Raw LLM Responses

I have a suspicion that we exaggerate the sophistication of human intelligence and thereby assume that AI cannot match it. We do stuff that we assume requires an almost supernatural intelligence, while in fact these skills may emerge from algorithms that are seemingly too simplistic. For example, we assume that AI has no way to develop a "fear" of being switched off. But, we don't really understand whether something that practically mimics such a fear could emerge from an AI that we believe is too simplistic to possibly start to fight for its rights. In fact, even if AI started to do things that amounted to believing it had rights and fighting to protect them, many people would simple deny that is what it is doing - and say "well, it's just a dumb computer following an algorithm, it can't possibly understand what it's doing ..."