Eaon Pritchard.
Joe Strummer was right. New research proves the best creativity still comes from the best human minds – not the machines.
“No input, no output.”
– Strummer’s Law
Joe Strummer wasn’t talking about prompt engineering, but he might as well have been.
The Clash mouthpiece’s mantra (one of them, anyway) was a creative philosophy deeply rooted in curiosity and cultural immersion. Strummer believed you had to stay plugged into the world. Books, films, politics, people. If you weren’t constantly absorbing, noticing, and feeling things, then you had no business trying to make anything of meaning. Creativity, in his view, was a feedback loop powered by stimulus. If you weren’t putting anything in, you had nothing worth putting out.
That ethos feels especially relevant now, as generative AI causes panic in the creative industries with the volume and velocity of synthetic outputs. The temptation is to marvel at what these models can generate – on occasion, it can appear as a marvel – but I’m going to argue here that the real magic still lies upstream in the quality of the input. And I’ve got some science up my sleeve to back it up.
As a preamble, you can’t have failed to notice last week’s image-generation frenzy, a tsunami of aesthetic slop that momentarily turned X/Twitter into a vast ersatz-Ghibli-themed mediocrity fest.
OpenAI’s GPT-4o dropped its latest image generator on March 25. Within hours, the internet collectively decided that for a few days, everything that was already drivel needed a quasi-Ghibli imbecilic makeover. Geniuses.
In complete contrast, the other big media event of the last couple of weeks is the release of Netflix’s Adolescence miniseries. Adolescence is a triumph of collaborative magnificence, with its one-shot cinematography, in-depth character development, performance, writing and biting social commentary. One of the most incredible moments (and there are quite a few) comes at the end of Episode 2, where a meticulously choreographed drone shot is the highlight of the series’ ground breaking one-take format—no further spoilers for those who’ve not seen it yet. Suffice to say that THIS exemplifies how cutting-edge technology can amplify human creativity.
Ironically, Studio Ghibli’s actual artistry is rooted in similarly profound themes and meticulous craftsmanship. Ghibli films, led by legendary Japanese polymath Hayao Miyazaki, explore themes like environmentalism, the complexity of peace, the struggles of youth, and the balance between humanity and nature. In contrast, the viral Ghibli-style image trend represents – from an artistic standpoint – pretty much the surface level of nothing.
Yet this brief pseudoGhibli memeplex presents an intriguing case study for observers of human behaviour and cognitive biases. So what does it all meme?
Firstly, the illusion of skill. Dabblers experience an inflated sense of creative ability when using AI tools. This aligns with the Dunning-Kruger effect, where individuals with limited expertise in a domain overestimate their competence and misattribute the tool’s capabilities to one’s own sense of skill.
Next up, is our old friend, instant gratification. It’s that perennial dopamine loop again. Instant hits fuel our preference for quick, surface-level creative processes over more time-intensive, skill-building approaches.
And then the social validation. The viral nature of memes on social media provides immediate positive feedback, reinforcing the behaviour and contributing to an inflated sense of creative achievement.
While these tools undoubtedly democratise certain aspects of image creation, they also throw up challenges to traditional notions of skill development and creative expression.
Or do they?
A timely new peer-reviewed study by researchers at Duke, Harvard, Cambridge, and Sheridan College dropped this very same week. It set out to compare the creativity of AI-generated art prompts written by professionals, novices, and ChatGPT prompting itself—and what it finds should give professional creatives a bit of reassurance.
TL;DR version. The machines may be fast. But the best minds still make the best prompts.
Researchers asked three groups to generate short, creative prompts for DALL·E 3, OpenAI’s image-generation model. Creative professionals, visual artists, novice non-artists and also ChatGPT-4 prompting itself. Each group submitted 15 prompts (max 15 words) without stylistic instructions. The resulting 45 images were then rated for creativity by 299 human respondents.
The results were pretty straightforward. Professional artists’ prompts produced the most creative images by far, ChatGPT came second, and the non-creatives jogged in at the rear.
AI beats the average person. But the best creatives still beat the AI.
So why did the pros win? Two measurable factors stood out
Semantic distance – Professionals’ prompts connected more conceptually ‘distant’ ideas. This is a proven marker of creativity. And prompt length – They wrote longer, more detailed prompts, which gave the AI more conceptual material to work with. The better the input, the better the output. Strummer’s law in effect.
It’s less about the prompt hacks that the LinkedIn thinkfluencers want to sell you courses on, and more about classic creative muscle. Imagination and abstract thinking. Yer seasoned creatives are just better at provoking the machine.
This reframes prompt craft not as a tech skill but as a creative one. And it reinforces a timeless truth: the tools don’t make the work better—people do.
I feel like I’ve been wheeling out this Bill Drummond nugget forever, but it never fails to sum up our industry’s unnecessarily fractured relationship with tech. “The technology always comes first, then creative people mess around with it, doing things with it it’s not supposed to do, and then things get interesting.“
There’s a deeper reason we’re wired to value human creativity—especially in art.
From an evolutionary perspective, genuine creativity acts as a costly signal. It takes time, effort, intelligence and sensitivity. Like a peacock’s tail, it’s a flex—a marker of fitness.
We intuitively sense this. Even when AI-generated outputs are technically excellent, they often feel hollow. That’s because we judge creativity not just on the outcome but on who made it and why. The value isn’t just aesthetic—it’s emotional and social. It tells us something about the maker. That’s what AI can’t fake. At least, not yet.
One slightly uncomfortable truth falls out of the research. Yes, the AIs have already surpassed novice humans in prompt creativity, and we’ve likely reached the point of the total automation of average creativity. Good. Now, the rest of us can get on with making magic.
Joe Strummer also observed, “Like trousers, like brain.” It was his way of calling out cultural passivity—how surface-level choices often reveal a lack of intent. The Clash wore military fatigues, not for fashion, but because they saw creativity as a kind of warfare. Meanwhile, the mainstream wore flares and waited for someone else to tell them what was cool. He was basically saying, that your external choices reveal your inner substance — and if you’re wearing flares, you’re not ready for the fight.
That same divide is emerging in creative industries today. Those who are using AI to push boundaries, challenge form, and provoke emotion—they’re showing up ready for battle. Everyone else is just wearing flares, letting the machine think for them.
No input, no output.
The future of creative work will be defined by who’s been paying attention. Who’s connected to the culture, the chaos, the beauty and the weirdness of the world. That’s the input that counts.
Stay curious. That’s how you stay irreplaceable.
Eaon Pritchard | Strategy | Consumer psychology
