A study by the University of Montana found that ChatGPT 4.0 could match the top 1% of creative people on a standard creativity test. The sci...
A study by the University of Montana found that ChatGPT 4.0 could match the top 1% of creative people on a standard creativity test. The scientists used the Torrens Creative Thinking Test, a well-known tool that has been used for decades to assess a person's creativity.
One can have different attitudes to the very idea of checking non-standard thinking with standard tests. But the result is still impressive.
A study by the University of Montana and its partners found that ChatGPT 4.0 can match the top 1% of creative people on a standard creativity test.
The comparison was led by Dr. Eric Guzik of the University of Montana College of Business. He and his partners used the Torrens Creative Thinking Tests, a well-known tool that has been used for decades to assess an individual's creativity.
The researchers presented eight responses generated by ChatGPT, an application powered by the GPT-4 artificial intelligence engine. They also provided responses from a control group of 24 University of Montana students attending Guzik's entrepreneurship and finance classes. These results were compared with those of 2,700 college students across America who took the TTCT in 2016. All submitted materials were evaluated by the Scholastic Testing Service, which did not know that artificial intelligence was involved in the process.
The results made it possible to classify ChatGPT into the elite group in terms of creativity. The test assessed: Fluency (the total number of ideas that correspond to the problem being solved and created in response to a request), Flexibility (the number of different categories of answers), and Originality (statistical rarity of answers).
The chatbot ranked in the highest percentage of testers for Fluency and Originality and dropped to 97% for Flexibility. ChatGTP has outperformed the vast majority of college students across America.
What does ChatGPT think about its result?
Scientists have been testing the chatbot for a long time and expected ChatGPT to be good at generating a lot of ideas (Fluency). But even they were surprised by such a high score on the originality test. The ability to generate fresh ideas seemed to be the hallmark of the human imagination.
Guzik also emphasized ChatGPT's unusually strong progress: “The previous study was a year ago, it involved GPT-3. That version did not show results comparable to the results of people in tasks requiring original thinking. And GPT-4, which came out in March of this year, is in the top 1% of human participants.”
Guzik expects that as development accelerates, AI will become a key tool in the business world and a significant new driver of its development.
The scientists asked ChatGPT how he rates his performance on the TTCT test. Guzik considers the chatbot's response noteworthy: “ChatGPT told us that we may not fully understand what human creativity is. He also suggested that we will need more sophisticated tools to assess creativity and methods to distinguish between human-generated and AI-generated ideas.”
COMMENTS