AI didn't replace my skills. It multiplied them.

Every few weeks someone asks me the same question: "Aren't you scared AI will replace you?"

My honest answer: AI replaces the parts of the job I was never paid for. It does not replace the parts I am paid for. And used the right way, it makes me better at those, not just faster.

Multiplication needs something to multiply

Think of AI as a multiplier. Zero times ten is still zero.

If you don't know what good design looks like, AI gives you average design, fast. If you don't understand code, AI gives you code that runs today and breaks next month, and you won't know why. I've seen it up close: an app built almost entirely by prompting can look finished on day one. Then real users arrive, and you learn that looking finished and being finished are two different things.

But if you bring real skill, the multiplier works in your favour. The better you are, the more AI gives back.

For me, the skill that made the biggest difference wasn't a clever prompt. It was testing. AI builds a feature in minutes. I then try it in as many scenarios as I can think of, note exactly what breaks, and send that back. It fixes, I test again. That loop is what turns fast output into something that actually works, and it's the main reason AI became genuinely useful in my work instead of just impressive.

No single AI does it all

I don't have "an AI". I have a team of them, and like any team, each one is good at something different:

One for thinking and code. Planning a feature, reviewing an architecture, writing and refactoring components, finding the bug I've been staring at for an hour.
One for images and visual exploration. Moodboards, lighting references, texture ideas, ten directions before I commit to one.
One for research and writing. Summarising documentation, comparing tools, getting a first draft of copy that I then rewrite in my own voice.
AI inside the product itself. In my warehouse safety simulator, the instructor that talks to trainees runs on Convai inside Unity. There AI isn't a helper on the side. It's part of what the client experiences.

Choosing the right tool for the task is a skill in itself. Asking an image model to plan your architecture, or a chat model to judge typography, gives you bad results and the wrong conclusion: "AI is useless."

Where AI boosts me

Speed on the boring parts. Boilerplate, config files, repetitive layouts, renaming things across fifty files. That time now goes back into design decisions and polish.

Exploring more before choosing. Before, I could afford to try two or three directions. Now I can look at ten, throw away eight, and push the best two further. The final choice is still mine, but it's made from a much wider set of options.

Learning faster. This is the big one. When I moved into 3D, Unity and C#, AI was the teacher who never gets tired of "why?". I don't ask it to just write the code. I ask it to explain the code, then I rewrite it myself. That's how a tool becomes a skill.

A second pair of eyes. I ask AI to attack my own work: find the edge cases, the accessibility problems, the weak spot in the argument. It doesn't get tired and it isn't polite.

What still comes from me

AI can give me ten options. It can't tell me which one is right for this client.

Taste. Knowing when something is almost good and what small change makes it good.
Understanding people. A client says "make it modern". What they mean might be "make it trustworthy" or "make it look like my competitor's site, but better". AI doesn't sit in that meeting.
Judgment. What to cut. When to stop. When the clever idea is hurting the user.
Testing like a human. When code arrives in seconds, the real work moves to checking it. I click through scenario after scenario: the happy path, the impatient user, the small phone, the slow connection, the person who does things in the wrong order. Only when a feature survives all of that do I trust it works the way it should.
Responsibility. When something ships with my name on it, I have to understand every part of it. "The AI wrote it" is not an answer a client accepts, and it shouldn't be.

Testing has quietly become one of the most valuable parts of my process. A small example: the little game on this site passed every automated check and every simulation said it was hard. Then a real person played it and won on the first try. No test suite would have told me that. Playing it, again and again, did.

Take this site. A, the little yellow guide walking around it, was built with heavy AI help: code, testing, tuning. But the idea of a mascot that lives inside my logo, wears a costume for each project and refuses to be caught in a game? That came from a person. AI helped me build it in days instead of months. It didn't invent it.

How to use AI to grow, not to shrink

If you're a designer or developer starting out and you're worried, here's what I'd tell you:

Ask "why", not just "what". Make AI explain its answer. If you can't explain it back, you haven't learned it yet.
Write first, then compare. Try the solution yourself, then ask AI for its version. The gap between the two is your lesson.
Use the right tool for each job. Treat your AI tools like a team, not a magic box.
Never ship what you don't understand. Read every line. Check every claim. Then test it the way a real user would, including the ways they're not supposed to.
Keep building your foundations. Typography, layout, colour, logic, clean code. AI makes strong foundations stronger and hides weak ones only until something breaks.

The real competition

AI won't replace designers. Designers who use AI well will replace designers who don't, and they'll also beat people who use AI with nothing behind it.

The goal was never to type less. The goal is to make better work. AI is the best tool I've had for that, as long as I stay the one holding it.

Want a project built this way, with human judgment and AI speed? Tell A about it.