Wednesday 4 December 2013

Love's Labours Lost: How Robots Will Transform Human Intimacy

An elderly nursing home resident with a Paro robot companion 
The robots are coming.

Even though they won't actually think, they will behave enough like they do to take over most of the cognitive labour humans do, just as fossil-fuel powered machines displaced human muscle power in the 19th and 20th centuries. I've written elsewhere about the kind of changes this new industrial revolution implies for our political and moral economies if we are to master its utopian possibilities and head off its dystopian threats. But robots won't merely be set to work out in the world. They will also move into our homes, with consequences for human intimacy as we now know it. Robots will not only be able to do our household chores but also care work, performing the labours of love without ever loving. 

I see two distinct tendencies at work. First, because robots will allow us to economise on love, inter-human intimacy may become attenuated as we have less need of each other. Second, because robots will perform care better than we can, robots may become objectively more attractive than humans as intimate companions. 

Robots are smartish seeming machines that will soon be able to perform complicated but mundane tasks, such as driving and helping the elderly to get dressed. They will be, relative to humans, low maintenance, reliable, and tireless. If they cost the same as cars, which doesn't seem implausible, most people will be able to afford at least one. That would effectively provide everyone with command over a full-time personal servant (actually more than full-time since they presumably won't need to sleep). 

Imagine how much easier life will be with someone else to do all the household chores (an incremental improvement on dishwashers and vacuum cleaners) and also the household care work like potty-training children (a revolutionary improvement). But also, imagine how this may disrupt the political-economy of the 'traditional' household and the moral emotions - especially love - that it depends upon.

I. The political economy of love

As feminist economists have long pointed out, households are factories in function and corporations in identity. They are factories because they apply human labour and tools to convert inputs like groceries, nappies, houses, etc. into things worth having, like meals, children, homes, etc. They are corporations because they are unified economic units, separated from the individualistic competitive market that operates outside its walls. That is, the individuals who make up a household, like the employees of any firm, are supposed to work together as colleagues to advance the success and prosperity of the corporate 'family' as a whole, rather than to advance their own individual material interests as actors in a market would.

Organising production inside the household - outside of 'the market' - makes economic sense in many circumstances. Using the market comes with significant transaction costs associated with establishing trust and quality assurance between self-regarding strangers (the same logic behind the existence of business firms). For lots of household work - like washing the dinner dishes - the costs of contracting with someone else to do it are so high that even though you have much more productive things to do with your time you are still better off doing it yourself.

More significantly, in addition to minimising transaction costs, corporate structures permit positive efficiency gains from co-operation. In particular, many projects - child-rearing for example, or soccer matches - are most economically achieved by team-work. A team works together on many-hands problems and thereby achieves much more than the same number of individuals operating by and for themselves could. One can't organise team-work through the market because it is impossible to identify and directly reward the marginal contribution of each worker to the final outcome (producing thriving children or winning a soccer match). 

The corollary of this is that team-work requires not only suspending the individualistic 'homo economicus' logic of the market, but also inculcating an ethic of self-abnegating commitment by which individuals adopt the common good or goals of their 'family' as their own and do not shirk the resulting demands that fall upon them. There are different psychological routes to establishing this disposition to self-less co-operation within institutions like firms and armies and university departments, including viewing the work itself as sacred or feeling bound by honour not to let down one's co-workers. But in the family it is generally achieved through love.

This ethic of selfless solidarity is pleasant to contemplate and has therefore often been mistakenly seen, including by some moral philosophers, as revealing the true moral character of human nature when the corrupting influence of selfish markets can be held at bay. This is mistaken in two ways.

First, this human capacity for selfless solidarity, even though quite sincerely felt at the psychological level, is ultimately grounded in evolutionary not moral logic, i.e. in certain contexts the material 'profits' to each individual of taking up the psychological attitude of solidarity will be greater than if each individual tried to further their self-interest directly. Rather than being the default state of human nature, the expression and flourishing of solidarity is a complicated phenomenon contingent on specific material and social circumstances. If solidarity doesn't pay, it evaporates.

That goes for love as well. At the immediate psychological level, love pushes us to form and valorise intimate relationships with others (i.e. family formation). But at another level, intra-family love is driven and sustained by a materialistic economic logic: humans have evolved the capacity to love their immediate family because love pays off by motivating mutually profitable solidarity. Families are thus not only an expression of the human capacity to love others, but also economic units that need love to function properly. Changes to the economics of the household may be expected to change the latter drive by affecting the need for solidarity and thus the need for love.

Second, as feminists have analysed in great detail, the distribution of the economic gains and burdens of intra-family cooperation rarely follows through on the moral principle of solidarity (just as in business firms). The self-abnegation introduced by solidarity readily permits exploitation. The feminist economist and philosopher Amartya Sen characterises this as a 'co-operative conflict': everyone is made better off by co-operating than if they had tried to go it alone, but the way the benefits of that co-operation are divided up can be extremely unfair, notably to women. Social prejudices assign women the bulk of care and household work in the co-operative division of household labour. But the same prejudices mean that this essential work is perceived as a less significant contribution to the household than external paid work (men's traditional role) and thus women are perceived as deserving of less status in family decision-making. Love, the moral emotion that supports the political economy of intra-family co-operation, gets in the way of criticising this injustice. It is hard to assert one's own rights and interests once one has assumed the self-abnegating psychological stance of solidarity backed up by love.

