Two People, One Prompt

There's a version of the AI story that goes like this: the tools have gotten so good that the skill gap is closing. Anyone can generate a beautiful image now. Anyone can produce professional-looking content. The playing field is level.

I've been generating images professionally for long enough now to tell you that this version is wrong. Not slightly off: structurally wrong.

The experiment

Try it yourself. Take any AI image generator (Midjourney, NanoBanana, Firefly, whatever your tool of choice) and give the same brief to two people. A designer with ten years of craft experience and someone who has never studied visual composition in their life.

Same prompt. Same tool. Same starting conditions.

The outputs will not be equal. Not close to equal.

The designer's output will have something the other one doesn't. Not because they wrote a better prompt (although they probably did) but because of everything that comes before the prompt. The designer knows what they're looking for. They have a reference point, an aesthetic judgment, a sense of what works and what doesn't. They iterate toward something specific. They recognize when an output is 80% there and know exactly what to regenerate to close the gap.

The person without that background iterates too. But they're iterating in the dark: they know they don't love the output, they're not sure why, they're not sure what to change.

The gap doesn't close with better tools. It moves.

What the tool actually amplifies

AI image generation doesn't replace artistic judgment. It amplifies it.

If you have taste, these tools let you produce at a speed and scale that was previously impossible. Ideas that used to require a full production day can now be prototyped in an afternoon, refined the next morning, finalized by lunch. You can explore ten creative directions in the time it used to take to explore one.

If you don't have taste, the tools give you volume. You can generate hundreds of images. Most of them will be technically competent. Some will be impressive, in a generic way. None of them will be distinctively yours, because there's no distinct point of view driving them.

This is actually the most important shift the technology has created, and it's the opposite of what most people expected. The expectation was democratization: everyone gets access to the same quality. The reality is stratification: the gap between the person with an artistic eye and the person without has gotten larger, because the person with an eye can now move at scale.

The first-take fallacy

The AI image generation world has created a specific misconception that I run into constantly with clients: the belief that a good result should come on the first generation.

It never does. Or almost never.

The process works like this: you generate, you evaluate, you identify specifically what's wrong: the lighting is too flat, the composition is too centered, the expression is generic, the background is competing with the subject, and you regenerate with that adjustment. Then again. And again. Each generation gets closer to what you're looking for, but only because you know what you're looking for.

Clients who understand this are a joy to work with. They come in with references, they give specific feedback, they trust the process. What we produce together is genuinely better than what either of us would produce alone: their product knowledge and context combined with our visual judgment and tool fluency.

Clients who expect first-take magic are harder. Not because their expectation is unreasonable in itself, but because the pressure to deliver on the first generation pushes you toward the safe and the generic. Safe and generic is what AI does easily. Original and specific is what takes iteration.

What this means for the industry

I've been in creative production for a decade. I've watched the industry absorb digital photography, Photoshop, Instagram filters, stock photo libraries, 3D rendering. Every time, the conversation was the same: this will commoditize the craft, eliminate the professionals, make it all too easy.

Every time, what actually happened was that the baseline quality rose, the commodity layer expanded, and the people with genuine craft found a new ceiling to operate at. The mediocre players got replaced by the tools. The strong players got amplified.

This wave is bigger than the previous ones; I'm not dismissing that. The tools are genuinely more powerful, the pace of change is genuinely faster, and the volume of AI-generated content flooding every channel is genuinely going to make it harder for anything to stand out.

But the solution isn't to compete with volume. It never was.

The solution is to have a point of view. A specific aesthetic. A visual language that's recognizable across executions. That's what the best agencies and the best creative directors have always had, and it's what no generator, however powerful, produces from a neutral prompt.

The prompt is not the skill

Here's the thing I want designers and creative professionals to internalize: the prompt is not the skill. The prompt is the interface.

The skill is what comes before the prompt and what comes after it. The knowing what you want, built from years of looking at good work and understanding why it's good. The evaluating of what came out, the recognizing of what's off and why. The iterating with intention, not at random.

These are not skills the tool teaches you. They're skills you bring to the tool.

If you have them, right now is an extraordinary time to be in this industry. The leverage available to a creative person with good judgment has never been higher.

If you don't have them, generating more images won't help.

Same prompt. Two people. The artistic eye wins. Every time.