Researchers at the Université de Montréal conducted the largest comparative study of machine and human creativity, testing leading AI models against more than 100,000 human participants using the Divergent Association Task. The study found that models including GPT-4, Gemini Pro 1.5, and Llama 3 and 4 outperformed the average human, though the top-performing humans still significantly surpassed all tested models. Researchers also found AI creativity could be tuned through temperature adjustments and prompt instructions, and positioned generative AI as a potential collaborative tool rather than a replacement for human creators.
