There's a conversation I keep bumping into lately. It goes like this: UI designers who can't at least vibe code a working prototype are going to become irrelevant. The idea is that a modern designer should just be able to fire off a few prompts, and suddenly you have a running, nearly production-ready solution in your hands.
I'd like to offer my two cents to the conversation by talking about two forks.
First up, have a look at this one from the Fiskars Functional Form series, designed by the Norwegian maestro Tobias Jacobsen. I think it's one of the most beautiful, well-considered, and quietly enduring interfaces to eating your everyday meals that exists. In fact, I like it enough that I've bought a set for both the Taiste office and another for home.
I have no idea how long Tobias spent on it, or how long it took Fiskars to get it onto a shelf — but honestly, does it matter? I'm pretty sure I'll be happily using these things until the end of my days.
Now, let's compare this to a fork envisioned by ChatGPT. The prompt was deliberately lazy: don't think about the design aspect at all, just make me something simple that works and that I can 3D print quickly. A print-ready STL file ready in 2 minutes and 8 seconds and the printing took 33 minutes and 39 seconds. You could argue — without being technically wrong — that I had a working fork in production in less than an hour.
Of course, I'm being a bit cheeky here on purpose. It's true that we are also able to produce things with AI that are genuinely good-looking and functional. But – and it's a big but – only when we already have a clear idea on what we want to make and why. Asking and seeking answers to these questions have arguably always been the designers' real job. Before a tool finds its way into our clammy hands, someone needs to consider things such as:
- What problem are we actually solving here?
- Who's this for, and what frustration does it take away?
- What should this do — and what should it definitely not do?
- How should it look, feel, and behave?
- How do we help people understand why they want it?
A designer facilitates, thinks, sketches, prototypes — and yes, codes too, if that's their thing. But then they test, talk to users, iterate, and help navigate decisions inside messy, complicated groups of people.
All of that can now be done faster with AI. That's real, and it's exciting. But it doesn't mean every designer needs to be writing code — not unless they're drawn into that direction in their careers.
At Taiste, different designers have a role to play during different parts of the process. Some have a knack for prototyping and interaction, others for research or service design.
My point is that AI should strengthen the specialisation path you're already on — not pressure everyone into the same mould. Something that has genuinely surprised me in our internal development conversations is that alongside the interest in AI prototyping, many of our designers are expressing an interest in honing their service design skills even further. This is something I've been really glad to see.
To summarize: AI speeds things up, design points the way. If you haven't thought carefully about the design's direction, you'll just get lost faster. The people who come out ahead will be the ones still asking: why and for whom are we making this fork?
