Ai and humans creativity.
Photo credit: theFreesheet/Nano Banana Pro

Generative AI systems have a built-in “mathematical ceiling” that prevents them from ever matching the creativity of expert humans, smashing the myth that machines will replace society’s most creative minds.

New research from the University of South Australia indicates that large language models (LLMs) like ChatGPT will never rival the originality of the most creative individuals.

Professor David Cropley, an expert in engineering innovation, computed the creative ability of LLMs using standard mathematical principles. The results revealed that LLM creativity has a maximum score of 0.25 on a scale from 0 to 1.

“While AI can mimic creative behaviour – 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,” said Professor Cropley.

Learned patterns

The study argues that because LLMs are trained on vast amounts of existing content, they respond to prompts based on learned patterns. This results in outputs that are “expected and unsurprising” rather than genuinely original.

“Creative performance, however, is not symmetrical. Typically, 60% of people are below average when it comes to creativity, so it’s inevitable that a sizeable slice of society will think that LLMs like ChatGPT are creative, when they’re not,” said Cropley.

The research suggests that if industries rely too heavily on current AI tools, they risk producing formulaic, repetitive work. To reach expert levels, AI would require a new architecture capable of generating ideas not tied to past statistical patterns.

Leave a Reply

Your email address will not be published. Required fields are marked *

You May Also Like

Why the world’s ship captains are terrified of autonomous vessels

A new wave of autonomous ferries is set to hit the water…

The death of original thought as chatbots force humanity to conform

It’s not just your syntax that artificial intelligence is hacking — it’s…

Fighting global fraud with AI evidence chains instead of mass surveillance

Rather than expanding government oversight into our private lives, policymakers must harness…

US team grabs aliens.gov as Reddit talks “Project Blue Beam” and disclosure

The US government has officially staked its claim to “aliens.gov,” igniting a…

Mind-reading milestone lets scientists watch movies inside a mouse’s brain

It sounds like pure science fiction, but researchers have figured out how…

A medical taboo: Why neurologists must start talking to patients about faith

Wading into questions of faith, purpose, and mortality is usually left to…