AI and the taste gap

A while back we wrote that AI would expose the agencies phoning it in. That piece was about the industry. This one is about what the tools did to the work itself, because somewhere in the past two years the fundamental scarcity of creative work flipped, and most of the conversation is still arguing about the old one.

The gap that closed

For the entire history of the craft, execution was the gate. The distance between imagining a thing and holding it was measured in years: learn the software, train the hand, build the muscle that turns intention into pixels or paragraphs. That distance kept most ideas unmade, which was both a tragedy and a filter. Plenty of good ideas died there. So did an ocean of bad ones.

That gate is now mostly open. Type a sentence, receive a landing page. Describe a mood, get forty logo directions before your coffee cools. What comes back isn’t brilliant, but it’s competent, and competent used to be the expensive part. The execution gap, the one that separated people who could make things from people who could only want them, has effectively closed for the broad middle of creative work.

Notice what didn’t close. Nothing about the tools tells you which of the forty directions is right. They won’t tell you that all forty are wrong, which is frequently the case. And nothing in the box will tell you the question the client asked isn’t the question the business has, which is the discovery most projects actually turn on.

The gap that opened

When anyone can make anything, the question stops being can we make it and becomes should this exist. That’s taste, and we mean the word precisely, not preciously. Taste isn’t a fondness for expensive typefaces. It’s judgment about fit: this move, for this audience, at this moment, and not the other two hundred plausible alternatives. And it’s knowing what to leave out, which no generator will ever volunteer, because leaving out is the one thing infinite output can’t do.

The blank page used to be the hard part. Now the hard part is the full page.

Starting is free; stopping is a skill.

A team with these tools can produce more directions in a day than a nineties agency shipped in a year, and every one of those directions arrives looking finished, wearing the polish that used to signal care. Selection is the whole job now. Choosing became the craft.

And here’s the uncomfortable arithmetic: when the supply of competent work approaches infinite, the value of competence approaches zero, and everything that matters migrates into the choosing. The feeds are already demonstrating this. Scroll any platform and you’ll find more well-executed nothing than any previous generation of mediocre work could have dreamed of producing. Volume is the new mediocrity. “Make more, faster” was the lesson everyone took, and it was the wrong one.

Where taste actually comes from

Clients feel this shift before they can name it. The decks arriving in their inboxes got thicker, the options multiplied, and somehow the decisions got harder, because a wall of plausible directions isn’t service, it’s homework with a fee attached. Generosity now looks like subtraction. Showing three options you’d stake your name on beats showing thirty you generated, and knowing the difference is the entire value of having been at this a while.

You can’t prompt your way to it. Taste gets built the slow way: reps, exposure, shipping things, being wrong where people can see you, and caring enough to notice the difference between what you made and what you meant. It’s less a gift than a ledger, a long accumulation of specific nos, each one paid for. The designer who can kill thirty-nine decent options in an afternoon spent years earning the confidence to do it, usually by shipping a few of the wrong ones and living with the results.

Which is why we don’t treat the tools as a threat, and also why we don’t treat them as the work. We use them daily for the parts that were never the point: scaffolding, variations, the tedious middle. What they hand back is raw supply. Deciding what deserves to survive contact with an audience, that’s still done the old way, by people with a point of view and something to lose.

If you’re hiring creative help right now, this is the whole interview: everyone you talk to can make anything. Ask them what they’d refuse to make, and why. The quality of the refusals will tell you more than the portfolio, because the portfolio is increasingly evidence of access, and refusals are evidence of judgment.

The machines have answered “how” more thoroughly than any tool in the history of the trade.

“Whether” is still hiring.