The obsession with taste is setting designers back

Hey, Miranda here 👋

There’s a word all over my feed right now, and it’s being sold to you as the thing that saves your career from AI.

Taste.

The pitch is everywhere: AI can churn out “good enough” in seconds, so the one thing you’ve got that it doesn’t is taste. Your eye. Your instinct for what’s good. Build your taste and you’ll be fine.

But that is absolute BULL, and anyone selling it, a lot of them “design leaders” who should know better, are setting designers up to fail.

Because the obsession with taste isn’t protecting us. It’s dragging us back about a decade, right when we’d finally clawed our way out of the “make it pretty” corner.

I will die on this mountain so let me explain….

In this article:

• Why “taste” is most harmful advice being handed to designers right now

• Why AI slop is really getting shipped, and why taste does nothing to stop it

• The exact language to argue a design decision with evidence instead of your eye

• A copy-pasteable acceptance-criteria checklist for AI and vibe-coded work

• One challenge for your next design review

If a designer walked into one of my design reviews and defended a decision because it was more tasteful, I might laugh them out of the room.

I mean it. “It’s more tasteful” is not an argument. It’s a preference that I could rip apart in seconds.

And that’s when you’re up against designers, who at least share some vocabulary. Now take it into a room of non-designers.

You think you get a lot of opinions in your reviews now? Put taste at the front and watch what happens. Not a f*cking chance you walk away with a win. Every PM, every eng lead, every stakeholder in that room has “taste” too, or thinks they do, and the second you make taste the reason, yours is worth exactly as much as theirs. Just because you’re the designer doesn’t mean you have more authority on taste… anyone with even 6 months of experience in this industry knows that.

That’s the problem with taste as a career strategy. You can’t argue it and you can’t own it.

We spent years dragging design out of the “just make it pretty” box. Taste-as-strategy marches us straight back in, except now we’ve slapped a sophisticated label on it and called it a moat.

AI slop getting shipped into products is very much the reality we’re all living in. Stuff going out the door because anyone can prompt a screen now and it looks done. It’s a genuine problem and it’s getting worse.

But taste is not the answer to it.

Because “we just need more taste” is unmeasurable, subjective, and impossible to defend. You can’t scale taste across a team. You can’t put taste in a ticket. You can’t look at an engineer vibe-coding a feature and go “make it more tasteful” and expect a single thing to change.

Even the people building the AI think taste is a shaky place to plant your flag. Jenny Wen , who lead design for Claude over at Anthropic, said designers may be holding onto taste and judgment as a moat “a little too tightly,” and that AI is going to keep getting better at exactly that.

The designers who come out of this stronger won’t be the ones with the best eye. They’ll be the ones who can do the thing AI can’t: connect a design decision to something the business already cares about, and defend it with more than “trust me, I’ve got taste.”

That’s the skill. Evidence. Reasoning. The thing I bang on about constantly.

Here’s the difference, and it’s not subtle:

✘ “This layout feels cleaner.” → ✔ “This layout reduces complexity to improve bounce rate.”

✘ “This pattern is more elegant.” → ✔ “This matches the mental model people already showed us in testing, we’ll reduce support tickets coming from this flow.”

✘ “It’s more tasteful this way.” → ✔ “This version makes it easier to read the primary action, which is the thing costing us conversions right now.”

The left loses to a stronger opinion. The right column illustrates real reasoning. You’re not defending a preference anymore, you’re defending an outcome to benefit the business. Much harder to overrule that with “well, I don’t love it.”

So when I say taste is setting us back, this is what I mean. Every hour you’re pouring into “developing your taste” is an hour you’re not learning to make a real case. And that case is the whole difference between the designer who gets seen and heard in a room vs one who’s completely bypassed. Guess which one keeps their job through this wave of change?

Keep these somewhere you can grab them fast. The second you feel “I just think this is better” forming in your mouth, reach for one of these instead:

Anchor to a metric: “The reason I went this way is [metric]. Here’s the number it ties to and here’s what I expect it to move.”

Pull rank with evidence: “I hear the preference. What I’m weighing it against is [research or data]. If there’s evidence pointing the other way, we can change it.”

