Literature on machine ethics tends to position itself on a spectrum where at one end sits notions about whether machines can be moral patients and, at the other end, sits notions of whether machines can be moral agents. While majority of the literature concerns itself with the latter - for example see [1], [2], [3] - my paper fits into the debate by focusing on the capacity for android companion robots to be moral patients in relation to their capacity to be phenomenally conscious in the sense of feeling what it is like [4] to be a moral patient and be morally wronged. It questions whether these robots should be treated morally well by human interactors and ultimately, whether there should be ethical boundaries established for their utilisation in the form of granting robots negative rights. As such, it grapples with both the descriptive and normative aspects of the topic, which seems to often be amiss in current machine ethics literature - an issue that Gunkel [5] has raised. I start my argument in favour of reflecting on the moral treatment of robots from a human perspective, by considering Turner's statement that “protection in law often follows shortly after society has recognised a moral case for protecting something” [6, p. 170] and critically analyse the concepts of there being a moral case to consider, as well as the need for protection, to formulate my argument for why we should establish ethical boundaries, but that we must, first and foremost, consider establishing ethical boundaries from a human perspective. Even though social robots of the kind we are considering here may not actually be conscious, human interactors may anthropomorphise them since they are, as Darling [7] and Scheutz [8] note, designed to elicit the tendency to anthropomorphise in human interactors. This creates the possibility for the formulation of 'real bonds' with these robots, despite robots not having the capacity to genuinely reciprocate human emotion - at least not currently, nor in the near future. The possibility for the formulation of 'real bonds', however, raises the question of whether treating them immorally may lower the moral standards of human interactors [9], [10]. Taking on a human perspective is thus a relational account in terms of the moral consideration of robots [11], [7], [12] since what matters is not whether robots are actually phenomenally conscious and can actually be moral patients who can feel what it is like to be morally wronged, but whether we view them as possessing the property of phenomenal consciousness due to the way in which we relate to them. Given this, and going back to Turner [6], if we take on a human perspective, there is a moral case for protecting something - this something being the human interactors - and the protection is from themselves. Thus, I argue that we must establish ethical boundaries but that the establishment of ethical boundaries must
firstly be considered from a human perspective so as to protect human interactors from their own potential immoral behaviour.
On the other hand, I suggest that in arguing for ethical boundaries from a ‘robot
perspective’, one implies that there is a moral case to consider for the sake of robots,
and that robots need protection from being treated immorally by human interactors –
that they warrant moral consideration because they can be moral patients and can feel
what it is like to be morally wronged. This is problematic. Firstly, if robots are not
actually conscious and cannot then feel what it is like to be morally wronged, from
what are we protecting them? Secondly, even if they were, or could one day be
conscious, we must still approach the topic at hand from a human perspective first and
foremost, because even if we cannot disprove the possibility of robot consciousness,
or even if robots can mimic consciousness, our misdeeds towards them may still
negatively impact our moral standards. Therefore, taking on such a ‘robot perspective’
would mean taking on a property account towards the moral consideration of robots
[
I then consider granting rights to robots. Given the overlap between morality and
legality [13, p. 169] I suggest granting negative rights to robots so as to inhibit the
maltreatment of robots by human interactors, thereby preventing negative impact
upon the moral fibre and quality of human societies. In literature on robot rights, few
authors clarify which kind of rights should, or should not, be granted to robots –
something with which Tavani [
I thus conclude that regardless of the case for treating robots well from their own perspective, the case from the human perspective is enough to argue that robots should be treated morally well. Taking on this humanistic perspective will guide the way as to how we should go about formulating much needed ethical boundaries for our interaction with social robots, as well as what these ethical boundaries should ultimately be.
Keywords: android, companion robot, social robot, robot rights, robot ethics, moral patiency, HRI.