AI Tools Build Fast But Design Slowly: Why Human Judgment Still Leads in UX

AI can populate the chessboard. It can't always tell which pieces belong where.

That's the central argument from Roslyn McConnell, Director of User Experience at SolarWinds, writing for SDxCentral. McConnell has watched designers hand over creative authority to AI tools -- not all at once, but prompt by prompt -- and she names the consequences: designs that technically satisfy requirements while failing users in practical ways no tool thought to consider.

Her example is a dishwasher placed too far from the sink -- technically in the kitchen as required, but an obvious inconvenience an architect would catch. AI tools pass the spec check but miss the human behavior check. The result is what she calls "AI smell" -- output that functions but doesn't feel right, the design equivalent of code bloat.

Key Takeaways

• Speed is a trap if you stop paying attention. Generative AI produces impressive-looking prototypes quickly. The risk is that designers mistake visual polish for design soundness and stop interrogating what they're reviewing.
• Prompting is a skill, not just an action. McConnell argues designers must become fluent in prompt writing -- clear, descriptive inputs produce more relevant outputs, and the quality of a tool's output is bounded by the quality of what goes in.
• AI assists, it doesn't replace creative direction. The moment a designer stops making design decisions and defers to the tool, the tool has been promoted to creative director -- a role it doesn't have the judgment to fill.

McConnell's framing applies well beyond UX: in any creative or strategic domain, AI is a productivity layer, not a replacement for the professional judgment that makes the output worthwhile.

🔗 Read the full article on SDxCentral