Characters generated by artificial intelligence lack the mystery and ambiguity that make human-written fiction memorable, according to a study from researchers at the University of North Carolina at Chapel Hill. The finding arrives as more novelists, screenwriters, and creative professionals turn to AI tools to draft storylines and dialogue, raising questions about what gets lost when machines help shape narrative.
Measuring character depth with CASPER
The research team developed an automated framework called CASPER to analyze thousands of stories and measure eight character traits, including realism, evolution over time, and whether a character remains mysterious or fully understood by the end. Drawing on literary theory, the framework evaluated AI-generated fiction in a way no prior study had attempted at scale.
CASPER allowed the researchers to spot patterns that would be difficult to detect manually. The analysis showed that AI-created characters often land on familiar archetypes and close with tidy resolutions, while human authors more frequently leave questions unanswered or let their characters stay contradictory.
AI models 'play it safe' with neat endings
"We found that AI models tend to 'play it safe' with their characters, in the sense that they wrap up storylines neatly," said Anneliese Brei, a graduate student in computer science at UNC-Chapel Hill and lead author of the study. "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."
The study examined fiction produced by a range of language models and consistently found a preference for closure. Characters created by AI were less likely to show internal contradiction or resist easy interpretation, qualities that literary critics have long tied to lasting fictional works.
Bigger models don't fix the problem
Surprisingly, scaling up model size did not noticeably improve character variety. "One of our most surprising findings was that bigger and more powerful AI models don't necessarily create more varied characters than smaller ones," said Nicholas Sanaie, an undergraduate student in computer science and co-author of the study. "That tells us the challenge isn't just about scale. It's about how these models understand storytelling itself."
This insight suggests that simply training larger neural networks on more text will not automatically yield richer fictional people. The issue lies deeper-in how the models process narrative patterns and prioritize coherence over complexity.
A tool to track progress in AI storytelling
The research comes at a time when creative writing platforms such as Sudowrite and Squibler are gaining traction, and surveys show many fiction authors already use AI in their process. By offering a systematic way to benchmark character depth, CASPER could help developers judge whether newer storytelling systems truly improve on narrative complexity rather than just fluency.
"As more people collaborate with AI to write novels, screenplays and other creative works, we need ways to understand both what these systems do well and where they fall short," said Snigdha Chaturvedi, associate professor of computer science at UNC-Chapel Hill and senior author of the study. "CASPER gives us a lens for evaluating character depth and diversity, which can ultimately help developers build storytelling systems that better reflect the complexity of human experience."
Why this matters for writers
For working writers experimenting with AI tools, the study delivers a concrete message: AI can accelerate drafting and generate functional plot structures, but it still struggles with the ambiguity that makes characters feel real. A character who remains slightly unknown or refuses a tidy arc is often what stays with a reader-and that effect remains hard to automate.
The findings reinforce that the most compelling stories require a willingness to embrace uncertainty and contradiction, traits human authors bring naturally. Writers who use AI for Writers tools can treat them as capable assistants for brainstorming and first drafts, but should plan to revise heavily when it comes to character depth and resolution. Letting a character stay unresolved or contradictory may be the deliberate choice that keeps an audience thinking long after the last page.
Your membership also unlocks:
