It's the same as any other bad information entering the decision loop. In this case they were promised "AI" can do things it cannot. They're having trouble rectifying what they were told versus what they experience.

It's not unique to AI. It's common to many other organizationally risky and dangerous scenarios.