Every few years, technology holds a funeral for something that isn't dead.
In 1981, MTV launched with "Video Killed the Radio Star." Radio skipped its own funeral, went on to become podcasting, and now everyone you know has one.
In 1995, as the web arrived, the designer David Carson published a book called The End of Print. In print. It's hard to imagine a more elegant rebuttal.
Now it's design's turn for the eulogy.
A product partner at Oracle Health once asked me, "Why don't we just prompt it and have AI spit out the UI?" It's a fair question, if you believe design is the interface. It isn't. It never was. Replacing designers with a prompt is a bit like letting the chef go because you bought a very good microwave.
Design isn't dead. In fact, it's now more important than ever.
A brief history of premature obituaries
Every major technology shift follows the same script.
Act one: a confident prediction of the end. Act two: widespread panic. Act three: the value quietly moves somewhere else, and the people who noticed win.
The internet was going to kill the store. Instead, it turned shopping into a competition of experience. Once customers could compare everything in a browser tab, the company that was easiest to buy from won. The internet also invented jobs that didn't exist yet, from information architecture to interaction design to UX. For an ending, it created a remarkable number of beginnings.
ATMs were going to kill the bank teller. Economist James Bessen found that the number of U.S. tellers actually grew as ATMs spread. The machines took over counting cash. The people took over the part customers remembered.
Canva was going to kill graphic design. Instead, it gave everyone the power to make a polished flyer in five minutes, and the world now has a great many polished flyers that look exactly alike. When a company needs to stand out, it still calls a designer, usually right after the fifth identical flyer.
The pattern hasn’t changed.
Production gets cheaper. Judgment gets more valuable. And every time, the obituary turns out to be premature.
AI is the whole box set
AI feels different because it's every one of those shifts in a single release: a new medium, automated production, commodity infrastructure, and an infinite supply of polished output.
That last part is the trap. The polish is the problem.
AI-generated work looks finished, so leadership sees speed and assumes the hard part is done. It's a beautifully frosted cake, and nobody has checked whether it's baked. Anyone who has watched teams vibe code without shared standards knows what comes next: quality drops, and every experience starts to look like it was designed by a different committee, because it was. At Oracle Health, where our users are clinicians and the staff who keep health systems running, that isn't a branding problem. A confusing screen can put patient care at risk.
So the question is back, and this time it's pointed at us: if AI can generate the interface, what exactly are designers for?
Fair question. Here's how I see it.
Design has never been the interface
Design has always been the work of understanding people, framing the right problem, working within real constraints, and creating something that sets a business apart. The screen was never the design. It was just the part people could see.
The real work happens before anything is drawn. It's deciding which problem is worth solving and which is a very expensive distraction. It's seeing the whole journey a person takes, not just the three minutes they spend in your product. It's understanding how people, their context, their environment, and the technology all interact, and what happens when any one of them changes.
The same lab result means something very different to a nurse at the end of a twelve-hour night shift than to a patient reading it on the couch at home. It's holding business goals, technical limits, regulation, and human needs at the same time, and finding the solution that doesn't sacrifice one for the others. None of that has changed with any tool, from the pencil to the prompt.
Some of the best design decisions remove the interface entirely. Tapping a card replaced signing a receipt nobody would ever read. Autosave ended a generation of people shouting at frozen computers. A good default means most people never have to make a choice at all. Nobody has ever admired a screen that isn't there, but someone designed its absence.
Healthcare makes this even clearer. Some of the most valuable design work in a hospital is invisible: making sure a nurse never types the same information twice, or that an alert fires only when it truly matters, because clinicians, like the rest of us, learn to ignore anything that beeps all day. Those are decisions about workflows, data, and timing, not pixels. They shape care more than any screen layout ever will.
The interface is only one place design shows up, and often not the most important one. Design also shows up in a workflow that just works, a default nobody has to change, an alert that stays quiet until it matters, and a step that no longer exists. It has never lived only on the screen. Judging design by its screens is like judging an architect by the paint color.
But the screen was the most visible part of the work, so for decades that's what design was judged by. That's how a discipline built on systems thinking earned a reputation for "making it pretty." And to be fair to our critics, we sometimes encouraged it, measuring ourselves by the craft of our artifacts rather than the outcomes they produced.
