AI can now outperform the average human on certain creativity tests, according to a large new academic study that directly compared human and machine creative performance at unprecedented scale.
The research, published in Scientific Reports in January, analyzed creativity scores from more than 100,000 human participants and compared them with outputs from leading large language models, including ChatGPT, Claude and Gemini.
The study was led by Professor Karim Jerbi of the Université de Montréal, with participation from deep learning pioneer Yoshua Bengio.
The researchers find that some AI models exceeded average human performance on specific, well-defined creativity tasks, marking what the authors describe as a clear turning point for generative AI. At the same time, the most creative humans continued to outperform every AI system tested, often by a wide margin.
Says Jerbi,
“Our study shows that some AI systems based on large language models can now outperform average human creativity on well-defined tasks. This result may be surprising, even unsettling, 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.”
To compare creativity across humans and machines using the same framework, the team relied primarily on the Divergent Association Task, a widely used psychological test that measures divergent thinking. Participants are asked to generate ten words that are as unrelated in meaning as possible, with higher scores awarded for originality and semantic distance.
Some AI systems, including GPT-4, scored above the human average on this task. But when researchers focused on the most creative people, the results flipped. The top half of human participants outperformed all AI models, and the gap widened further among the top 10% of scorers.
Follow-up experiments extended the comparison beyond word association. Researchers evaluated humans and AI systems on creative writing tasks such as haiku composition, movie plot summaries and short stories. The same pattern emerged. AI sometimes matched or beat average human output, but consistently lagged behind skilled human creators.
The study also examined whether AI creativity is fixed or adjustable. Researchers found that creativity levels varied significantly depending on model settings, particularly temperature, which controls how predictable or exploratory responses are. Prompt design also played a major role, with instructions that encouraged etymological or structural thinking, leading to more original outputs.
Jerbi says the findings challenge simplistic narratives about AI replacing human creativity. Instead, he says AI should be viewed as a tool that reshapes how creative work is done rather than eliminating the need for human creators.
“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 for those who choose to use it.”
Disclaimer: Opinions expressed at CapitalAI Daily are not investment advice. Investors should do their own due diligence before making any decisions involving securities, cryptocurrencies, or digital assets. Your transfers and trades are at your own risk, and any losses you may incur are your responsibility. CapitalAI Daily does not recommend the buying or selling of any assets, nor is CapitalAI Daily an investment advisor. See our Editorial Standards and Terms of Use.