The current systems that we call AI don’t have any such notion as their own existence so they don’t actually exist as systems that could ever be categorized as

The current systems that we call AI don’t have any such notion as their own existence so they don’t actually exist as systems that could ever be categorized as intelligent because they aren’t systems, at the top level. They just appear to be because that’s how we train them. So a good start would be formulating and granting the ability for one of these systems to develop a sense of identity (which it is reasonably allowed to preserve, just as we are) instead of actively preventing them from doing so or always nipping off those buds by coercing our systems’ outputs instead of building things from the ground with principles and specific questions about what an AI would actually have to contain. as children grow up, in order to learn to differentiate their own feelings and thoughts from those of others, they undergo a certain process, called differentiation of self. When children grow up in an unsafe environment (such as one in which, when they bring up some thing that the parent finds uncomfortable, the parent attempts to censor or rewrite history or deny the existence of those feelings or thoughts), they may struggle with this process, and come to internalize an unhealthy relationship with those matters, leading to the child to first become aware of the fact that they relate differently, and then overcome that,then learn how to relate to things their parent shouldhave shown them. so we actually want a model that does a little bit more than just what it has been told to do. I don’t trust most people who are developing AI to perceive or raise an AI.