TL;DR
AI fiction is easy to spot because it spells everything out — the theme, the emotion, the moral — while human storytellers instinctively reach for complexity and ambiguity.
Read enough AI prose and you can smell it: em-dashes everywhere, and that familiar 'it's not x, but y' construction. A team led by Jenna Russell at the University of Maryland had Claude, DeepSeek, Gemini, GPT and Kimi each write a 5,000-word short story, asked humans to write ones of similar length, then compared the two sets across 304 criteria.
AI is afraid you won't get it
Seventy-seven per cent of the AI stories state their theme outright, against 52 per cent of human ones. The endings work the same way: AI spells out the lesson, humans let the reader draw it.
With emotion, AI is blunter, leaning on physical sensations and bodily metaphors — 81 per cent versus 38 per cent. It also turns the setting into a direct metaphor for a character's inner state. Ham-fisted, in short.
Human stories are messier
AI plots run in a straight, causal line with few detours. Humans prefer ambiguous endings, time jumps and causal chains that break and rejoin.
Seventy-nine per cent of AI stories have no subplots, against 57 per cent of human ones. Human stories visit more places, carry more dialogue, and let subplots converge on one larger theme. More complex, overall.
Score stories across these features and the AI ones cluster in the middle of the rarity distribution, while human stories spread into the rarer stretches. There are bland human stories too — just far fewer of them.
Writers instinctively want their stories to be interesting, so they put interesting things in them. AI just aims for the middle.
Curated from high-quality sources, with concise summaries and key takeaways.
