AI Outperforms Humans in Creativity Tests, Study Reveals

A comprehensive study has revealed a significant milestone in the ongoing exploration of artificial intelligence and creativity. Researchers found that advanced generative AI systems, such as GPT-4, can outperform the average human in specific creativity assessments, marking a notable shift in how creativity is measured and understood. The study, led by Professor Karim Jerbi from the Université de Montréal, involved more than 100,000 human participants and is considered the largest direct comparison between human and AI creativity to date.

While the findings indicate that some AI models can exceed average human scores on tasks designed to assess divergent thinking, the most creative individuals—particularly the top 10%—still maintain a clear advantage in more complex creative endeavors. Tasks such as poetry and storytelling remain areas where human creativity shines.

AI Matches Average Human Creativity

The research evaluated various leading large language models, including ChatGPT, Claude, and Gemini. The results signify a turning point in AI capabilities. For instance, tasks aimed at measuring divergent linguistic creativity showed that models like GPT-4 frequently surpassed typical human responses.

“Our study shows that some AI systems based on large language models can now outperform average human creativity on well-defined tasks,” Professor Jerbi stated in a research brief. “This result may be surprising—perhaps 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.”

Measuring Creativity: Humans vs. AI

To ensure a fair comparison of creativity across both humans and AI, the research team employed several methods. The primary instrument was the Divergent Association Task (DAT), a psychological test that measures the ability to generate diverse and original ideas from a single prompt. Participants were tasked with listing ten unrelated nouns, with the aim of showcasing the breadth of their creative thinking.

The DAT engages cognitive processes integral to creative thought and serves as a reliable indicator of performance on other established creativity tests. While AI systems demonstrated proficiency in this task, the researchers sought to determine if this success translated into more sophisticated creative writing tasks, such as composing haiku, summarizing movie plots, and developing short stories.

Findings indicated a familiar trend: although some AI systems matched or even exceeded average human performance, skilled human creators consistently produced more original and compelling work.

Shaping AI Creativity

An intriguing aspect of the study was its examination of whether AI creativity is static or if it can be influenced. Researchers discovered that adjusting parameters—particularly the model’s temperature—could enhance AI creativity. Lower temperature settings yield more predictable outputs, while higher settings encourage more varied and exploratory responses.

Additionally, the way prompts are crafted significantly impacts AI creativity. For example, prompts that encourage exploration of word origins and structures lead to more innovative associations and higher creativity scores. This underscores the notion that human interaction and guidance are essential in shaping AI’s creative potential.

Despite the impressive advancements in AI, the study provides a balanced perspective on concerns regarding the replacement of human creators. While AI can meet or exceed average human creativity in certain tasks, it remains dependent on human direction.

“Even though AI can now reach human-level creativity on certain tests, we need to move beyond this misleading sense of competition,” Jerbi emphasized. “Generative AI has 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 research, published in Scientific Reports, suggests a future where AI acts more as a creative assistant, enhancing human imagination and expanding the possibilities of creative expression. This perspective fosters a collaborative relationship between human creators and AI technologies, potentially leading to new forms of artistic innovation.