Robots’ Legal Personality

Smart robots appear capable of purposive actions, and they exhibit ‘moral agency’: they seem to understand the consequences of their behaviour, and they have a choice of actions. Further, manufacturers of smart robots often cannot fully foresee their behaviour because of machine learning. Should we, therefore, accord smart robots legal personality?

This would have significant practical consequences. For example, think of a fully autonomous self-driving car that causes an accident. Should we hold the car liable instead of its owner or driver? If so, then we might conclude that the car should also have the power to acquire property, conclude contracts, etc.

Societies will answer these fundamental and difficult questions differently, depending on their respective ‘deep normative structure’. By ‘deep normative structure’, I refer to the shared value judgments and conceptions that shape the social fabric of a particular society. If this structure is utilitarian, according smart robots legal personality does not seem to be utopian.

To begin with, smart robots act similarly as humans would in similar situations. Second, we treat anthropomorphic robots like humans. Should we not then give them rights to send a signal against mistreatment of humans generally? Third, we accord corporations legal personality. There is a long debate amongst legal scholars whether corporate personality is based on a fiction (as von Savigny argued) or whether there is something ‘real’ about corporations, whether there is something that could be called a ‘real group-person’ (as von Gierke argued). Smart robots seem no less real than corporate persons.

Most of us probably feel uneasy when considering whether smart robots should be accorded legal personality. The case against treating robots like humans rests on epistemological and ontological arguments. Legal personhood, one can argue, is tied to humans because only humans understand the meaning of rights and obligations. Thinking is more than formal symbol manipulation (syntax). It involves sensitivity to the meaning (semantics) of these symbols, and robots don’t have that. Robots can be programmed to conform to rules but they cannot follow rules. Rule-following presupposes an understanding of the meaning of these rules. Robots are not capable of such understanding. Robots are not active in the discipline of hermeneutics – and they will never be.

The second argument against treating smart robots like humans is an ontological one. The laws of a particular society in general, and the rights and obligations accorded to members of that society in particular, are an expression of the ‘human condition’. Laws reflect what we believe is a precondition for an orderly interaction between humans. But laws also reflect what we believe lies at the heart of humanity, at the heart of what it means to be human. Just think of fundamental human rights in general and of freedom of expression and speech in particular. But also think of such controversial issues as abortion or same-sex marriage. It simply and literally would be dehumanising the world if we were to accord machines legal personality and the power to acquire property and conclude contracts, even though such machines may be smart – possibly even smarter than we humans. So treating robots like humans would dehumanise humans.

Is corporate personhood a good counter-argument, weakening the case against treating robots like humans? I don’t believe so. Corporations always act ‘through’ humans. Humans sit on the boards whose actions are attributed to the corporation. It is true that also in firms, Artificial Intelligence is becoming more and more important. At the same time, it is still the case that final decisions are taken by humans – at least for the time being.

Be that as it may:  it seems to be clear that the question about the legal personality of robots raises deep philosophical problems, and robot law will be shaped by what I have called the ‘deep normative structure’ of a society. It very much matters whether a society is based on a utilitarian conception of ‘the good’ or whether it rather is based on a humanitarian/Kantian vision according to which not everything that is utility-maximizing is necessarily the better policy. What seems to be clear is that a utilitarian conception of ‘the good’ will tend to move a society in a direction in which robots eventually will take a fairly prominent role – by virtue of the law.

Horst Eidenmüller is the Freshfields Professor of Commercial Law at the University of Oxford.