A 2024 study in Science Advances by researchers Anil Doshi and Oliver Hauser found that AI-assisted short stories were more creative, better written, and more enjoyable than those written without AI support. The catch: those stories were also significantly more similar to each other. Every writer got sharper. The collective got blander. Think of it like the Spotify algorithm — your personal recommendations are eerily good, but somehow every playlist in the office sounds identical by Friday afternoon.
The Creativity Trap
AI is helping you win the battle while losing the war on original ideas.
Doshi and Hauser describe this dynamic as a “social dilemma”: individuals benefit while collective novelty quietly erodes. A 2025 meta-analysis confirmed that human–AI collaboration outperforms humans working alone — but idea diversity drops sharply when AI enters the room. Analysts at MIT Sloan and Wharton have observed that AI-assisted teams produce strikingly similar language and overlapping concepts, reducing the variety that actually drives breakthrough thinking.
Here’s what the data shows:
- A study of 4 million text-to-image artworks found AI boosted creative productivity by roughly 25% and increased work value — measured as the likelihood of receiving a “favorite” per view — by roughly 50%
- ChatGPT users generate more original and useful ideas than non-users, but their answers converge more
- AI systems exhibit fixation bias, clustering outputs around conventional categories and making it harder for users to break toward genuinely novel solutions
“With generative AI, writers are individually better off, but collectively a narrower scope of novel content is produced.” — Doshi and Hauser, Science Advances, 2024
Every startup pitch deck now reads like it was trained on the same data. Because increasingly, it was.
When the Brain Checks Out
Heavy AI use may be idling the cognitive machinery that makes your thinking distinctly yours.
Psychologists writing in Frontiers in Psychology coined a term worth knowing: AI chatbot–induced cognitive atrophy (AICICA) — the gradual erosion of critical thinking, analytical reasoning, and creativity through overreliance on conversational AI. A separate neurophysiological study found that excessive AI reliance correlated with reduced brain network connectivity during tasks. Your brain, essentially, stops showing up to work.
The relationship isn’t binary:
- Moderate AI use shows minimal cognitive impact
- Excessive reliance correlates with lower critical thinking, weaker self-monitoring, and worse long-term retention
A field experiment at a technology consulting firm reinforced this: ChatGPT only boosted creativity ratings for workers with strong metacognitive habits — people who actively planned, reflected, and revised their approach. Those who simply accepted the output didn’t get the same lift.
The tool doesn’t make you robotic. Passivity does.
The same technology that helps you write better can quietly outsource the thinking that makes your writing yours. The research isn’t telling you to quit using AI. It’s telling you to notice who’s actually steering — and whether that’s still you.
