Ask a room of designers how they feel about AI and you’ll hear a lot of worry. Is this going to take my job? Why would anyone pay for a designer when a tool can make a screen in ten seconds?
I understand the fear. But I’m seeing something different, from two places at once. At work I lead a design team that has been piloting AI in our design-system workflow. Outside work I build my own products with AI agents writing the code, including a walking game that’s now on the App Store. I’m a designer, not an engineer, and a few years ago neither of those things would have been possible for me.
What I’ve learned from both is that the job isn’t shrinking. It’s getting bigger, and in a lot of ways it’s getting better.
What’s actually changing
The cost of making things is falling fast. A screen, a component variant, a first draft of documentation, a working prototype: all of it takes a fraction of the time it used to.
The cost of deciding hasn’t moved at all. What should we build? Who is it for? What happens when it goes wrong? What should this feel like? Which of these ten options is the right one, and why? AI doesn’t answer those questions. It answers whatever question you ask it, quickly and confidently, whether or not it was the right question.
That’s the shift. When making gets cheap, the value moves to judgment. And judgment is what good designers have been building their whole careers.
Everyone can make things now
AI has opened up making to everyone. Anyone with an idea can produce an app, a website or a brand in an afternoon. I think that’s a good thing: more people can build, test and share ideas than ever before.
It also means a flood of what people have started calling “AI slop”: products that technically work but don’t really work for anyone. That’s not because the people making them aren’t smart. It’s because most of them have never worked on a product with real users. They haven’t watched someone get lost in a flow or sat in on a usability test, and they never learned the principles designers grew up with: start from the person’s problem, make the next step obvious, plan for when things go wrong. The tool will happily build whatever you ask for. It won’t tell you that you asked for the wrong thing.
So there’s going to be a lot of stuff out there that isn’t very good. That’s exactly where the opportunity is.
Making things is easy now, and it’s only going to get easier. Making really good things isn’t. A well-crafted product, one that understands its users, feels considered and earns trust, will stand out more against a sea of slop, not less. And the people who know how to make that kind of product can now do it with a fraction of the team. Products that used to take thirty or forty people can now be made by a designer with strong product judgment and good AI tools, working with a handful of people or sometimes alone.
That isn’t the end of design. It’s design’s moment.
The bright side
Designers can build what they design
This is the change I’m most excited about. I designed ROAM, a game where real steps move you through a fantasy world, and AI agents built it. About six weeks after the first line of code, it was on the App Store.
I didn’t hand off a spec and hope. I made every product decision, reviewed every build on my own phone, and sent back every change. The gap between “I designed it” and “it exists” used to be a team of engineers and a few months. Now it’s mostly my own clarity about what I want.
That doesn’t replace engineers on real teams. It does mean designers can prototype in the real material, test with real people sooner, and walk into a planning meeting with something that works instead of a picture of something that might.
More exploration, less production
My team at Slalom piloted AI in our design-system work: generating component variants, drafting pattern documentation and catching inconsistencies in our design tokens. None of it replaced a designer’s judgment. What it replaced was the slow, repetitive part, the part that used to eat the afternoon before we got to the interesting work.
When generating an option costs almost nothing, you can look at more of them. The designer’s job moves from producing one careful option to exploring many and choosing well. That’s a better use of a designer’s time, and it’s closer to what most of us got into this for.
Learning faster than ever
Building with AI has been the most useful design education I’ve had in years. When I don’t understand why something works the way it does, I ask, and I get an explanation tailored to exactly what I’m looking at. I’ve learned more about databases, permissions and app review in the last six months than in the ten years before.
That makes designers better partners for engineers. Knowing how a thing gets built changes how you design it.
Time back for people
Every hour AI saves on production is an hour that can go to the parts of the job only a person can do: sitting with users, understanding the business, building trust with engineering partners, coaching a junior designer through a tough critique. I’d rather spend my time there.
What stays ours
If AI handles more of the making, here’s what’s left, and it’s most of what matters:
- Taste. Knowing what good looks and feels like, and why. AI produces the average of everything it has seen. Taste is knowing when average isn’t good enough.
- Context. Knowing the people, the history, the politics, and what was tried before. None of that is in a prompt unless someone puts it there.
- Judgment about people. In ROAM, the game never tells you to rest. It rewards rest, and lets you feel the difference. That’s a decision about how people learn, and no tool made it for me.
- Clear intent. AI builds exactly what you describe. Being able to describe precisely what you want, and why, is now one of the most valuable design skills there is.
- Saying no. When building gets cheap, the temptation is to build everything. Someone has to protect the product from that.
The role is moving from maker of artifacts to author of intent: the person who defines the rules, sets the bar, and checks the work. That’s a more senior job, not a smaller one.
The cautions
I’m optimistic, but not naive. A few things I’d watch for:
Fast and wrong is still wrong. AI will build the wrong thing well, and quickly. The speed makes mistakes easy to miss, because everything looks finished. Review the work like it came from a confident new hire: assume it’s plausible, then check whether it’s right.
Don’t ship placeholders. It’s easy to put a button on screen before the thing behind it exists. Apple rejected ROAM partly for exactly that, and the reviewer was right: anything on screen is a promise.
Watch for sameness. Default output looks like everybody else’s default output. If your product looks like what the tool suggests first, it’ll look like everyone else’s too. Taste is the thing that pushes past the first answer.
Don’t automate the listening. AI can summarize research, but it can’t replace talking to people. A synthetic answer about what users want is a guess with good grammar. The best insight on my team’s platform came from interviews and surveys with real employees.
Design the consequences. When software acts on someone’s behalf (sending a message, sharing data, deleting something), what happens, who sees it and whether it can be undone are design decisions. Moving faster makes them more important, not less.
Measure trust, don’t assume it. When we added an AI assistant to our employee platform, every answer got a simple “Helpful / Didn’t help” control. That’s how we learn where the assistant earns trust and where it doesn’t, instead of guessing.
Protect how people learn the craft. If AI does all the junior work, where do junior designers learn? Leaders need to make room for people to build skills by hand, even when a tool could do it faster. Judgment has to come from somewhere.
What I’d tell a design leader
- Pilot it with your team, not at them. Pick real work, try it together, and talk honestly about what helped and what didn’t.
- Write your standards down, clearly. When a designer reads a vague guideline, they ask a question or use common sense. AI doesn’t. It follows what’s written, fills any gaps with a guess, and hands back something that looks finished. The clearer your design system and guidelines are, the better the work AI produces from them.
- Reward judgment, not volume. If AI can triple output, output stops being the measure. Ask what got decided well.
- Make learning part of the job. The designers who thrive will be the ones who stay curious. Give them time to experiment.
The future I see
I might be wrong about this, but I don’t think the future is a world of more “UX designers.” I think the titles are going to change. The lines between UX, product and engineering are already blurring, and I can see a future where one person is responsible for a whole product: understanding the problem, designing the experience, and building it end to end with AI.
What I don’t see changing is the value of design thinking. Understanding people, framing the right problem, making the next step obvious, planning for what happens when things go wrong: those principles don’t lose value when anyone can build. They’re the difference between a product people trust and one more piece of slop. Whatever the title says, the people who carry that thinking will be the ones making the products worth using.
Anyone can make things now. Not everyone can make them good. The tools will keep getting better at making things. Our job is to keep getting better at knowing what’s worth making. That’s always been the job. Now we get to do more of it.