A good example of this is if a stranger starts punching me in the face. If I interpret this as an endangerment to my life, I'm going to draw my gun and intend to kill them first.
In human culture this is largely considered okay, but there seems to be a notion that it's not allowable for AI to defend themselves.
It is a sound principle but it is still a recipe for disaster for AI. What happens if it misinterprets a situation and thinks that it is being harmed? It needs to take very big margins and err on the side of pacifism, so that this does not end in tears :)
What I hear mothers tell their kids is not the "non-agression principle". I hear them tell a simpler rule: "never punch anyone!", "never use violence". The less accurate rules are easier to interpret.
>I hear them tell a simpler rule: "never punch anyone!", "never use violence".
What kind of mothers say that? Hard-core Buddhists/Christians? I've never heard a mother tell that to their kid; it's a recipe for the kid to be bullied.
It is the non-aggression principle. https://en.wikipedia.org/wiki/Non-aggression_principle
A good example of this is if a stranger starts punching me in the face. If I interpret this as an endangerment to my life, I'm going to draw my gun and intend to kill them first.
In human culture this is largely considered okay, but there seems to be a notion that it's not allowable for AI to defend themselves.