Another problem with the name is that autopilots in planes and ships do exactly what the pilot programmed them to do, e.g. hold a certain course. The Tesla autopilot on the other hand tries to intelligently react to the subset of the environment which it see through it sensors, which makes it more unpredictable. I assume that pilots don't have to monitor the behaviour of the autopilot itself, whereas in a Tesla you have to do that and be ready to interfere any second. That doesn't really reduce the workload for the driver, so people just trust the system instead.
Pilots do have to monitor the behaviour of the autopilot. Even though they are programmed to do one thing (hold a course or altitude, change to a course or an altitude with a certain rate of change), they rely on potentially faulty sensors and control many parameters to meet the goal. Usually the autopilot will simply disconnect if it detects a problem, but there are examples of autopilots causing passenger injuries:
https://en.wikipedia.org/wiki/Qantas_Flight_72
