I think the core concept is pretty simple, it's meta cognition. A feedback loop of observing oneself for the purpose of understanding which part of your observations is yourself and which is the environment. Just like inteligence, its scales from rudimentary in animals to overdeveloped in humans.
The only way to know if it's genuinely occurring or not is to observe the internal state... which is a bit problematic for biological organisms since we tend to die in the process. But for AI it should be fairly straightforward to verify once DNN analysis progresses enough.
The only way to know if it's genuinely occurring or not is to observe the internal state... which is a bit problematic for biological organisms since we tend to die in the process. But for AI it should be fairly straightforward to verify once DNN analysis progresses enough.