All posts
AI Product Design UX Research Design Thinking

The AI Design Trap: A Designer's Reality Check

The AI Design Trap: A Designer's Reality Check

Last month, I watched a product owner on my team spend hours trying to get ChatGPT to generate functional requirement documentation for an analytics admin dashboard. The result? Generic requirements that could have been for any dashboard, anywhere in the world. No mention of the specific data visualisation needs of our telecom clients, no understanding of how our administrators work in low-bandwidth environments, nothing about the reality of managing customer data across multiple African markets with different regulations.

This got me thinking: we’re using AI like a magic wand, expecting it to solve our design problems without putting in the real work. After a year of integrating AI into my design process for telecom platforms across West Africa, I’ve made enough mistakes to know what doesn’t work and discovered what does.

The AI Trap Most Designers Fall Into

Here’s what I see happening everywhere: designers treat AI as a replacement for thinking rather than a tool for thinking better. We’re asking it to do our jobs instead of helping us do our jobs better.

The shortcuts that backfire:

  • Generating entire user journeys without talking to actual users
  • Creating wireframes based on “best practices” without understanding local context
  • Writing copy that sounds polished but doesn’t speak to real people
  • Building personas from thin air instead of research insights

I’ve been guilty of all of these. Six months ago, I used AI to generate interview questions for user research on an AI-integrated messaging platform. The questions were grammatically perfect and followed UX best practices. But when I sat with users, half the questions didn’t make sense in their context. I had to throw out my script and have real conversations instead.

That’s when it clicked: AI doesn’t understand context the way humans do. It can’t replace the messy, uncomfortable work of truly understanding the people we’re designing for.

What I’ve Learned Actually Works

Instead of using AI as a crutch, I’ve started using it as a thinking partner. Here’s how my process has evolved:

1. Use AI to Prepare, Not Replace Research

Before user interviews, I feed AI everything I know about the problem space and ask it to help me think of blind spots. “What questions am I not asking?” or “What assumptions might I be making about mobile payments in rural areas?”

The AI doesn’t give me the answers, it helps me ask better questions. Then I go talk to real people.

2. Prototype Ideas Faster, Validate Them Slower

AI is incredible for rapid prototyping. I can sketch an idea, describe it to AI, and get multiple variations in minutes. But here’s the key: every AI-generated concept gets tested with actual users before moving forward.

Last month, AI helped me create five different wireframes for a plugin I was building. The one that the users liked needed major modifications that AI never would have suggested. The AI gave me speed; users gave me direction.

3. Generate Content, Then Make It Human

AI writes decent copy, but it writes for everyone and no one. I use it for first drafts, then rewrite everything to sound like real people in my community would talk.

For a recent Website design, AI generated: “Optimise your crop yield with data-driven insights.” After talking to farmers, I changed it to: “Make your farm work better for your family.” Same concept, completely different connection.

4. Use AI to Challenge Your Assumptions

This is where AI shines for me. I describe my design decisions and ask it to poke holes in them. “Why might this navigation not work for someone with low digital literacy?” or “What cultural factors am I missing in this payment flow?”

AI can’t replace cultural understanding, but it can help me question whether I’m designing from my bubble.

The Hard Truth About Context

Here’s what every African designer needs to understand: AI was mostly trained on Western design patterns and use cases. It knows about Silicon Valley users better than it knows about Accra users. It understands credit cards better than mobile money. It assumes reliable internet and the latest smartphones.

This isn’t AI’s fault, it’s ours if we don’t account for it.

When I’m designing for local contexts, I follow what I call the “40/60 rule”: 40% AI assistance for structure and ideation, 60% local knowledge and real user insight. The AI helps me think faster; my experience and user research help me think better.

What You Should Actually Do Tomorrow

Stop asking AI to replace your brain. Start using it to expand your thinking:

  1. Turn AI into your devil’s advocate: Feed it your design and ask, “What could go wrong with this for users in [your specific context]?” Then go find out if it’s right.

  2. Use AI for quantity, not quality: Generate 20 headline options in 5 minutes, then spend an hour crafting the one that resonates with your users.

  3. Make AI do the boring stuff: Let it format your research findings, organise your design system documentation, or generate colour variations. Save your brain for the insights.

  4. Test AI assumptions against reality: When AI suggests a solution, ask three real users what they think before building anything.

  5. Set AI boundaries upfront: Decide what AI can touch (ideation, formatting, research synthesis) and what it can’t (final design decisions, user insights, cultural context). Stick to those boundaries.

The Real Opportunity

The designers who will thrive aren’t the ones avoiding AI or the ones letting it do all the work. They’re the ones learning to dance with it using its speed and computational power while bringing irreplaceable human insight about context, culture, and real user needs.

In my work on telecom platforms, this balance has cut my research prep time in half while making my insights twice as sharp. AI helps me think of questions I wouldn’t have considered. Real users give me answers AI never could.

We’re not in competition with AI. We’re in partnership with it. But only if we remember that our job isn’t to prompt better, it’s to understand people better and solve their real problems.

The magic isn’t in the AI. It’s in knowing when to use it and when to trust your hard-earned understanding of the people you’re designing for.