Break a taste-vs-taste stalemate: “We could go around in circles on which one feels better. Can we agree what we’re optimising for, and let that make the call which we think will be most effective?”

That last one is what lifts the whole room off taste, yours and theirs, and drops it onto the only ground where design reliably wins: what are we here to do, and which option gets us there.

This is the entire thing I teach in Why Before UI.

Become the designer who can defend any decision in the room. Find the why underneath a design decision and learn to defend it in a way non-designers listen to. Taste gets you overruled. A real, strong defence gets you recognised.

“This forced me to articulate my thinking and ask ‘why’ for every decision. It’s changing how I approach design entirely.”, Sammie, Sr. Product Designer

👉 Learn more →

Got a team that keeps defaulting to “I just like it better”? It works well as a shared foundation.

So if taste isn’t the answer to AI slop, what is?

Guardrails. Acceptance criteria. The boring, defensible, scalable stuff nobody’s putting on a carousel because it doesn’t sound sexy.

Ai slop gets ships because vibe-coded and AI-generated work is going out with no quality threshold to clear first. No definition of done. So “looks finished” becomes the bar by default, and “looks finished” is the one thing AI is world-class at.

Taste tries to catch slop by feel, one designer at a time, after the thing’s already built. Acceptance criteria catches it by definition, across the whole team, before it ships. One of those scales to a company. The other one is you, tired, spotting problems in a review after the fact.

So here’s a starter checklist. Nick it, adapt it, make it yours. Before anything AI-generated or vibe-coded ships, it clears these:

• ✔ Named user goal. What’s the user actually trying to do here? If nobody can say it, it doesn’t get a pass.

• ✔ Real content, real states. Does it survive real data, empty states, errors, and the longest realistic string? Or does it only work in the happy-path demo?

• ✔ Accessibility floor. Contrast, focus states, keyboard nav, labels. This is a floor, not a polish pass you get to skip.

• ✔ Decision you can trace. Can someone point to why this pattern? Tied to research or a number, not “the model suggested it and it looked nice.”

• ✔ Failure handled. What happens when it breaks, hangs, or gets junk input? AI doesn’t check by default. A standard does.

• ✔ Fits the system. Does it reuse what you’ve got, or did it invent three new components someone now has to maintain forever?

Not one line of that needs taste. All of it is teachable, ticketable, and defensible to entire teams. A standard you can hand to someone else beats an instinct only living in your head.

That’s the craft moat, if we’re going to insist on having one. Not “my eye is better than the machine’s.” It’s “I know how to build the guardrails that keep what ships at quality, whoever, or whatever, built it.”

In your next design review, catch yourself before you defend anything with a feeling.

Every time you’re about to say “cleaner,” “nicer,” “more elegant,” “more tasteful”, stop. Find the outcome sitting underneath it. What is that cleaner layout actually doing for the user or the business? Say that instead.

Just once. One review. And watch how differently the room treats you when you stop handing them your taste and start handing them a reason.

And if you’re one of the “design leaders” out there telling your people to go build their taste: sit down and think long and hard about what you actually need from them. The fix isn’t hiring a better eye. It’s defining the quality bar everything needs to pass before it ships. That’s your job. Do that instead.

I get why taste is the comfortable answer. It flatters us. It tells us the thing we’ve always been proud of, the eye, is the thing that saves us. Of course that’s the version everyone wants to hear.

But it isn’t, and I’d rather tell you that now than watch you lose your next design review. The designers I’ve watched become genuinely impossible to overrule didn’t get there on taste. They got there because they could tell you exactly why, tied to something the business already cared about, and stand behind it when the room pushed back. That’s harder than having a nice eye. It’s also the only version of this job I don’t think AI is coming for.

Build that one.

I’m happy you’re here.

-Miranda

💬 Chat with Miranda — does your portfolio lean too much on taste andnot enough on strategic judgement? Drop it in and I’ll help you swap the taste argument for the evidence one Try it →

Know a designer who’s been sold the taste gospel? Send them this. Better they hear it from me now than lose the argument later.

Share