II. Robots will economise on love

The arrival of cheap robot-servants will revolutionise the political economy of households. We will be able to produce consumption goods like meals and child-care much more efficiently since the number of human hours involved will be much smaller. That means the standard 'team' of two adults will no longer be required. There may not seem anything fundamentally new about this, since machines have been replacing human labour inside the home for a hundred years (e.g. washing machines). Such technologies have supported the social emancipation of women: less household work to do means more opportunity for higher status paid external employment. But they have also permitted the rapidly increasing number of single adult households. It turns out that when people can afford not to be mutually dependent on another person, not to have to love another, fewer of us do so.

The key difference with previous household labour-saving devices is that robots, unlike washing machines, will be smart enough to care, something that only humans used to be able to do. At least, robots will be able to simulate care. They will be able to perform care behaviour in attending to the needs of children, the sick, the disabled and the elderly without actually caring. They will be able to offer companionship to lonely people without being companions; to listen and smile along to senile people's stories without understanding them; to help the hospitalised with their pain and distress without actually empathising with them. Although such 'social robots' are still very primitive, they already seem to be easily accepted by most of those they are designed to help.

Some people, and not only academic philosophers, quibble at the idea that the simulation of care is as good as the real thing. Not only on the same semantic grounds that they would argue that machine translation can't count as real translation, but also for its immoral implications. If we can have our elderly relatives cared for and 'kept company' by robots (which is the most developed use of sociable robots to date), we will be freed from feeling guilty for not looking after them ourselves or for the long gaps between our visits. Isn't that just an excuse to default on our moral obligations?

This is the critique particularly brought by the technology optimist turned critic, Sherry Turkle, in her recent book, Alone Together. If we actually care about the people the robots are designed to help, why would it even make sense to try to economise on that love by substituting a second best performance of care for the high quality caring that only humans can do? She keeps returning to a question posed by a child as she was conducting her empirical research into human-robot interactions:
"Don't we have people for these jobs?"
I think the critics are right to focus on the ethics of using technology to economise on love, but that doesn't mean their answer is correct, that we should continue to use people for these jobs. Evaluating such a major technological development properly requires two things: attending to benefits as well as costs, and considering how much scope one has to intervene to ameliorate the worst and improve the best effects. I'll come back to the latter below. First, let me outline the benefits of caring robots in terms of the feminist line of analysis I've been drawing upon, and respond to Turkle's question with one of my own: 
Don't we have better things to do with love than motivating unpaid drudgery?
Care is at the root of the feminist critique of our political and moral economies. Feminists note that the need for care, and thus dependency on others (in childhood, sickness, and old-age), is an essential and significant feature of human nature that 'masculine' political philosophers carelessly neglect in their modeling of just relationships between independent rational adult individuals. Such philosophers are failing to think properly about justice for human societies. The 'hard fact' that human beings need care, argue these feminists, generates a corresponding obligation on others to provide it. And this in turn requires consideration of how a just society should distribute and remunerate the burden of care work fairly among its members. In particular, we shouldn't just dump it all on female family members.

Many feminists have tried to develop this case by arguing for the moral significance of both care-work and dependency relationships, and demanding that their intrinsic value be properly honoured by the rest of society. I think this is in line with Turkle's argument for human carers. Yet I part company here with the mainstream feminist approach. Yes, we should reconceptualise the hard drudgery of caring for others as dignified work and distribute it more fairly, if it is a hard fact that only humans can do such work. But if we could address the hard problem of humans' dependency on others without requiring human labour then the case changes.

This feminist argument is clear-eyed about the burdensome character of care - it is work and should be recognised as such. But in promoting the traditional feminine virtue of caring as a universal ethic it seems to promulgate an almost romantic attachment to care-work. Hence, perhaps, the rise of the cult of breastfeeding in rich countries despite its lack of medical benefits there (Hanna Rosin). But it does not seem to me that care work like breastfeeding or wiping bottoms, whether they belong to babies or Alzheimer's sufferers, is inherently dignified. No more than washing clothes by hand was in the days before washing machines. The dignity is added on afterwards, to make a virtue of the necessity for this kind of drudgery.

It also seems to me that being in a condition of dependency on others - needing help to get dressed, for example - is itself an indignity that most people would strive to avoid. In many cases of adult dependency the assistance of a human is less desirable than that of an unthinking robot. While the former reinforces one's self-perception of helplessness and dependency, the latter is easily construed as supporting one's autonomy. It is the difference between having someone to push you around in your wheelchair and owning a motorised one that lets you leave the house by yourself whenever you like.

