University of Arkansas: AI outperforms humans in standardized tests of creative potential, new study
FAYETTEVILLE, Ark. (KNWA/KFTA) — A study done by University of Arkansas PhD students in psychological science pitted artificial intelligence (AI) against humans to test creative potential, according to a U of A news release.
The study tested 151 human participants’ divergent thinking, which is considered to be an indicator of creative thought.
Their scores were compared to those of ChatGPT-4, an artificial intelligence chatbot.
According to the news release, GPT-4 “provided more original and elaborate answers than the human participants.”
Three tests were utilized in testing the creative potential of humans and AI.
The first test was the Alternative Use Task, which asked participants to come up with creative uses for everyday objects like a fork or a rope.
According to the news release, the Consequences Task test invited participants to imagine possible outcomes of hypothetical situations, like “what if humans no longer needed sleep?”
The last test utilized was the Divergent Associations Task that asked participants to generate 10 nouns that are as semantically distant as possible. According to an example in the news release, there is not much semantic distance between “dog” and “cat” while there is a great deal between words like “cat” and “ontology.”
Human and AI responses were evaluated for the number of responses, length of responses and semantic differences between words.
“Overall, GPT-4 was more original and elaborate than humans on each of the divergent thinking tasks, even when controlling for fluency of responses. In other words, GPT-4 demonstrated higher creative potential across an entire battery of divergent thinking tasks,” according to the study.
The authors did say that the purpose of the study was to examine human-level creative potential, not necessarily people who may have established creative credentials.
UA plans to renovate the exterior of Yocum Hall, current students express feelings
“It is important to note that the measures used in this study are all measures of creative potential, but the involvement in creative activities are another aspect of measuring a person’s creativity,” the authors said.
The news release added that researchers did not evaluate the appropriateness of GPT-4 responses. While the AI may have provided a higher number of original responses, human participants may have felt that their responses needed to be grounded in the real world.
The point of the findings of the study is that large language models are rapidly progressing and outperforming humans in ways that have not been seen before, according to the news release.
“Moving forward, future possibilities of AI acting as a tool of inspiration, as an aid in a person’s creative process, or to overcome fixedness is promising,” the authors said.
The study was published in Nature’s Scientific Reports and authored by U of A PhD students in psychological science, Kent Hubert and Kim Awa, as well as Darya Zabelina, an assistant professor of psychological science at the university and director of the Mechanisms of Creative Cognition and Attention Lab.
For the latest news, weather, sports, and streaming video, head to KNWA FOX24.
