Why creative people matter more now

OpenAI says people now create more than 3 billion images a week with its tools. Add the text, music and video made by other models, and the conclusion seems obvious: if a machine can produce competent creative work in seconds, the people who used to make it matter less.

The evidence points the other way. AI raises the floor of creative work and pulls it towards the middle. That makes the people who can lead it somewhere else more valuable, not less.

Better alone, more alike together

The clearest study comes from Anil Doshi and Oliver Hauser, published in Science Advances in July 2024. Writers who were given story ideas by a language model wrote stories that readers rated as more creative, better written and more enjoyable, with the largest gains among the less creative writers. But the stories written with AI were more similar to each other. The authors describe “an increase in individual creativity at the risk of losing collective novelty”.

A 2025 meta-analysis by Niklas Holzner, Sebastian Maier and Stefan Feuerriegel, covering 28 studies with 8'214 participants, found the same pattern at scale. Generative AI on its own performed about as well as humans on creative tasks. People working with AI did better than people without it, by a modest margin. And AI had “a significant negative effect on the diversity of ideas”, an effect about three times the size of the gain.

The jagged edge

Sameness is not the only risk. In a field experiment with Boston Consulting Group, now published in Organization Science, 758 consultants worked on realistic tasks with or without GPT-4. On tasks within the model's abilities, those with AI completed 12.2 percent more tasks, 25.1 percent faster and at clearly higher quality. On a task chosen to lie outside them, they were 19 percentage points less likely to reach a correct answer than colleagues without AI.

Ethan Mollick, one of the authors, noted that the answers produced with AI were “a bit homogenous and same-y in aggregate”. A survey of 319 knowledge workers by Microsoft Research found that higher confidence in AI went with less critical thinking, and that the work itself shifts towards checking, combining and taking responsibility for what the machine produces. Knowing where the edge of a tool lies is not a technical skill. It is judgement.

What people bring

A model is trained on what already exists and, left alone, returns its most likely version. Ted Chiang put the difference in the subtitle of his essay for The New Yorker: “an artist makes choices that are fundamentally alien to artificial intelligence.” A brief, a prompt and a reference board are each a set of choices. The more of them a person makes deliberately, the less of the result is left to the average.

Those choices draw on what a model does not have: a point of view, knowledge of one particular audience, a memory of what was tried and failed, a sense of what a brand would never say. Employers seem to see it. In the World Economic Forum's Future of Jobs Report 2025, creative thinking is among the skills expected to keep rising in importance until 2030, next to AI and big data.

Practitioners notice when a company forgets this. In February, The Register reported, the video platform Higgsfield removed a post boasting that its motion design tool had ended more than 20 creative jobs, after a backlash from the artists who use it. Aharon Rabinowitz, who runs Motion Management, wrote that celebrating the end of artists' careers was “just super dumb and shortsighted”, and added that the tools were not good enough to put anyone out of work.

How good teams work with AI

The research also suggests how to keep the gains without the sameness. Think before you prompt: Doshi and Hauser describe the risk of writers anchoring on the machine's ideas, so the first ideas, the angle and the references should come from people. Mollick describes two working styles that both did well: centaurs, whose work has “a clear line between person and machine”, and cyborgs, who weave the two together. Neither hands over the decision.

In practice, good teams go wide with the machine and narrow with people. They feed it references instead of adjectives, ask for many options and throw most of them away. They compare their work with what competitors publish, because the danger is not bad output but output that looks like everyone else's. And a named person checks the facts and signs off, because the moment of greatest trust in a tool is when critical thinking drops.

Where crue stands

We use AI for the making, the variants and the routine, and we want all of it to get faster. In Reel, the ideas, the references and the styles to lock are agreed with a person before anything is made, the best model is chosen for each job, and a person picks every image and every clip that is used. The machine produces the options. People supply the direction and the no.

AI has made competent work abundant. That is exactly why the people who can make it specific, surprising and right are worth more than before.