Are generative artificial intelligence systems inclusive of ChatGPT capable to real creativity? A new large-scale study headed by Professor Karim Jerbi from the Department of Psychology at the Université de Montréal set out to answer that query. The research team also included Yoshua Bengio, a leading AI pioneer and professor at the Université de Montréal. Together, they performed the most considerable comparison to date between human creativity and the creative abilities of large language models.
The observation, posted in Scientific Reports, factor to a huge shift. Generative AI systems have now attained a degree where they are able to exceed the average human on specific creativity measures. At the same time, the study makes it clear that the most creative people nonetheless surpass the performance of even the strongest AI models.
AI Reaches Average Human Creativity Levels
The researchers evaluated several foremost large language models, such as ChatGPT, Claude, Gemini, and others, and as compared their results with data from 100,000 human participants. The result marks a clean turning factor. Some AI systems, which include GPT-4, earned higher than the average human on tasks designed to measure divergent linguistic creativity.
“Our study demonstrates that some AI systems based on large language models can now surpass average human creativity on well-described tasks,” stated Professor Karim Jerbi. “This outcome can be surprising — even unsettling — but our study also emphasize an equally important observation: even the best AI system still fall short of the levels got to the most innovative people.”
How Creativity Was Measured in Humans and AI
To make a fair comparison among people and machines, the research team used numerous techniques. The primary tool was the Divergent Association Task (DAT), a psychological test formed to measure divergent creativity, or the ability to form many original and varied ideas from a single prompt.
Created by study co-author Jay Olson, the DAT asks participants, whether human or AI, to create ten words which are as exclusive in meaning from one another as possible. A highly creative response might include words along with “galaxy, fork, freedom, algae, harmonica, quantum, nostalgia, velvet, hurricane, photosynthesis.”
Performance in this task in humans intently mirrors outcomes on different well-established creativity tests used in idea generation, writing, and creative hassle solving. Although the task is language-based, it does not certainly test vocabulary. Rather than, it taps into broader cognitive tactics include in creative thinking across many domains. Another benefit of the DAT is its speed and accessibility, as it takes 2-4 minutes to complete and is available online to the general public.
From Simple Word Tests to Creative Writing
Building on those outcomes, the researchers tested whether AI performance on this basic word association task could translate into more complex creative activities. To test this, they directly compared AI systems and human participants on creative writing tasks.
These consisted writing haiku (a short 3-line poetic form), creating movie plot summaries, and create short stories. Once again, the pattern was clean. While AI sometimes to time outperformed common human participants, the most skilled human creators endured to show a clear benefit.
Can AI Creativity Be Adjusted?
The findings increased an essential follow-up inquiry. Can AI creativity be shaped or controlled? As per the study, it can. One main issue is the model’s temperature, a technical setting that impacts how predictable or adventurous an AI’s responses are.
At lower temperature settings, AI structures tend to generate more secure and more predictable outputs. At higher temperatures, the responses end up more varied and less constrained, encouraging risk-taking and more unique associations.
The researchers also found that the way prompts are written plays a big role. For example, instructions that inspire AI models to consider the origins and structure of words using etymology cause more surprising ideas and higher creativity score. Together, these outcomes demonstrates that AI creativity relies closely on human input and instructions, making interaction among people and machines a central part of the creative process.
Will AI Replace Human Creators?
The study provides a balanced angle on fears that artificial intelligence could replace creative professionals. While some AI systems can now rival human creativity on precise tasks, the research also emphasize clear barriers and the continuing significance of human creativity.
“Even although AI get to human-level creativity on certain tests, we want to move beyond this misleading sense of competition,” stated Professor Karim Jerbi. “Generative AI has above all an highly powerful tool in the service of human creativity: it’s not going to replace creators, however intensely change how they imagine, explore, and create — for those who choose to use it.”
Instead of anticipating the end of creative careers, the findings inspire a new way of thinking about AI. The technology may also serve as a creative assistant that increase possibilities for exploration and inspiration. The future of creativity may also rely less on humans as opposed to machines and more on new sorts of collaboration, where AI assists and enhance human imagination.
“By directly facing human and machine capabilities, studies like ours push us to reconsider what we mean by creativity,” concludes Professor Karim Jerbi.
