AI isn’t replacing designers — it’s replacing the parts that weren’t designed
The designers who panicked were the ones who didn’t like what the mirror showed.
In this article:
Why the designers most threatened by AI were already doing the most replaceable work
The difference between taste and execution — and why only one of them scales
What AI actually can’t touch, and why that’s the only part that ever mattered
A designer I respect sent me a portfolio last month. Thirty-two screens. A checkout redesign for a fintech app. Every corner radius consistent, every component aligned to an 8-point grid, every interaction annotated in a way that would make any design system lead nod approvingly.
I couldn’t find a single decision in it.
Not one moment where someone had weighed two reasonable options and chosen the harder, less obvious one. Not one place where the design solved a problem that couldn’t be described in a Jira ticket. It was technically flawless and strategically empty. The kind of work that AI now produces in eleven minutes.
That portfolio didn’t become less valuable because AI arrived. It was already less valuable. AI just made it obvious.
The mirror nobody asked for
There’s a LinkedIn post circulating that frames designer reactions to AI as five stages — denial, fear, experimentation, integration, and building. It’s neat. It’s wrong.
The real divide isn’t a progression. It’s a binary.
Some designers watched Figma AI generate a layout and thought, “That’s going to take my job.” Others watched the same demo and thought: “That’s going to take the part of my job I never liked anyway.”
The difference has nothing to do with how quickly you adopted the tools. It has everything to do with what you thought your job was in the first place.
If your work was arranging elements on a screen — choosing typefaces, nudging padding, polishing transitions — then yes, AI is coming for you. Not because it does that work better, but because it does it well enough that “well enough” is now the baseline. Figma’s 2025 AI report found that 78% of designers say AI makes them more efficient. Only 47% say it makes them better at their role. That gap is the whole story.
Efficiency isn’t quality. Speed isn’t judgment.
Taste is the new bottleneck
Here’s what I’ve seen happen in practice. A product team uses AI to generate four layout options for a dashboard. All four are competent. All four follow the design system. All four would pass a design review if nobody asked hard questions.
The question is which one to ship. And why. And what happens to user trust when you choose wrong?
AI generates options. It doesn’t know which option serves the person on the other end. It doesn’t know that the third layout buries the one metric the operations team watches every morning. It doesn’t know that the bold card treatment will train users to expect urgency where none exists.
Nielsen Norman Group defined taste as the ability to “orchestrate hundreds of small decisions around a central vision focused on both users’ needs and strategic business goals.” That’s not a skill you learn in a boot camp. It’s earned through years of watching what happens when your designs ship and measuring the distance between what you intended and what actually occurred.
The designers who have that skill aren’t worried about AI. They have more options now than they’ve ever had. The ones who don’t have it are discovering that execution was never the hard part — they just didn’t know because execution used to take long enough to feel like the whole job.
I’ve seen this play out in hiring conversations too. A design director I work with told me she stopped reviewing portfolios for polish six months ago. “I look for the decisions now,” she said. “Show me a screen you argued against shipping. Show me a feature you killed. Show me where you pushed back and why.” Half the candidates couldn’t answer. Their portfolios were galleries, not case studies. Beautiful surfaces with no visible thinking underneath.
That’s what taste looks like in practice. Not refined aesthetic preferences — though those help. Taste is knowing what to leave out. Knowing when a stakeholder request will degrade the experience, even though the data supports it. Knowing that the fourth iteration isn’t better than the second, just different.
Designative put it precisely: “Taste is ‘this looks good.’ Judgment is ‘this will lose users at step three, and here is why.’” When AI can generate a hundred options in minutes, the person who can tell you which three are worth testing becomes the most valuable one in the room.
What AI can’t do for you?
AI can generate a wireframe in seconds. It can’t tell you that the wireframe solves the wrong problem.
AI can write microcopy that tests well. It can’t sense that the tone will feel patronizing to the specific audience you’re designing for, because it’s never sat across from them and watched them struggle.
AI can ship a prototype overnight. It can’t hold the tension between what users say they want, what the data suggests they need, and what the business is willing to sustain — and make a call that balances all three without collapsing into the easiest option.
I’ve watched junior designers use AI to produce work that looks senior. The gap shows up three weeks later, when the design ships and the edge cases start surfacing. When the flow that looked elegant in a prototype breaks under real data. When the component that tested well in isolation creates confusion in the context of the full product.
Decoration scales. Judgment doesn’t.
The question that was always there
The anxiety around AI in design isn’t really about AI. It’s about a question that’s been sitting under the surface for years: how much of what you do is actually design?
Not layout. Not styling. Not the craft of making things look polished. Those matters — but they were never the core. The core is the thinking that happens before a pixel moves. The decision about what to build, for whom, and why this solution instead of twelve others.
If you spent the last five years refining that skill — studying users, debating product strategy, learning to articulate trade-offs — AI is the best thing that ever happened to you. It removed the slowest part of your workflow and gave you more time for the work that compounds.
If you spent those years perfecting the output layer without developing the judgment underneath, AI didn’t take anything from you. It revealed what was already missing.
So here’s one thing you can do this week. Open your last shipped project. Not the Figma file — the live product. Find one decision you made that shaped the outcome beyond aesthetics. One place where your judgment changed the direction, not just the surface. If you can point to it clearly, you’re in a stronger position than you think. If you can’t find one, that’s not a failure. It’s a signal. And now you know where to aim.
That’s not a five-stage journey. It’s a mirror. And mirrors don’t lie.
Sources: Figma 2025 AI Report; Nielsen Norman Group, “Design Taste vs. Technical Skills in the Era of AI” (2025); Designative, “Taste Is the New Bottleneck” (2026); Andrea Grigsby, “AI is coming for our design jobs, but it can’t touch taste” (UX Collective, 2025).
Follow Nurkhon for more on AI, design practice, and the skills that compound when tools get cheaper.