Trust Calibration & UX Patterns

The goal isn't 'make users trust the AI' — it's calibrated trust: high when the system deserves it, skeptical when it doesn't.

Over-trust is the dangerous pole: users who rubber-stamp a usually-right system produce the worst incident outcomes, because nobody's checking the day it's wrong. Every UX choice either calibrates or inflates trust — frictionless accept-all buttons inflate; showing work calibrates.