A new study from the Université de Montréal has measured something researchers thought would take much longer to happen: AI systems now perform as well as the average person on certain creativity tasks.
Professor Karim Jerbi and his team tested several major language models—ChatGPT, Claude, Gemini, and others—against data from 100,000 human participants using the Divergent Association Task (DAT), a standard psychology test that asks people to generate ten words as different in meaning as possible from each other. Some AI systems, particularly GPT-4, scored higher than the median human score.
But here's the part that matters: the most creative half of humans still outperformed all the AI systems tested. And the gap widened dramatically at the top. The most creative 10 percent of people scored substantially higher than any AI.
We're a new kind of news feed.
Regular news is designed to drain you. We're a non-profit built to restore you. Every story we publish is scored for impact, progress, and hope.
Start Your News Detox"This result may be surprising — even unsettling," Jerbi acknowledges in the study, published in Scientific Reports. "But our study also highlights an equally important observation: even the best AI systems still fall short of the levels reached by the most creative humans."
What this actually means
The researchers didn't stop at word association games. They tested whether AI creativity on that basic task translated to more complex work: writing haiku, creating movie plot summaries, drafting short stories. The pattern held. AI sometimes beat average humans. The skilled creators among humans consistently won.
This matters because it reframes a conversation that often gets stuck in "humans versus machines." The real story is more textured: AI has crossed a threshold where it's genuinely creative by human standards—just not exceptional creativity. That distinction changes what the technology can actually do.
The study also discovered that AI creativity isn't fixed. The researchers found two levers that shape how inventive an AI becomes. The first is "temperature," a technical setting controlling how predictable or adventurous the model's responses are. The second is prompt engineering—how you ask the question. Instructions that encourage the AI to consider the origins and structure of words led to more unexpected ideas and higher creativity scores.
That's the part that hints at the actual future. Jerbi puts it plainly: "Generative AI has above all become an extremely powerful tool in the service of human creativity. It will not replace creators, but profoundly transform how they imagine, explore, and create."
The competition narrative—AI versus human creators—was always the wrong frame. The research suggests the real shift is toward collaboration: AI handling the expansion phase (generating many ideas, exploring variations, combining concepts), humans handling the judgment phase (which ideas matter, what's worth developing, what's true). That's not replacement. It's augmentation.
The future of creativity may depend less on who's better and more on how they work together.
