A new study pitted ChatGPT's creativity against that of humans (specifically university students). Using the classic Torrance Test of Creative Thinking, it probes creativity through questions like "What if..." (we had 7 fingers instead of 5) and alternative uses ("What else could a pen be used for besides writing?"). Participants try to generate many novel, unexpected answers, requiring mental flexibility, imagination and originality. The research was led by Dr. Eric Guzik of the University of Montana. He gave the test to both ChatGPT and his top students.
The results? ChatGPT scored in the top percentile overall, outperforming the students in most cases.
So what gives?
I spoke with Eric, still processing the implications like the rest of us. After getting these startling results, he told me he asked ChatGPT itself what it thought they meant. It replied that humans need to redefine creativity in a way that separates it from artificial intelligence...and find new ways to measure it.
A fair point, it must be said.
There's no denying creativity encompasses far more than churning out ideas against the clock. Reducing it to that would be absurd. But this certainly gives pause for thought.
The machines aren't making it easy on us. We'll have to reexamine our definitions and what uniquely human traits remain as AI advances. What does "creativity" really mean going forward? And "cooperation" (given how well AIs now collaborate)?
Where do humans still have the advantage?