Back in 2023 I used AI the way most designers did, just to ideate. It helped me get unstuck, draft a line of UX copy, throw ideas at the wall and see what stuck. Useful, but nothing I would have called important.

A few years later it sits inside almost everything I do as a freelancer. Research synthesis, data analysis, the rough first pass of my wireframes. GPT and Claude took the parts of the job that used to eat whole weeks and made them quick, and I don’t think I’m overstating it when I say they changed what one person can ship.

Two changes stand out. The first is research. I can read across what feels like a thousand papers, pull out the patterns, and put together a few different versions of a user journey before I have drawn anything at all. I understand the product earlier, and more honestly, than I ever could back when research meant five interviews and a lot of hoping.

The second is the wireframe. If I build an interaction-first version before I open the visual file, I can settle the flows and the components while they’re still cheap to change. The bones are set before anyone argues about color.

All of that gave me a kind of confidence I didn’t have before. The honest version is that I ship work now that feels twenty or thirty times better than what I made on my own a few years ago. So no, I’m not here to tell you AI is killing design. It made me better. That’s where I have to start.

And yet. Open any design group right now and you can feel the panic. The loudest posts say the same thing in different fonts: AI makes gorgeous interfaces in minutes, the job is finished, go home. In my own country I’ve watched senior designers tell people who haven’t even started yet to quit. To pick something safer. To give up before the first real attempt.

Some of that is fair. A founder who once needed a small tech team to get an MVP out the door can now do a good chunk of it with AI. That’s genuinely happening and pretending otherwise would be silly.

But most of the doom isn’t analysis. It’s reach. The post does well precisely because it scares people, and the cost of that lands on someone none of us ever meet: the beginner who quietly shuts the laptop and decides the door is closed. We are talking people out of a future to win a little attention, and the argument was never that clean to begin with.

Yes, you can make something beautiful in minutes now. I won’t argue with that. But a beautiful screen and a working product are not the same thing, and anyone who has actually shipped one knows the size of the gap between them. Products get good slowly. Someone notices a small friction, the team argues about it, gets it wrong, tries again. Do that a few hundred times over a few months and you end up with something more than a layout. You end up with judgment sitting inside the interface.

That’s the part I kept calling soul, though I think I had the wrong word for it for a long time. It was never that AI work looks bad. A lot of it looks great. Clean, fast, polished. What it skips is the why. Why this tradeoff and not the other one. Why this flow makes sense for the specific, distracted person who is going to use it on the worst day of their week. AI will hand you a thousand reasonable answers in an afternoon. It still can’t tell you which one is true for these people, in this place. That call comes from someone who has sat with the problem long enough to have an opinion worth trusting.

For years you needed a trained eye to feel the difference between work made with care and work made without it. What I notice now is stranger. People want the human signal back. I keep seeing teams quietly drop “AI-powered” from the way they describe themselves, because the phrase started to read as the opposite of considered. The market is asking for fingerprints again, and to me that’s the clearest sign yet that the human part of this work is going up in value, not down.

So here is what I think AI actually did. It didn’t take the work away from us. It raised the floor under it. It took the slow, expensive parts and made them nearly free, which means you now have almost endless ways to fix a person’s experience, and you can reach most of them fast. But the soul of a design never lived in those parts. It lives in the conversation, the slow back-and-forth between people who see the problem from different sides and argue their way toward something deliberate. A design that looks like it was made with attention usually was. There is no shortcut into that room, and AI hasn’t found one yet.

Which is why the panic is pointed the wrong way. AI got better at everything around the human core, so the human core matters more now, not less. The beginners being told to leave are being pushed out at exactly the moment their taste and their willingness to actually care became the rare thing in the room.

UX is better than I ever imagined it could be when I started messing with these tools in 2023. And the soul of design hasn’t gone anywhere. It just turned into the part worth protecting. If someone tells you there’s no room left for you, listen to what they’re really saying. The easy half got automated. The half that was always the point is wide open, and it’s waiting for you.