Is unplugging a robot from power the same as killing a person? As the commenters said this does not quite work because robots can be plugged back in. So let's replace robots with philosophical zombies, they behave just like humans except for not having consciousness. In particular, once killed they are not coming back. Is killing them the same? In both cases you deactivate them permanently. If it is still not the same, why is harming something with consciousness ethically different from harming something without consciousness?
That depends on what the problem is with killing people:
- A Deontologist could argue that the zombies have no inherent duty of care, being entirely imaginary entities, and so declare Open Season without qualm.
- A Consequentialist could notice that killing philosophical zombies has no effect IRL, and grab a shotgun.
- A Virtue Ethicist could acknowledge the degradation of character inherent in any killing and so have his philosophical brainz eaten.
And whether the morality is to be judged within the Thought Experiment:
- A Deontologist should enquire whether there is a way to distinguish the zombies from normies, allowing different duties toward each.
- A Consequentialist ought consider both killings the same, insofar as the effects of the killing on loved ones, in amount of pain during the process, in GDP,... would be very similar
- A Virtue Ethicist is even more likely to be food for a Zombie Thought Experiment.
The teleporter paradox can help us understand. If we can have continuity with the new copy, we may not object to vaporising the old. Irreversibility, is a big issue. We have the same issue with losing biodiversity, and species extinctions.
If we consider replacing the function of a human brain neuron by neuron with computer components, we have to accept a robot can in principle be a person. If we simulate the function of those components digitally, we have to admit a programme can be a person. Turing proposed hus famous test, to shift away from concern about invisible 'essences', to focus on what intelligence does. 'Passing the Turing test' is at best ambiguous evidence, & there are many versions, but the point stands to focus on the functionality of intelligence, and personhood.
Peter Singer argues for animal rights on the basis of capacities, arguing that the 'imaginary circle' we draw around humans is not coherent or consistent, if say we prioritise a brain dead human's well-being over a dolphin that we know pass the mirror test and have complex language, and can we think suffer in ways the brain dead human can't.
Bostrom proposes 'mindcrime', that we extend moral considerations to complex programmes and machines, in a parallel way to our concerns for animals, capacity to suffer & capacity to take up moral duties - and crucially, consider how deeply alien their experiences, potentially causing their suffering to be difficult for us to understand or see.
Has a philosophical zombie 'grown', in a complex and unique way through interaction? Or, is it like a Boltzmann brain, or 'beamed down' by a teleporter, and so potentially replaceable with perfect fidelity? We should be more careful about what we can un-do, than what we can't. More broadly than that, it depends on your framing of What is one’s incentive to be moral?