When AI first hit the market, the industry predicted a crash of designers. Many were questioning the point of having a design team at all. ChatGPT was generating usable social media posts, UX/UI was done by Claude in a matter of minutes.
And yet - we are still getting our paychecks regularly. We would even argue designers are more needed than ever.

Back to basics: how does a LLM actually work?

Human creativity cannot be easily replaced. Let alone by large language models.
Let's go back to basics. How does an LLM function? It predicts your next word and generates content based on its training data. The output will vary based on your instructions, your skill level, and your creativity.

Read that again: your instructions, your skill, your creativity. The machine is only ever as good as the human driving it.

The pattern problem

If you live on LinkedIn long enough, you will start spotting AI-generated content within seconds. Instagram is flooded with posts signed by AI. They all look alike, but still a bit different.

Try it yourself: delegate Claude to make 10 posts about any topic, and you will recognise the signature UX, the component style, the font. Same handwriting, different words.

There is a way around generic output, of course. Import your fonts, your design guide, your style, every icon you use - train it on everything visual, and you will be almost happy with the result. Until you run out of tokens.

But who will make that design guide?

The designer will.

Okay, but the ideas?

Perfect, we have resolved the visual part. Now you only need ideas - for posts, slides, and everything else. Because left on its own, Claudy will fall back on its favourite forms, and your profile will blend right in with everyone else's.

But who will come up with the ideas and instructions?

The designer will.

The part AI really can't fake: emotion

Visuals sorted, ideas sorted. But emotions are not something AI can brag about.
Design is all about transferring emotion, communicating through visuals. If your design doesn't make people feel something, customers will dismiss it and file you into the spam folder in their head. In a world where everyone is fighting for attention, you don't have the luxury of ending up in someone's mental spam folder.

To break those generic patterns, we have to roll up our sleeves and do things manually - or as we like to say "pješke" to finally reach that authentic "made by human" moment.

But who will come up with a plan that makes people feel something?

You know the drill by now.

So how do designers actually use AI?

Plot twist: designers use AI every single day. And we love it.

It made the design process so much faster, and it took over the most boring part of our work - the endless testing of different ideas. Ten variations of a layout? Minutes, not days. Testing now costs less, and that allows us to be what we actually are: creatives.

On the other hand, AI resolved one long lasting problem: communication between designers and developers. By prototyping, developers can see how we imagined design to be animated, where certain buttons lead etc. We can feel our creation long before staging release, and so does the client.

Drawing the line

So when we draw the line, AI did to designers the same thing it did to every other industry: it automated the boring tasks. We treat it as exactly what it is: a powerful tool. In areas where it is much faster- specifically, those tedious, repetitive tasks that bore us to tears, we happily let it take the wheel.

What's left is our core. We remain the decision-makers. We dictate where the algorithm stops and where the human touch begins. AI didn't replace us - it just let us do what we fell in love with in the first place.

Creation.