This BrainFacts article explores whether AI can genuinely replicate human creativity, drawing on insights from neuroscience, psychology, and computer science. It explains that creativity in the human brain is not housed in a single region but relies on collaboration between multiple neural networks. The article outlines Graham Wallas's four-stage model of creative thought — preparation, incubation, the "aha" moment, and verification — and maps each stage to specific brain networks, including the executive control network, the default mode network, and the parietal cortex. Neuroimaging research suggests that the ability to dynamically switch between these networks is a key marker of creative capacity.

The piece examines efforts to measure creativity, including psychologist Jay Olson's divergent association task (DAT), which tests how semantically distant a person's word associations are. When AI models like GPT-4 were tested using the DAT, they outperformed the average human — but still fell short of the most creative individuals. Skeptics like Selmer Bringsjord argue that AI outputs are fundamentally remixes of training data and fail his Lovelace Test, which requires an AI to produce something unexplainable by its own design. Psychologist Mark Runco emphasizes that intentionality is a crucial ingredient of authentic creativity that AI lacks entirely.

The article concludes on a sobering note about the implications for creative professionals. Experts warn that even if AI isn't truly creative, public perception may not distinguish between human and machine-generated content, potentially threatening careers in creative fields. Both Bringsjord and Runco express pessimism about society's ability to navigate this challenge ethically, with Runco cautioning, "I'm afraid we're not going to make the right decisions."