The supposed “creativity” of artificial intelligence (AI) is subject to strict mathematical limits, according to a study published Nov. 11 in the Journal of Creative Behavior.
David Cropley, a professor of engineering innovation at the University of South Australia and sole author of the study, found the limit of AI’s abilities lies somewhere between the amateur and professional level in human beings, meaning AI will never outstrip the creativity of the most talented human artists.
Cropley’s finding, however, has done little to quell concerns that AI will cause creative sectors of the economy to disappear. Experts continue to debate the creative potential of AI, and how we define creativity is one of the biggest stumbling blocks. Just like “smart” or “attractive,” “creative” is a very human descriptor that can mean different things in different domains, and that defies easy or quantitative measurement.
Cropley applied the Standard Definition of Creativity to the outputs of various large language models (LLMs), such as ChatGPT. He found that “while AI can mimic creative behavior — quite convincingly at times — its actual creative capacity is capped at the level of an average human and can never reach professional or expert standards under current design principles,” according to a statement.
AI is not human
Jack Shaw of Shawfire Media, an e-commerce strategist who deploys and benchmarks LLMs to generate and test marketing content, said that under some definitions, the assertion of the study is correct. “If creativity means reframing a brief, setting new cultural cues, and taking responsibility for risky choices that could fail, then humans lead. Models synthesise patterns optimized for likelihood; they do not carry intent, lived context, or stakes, and they do not originate goals.”
The biggest gap in AI’s creativity is that it will never have a human experience, said Alesha Brown, founder and CEO of Fruition Publishing Concierge Services and Alesha Brown Productions, companies that help authors, thought leaders and brands turn lived experiences into books, films and campaigns.
“No LLM wakes up with a childhood trauma, a cultural lineage, or a moral conflict and decides, ‘I’m going to make a film or write a book that could cost me relationships but might free other people’,” she said. “That ‘why’ behind the work — the willingness to risk reputation, income, or belonging for an idea — is a big part of what we intuitively count as creativity, and AI doesn’t have that. It’s an argument about agency and depth, not an ironclad mathematical proof that AI can never match or surpass us.”
But AI is creative
But by other benchmarks, AI can be considered creative. Gor Gasparyan, co-founder and CEO of Passionate Agency, a digital intelligence agency that offers digital experiences with a focus on AI engineering, thinks the idea of a mathematical limit to AI creativity is based on an old-fashioned definition of the word that discounts the value of novelty.
“In my practice, AI models produce keywords and theme connections that are novel to our human SEO experts 80% of the time, which leads to content strategies that haven’t been considered before,” Gasparyan said.
To Iliya Rybchin, founder and principal of AI consulting firm Vorpal Hedge, AI generates creative materials in a very similar way to humans. “Both humans and LLMs rely on the same underlying mechanism — recombining stored patterns under constraints. The real problem isn’t that AI ‘lacks creativity’, it’s that we keep packaging creativity in mystical language that collapses as soon as we look at how human creators actually work,” he said.
“We romanticize the novelist staring at a blank page or the chef conjuring a dish nobody has ever imagined before,” he explained — but talented creators pull from their lived experiences, having tasted foods, read literature and learned skills that they then recombine into new variations.“None of this is ex-nihilo creation, it’s high-fidelity remix. In reality, creativity is almost exclusively combinatorics.”
He added that the claim AI has a mathematical ceiling lower than humans is a math error. “If creativity is the ability to connect unconnected dots, the entity with the most dots wins.”
That principle is why AI could match human creativity, said James Lei, CEO of legal class action platform Sparrow. “Creativity is generation plus selection against a purpose,” he said. “Generation is the ability to produce many candidates, while creativity requires novelty, usefulness and acceptance by an audience or domain gatekeepers. This is why AI already works for ad concepts, onboarding flows, contract clause options and musical motifs, where quality is measurable and the brief anchors direction.”
You get out what you put in
Some experts believe any perceived shortcomings of AI are caused only by a lack of input from humans. For instance, if you can give clear instructions, set ways to judge results and keep improving them through human feedback and tests, Lei adds, AI meets the standard because it generates new options and the process keeps those that add value. “Where it struggles is open-ended, long-horizon agenda setting that draws on lived experience, embodied context and cross-domain judgement.”
Vague prompts may also result in AI that outputs a fairly bland idea, said Amit Raj, founder of The Links Guy, an SEO consultancy that uses AI workflows in content marketing tasks. “But give it context, challenge it, refine it and debate with it, and creativity emerges.”
Ultimately, the definition of creativity will continue to evolve as long as the debate over the creative ability of AI persists, said Paul DeMott, Chief Technology Officer at Helium SEO. “The argument about how creating something is not equivalent to being creative is the indication that we’re shifting goalposts,” he said. “Critics claimed AI was devoid of intent, then emotional richness, then originality. We conceptualize creativity as anything humans can accomplish that machines can’t, then redefine it when machines break that barrier.”