AI is transforming design's production layer, making it harder than ever to confuse the artifact with the value behind it.
What remains is the part that always mattered most: understanding people, framing the right problem, and making the judgment calls no prompt can make for you. In a world drowning in generated output, that part matters more than ever.
When the interface disappears
The interface is becoming less visible, more dynamic, and increasingly shaped by AI.
Agents book, buy, and decide on people's behalf. Conversations replace forms. AI generates screens on the fly that no designer ever drew. Somewhere, a pixel-perfect spec is quietly weeping.
Healthcare shows where this is heading. With Oracle Health's Clinical AI Agent, a physician talks with a patient while the agent listens, drafts the visit note, and suggests follow-ups such as labs and referrals. The best version of that experience is one the physician barely notices, so their eyes stay on the patient instead of the screen. But barely noticeable is not the same as simple. What must the physician confirm before an order is placed? How does a draft show what the AI isn't sure about? How fast can a wrong detail be fixed? There's no screen to point to for any of those answers. Every one of them is design.
When there's no fixed screen, design's medium becomes behavior, and the questions get harder, not easier. When should the AI act on its own, and when should it ask? How does it earn trust, and how does it recover when it's wrong? What does it need to know about the person in front of it? What must always be true, no matter what it generates: safe, accessible, compliant, on-brand? "Just prompt it" doesn't answer any of these. It just makes them someone else's problem, usually the customer's.
Conversation is not the experience
There's a related trap: assuming conversation is the whole experience.
Right now, it seems every company is rebuilding its product around a ChatGPT-style conversation, as if one mode of interaction could serve every person, task, and moment. It can't. If people have to ask for every single thing they need, we haven't removed the work. We've traded clicking for typing and handed them the design job.
Most of what someone needs can be anticipated: who they are, what they're doing, what usually comes next.
Good design puts that in front of them before they think to ask, blends conversation with everything else the experience can do, and saves the asking for what is genuinely new, unusual, or an exception. A physician shouldn't have to ask for a patient's latest vitals before every visit; they should already be there.
The best AI experience isn't the one that can answer every question. It's the one that leaves fewer questions to ask.
None of this shows up as a screen, and all of it is design. As the interface disappears, the work doesn't shrink. It moves to where it always mattered most: deciding how an experience should behave, what it should anticipate and what must always be true.
The mandate
For design leaders, the job just got clearer. Shaping how products look was never the only mandate. It only seemed that way because the screen was the one part of our work everyone could see. The mandate has always been to shape how people experience the business, and as the interface disappears, that means shaping how the enterprise behaves with its customers. In practice:
Move upstream. Design must help shape the decisions that determine an experience, not simply the interface that delivers it. That means influencing product strategy, workflows, business processes and the boundaries of AI autonomy before solutions are built.
Codify judgment. Turn what your teams know about users, patterns, and standards into something AI can apply, and own what goes into it.
Decide where AI belongs. AI isn't free; every inference costs money, time, and predictability. Some tasks need a great interface and no AI at all. Some can be handed off entirely. Some need people and AI together. Knowing the difference is design.
Prove outcomes, not output. Agree with your partners on what success means before the work starts, then show what moved. At Oracle Health, that means clinical, business, and experience metrics agreed up front and built into the system, so each experience can be measured against them.
If anything, designers are built for this moment. Reinvention is part of the discipline. We've hand-coded websites line by line, wrestled layouts into divs, then moved from Illustrator to Sketch to Figma. We went from command line interfaces to graphical user interfaces (GUI) then touch and ambient, internet, cloud and now AI. Each shift meant new tools and new ways of working, and each time, designers adapted. What never changed was the work itself. AI is the next reinvention, and it asks more of that work, not less.
Now the job is understanding relationships, behavior, and the rules of engagement between people and the systems that act on their behalf.
That is design with a capital D.
So ignore the obituaries, especially the ones written by people who mistook design for decoration in the first place. As any good investor knows, the noise fades and the fundamentals compound.
The tools will keep changing. Design never has. And it has never mattered more.