I think continual evaluation is important in the things that carry risk. We all use heuristics most of the time though.
but then data changes, so I have to continually re-evaluate that my trust is still valid...
This is something that AI and robots don't need to do. Once a new rule is better, they can adopt it instantaneously across the board. We were talking about SD cars the other day. Once a new road law comes in, without announcement, they all start to follow it. People just can't do the same.