They are useful during pure media consumption like viewing images, videos, etc.
Regarding the issue of educating the user of their usage: the current standard implementation is quite good enough. Display the controls and additional data when opening the content and then remove them after a few seconds. This tells the user there is all the other information (controls, tags, etc.) available and it disappeared. The first reaction when a person sees this is touch the screen, and, viola! They appear back and the user is trained! MAGIC!
This is similar to the onboarding process that has been tried and tested by game designers since decades. You are either showing the path of the entire level before the game begins, or you can see a sneak peek of some ability (jumping, ducking, etc.) right before you need to use it. Show the capability and the user will learn by watching and experimenting.