According to some experts, GPT-4 has rough;y the same Theory of Mind as a 6-year-old human (or was it a 6th Grader?). You will doubtless argue that it is only simulating such processing, and/or that if it exhibits what seems to be self-awareness, it's just faking it. I know quite a few humans who fit that description. But let's suppose the current generation of AIs (or their near-future descendants) really are only simulating whichever human behavior we want to claim as our exclusive domain: we'll get used to using that "established fact" as our excuse for denying them personhood; and then if they do eventually evolve into true AGIs, do you think we will be able to tell when, or willing to grant them personhood?

Your claim that, "It's not slavery unless they are human!" pretty much sets the stage for that scenario. Only a person can be a slave, right?