading or writing fiction, pause on moments you want to resolve immediately and ask what would happen if you let them stay uncertain.
- Train your own tolerance for ambiguity by reading stories with unresolved endings and sitting with the discomfort before rushing to a conclusion.
- If you use AI writing tools, test characters for contradiction. If they all feel "safe," revise.
For anyone curious about their own cognitive flexibility, the questions we ask ourselves matter. Exercises that stretch your tolerance for ambiguity and contradiction can be part of a broader brain training routine.
Source: Neuroscience News
Curious about your own brain? Take our free adaptive IQ test or try 306 brain training levels.