AI is often portrayed as a miracle tool that “does the job for you.” In reality, it’s more like a very skilled assistant—as long as you know how to guide it. Because AI is not a person. It won’t read between the lines. It won’t ask follow-up questions. And every prompt you write is a brand-new start.

From my perspective as a UX designer who works with AI daily, I want to share what this collaboration actually looks like—what it can do, what it can’t, and where human expertise becomes irreplaceable.

🤖 AI Is a Tool, Not Magic

While reviewing the internal presentation Accessibility and AI by Joanna Domagalska, one point really stood out: AI can indeed support accessibility-centered design—but only if you know what you’re doing.

AI can:

• clarify WCAG 2.2 requirements,
• suggest accessible color palettes,
• generate test scenarios for specific components,
• help design for screen reader compatibility.

But it won’t do any of this on its own. You need to ask the right questions. You need to know your context. And you need to drive the conversation as if the other side knows nothing about your project. Because it doesn’t.

🧠 You Need Expertise to Ask the Right Questions

Working with AI in design is entirely dependent on your own knowledge—not something AI replaces. Without that foundation:

• you won’t know what to ask,
• you won’t notice when the AI gives you poor advice,
• you won’t realize when something looks good but breaks accessibility rules.

It’s not enough to ask, “Is this screen accessible?” You have to go deeper:

• “Does the color contrast meet WCAG AA standards?”
• “Are interactive elements large enough for mobile use?”
• “Do button labels provide meaningful context for screen readers?”

Only then will the answers be useful.

🧩 I Have Dozens of Projects. And a System That Works

I love coming up with UX projects and refining them with AI’s help.
📓 Everything starts in my notebook—filled with ideas, sketches, and structure.
📤 I bring those ideas into ChatGPT, where I explore possibilities, test ideas, and generate concepts.
📐 Once they take shape, I copy the results—together with my reflections and decisions—into Figma, where I build out user flows and early layout designs.

It’s not a textbook workflow. But it works.
And I learn something new with each iteration.

✅ A Practical Example: Accessibility Review with AI

When using AI to check accessibility, I follow a structured approach. Even based on a single screen, I can:

• validate key criteria from WAI Accessibility Fundamentals,
• ensure the POUR principles (Perceivable, Operable, Understandable, Robust) are respected,
• assess how the layout considers users with different impairments—not just visual ones.

If you upload a screen and clearly explain its purpose, AI can return highly relevant feedback. But only if the prompt is clear and complete. This isn’t a conversation—it’s a single shot. You don’t get automatic follow-ups. That’s your job.

🧭 Final Thoughts

AI isn’t a magic wand. But in the hands of a designer who knows the rules, understands the user, and can ask the right questions—it’s a powerful extension of our process.
It doesn’t replace us. It amplifies us.
It doesn’t lead. It helps us steer.

That’s my kind of collaboration.
I don’t need another human—I need thinking space, fast iteration, and a tool that lets me explore better solutions.

How do you use AI in your design work?