Exploring the Boundaries of Creativity: AI vs. Human Innovation in a Groundbreaking Study

Can AI Truly Create Original Ideas? A Deep Dive into Creativity and Generative AI

The question of whether generative artificial intelligence systems, such as ChatGPT, possess the capability to create genuinely original ideas has sparked considerable debate. A landmark study led by Professor Karim Jerbi from the Université de Montréal, in collaboration with renowned AI researcher Yoshua Bengio, dives deep into this complex issue, presenting findings from the largest direct comparison between human creativity and that of large language models to date.

A Turning Point in AI Creativity

Published in Scientific Reports (Nature Portfolio), the study suggests a significant shift in the landscape of creativity. For the first time, it appears that generative AI systems have reached a level of creativity that allows them to outperform the average human in certain well-defined tasks. However, it’s important to underscore that the most creative individuals still demonstrate a clear advantage over the highest-performing AI models.

Evaluating Creativity: The Study’s Framework

Jerbi and his team evaluated various leading AI models, including ChatGPT and others, against responses from over 100,000 human participants. Their findings highlight a pivotal moment: some AI systems, particularly GPT-4, exceeded the average scores of humans on tasks measuring divergent linguistic creativity.

"Our study shows that some AI systems based on large language models can now outperform average human creativity on well-defined tasks," says Jerbi. However, this should not overshadow the fact that peak creativity remains a distinctly human trait. When exploring the top half of the most creative human participants, their average scores eclipsed those of the AI models, widening further among the top 10 percent of creatives.

Understanding Creativity Measurement

To ensure a fair evaluation of creativity across both humans and AI, the study employed multiple methodologies. One of the primary instruments was the Divergent Association Task (DAT). In this widely recognized psychological test, participants—whether human or AI—are asked to list ten words that are as unrelated as possible in meaning. A particularly creative response might include words like "galaxy, fork, freedom, algae, harmonica, quantum, nostalgia, velvet, hurricane, photosynthesis."

Success in this task correlates strongly with other established creativity assessments in writing, idea generation, and problem-solving, showcasing that creativity transcends just vocabulary. Moreover, it’s a time-efficient and accessible task, taking only a few minutes to complete.

From Simple Tasks to Complex Creativity

After the initial tests, researchers sought to determine whether the AI’s performance on simple word associations could translate to more complex creative endeavors. They compared AI systems and humans in creative writing tasks—like composing haiku, writing movie plot summaries, and crafting short stories.

The results echoed the earlier findings: while AI sometimes matched the performance of average humans, the most skilled human creators consistently produced stronger and more original work.

The Flexibility of AI Creativity

The implications of this research raise interesting questions about the nature of AI creativity. Can it be molded? The study indicates that AI creativity isn’t fixed; it can be fine-tuned through various technical adjustments, particularly the model’s "temperature." Lower temperature settings lead to more predictable outputs, while higher settings encourage more adventurous and varied responses.

Additionally, the ways in which instructions are phrased greatly impact AI creativity. Prompts that invite exploration of word origins and structures can yield higher creativity scores. Such findings underscore the vital role of human direction in shaping AI creativity, suggesting that collaborative interaction is key.

A Symbiotic Relationship: AI and Human Creators

The prospect of AI as a replacement for human creators is commonly debated. However, this research provides a nuanced perspective. While AI can match or even exceed average human creativity on certain tasks, it remains reliant on human input.

"Even though AI can now reach human-level creativity on certain tests, we need to move beyond this misleading sense of competition," says Jerbi. The findings suggest that generative AI is not so much a competitor but a formidable tool that can transform the creative process. Rather than heralding an end to creative professions, we might be on the brink of a new era where AI aids in amplifying human imagination and exploration.

Concluding Thoughts

As we grapple with the definition of creativity in light of evolving technology, studies like Jerbi’s invite us to reconsider our understanding of creativity itself. By challenging and comparing the capabilities of both humans and machines, the research not only sheds light on the strengths and limitations of generative AI but also highlights the potential for a collaborative future.

The paper titled "Divergent creativity in humans and large language models," published on January 21, 2026, brings together experts from Université de Montréal, Université Concordia, University of Toronto Mississauga, Mila (Quebec AI Institute), and Google DeepMind, marking a significant milestone in the intersection of psychology, creativity, and artificial intelligence.

As we continue to explore these themes, one thing is clear: the journey of understanding creativity in humans and machines is just beginning, and the possibilities for collaboration are vast.