Originality with intention

AI recombines patterns from its training data. It is very good at identifying what tends to follow what, what styles pair with what moods, what structures readers expect. That makes it fluent.

What it cannot do is break patterns on purpose, with a reason. Human creative originality is not just about generating something unexpected. It is about asking: what is worth making, and why?

That question involves values, cultural specificities and cohesive brand thinking. It involves knowing what has already been done and choosing to go somewhere different, because the work calls for it.

Taste and curation

This is one of the most underrated creative skills, and it barely features in most AI vs human creativity discussions.

The hard part of a lot of creative work is not generating ten ideas, especially not with all the tools available these days. It is knowing which one has the right tension, the right timing, the edge that makes it land. Taste is cultivated through experience and years of honing your craft, and there’s something un-algorithmic about it.

AI can produce ten options. It cannot tell you which one feels the best.

Lived meaning

AI can imitate the language of grief. It has read enough elegies to produce something that sounds mournful. It can write sentences that look like longing, irony, or joy.

It does not live any of those things. It has no accumulation of experience that gives words their weight.

That gap matters most in work that is supposed to evoke meaning: literary fiction, memoir, personal essay, brand storytelling built around identity, satire that cuts because someone actually lived through what they are mocking. 

Creative judgment

AI works best when the task is clear. Give it a well-defined brief with specific constraints, and it performs well. But when direction is vague, as is the case with many creative projects, AI output falls short of capturing what’s needed. Something is off in a way that is hard to articulate.

Evidence that supports this

The University of Barcelona study published in Advanced Science, is worth returning to here. Researchers designed a visual creativity task using abstract stimuli and had participants draw from imagination. They then had both human evaluators and AI systems assess the results across five dimensions: how much they liked the drawing, its vividness, originality, aesthetics, and the curiosity it provoked. Simply put, the experiment was an AI art vs human art comparison.

The ranking was consistent across all five measures. Visual artists scored highest. The general public came second. Human-guided AI came third. Unguided AI came last. Co-lead researcher Xim Cerdá-Company noted that AI showed poor performance in the production of creative images, and did even worse when deprived of human assistance.