In the example of the child and the coal, what you are trying to do is maximize the child's expected utility.

On topic - I have developed a simple answer to most of these questions: use your own dang brain. If you would prefer poor people to be fed, then work towards that. If you would prefer it if people had wires put in their brains against their will, desire that. If you want people to control their own lives, program it into an AI. But nobody can do the deciding for you - your own preferences are all you can use.