Code Got Automated. Taste Didn’t.

For the last ten years I’ve worked on both sides of the screen, designing interfaces and building them. I always believed you can’t do one well without understanding the other. AI has now shaken both crafts, though not equally, and the difference says a lot about what each one really is.

Two halves of one thing

Design and development aren’t two separate crafts. They’re two halves of the same work, and each one falls apart without the other. AI pushed programming up one level of abstraction, and it could do that because code is deterministic logic with one clear measure of success: efficiency. Design has no single measure like that. It runs on perception, emotion, mood and timing, and all of them keep moving, in directions that even the people who create it struggle to predict.

A designer who doesn’t know how a system works ends up drawing pictures of software instead of designing it. Before you draw a screen, you have to understand it from the inside. What information and functions does it carry? Which entities make up the system, and how do they relate? What data moves through it, and which user needs does it meet? Only after that comes the visible question: how do you put all of it on a screen so a person understands it at a glance?

Good design makes something complicated feel easy, and you can’t simplify what you don’t understand. Development has the same problem the other way round. A developer who doesn’t see why an interface is built the way it is will “optimize” away the parts that made it work.

Code was always logic in a physical form

AI raised programming by one level of abstraction. We used to write the loops ourselves; now we describe what we want and check what comes back.

Looking back, that isn’t surprising. Programming was always the concrete form of business logic: strictly deterministic, where the same input gives the same output. Code has creativity in it, but it’s rigid and formal. To a machine, a for loop and a while loop, or one framework and another, differ only in efficiency, meaning how much time and how many resources they use to achieve the aim. Anything with a clear measure of efficiency can eventually be optimized by something that isn’t human.

That’s exactly where AI is strong. It works on rules, structure and inference, the formal side of thinking.

Design doesn’t have a single metric

Design carries a psycho-emotional layer that code mostly lacks, and it’s often hard to predict. Why was a blue-purple gradient on every landing page last year, while this year everything is orange? Maybe because of a sudden season of funny talking oranges. Nobody can say for sure.

This isn’t only about how things look. It goes down to structure. Navigation used to sit comfortably in a sidebar, then moved to the top, and sidebars still works. The hamburger menu went from clever trick to desktop standard to something people now call bad for discoverability. On phones, the main navigation drifted from a top drawer to a bottom tab bar where thumbs can reach it. Infinite scroll beat pagination in feeds, yet people still want pagination when they need to find their place again. None of these was simply “more efficient.” Each matched how people were holding devices, reading and feeling at that moment.

So a designer’s job isn’t to compute the best layout. It’s to take experience and knowledge, run them through imagination, and turn them into a concrete interface. On a website, a formal function has to become desire. In a product, it has to become a tool that fits the human mind. And you do this knowing that how people see things keeps changing, often because of mood and inspiration rather than efficiency.

One of my university professors understood Hegel’s Science of Logic thoroughly and came to lectures dressed like he was about to play a rock show. He was rigorous logic and a deliberately chosen style in one person, and neither cancelled the other. His style allowed us to feel closer to him and we understood such complex material more easily.

That’s roughly the split I see now. AI belongs to logic: reasoning, structure, the computable side. Design belongs to creativity and artistic vision, to the choices nobody can fully justify yet everybody feels.

If we saw a car only as transport, only its engineering would matter: how efficiently the mechanisms are built and packed under the hood. That part can be calculated, improved and eventually automated.

But nobody buys a car purely as transport. Even trucks, whose shape is dictated almost entirely by function, have plenty to show off to each. That’s where design lives, and it’s why AI has touched it less than code. The machine is getting very good at the mechanism, and I’ll happily hand it the loops. But when I drive to a friend’s place for a barbecue, I’d much rather do it in that black car of my favorite make.