AI is not just another shiny thing

When ChatGPT exploded, so did a new wave of optimism (and FOMO) across the tech world. Suddenly, every product needed to “have AI.” I’ve been in more than one meeting where someone suggested “we could add a chatbot here” before we even knew what problem we were trying to solve.

And that’s the thing: AI isn’t magic. It’s not decoration. It’s a tool—powerful, yes—but like any tool, it needs context, clarity and care.

As UX designers, our job isn’t to make something “look smart.” It’s to make it feel right. To ensure it solves a real pain, at the right moment, in a way that feels useful, honest and trustworthy.

Traditional Discovery vs. AI Discovery: it’s not the same beast

If you’ve ever led a product discovery, you know the drill: map the goals, talk to users, sketch ideas, test them, iterate.

But when AI enters the scene, that script doesn’t hold up.

Here’s how it usually changes:

It’s messier. More technical. And you can’t separate design from data anymore.

Four UX Focus Areas That Matter (A Lot)

Let’s break it down. If you’re working on an AI project, these are the four lenses you cannot skip:

1. 📍 Problems + Data

Sometimes a problem sounds great… until you realize there’s no data to support it. Or the data exists, but it’s outdated, biased, or incomplete. That’s why you need your Data Scientist friends at the table from day one. Not after the interface is done.

2. 🤖 Expectations of Intelligence

Here’s a truth we often forget: users don’t care if your model is 94% accurate. What they really care about is: “Did it help me?”

It’s not about numbers. It’s about the feeling of being understood. Design for that feeling.

Example? A user asks for help, and the AI says: “I’m still learning, but here’s what I found. Was this helpful?” That one sentence changes everything. You’re not a machine. You’re a partner.

3. 🔐 Trust & Control

Let’s face it—AI makes mistakes. And when it does, the worst thing you can do is hide it.

Be transparent. Show when something was AI-generated. Let users edit. Let them undo. Trust grows in the little details.

4. 💬 Conversational or Assisted UX

If your AI talks, the UX is the conversation. And that means:

• Avoid robotic answers.

• Design smart prompts.

• Use a tone that matches your brand and your user’s mood.

An assistant that answers like a spreadsheet? Not helpful. One that’s concise, clear, and maybe even warm? Way better.

Tools that make sense (and save time)

So, how do we navigate all this complexity? With a few well-placed tools:

Lean UX Canvas (for AI)

Yes, the classic canvas—but updated. Add blocks like “AI capability,” “dataset bias,” or “success criteria for the model,” and you’ll have a much richer picture from the start.

📊 AI Opportunity Map

Think of it like a prioritization compass:

• High user value + High feasibility → Do it now.

• Low value + Low feasibility → Delete that idea immediately.

  
  

🧭 Decision Maps + Confidence

Does your AI act on its own? Ask for confirmation? Show confidence levels? This map helps you decide what happens when things go wrong—and they will.

🔍 Dataset Evaluation (from a UX lens)

This is not just data science territory. As UXers, we must ask:

• Are all user groups represented?

• Are there blind spots or harmful biases?

• Will this dataset lead to fair experiences?

  
  

If the data’s broken, your product’s empathy is too.

One last thought

“We design for people. And most of them aren’t looking for artificial intelligence. They’re just trying to get something done without getting frustrated.”

AI should never be the headline. It should be the quiet engine making things smoother, smarter, simpler.

Our role is to shape those experiences—not with fear or hype, but with intention, creativity, and a strong sense of responsibility.

Let’s keep asking better questions. And let’s keep designing for the people on the other side of the screen.