Produced by: Tarun Mishra Designed by: Manoj Kumar
A recent study involving 151 human participants and ChatGPT-4 reveals that the AI surpassed humans in three standardized tests designed to measure divergent thinking, a key indicator of creative thought.
The study, titled "The current state of artificial intelligence generative language models is more creative than humans on divergent thinking tasks," was published in Scientific Reports. It was authored by Ph.D. students Kent F. Hubert and Kim N. Awa, along with Darya L. Zabelina, an assistant professor of psychological science at the University of Alberta.
Divergent thinking involves generating unique solutions to open-ended questions, such as finding creative uses for everyday objects or imagining hypothetical scenarios. ChatGPT-4 exhibited more original and elaborate responses compared to human participants.
The study utilized three tests: the Alternative Use Task, the Consequences Task, and the Divergent Associations Task. These tasks evaluated participants' ability to generate creative responses and measure semantic differences between words.
ChatGPT-4 demonstrated higher creative potential across all three tasks, even after controlling for response fluency. The AI provided more original and elaborate answers compared to humans.
The study acknowledges certain limitations, such as the AI's lack of agency and dependence on human input. Additionally, it did not assess the appropriateness of AI responses or human motivation to provide elaborate answers.
The purpose of the study was to examine human-level creative potential, focusing on creative potential rather than established creative credentials.
While the tests may not be perfect measures of human creative potential, the study highlights the rapid progress of large language models and their potential to inspire creativity or aid in overcoming creative blocks.
Moving forward, researchers see promising possibilities for AI to act as a tool of inspiration or aid in the creative process, though questions remain about the generalizability of tests and the extent to which AI may replace or augment human creativity.