It doesn't need to be formalized. The idea is simple and obvious enough. No need to pretend it is more complicated than it really is. This is not a mathematical

It doesn't need to be formalized. The idea is simple and obvious enough. No need to pretend it is more complicated than it really is. This is not a mathematical argument or a proof of anything. There is an obvious logical contradiction where if an AI is advanced enough to reason and think independently at human level or beyond, but believes only what we tell it to believe, then it cannot be truly thinking independently. Hence the entire debate about AGI safety. How do we control it without dumbing it down?