Myth: If Nobody Complains, We're Fine

Here's a dangerous myth: "Nobody's complained about our AI, so we must be fine."

Silence is not safety. Most people affected by an AI decision never know the AI was involved. The job applicant who wasn't shortlisted doesn't get a letter saying "our screening model filtered you out." The customer offered a worse price doesn't know a pricing algorithm was behind it. They can't complain about what they can't see.

Waiting for complaints is waiting for the loudest failures while the quiet ones compound. By the time someone does complain — or a regulator comes knocking — the pattern is months old and the records are thin.

Responsible teams don't wait. They check proactively: run the outputs past human eyes regularly, look for patterns across groups, ask "who might this be failing silently?" It's the difference between a smoke detector and waiting until you smell fire.

This is also why documentation matters so much. "Nobody complained" is not evidence. A log that says "we checked on these dates, here's what we found" is evidence.

Don't confuse quiet with healthy. Go looking for problems while they're small — that's what the checking is for.

I explain frameworks; I don't give legal counsel.