AI is good at the mathematical half of this job, which is code. It's not good at the part that makes a product feel like something. Here's where I think that leaves designers by 2028.
Muhammad Abdullah Ali
Creative sense is a subjective thing, built out of taste, reference, and judgment that doesn't reduce cleanly to a dataset. There's an enormous amount of design reference floating around online, but it's scattered, unlabeled, and contradictory, and not everyone on earth has the time or inclination to sit down, collect it properly, and build something coherent and on point out of it. That scatteredness is the trade-off AI in design keeps running into. It can remix what already exists convincingly. It struggles to originate a genuinely new aesthetic direction on purpose.
The current AI design stack makes this obvious if you're paying attention. Tools like Claude Design, Lovable, Aura, & Emergent are excellent at generating a clean, functional first draft of an interface. What they can't do is decide that draft is wrong for reasons that have nothing to do with function, reasons rooted in brand, in context, in a hundred small decisions a person with taste makes without being able to fully explain why. That's not a temporary gap waiting for a bigger model. It's a different kind of problem entirely.
Where the Real Shift Is Actually Happening
The bigger change isn't in taste, it's in speed to execution. Tools like MCPs are making it faster than ever to move from a design decision straight into working code, and that's the muscle designers should actually be building right now. Not to become full stack developers. To stop being the bottleneck between an idea and something real running in a browser.
What that looks like day to day is a designer who understands business logic and how systems fit together well enough to move fast without an engineer translating every decision for them. That's a different skill than knowing how to draw a beautiful interface. It's closer to fluency.
Not Every Kind of Designer Needs the Same Answer
I don't think this shift lands the same way across every corner of design, and I'd push back on anyone treating it as one universal mandate.
Brand designers, in my opinion, don't need to become semi-technical. Their entire job is creative output, and their actual leverage is learning the creative AI tools well enough to move faster on that output, not learning to code.
Web designers are a different story. I think by the end of 2028, most competent web designers should be able to design and ship an entire website with AI assistance, start to finish, without a developer in the loop for anything standard. The tools are already most of the way there.
Product designers should be the ones using AI the most. Even if they never ship production code themselves, they should understand how to use AI inside their actual workflow, know enough to build a working prototype in code, and have a working grasp of system level thinking. Not to replace engineers. To stop proposing things that are quietly impossible and contribute to fast-paced growth of businesses and products.
So no, I don't think AI in design replaces design, and I don't think that's even the interesting question anymore. The interesting question is which designers close the gap between deciding something and shipping it, and which ones keep waiting for someone else to translate their Figma file into something real. By 2028 I think that gap is the actual dividing line in this profession, not AI versus human.
