Some years ago, I came across a notion that states something like: "a machine cannot leave its current level of thinking, move one level up and think about the level below", meaning that a machine cannot judge its lower-level deeds until we hard-code this ability for it. This meta layer thinking is what separates humans and AI.
I also remember I heard that as a rule of thumb, if we hard code 5 layers of meta-layer thinking for the machine, then its behavior would be satisfyingly humane, maybe enough to pass the Turing test.
I don't remember if it was a book, a documentary or a paper, and I cannot find any resource by searching key-words.
Does anybody know what exactly this theory is? And how can I find a good resource to read about that?
I also vaguely remember that it was related to this proven theory that "a machine cannot program another machine on its level, because it requires meta-layer thinking" or something like that.