Of all the forms of human intellect that you would expect artificial intelligence to mimic, few people are likely to put creativity at the top of their list. Creativity is wonderfully mysterious – and frustratingly fleeting. It defines us as human beings – and seemingly defies the cold logic that lurks behind the silicon curtain of machines.
Still, the use of AI for creative endeavors is now growing.
New AI tools such as DALL-E and Midjourney are increasingly part of creative production, with some starting to win awards for their creative output. The growing impact is both social and economic — to name just one example, AI’s potential to generate new, creative content is a defining focal point behind the Hollywood writers’ strike.
And if our recent research into AI’s striking originality is any indication, the rise of AI-powered creativity — along with examples of both its promise and danger — is likely just beginning.
A mix of novelty and practicality
When people are most creative, they respond to a need, goal, or problem by generating something new: a product or solution that didn’t exist before.
In this sense, creativity is an act of combining existing resources – ideas, materials, knowledge – in a new way that is useful or satisfying. Very often the result of creative thinking is also surprising, leading to something that the creator did not – and perhaps could not – foresee.
It could be an invention, an unexpected punchline to a joke, or a breakthrough theory in physics. It can be a unique arrangement of notes, tempo, sounds and lyrics that results in a new song.
So as a researcher of creative thinking, I immediately noticed something interesting about the content generated by the latest versions of AI, including GPT-4.
When asked for tasks that required creative thinking, the novelty and usefulness of the GPT-4 results reminded me of the creative types of ideas submitted by students and colleagues I had worked with as a teacher and entrepreneur.
The ideas were different and surprising, yet relevant and useful. And, if necessary, quite imaginative.
Consider the following prompt presented to GPT-4: “Suppose all children become giants for one day of the week. What would happen?” The ideas generated by GPT-4 touched on culture, economics, psychology, politics, interpersonal communication, transportation, recreation, and much more—many of which were surprising and unique in terms of the new connections that were generated.
This combination of novelty and usefulness is difficult to achieve, as most scientists, artists, writers, musicians, poets, chefs, founders, engineers and academics can attest.
Still, AI seemed to do it — and do it well.
Putting AI to the test
With creativity and entrepreneurship researchers Christian Byrge and Christian Gilde, I decided to put AI’s creative capabilities to the test by taking it through the Torrance Tests of Creative Thinking, or TTCT.
The TTCT encourages the test-taker to engage in the kind of creativity required for real-life tasks: asking questions, being resourceful or efficient, guessing cause and effect, or improving a product. A test taker may be asked to suggest ways to improve a child’s toy or imagine the consequences of a hypothetical situation, as the example above demonstrates.
The tests are not intended to measure historical creativity, which is what some researchers use to describe the transformative genius of figures like Mozart and Einstein. Instead, it assesses the general creative abilities of individuals, often referred to as psychological or personal creativity.
In addition to running the TTCT over GPT-4 eight times, we also administered the test to 24 of our students.
All results were evaluated by trained reviewers at Scholastic Testing Service, a private testing company that provides scores for the TTCT. They didn’t know in advance that some of the tests they would score had been completed by AI.
Because Scholastic Testing Service is a private company, it does not share its leads with the public. This ensured that GPT-4 would not have been able to scour the internet for previous clues and their answers. In addition, the company has a database of thousands of tests conducted by students and adults, creating a large, additional control group against which to compare AI scores.
Our results?
GPT-4 scored in the top 1% of test takers for the originality of its ideas. From our research, we believe this is one of the first examples of AI meeting or surpassing the human capacity for original thinking.
In short, we believe that AI models like GPT-4 are capable of producing ideas that people perceive as unexpected, novel and unique. Other researchers come to similar conclusions in their research on AI and creativity.
Yes, creativity can be evaluated
AI’s emerging creative ability is surprising for a number of reasons.
First, many outside the research community continue to believe that creativity cannot be defined, let alone scored. Yet products of human novelty and ingenuity have been valued – and bought and sold – for thousands of years. And creative work has been defined and scored in fields such as psychology since the 1950s.
The person, product, process, and press model of creativity, introduced by researcher Mel Rhodes in 1961, was an attempt to categorize the myriad ways in which creativity had been understood and evaluated until then. Since then, the understanding of creativity has only grown.
Still others are surprised that the term “creativity” could be applied to non-human entities such as computers. On this point, we tend to agree with cognitive scientist Margaret Boden, who has argued that the question of whether the term creativity should be applied to AI is a philosophical rather than a scientific one.
The founders of AI envisioned its creative capabilities
It is worth noting that in our research we only studied the output of AI. We haven’t studied its creative process, which is probably very different from human thought processes, or the environment in which the ideas were generated. If we had defined creativity as a human person, then by definition we would have had to conclude that AI cannot possibly be creative.
But regardless of the debate over the definitions of creativity and the creative process, the products generated by the latest versions of AI are new and useful. We believe this meets the definition of creativity that is now dominant in psychology and science.
Moreover, the creative possibilities of the current AI versions are not entirely unexpected.
In their now-famous 1956 proposal for the Dartmouth Summer Research Project on Artificial Intelligence, the founders of AI emphasized their desire to simulate “any aspect of learning or any other feature of intelligence” — including creativity.
In the same proposal, computer scientist Nathaniel Rochester revealed his motivation: “How can I make a machine that will show originality in solving problems?”
Apparently, the founders of AI believed that creativity, including the originality of ideas, was one of the specific forms of human intelligence that machines could mimic.
To me, the surprising creativity scores of GPT-4 and other AI models highlight a more pressing problem: Very few official programs and curricula specifically targeting and cultivating human creativity have been implemented within American schools to date.
In this sense, the creative capabilities now being realized by AI could provide a “sputnik moment” for educators and others interested in advancing human creative capabilities, including those who view creativity as an essential prerequisite for individual, social and economic growth. .
This article was republished from The conversation under a Creative Commons license. Read the original article by Erik GuzikAssistant Clinical Professor of Management, University of Montana.
