If you think designers will disappear in the AI age, you misunderstand what design is
"AI will replace designers" is trending and I've said it myself in pitches (coming from a designer, it grabs attention pretty fast). But most people repeating it don't understand what it really implies. It reduces design, and UI specifically, to something it isn't, and exposes how little most people, even in tech, understand what designers actually do.
So here is a quick intro to software/interface design:
An interface isn't paint. It's the layer that lets two sides communicate. In French, interface can even mean the person translating between spoken and sign language: the mediator who makes meaning flow between systems. User Interface plays that role between humans and software, and this is what designers work on. Buttons and colors are the visible surface; the interface is the structure that makes information understandable and usable.
Why are interfaces useful?
Because humans don't handle unstructured information well.
We need stable patterns to grasp complexity.
Cognitive scientists call this Distributed Cognition: our intelligence doesn't live in our heads alone. It lives in the tools and structures we think with.
Text and voice force a fixed order. We think faster spatially, by scanning, comparing, spotting anomalies at a glance. Most of the time, we don't need to be told; we need to be shown.
That's Pre-attentive Processing: the brain detects color and spatial shifts in under 250ms, long before we've "read" a single word.
The more complex the system or information, the more we rely on visual stability to know what's normal and what's off.
This is why the fantasy of "AI fully replacing UI" or "AI-generated UI on the fly" falls apart in most use cases.
I don't think AI will fully replace designers¹. If it does and ends up generating "designs" on the go, it'll be a terrible experience. Here's why:
Imagine living in a house with a room that changes based on what you need: the Multipurpose Intelligent Room™ (perfect since we couldn’t afford our dream house)
On Monday morning, the room is a kitchen. A table sits near the window that gets the morning light. The coffee maker has already started your coffee, it smells good and you’re saving $23 on Starbucks. A toaster waits on the table, plugged into an outlet installed right there so you don't have to stand up. Ideal setup. You like it.
The next day, it appears the same. You develop muscle memory: walk in, pour coffee, sit, eat toast.
On Wednesday, the AI optimizes the room. You mentioned you had a hard time falling asleep last night, so it decides you should avoid coffee today. The counter and the coffee maker are gone. The kitchen has become a reading nook designed to help you relax. The window is larger because the Multipurpose Intelligent Room™ wants you to catch more morning light for melatonin timing. The outlet has disappeared, because the room doesn't want you to be able to charge your phone.
Technically, the room is correct. The walls are solid. The physics are sound. But the room isn't what you were expecting, and now you're standing in the doorway trying to remember how to start your day.
This is what "Generative UI on the go" actually feels like. Interfaces that constantly shapeshift to match inferred intent sound efficient but they break a fundamental biological need: spatial memory.
When the interface keeps shifting because AI is trying to "optimize" the layout, humans lose the one thing that makes them fast: pattern recognition. You violate Hick's Law and destroy motor memory. If the button moves, your brain has to stop and search for it instead of just act.
A constantly generated UI means:
- You don't know where to look.
- You can't tell what's unusual.
- You're always re-learning the product.
Novelty becomes noise. And according to the Cognitive Load Theory, noise kills expertise, speed, and trust.
We don't want a new map every time. We want a familiar map with new insight inside it.
Familiarity isn't nostalgia. It's a cognitive requirement.
And that's why UI won't disappear. Not even close.
What disappears is the annoying UI: the parts that existed only because software couldn't think. Endless forms. Configuration screens. Toggles for features software couldn't decide on its own. Rigid workflows. Repetitive searches and queries. This is incidental complexity, and AI is perfect for automating it away.
But the UI itself? The spatial map? The stable surface that lets humans see the world at a glance?
That's the part AI makes more valuable, not less.
The future isn't "no UI" or “AI-generated UI that changes every day.” It's a stable UI layer with AI-generated content. It might not always live inside a window or a browser, but it will still have to act as the interface between systems and the human brain: reducing cognitive load rather than increasing it.
A consistent structure where:
- The hierarchy doesn't shift.
- The navigation stays familiar.
- The anchors remain in place.
And inside this structure, the information becomes richer and sharper: contextual, prioritized, analyzed, highlighted.
The kitchen stays a kitchen. The coffee gets better.
AI handles the complexity. UI handles the cognition.
The winning products of the AI era won't be the ones that remove UI. They'll be the ones that respect how humans actually think and use AI to elevate the content inside the interface, not destabilize it.
AI doesn't kill UI. AI kills the parts of UI that were only ever compensating for our software's limitations.
The rest stays. Because humans still need a place to look.
And since designers don’t have to waste time thinking about forms and explaining how to vertically align content to engineers anymore (thanks to AI), they can finally bring some delight to the experiences they design… so when you look, it feels a bit like the magic AI products said they’d bring in the first place.
References
- Hutchins, E. (1995). Cognition in the Wild. (Distributed Cognition)
- Treisman, A. (1985). Preattentive processing in vision. (Visual search and patterns)
- Sweller, J. (1988). Cognitive load during problem solving. (Mental effort and noise)
¹If AI replaces designers, it'll be by changing the way they work, which might lead to changing their job titles. You might call them Product Engineers, or Design Engineers, or Experience something. I don't know where we will land, but one thing is for sure, we'll always need someone to create the surface between humans and machines, however smart the machines become.