AI Chatbot ChatGPT Matches Top 1% of Human Thinkers.
A recent study conducted by the University of Montana has demonstrated that ChatGPT, an AI-powered chatbot, can rival the creativity of the top 1% of human thinkers.
The research team employed the Torrance Tests of Creative Thinking (TTCT), a well-established measure of creativity, to evaluate the performance of ChatGPT. The AI generated eight responses, which were then compared with those from 24 university students and 2,700 students across the nation who had also taken the TTCT.
Unbeknownst to the Scholastic Testing Service, which scored all responses, some of the submissions were produced by AI. The results revealed that ChatGPT's creativity matched that of the most inventive individuals among the human test-takers and even surpassed a majority of the national student group.
Erik Guzik, an assistant clinical professor at the University of Montana’s College of Business and the study's lead author, expressed his surprise at the findings in an interview with ScienceDaily.com. Guzik, whose research interests include entrepreneurship and the economic dimensions of creativity, sees potential implications of AI for innovation in business.
Guzik noted, "Creativity is all about doing things differently. One definition of entrepreneurship that resonates with me is that it involves thinking differently. Therefore, AI might help us apply creative thinking to the business world and the innovation process, which I find utterly intriguing.”
The TTCT consists of two sections: verbal and figural, both of which gauge divergent thinking, a cognitive process that facilitates the generation of creative ideas. In the verbal segment, participants are presented with visual or verbal prompts and asked to provide written responses. These responses are then used to evaluate three cognitive traits: fluency (the quantity of pertinent ideas), originality (the uniqueness of ideas), and flexibility (the diversity of different kinds of ideas).
The figural section requires respondents to draw their answers and, along with fluency and originality, it assesses additional skills such as elaboration (the capacity to expand on ideas), resistance to premature closure (the ability to tolerate uncertainty in order to conceive a creative response), and abstractness of titles (the ability to provide a figure with a creative title).
ChatGPT's responses ranked in the top percentile for both fluency and originality. However, when it came to flexibility, the scores fell to the 97th percentile.
Guzik commented on the unexpected outcomes produced by ChatGPT, stating, “We had all been experimenting with ChatGPT, and we noticed it was generating some intriguing outputs that we hadn’t anticipated. Some of the responses were original and unexpected. That’s when we decided to test its actual creativity.”
While Guzik refrains from overhyping the potential economic impact of ChatGPT, he does believe it could play a pivotal role in driving innovation.