Divergent thinking, often considered a cornerstone of creativity, is a hallmark of human creative expression. If models like ChatGPT-4 demonstrate divergent thinking surpassing that of humans, does it imply that this uniquely human trait has been acquired by artificial intelligence (AI)?
Recently, a research team from the University of Arkansas has found that ChatGPT-4's performance in creative thinking tests has surpassed that of humans. The relevant research paper titled "The current state of artificial intelligence generative language models is more creative than humans on divergent thinking tasks" has been published in the Nature sub-journal Scientific Reports.
The team stated, "Overall, even when controlling for fluency of responses, GPT-4 demonstrates greater originality and complexity than humans on each divergent thinking task. In other words, GPT-4 exhibits higher creative potential across the entire divergent thinking task."
Is ChatGPT-4 really better than humans at divergent thinking?
According to the paper, the research team tested 151 human participants and ChatGPT-4 on three creative thinking tasks: the Alternative Uses Task, the Consequences Task, and the Divergent Associations Task.
These tasks aim to evaluate individuals' creative thinking abilities, including innovative uses for everyday items, potential consequences of hypothetical scenarios, and understanding of semantic differences in words. These criteria help quantify each participant's performance in creative thinking tasks and provide a foundation for subsequent analysis.
The test results showed that in all three creative thinking tasks, ChatGPT-4 significantly outperformed human participants. Whether in terms of the number of generated responses, response length, or semantic differences, ChatGPT-4 demonstrated a clear advantage, highlighting its potential in the field of creative thinking.
Image: In the Alternative Uses Task, GPT-4 demonstrates greater success in generating diverse answers and exhibits higher originality compared to humans when given the same number of opportunities to produce responses, albeit within specific prompts. (Source: The paper)
Chart | GPT-4 Outperforms Humans in Consequences Task
In the Consequences Task, GPT-4 demonstrates greater success in generating a variety of responses compared to humans, given the same number of opportunities. Moreover, it exhibits higher originality based on prompt types. (Source: the paper)
To eliminate the influence of other factors, the study also controlled for response fluency. Results indicate that even when controlling for response fluency, ChatGPT-4 still demonstrates higher creative potential, further confirming its superiority in creative thinking tasks.
These findings sparked discussions on the creative potential and practical applications. While ChatGPT-4 shows significant advantages in creative thinking tasks, the research team also noted limitations in its applicability.
For instance, compared to humans, ChatGPT-4 lacks autonomy and practical experience, relying on guidance and input from human users for its creative potential. Therefore, despite its ability to generate creative ideas, its actual application remains constrained.
Additionally, when assessing ChatGPT-4's creative potential, the relevance and practicality of its responses need consideration. Although ChatGPT-4 can generate numerous original ideas, further research and validation are required to determine their applicability in real-life scenarios and whether they can yield meaningful outcomes.
Lastly, the research team emphasized the uniqueness and diversity of human creativity. Compared to ChatGPT-4, humans possess broader experiences and emotional insights, allowing them to think and create in more flexible and diverse ways. Thus, when discussing creative potential, it's essential not to simply equate ChatGPT-4's performance with humans but also consider its applicability and limitations in practical settings.
AI Replacing Humans? Still Subject to Observation
Compared to previous research, the test results of this study reveal ChatGPT-4's significant advantages in multiple creative thinking tasks, offering new insights into AI's role in creative thinking.
However, this also raises more questions and avenues for future exploration. Further research is needed to delve into the relationship between artificial intelligence and human creative thinking and explore how to better utilize AI to foster innovation.
Future studies will focus on developing more intelligent and comprehensive measurement methods to better understand and evaluate various aspects of creative thinking. As AI technology continues to evolve, our understanding of human creativity will deepen, laying a solid foundation for building a more creative and intelligent future.
The research team believes that whether these tests can perfectly measure human creative potential is not the key issue. The key lies in the fact that large models are rapidly evolving and surpassing humans in unprecedented ways, and whether they could replace human creativity remains to be observed.
They also suggest that in the future, AI could serve as an inspiration tool, as an auxiliary tool in personal creative processes, or as a tool to overcome fixedness.
Disclaimer: Sourced from the internet, exercise caution in discernment.
