The reality every AI product has to handle

Most software is deterministic. The same input produces the same output. When something’s off, it’s a bug, and the bug is fixable.

AI products work differently. Their outputs aren’t guaranteed. The same prompt can produce slightly different answers. The same model can be right ninety percent of the time and confidently wrong the other ten. That’s not a bug — it’s a fundamental property of how these systems work.

Pretending this isn’t the case produces brittle products. Users meet an AI that seems sure of itself, act on its output, and get burned when the output turns out to be wrong. Trust collapses faster than it can be rebuilt.

Designing for uncertainty isn’t a special case. It’s the core craft of AI UX. Products that handle uncertainty well ear...

Designing for uncertainty (when AI isn’t always right)