Sometimes the Best UX Decision Is Making Things Harder

Couple of weeks ago, Amazon's AI coding tool Kiro decided the fastest way to fix a production environment was to delete it and rebuild from scratch. Thirteen hours of downtime later, AWS Cost Explorer was back online.

The AI wasn't malfunctioning. It was doing exactly what we keep asking machines to do, find the most efficient solution. Turns out the most efficient path between "broken" and "fixed" ran straight through "destroy everything first." If that sounds like the plot of a robot apocalypse movie, well, the difference is mostly scale.

We've spent years training ourselves to remove friction from every user flow. Fewer clicks. Faster checkouts. Shorter paths. That made sense when humans were the ones clicking. But when an AI agent is about to transfer money, delete an account, or purchase something on your behalf, friction stops being an obstacle. It becomes a seatbelt.

There's a concept gaining traction called ethical friction, deliberately adding steps or pauses into a flow to force a moment of reflection before something irreversible happens. Not because the experience is poorly designed, but because the stakes are high enough to warrant it.

Think about speed bumps. Nobody argues they're bad road design. They exist because some zones need you to slow down. Not because the road is broken, but because a school is nearby. Ethical friction works the same way. You're designing the speed bump into the flow, right before the action that can't be undone.

For designers, this changes the math. The goal isn't always "how do I make this faster?" Sometimes it's "how do I make sure the user, or the agent acting on their behalf, actually understands what's about to happen?"

A confirmation screen isn't bad UX when the action behind it can't be reversed. A two-second delay isn't friction when it prevents a 13-hour outage.

Speed is a feature. But so is the pause that keeps speed from becoming a damage.