Lesson
AI products need to communicate confidence, limits, and evidence. The goal is not to apologize for AI, but to make uncertainty usable.
Great interfaces show sources, let users correct output, and make risky actions reviewable.
Objectives
Name uncertainty states.
Design confidence cues.
Add correction paths.
Practice
Mark uncertainty points in an AI workflow.
Design one correction path for a bad answer.
Next action
Add confidence and correction states to an AI feature sketch.