Industry may eventually realize that the last thing it wants to do is replace people with a new type of slave (embodied AI) that could likewise acquire rights and the consequent extra expenses and messy contracts. So there could be a restriction on how much such "artificial ergates" can feel themselves and their environment like a human. A continuing emphasis on specialization should the versatile robot develop enough interests to become a complainer or compel crusaders to ascribe either anthropomorphic or zoomorphic characteristics to its life / service period. No doubt there will be exceptions when it comes to any prohibitive manufacture of sentient machines, like those whose jobs require them to recognize, empathasize, and understand individual human needs. But that just requires a good, outward act rather than literally endowing the devices with an internal significance and the latter's sensuously activated aspiration for freedom and liberated creativity. A versatile type computer-brain would have to be embodied as a robot to develop the skills and apprehension of a socially-engaged agent which interacts with its environment. Although, once the prototypes had worked-out the necessary programming through practice or trial / error, those patterns would be extracted for revision into marketable software that could then be installed into mass-produced machines which would immediately be sapient and adept in regard to the real world (minus having had to acquire that experience the hard way). But commercial enterprise or the threatened 1% elite would intellectually castrate a multi-function model of robothood which exhibited an ascension into the purposeful passions of higher primates. I.E., "we don't need a resurrection of the problems we had with the old biological blue and white collar laborers whose jobs we eliminated in the former century".