A study from the University of California at Berkeley explored how artificial intelligence compares to human imagination in creative tasks. The research, published in Perspectives on Psychological Science, examined whether even the most sophisticated AI models can match a child’s or an adult’s ability to think creatively when confronted with unusual uses for common objects. The comparison included OpenAI’s ChatGPT as the AI agent, children aged 3 to 7, and adults solving nonstandard problems.
The findings highlight a clear gulf between human and machine creativity. People demonstrated a strong capacity to invent solutions that are both meaningful and visually distinct, choosing tools or objects for purposes beyond their traditional roles. This shows not only that humans can recognize established connections but also that they can break away from them, using abstract thinking to uncover hidden functional properties and to repurpose items in novel ways.
On the other hand, AI systems tended to map surface relationships between items but often skipped the leap to genuine novelty. When asked to assign objects to new uses, the algorithms frequently defaulted to conventional pairings and missed opportunities where fresh applications could emerge.
One of the study’s authors described the results in a practical frame: young children can offer intelligent responses that AI models do not grasp yet. Rather than casting AI as intelligent agents on par with humans, she suggested viewing them as a new type of library or search engine that can relay information that humans then remix and interpret. In this view, AI passes knowledge along to people, who complete the creative process.
The research also touches on the broader perception of AI output. Prior work has suggested that people may show preference or bias toward work produced by artificial systems, which makes the current findings especially relevant for educators, designers, and policymakers interested in how humans and machines can complement each other in creative tasks.
From an applied standpoint, the Berkeley work invites ongoing discussion about how AI can be harnessed to support creativity without overestimating its current capabilities. It underscores the distinct strengths that people bring to problem solving, including flexible thinking, contextual judgment, and the ability to reinterpret tools in fresh, purpose-driven ways. The study thereby contributes to a nuanced view of AI as a powerful tool that expands human potential rather than replacing it. (PPS study)