I think this has to do more with in/outgroups than with any objective criterion of "humanness". As you said, AI will have an extremely hard time arguing for personhood - because people will consider it extremely dangerous to let machines into our in-group. This doesn't mean they could sense an actual difference when they don't know it's a machine (what the Turing test is all about)

It's the same reason why everyone gets up in arms when an animal behaviour paper uses too much "anthropomorphizing" language - whereas ignoring the bits of animal cognition and social behaviour that we know exists and just treating them as biological machines is fair game.