'Merely a means to an end'

Philosophers who are impressed by Immanuel Kant’s moral philosophy often struggle with his view of animals, neatly but unfortunately expressed in this passage:

The fact that the human being can have the representation “I” raises him infinitely above all the other beings on earth. By this he is a person….that is, a being altogether different in rank and dignity from things, such as irrational animals, with which one may deal and dispose at one’s discretion. (Kant [1798] 2010: 239)

The typical response is to argue that some non-human animals have so-called ‘moral standing’ despite being ‘irrational’ in Kant’s sense. The most heroic of these attempts is Christine Korsgaard’s book Fellow Creatures which argues that some animals have final ends and thus fall under Kant’s injunction to treat them always as end in themselves and never merely as a means.¹

I am inclined to agree that many animals have much richer mental lives than Kant was prepared to allow, but that is not, I think, the biggest problem here. While Kant may give excellent reasons for treating humans and by extension animals with respect and dignity, his attitude to other things in the world is more worrying and more perniciously pervasive in western, and thus in many cases global, society.

Consider something which we can be pretty certain doesn’t have the sort of rich mental life which might merit an attribution of ‘moral standing’, perhaps a natural thing like a hill or an artefact like a piece of furniture. Does that lack of moral standing mean that we can - morally speaking - do what we like with such a thing to the point of abusing and maltreating it? The Kantian view, even when extended to include higher animals, seems to have this consequence.

The view seems obvious when we make an unremarked inference from doing wrong when interacting with X to wronging X. If X is the kind of thing that has no moral standing, it cannot be wronged. And if X cannot be wronged, what I do to X cannot be wrong.

Those inclined to accept such an inference are then faced with a dilemma when confronted by wanton acts of destruction of hills or furniture: either they have to say that such wanton destructiveness is entirely morally permissible, or they have to find some story about it wronging some other thing which can be wronged, perhaps a distant or future person. However, when we recognise a wanton act of destruction for what it is, that it is morally ‘off’, then we rule out the former and make the latter unnecessary.

Of course, I need to say a little more about what wanton destruction is, but it is a familiar concept and - absent self-interest - most of us are comfortable applying it to certain treatments of hills and furniture and a great range of other things without worrying whether they have the moral standing to be wronged.

However, I do want to resist the demand to say why wanton destruction is wrong. That demand comes from assuming that we should be able to reduce moral features to some base layer of moral properties and thus justify a moral judgement in terms of those base properties, be they happiness or well-being or being an end in itself or whatever we get down to. Such reductive explanations may be attractive precisely because they fit into a generally satisfying model of explanation in use elsewhere. Yet it would be wrong to conclude from the desirability of such explanations, should they be available, to their unavailability being a reason to reject a moral judgement as mistaken.

The demand for an explanation of why something is wrong presupposes a specific burden of proof: that any restriction on unconstrained human freedom needs justification. And the form of justification demanded, presupposes that human freedom is a property of the individual in isolation. Both these presuppositions can be, and have been, questioned. It is, however, hard to question that there is something wrong with wanton destruction without appealing to some such doctrine of the priority of individual self-interest.

That wanton destruction is wrong even if it wrongs no creature with the moral standing to be wronged may seem hard to accept for someone who has done some moral philosophy, but not for one whose views are undistorted by theory, whether explicit or in the form of a cultural endorsement of the fundamental assumptions underlying that theory.

And so to AI

People agonise over whether it might be wrong to turn off some suitably sophisticated AI but not whether it is wrong to turn of their TV. There is, of course, a difference here, but if what I said above is heading in the right direction, we shouldn't be unpacking that difference in terms of whether the AI has moral standing, whether it has some set of properties such as sentience or freewill which ground that moral standing.

Rather we should be thinking about what is appropriate behaviour towards TVs or AIs given what kind of thing they are. Switching off a TV when not in use is not merely permissible but to be recommended. Smashing it with a sledge hammer looks morally 'off'. Maybe if it no longer works and cannot be repaired and re-used, then destruction (for recycling) is fine. But anyone who says 'It is mine to do what I like with' is expressing a 'wanton' attitude.

So when it comes to 'turning off' an AI we need to know what kind of behaviour that is on the part of the agent. If it is self-centered, immature or destructive, then we can bring ethical considerations to bear without even considering the cognitive and conative properties of the AI. In a slogan: Don't be a dick applies to how we treat things as well as people.²