
Caroline Bassett
University of Cambridge
Are algorithmic voices friendly?
Friends… ‘enhance our ability to think’ (Aristotle: 1155a14).
Boundaries between the artificial and the human are less absolute than they once were. We have had human machine interactions; can we now have human machine relations? If so, may these be defined in terms of sex, or love, or friendship?
The sex industry normalised artificial relations long ago. In this sphere newly smart alien bodies whose puny brains are giant compared to their mechanical cousins, change everything and nothing; they provide a new means through which a relationship that is by definition thoroughly alienated is consummated – love or friendship is never part of the sales plan. Which doesn’t mean the categories don’t leak, or that they won’t leak more, or in new ways in the future. Fiction lately has played with how. Consider Westworld with its violent delights and violent ends, where at least some of the artificial ‘livestock’ of a theme park selling them as the main attractions revolt from sex (work) in the name of (free or freely given) love. Desire is at the heart of these players’ assumption of personhood, an assumption that is at once undeniable and denied by their owners.
Westworld however, is also about friendship, and this is another kind of measure, since friendship may entail love, or even include sex, but is different from both of them. Friendship raises questions about equality, quantifiability, intention, reciprocation, fungibility, and context, that may easily be routed around in discussions of sex or love – consider primary narcissism and the love of the self that is reflected in the other for instance.
These kinds of questions become urgent as emerging AI technologies embed into everyday life and as AI agency becomes an integral, if radically unevenly distributed, element in global cultures. They are also questions that are somewhat neglected. Mediated human-to-human friendship in an era of machine supported sociality (platforms) and forms of prosthetic friendship (influencers, celebrity) is much studied, but that other question – can the algorithm (or the cyborg) be my friend – which might amount to asking if friendship is computable is posed less often.
What is friendship? Friendship is a relationship that is not fungible, it matters that I am friends with you. It is also reciprocal, although it may be offered without the hope of return. It may also be virtuous. Aristotle defined friendship in general as ‘the condition in which two persons will the good for one another and are aware of one another’s good will’ (1156a3-5). He also broke friendship down into three categories. There are two lesser forms; the first defines friendships based on utility in which we will the good of the friend because they’re useful to us – we feel affection for the useful and not for the person. The second category of friendship is based on pleasure. We will the good of the friend because they give us pleasure – that is we feel affection for the pleasant and not the person. The greater form of friendship is defined by Aristotle as virtuous and disinterested; ‘in perfect friendship the friends are viewed by each other as ends in themselves rather than simply means to other ends more desirable’ (Ward, p.35).We will the good of the friend because ‘they are good’. Moreover, ‘[t]hose who wish for their friends’ good for their friends’ sake are friends in the truest sense, since their attitude is determined by what their friends are and not incidental considerations.’
One way to think about this is to suggest that friendship may not be simulated.
This reading of friendship, particularly its insistence on non-fungibility, is interesting to explore in relation to human-computer relations in a time of rising cognitive agency perhaps as a limit point. There may be human/machine sexual relations and given that it has been acknowledged that in relation to human to human forms sexual desire may already come from one direction and may not demand a sustained relationship, nor a particular person (standardized depersonalization being the modus operandi of the sex industry), this seems a justifiable claim, but it is far less clear that there may be human machine friendships.
And yet such relationships are being built.
Friendship, that is, is beginning to be offered as a purchasable and implementable human machine relation. Care robots and companions and the automated ‘friendship’ they are advertised as providing are obvious examples. They constitute an expanding market and are targeted as a solution for rising health and care costs. The assistance they give may be welcome but the invocation of friendship in these contexts is invidious nonetheless. Don’t get old, isolated, demented, poor, or outlive your relatives if you wish to avoid the queasy friendship based on machine learned empathy and the form of artificial ‘presence’ that the care robot provides (at best). Artificial friends might be considered a good enough replacement for those who lack ‘real’ friends – or for those who are not be-friended in the name of the community of which they are a part, particularly by governments who do not wish to provide the resources for such be-friending.
At this point it becomes clear that friendship scales up. It can be, and has long been, explored as political relation as well as a relation between persons. Aristotle understood friendship as that ‘concord’ necessary to hold political communities together, Ann Ward glosses (Aristotle’s sense of) ‘concord’ as a form of political friendship entailing ‘unity of mind among citizens in deliberation and action with regard to important matters affecting the public interest’ (Ward, p.32). Feminism might question whose public interest is being invoked here, or which public, and at this point a suspicion of friendship as entailing recognition between those who are the same might arise. Switching frames one more time then, friendship and its claims might then be regarded as entailed in disputes around (self and) sameness and difference. Aristotle claimed friends flocked together. Derrida in a deconstruction of friendship’s genealogy (or the line of argument coming down from Aristotle in a tradition of political thought that includes Arendt) comes closer to arguing for a form of orphan friendship.
Can friendship then, be between the very unalike, the radically different; the human and the machine? My own take would still be no – since the demand for (the recognition of) personhood, the non-fungible, remains central to friendship, and recognition of personhood would be an important way to gauge care too. This side of a singularity there can be no robots who can be my friend, but there is a politics of friendship that pertains to human machine relationships in advance of that friendship being possible.
Should I leave it at that? Perhaps, but there is another turn that might be taken; if we abandoned friendship in its most virtuous form, and also refused to accept its inferior instrumental versions as bad replacements, could we instead explore friendliness as a category that could do a different kind of political work in exploring how to build relations between humans and machines? If so who will do that work?
Who does the work?
References
Ward, Ann (2008) ‘Mothering and the Sacrifice of Self: Women and Friendship in Aristotle’s Nicomachean Ethics’ thirdspace 7/2 pp.32-57.
Aristotle, Nicomachean Ethics, Martin Ostwald, trans. (New York: MacMillan Publishing Company, 1986).