Here’s a breakdown of the key facts from the provided text, answering potential questions about the study:
1. What was the main goal of the study?
The main goal was to develop a way to fairly compare creativity in humans and AI, and to understand the nature of AI creativity. They wanted to move beyond simply saying AI is “creative” and actually measure it in a comparable way to human creativity.
2. How did the researchers measure creativity?
* Divergent Association Task (DAT): This was the primary tool. Participants (both human and AI) were asked to list ten unrelated words. The more diverse and original the words, the higher the creativity score.
* Creative Writing Challenges: They also tested AI and humans on tasks like writing haiku, movie plot summaries, and short stories.
3. Who was involved in the study?
* Professor karim Jerbi (Université de Montréal & Mila) – lead researcher
* Jay Olson (University of Toronto) – co-author and creator of the DAT
* Scientists from Université concordia (mentioned in the “About the Study” section)
* Over 100,000 human participants
* various AI systems (specifically Large Language Models)
4. What were the key findings?
* AI can match average human creativity: AI systems sometimes performed as well as or better than average humans on the DAT and some creative writing tasks.
* Highly skilled humans are still more creative: The most skilled human creators consistently produced more original and stronger work than AI.
* AI creativity is adjustable: Changing the “temperature” setting of AI models (controlling predictability) impacts creativity. Higher temperatures lead to more varied and exploratory outputs.
* Prompting matters: How instructions are written substantially influences AI creativity. Prompts focusing on word origins (etymology) led to more creative results.
* AI is a tool, not a replacement: the study suggests AI will likely augment human creativity rather than replace it, acting as a powerful assistant.
5. When and where was the study published?
The study, titled “Divergent creativity in humans and large language models,” was published in Scientific Reports on January 21, 2026.