Skip to main content

Design in the Age of AI

AI generates content, images, code. What does that mean for designers? An honest perspective on craft, tools, and what makes design human.
10 April 2023·5 min read
Rainui Teihotua
Rainui Teihotua
Chief Creative Officer
Midjourney can generate a brand identity in seconds. ChatGPT can write UX copy that reads better than most first drafts. GitHub Copilot writes code from natural language descriptions. So where does that leave designers?

The Honest Reaction

I've been designing interfaces for over a decade. When I first saw what Midjourney could produce, my reaction was a mix of fascination and something I'm still processing. These tools are extraordinarily good at producing visual output. They're fast, they're prolific, and they don't have bad days.
But here's what I keep coming back to: they're producing outputs, not solving problems.
A beautiful hero image doesn't make a product usable. A well-written heading doesn't mean the information architecture works. A pixel-perfect mockup doesn't tell you whether the user can actually complete their task.
Design was never really about the artefacts. It was about the thinking that produces the artefacts. And that thinking - understanding people, defining problems, navigating trade-offs, making judgement calls about what matters most - that's still very much a human activity.

What AI Does Well

I want to be fair about this. These tools genuinely help.
Exploration. When I'm in the early stages of a project, AI can generate a dozen visual directions in the time it takes me to sketch one. Not as solutions - as starting points. Conversation starters with the team. "What if it felt like this?" is a useful question to ask quickly.
Production. The tedious parts of design - resizing assets, generating copy variations, creating responsive variants, writing documentation - AI handles these competently. This is time I'd rather spend on the harder problems.
Accessibility. AI is already better than most designers at flagging contrast ratios, suggesting alt text, and identifying accessibility violations. I'll take that help gladly.
Prototyping. With code-generating AI, the gap between design and implementation is shrinking. I can describe an interaction pattern and get a working prototype faster than I can build it in Figma.

What AI Doesn't Do

Understand context. AI doesn't know that the user of this insurance claims system is a stressed adjuster with 47 open cases and a headache. It doesn't know that the data entry screen needs to work on a tablet in a warehouse with poor lighting. Context is everything in design, and AI doesn't have any unless you give it.
Navigate trade-offs. Every design decision is a trade-off. More features or more simplicity? Speed or thoroughness? Consistency or flexibility? These aren't technical decisions - they're judgement calls that require understanding the business, the users, and the constraints. AI can generate options. It can't weigh them.
Challenge the brief. The best design work I've done started with questioning the assumption behind the request. "You asked for a dashboard, but what your users actually need is a notification system." AI takes the brief as given. Designers interrogate it.
Build relationships. Enterprise design happens in the messy space between stakeholders with conflicting priorities, users with unarticulated needs, and engineers with technical constraints. Navigating that space requires empathy, patience, and communication skills that have nothing to do with pixels.

Where I Think This Goes

The designers who thrive in the AI era will be the ones who were already focused on problems rather than outputs. If your value was in pushing pixels, yes, AI is coming for that. But if your value was in understanding people, defining the right problem, and making complex things feel simple - that value just increased.
The tools are changing. The craft is evolving. But the core of design - giving a damn about the people who use what you build - that doesn't change.
What changes is speed. We'll move faster from insight to prototype. We'll explore more options in less time. We'll spend less time on production and more time on the thinking that precedes it. That's a better version of the job, not a worse one.
I'm cautiously optimistic. And I'm paying very close attention.