Apropos of nothing, I keep circling back to Bjarke Ingels’ throwaway observation he shared recently on Instagram about Ai prompting feeling a bit like briefing a team. You describe an intention, something comes back, you adjust. Anyone who’s spent time in a design or strategy studio recognises that scenario instantly, the loose sketch of an idea, the return volley, the shrug, the “maybe try it with less… erm whatever that is.” It felt like it was a nice clean analogy and I was nodding along.
But a studio isn’t a stochastic mirror. It’s a small society of taste and memory. People remember your last terrible idea. Someone raises an eyebrow when a line of copy looks off or a Figma file has gone fully feral. Someone else brings up the project you swore you’d never repeat. The feedback loop is human, textured, occasionally bruising. There’s judgement, shared reference points, work blue-tac’d up on the walls and a quiet sense of “let’s not embarrass ourselves again.”
This is where Ingels’ analogy starts to wobble. When you brief a human, you’re drawing on their judgement, experience, and the unspoken etiquette of a team. When you brief Ai, it behaves nothing like a junior designer and everything like a very confident autocomplete. It gives you the shape of participation without the substance. A colleague can resist you, encourage you to slow you down, challenge the premise. A (poorly prompted) model can’t. It just accelerates whatever direction you gesture toward, even when said direction is wafer thin.
And that, dear reader, is the seduction.
You type a mood, an intention, a half-formed thought, and it hands you an almost-finished artefact that looks uncannily like something you might have made if you’d only had more time or fewer meetings. The danger isn’t metaphysical (“is it creative?” please f- off, of course it is, have you actually looked up the definition of creative?1). The danger is how easy it becomes to confuse fluency with thought.
In my view, craft survives when you know what good feels like before you’ve picked up the pen or clicked New Document. And that’s the bit people don’t want to hear. The reasoning, the taste, the internal guardrails — they’re all invisible, they take years. Instead we’ve bred a culture (particularly evident on LinkedIn) where commentary stands in for competence, and Ai’s instant coherence makes that substitution feel almost legitimate.
I’m genuinely unmoved by the theological wrangling over whether Ai creates. If it’s parrot or Picasso. It’s a probabilistic parlour trick. What matters is simpler: whether the person using it can spot when the output stops making sense and is in fact bullshitting. Shallow but shiny. A calculator is harmless until someone who never learned to add starts doing the accounts. As the kids say “Same energy”.
Used properly, Ai is a fast way to think aloud, and as a sole practitioner, it’s become one of my favourite ways to work. It’s my pressure valve. A drafting companion. It pushes out variations I’d never have the patience to make by hand. But it only works because I already have a decades long sense of structure and gut instinct, the bit that quietly mutters “nope, that’s wrong”, or more accurately “what the actual fuck?” before I can articulate why. Without that, the tool becomes the teacher, and its blind spots become your worldview. I keep thinking about graduates walking straight into roles heavily emboldened by Ai before their judgement has even started to calcify and in that sense it’s a bit like giving a BMW M3 to someone who’s just passed their test. The horsepower arrives long before the skill that stops you putting in a hedge.
This is why the creative and consultancy industries feel brittle. So many people want the polished thought without the unglamorous labour that gives it heft. They want the sketch without the sketching. The judgement without the years that make judgement possible. And Ai, obliging thing that it is, makes that performance look convincing enough to fool the untrained eye, and sometimes even the trained one.
None of this makes the technology good or bad. It just makes it pretty shouty. And once a tool starts talking back, the responsibility shifts to the person holding it. Which is really just to say: the work doesn’t get better because the software is clever. It gets better because someone in the room still knows what good feels like.
AI: This piece was assisted with Ai. I used it for the tags, excerpt, the image generation and a little sub-editing. The ideas, references, and anecdotes were, however all mine.
- If you want the long version of why the question “Is Ai creative?” is a trap, Lisa Talia Moretti does a tidy job of dismantling it. She walks through the mess of competing definitions, points out the extent of human labour and data sit behind every so-called “creative” output, and ends up arguing that generative Ai is better understood as a medium than a tool. ↩︎
John Gibbard 