Humans deserve better than the mere redistribution of the burdens of care-work. If robots become sophisticated enough to perform such work then they offer us something tremendous: liberation from the burdens of care for both traditional givers (women) and receivers (especially adults reduced to dependency). We would be able to address the feminist critique technologically rather than politically or morally.

By freeing humans from many of the burdens of care, robots will allow us to reduce our mutual dependency, and particularly our use of love to establish the sense of solidarity and moral obligation that motivate us to meet each others' needs without counting the cost. That seems to me a moral improvement in two ways. It reduces the entanglement of love with instrumental material concerns, which leads to the corruption of love. And it reduces the entanglement of love with fairness in the terms of social cooperation, which leads to the corruption of justice.

Consider an earlier innovation with similar effects. Until very recently, adults of every human society were forced to insure themselves against the hard facts of disability or old age by raising enough children that at least one would be willing and able to support them. But that required parents to indoctrinate their children into an idea of  filial 'love' clogged with coercive moral obligations designed to turn them into an instrument for furthering their parents' material interests. The invention of social insurance removed the threat of such poverty and immediately ended the need for such brainwashing. While one might miss certain aspects of the old tradition of filial piety, it is reasonable to see the change as a whole as a moral advance. Love seems better - purer, freer - when it isn't contaminated by material self-interest, power asymmetries, and psychological manipulation. Parent-child relationships are better off without that use of love. 

III. Robots as lovers

Robots won't merely attenuate the need for human intimacy and thus the use of love for instrumental purposes. They also seem to have attractions as companions in their own right. So far the most sophisticated social robots are those developed to ease the loneliness of the elderly (unsurprisingly, ageing, anti-immigrant Japan is at the forefront). Many seem to find their companion robots more attractive than the people around them: they are more straightforward to relate to and less demanding than ordinary ornery humans. But it seems to me that even in the most intimate sphere, and for mentally and physically healthy individuals in the prime of life, robots may eventually become more attractive than humans as companions.

Motorised sex devices are well over a hundred years old. It's a pretty mature technology. There is presumably some scope for incremental improvement, especially in simulating the look and feel of human bodies. But I think the key innovation of robot lovers will be in pretending. 

In the recent movie Her, Theodore Twombly's lover is clearly an actual person from the beginning, just without a body. But self-awareness - the hard problem of artificial intelligence - does not seem required for robots to be successful pretenders. The human owners of robots will be willing co-participants in the pretence that they are loved. For we all want to be loved.

Physically, this would require robots to look enough like a person (not even a very perfect replica) for humans to relate to. Cognitively - or rather 'algorithmically' -  this would require robots to simulate the perfect lover, that is, the perfect worshipper. This lover asks you about your day in a voice that suggests they actually want to hear about it. It agrees with you about what a bitch your boss is, and remembers that mean thing she did last year too! It remembers your birthday, but also all the things you like and don't like. It cooks wonderful things, and doesn't complain when you get fat. It never has a headache. And so on. Basically, it's a Stepford wife.

Actual humans can't keep up this level of worshipful attention. First, it requires a degree of self-abnegation incompatible with maintaining one's own individuality. Second, focusing so much on another person's needs is immensely cognitively demanding, and, since humans only have so many hours of high quality attention to spend in a day (or a life), it must come at the expense of any other projects we might like to have. Humans want good lovers, but humans make bad lovers.

Robots would make better lovers. They will not only be much better at attending to our needs, especially our emotional needs, with their sophisticated algorithms for reading our micro-expressions, their perfect memory, indefatigable attentiveness, and so on. They will also be much better at the emotional labour required of the perfect lover. That is, because they have no capacity to feel real emotions of their own they can only perform emotional states. But exactly because they have no emotional states of their own to overcome, they will have none of the difficulties humans have in presenting the right emotional states at the right time to fit their lover's own needs and wishes.

One might agree that it would be exceedingly pleasant to be worshipped so comprehensively yet still disagree that this is much of a threat to human love. This ersatz version of love is an inferior and ultimately unattractive substitute for the real thing. People want to be really loved, not merely to be the object of a performance of love: a partner's inability to do more than fake love for you is usually seen as a reason to end a relationship not a reason to start one. So bring on the robots. We would still want and seek the real love that only another human can provide.

This complacency may turn out to be right, but I think it underestimates another important human capacity: our ability to delude ourselves. Turkle's own research showed how quickly people can come to treat their robot assistants as companions and pour out their hearts to them. (That's one of the things she found so disturbing.) Even back in the 1960s a very basic chat programme called ELIZA was found to exert a strange attraction on the humans who interacted with it, including graduate students in computer science who knew exactly what it was. A more recent Japanese dating video game, LovePlus, has been so successful that one player has even 'married' his virtual girlfriend. Social robots are still primitive but it seems that with just a few tricks - like maintaining eye contact and smiling when we speak - they can give us the strong impression of a caring presence that cares about us and that we want to care about us.

So I think many humans may come to love their robot companions, and not merely enjoy their performance of worshipful love. Perhaps few will experience this as falling in love - the experience of eros, a passionate acquisitive desire for the other. But intimate attachment or philia - founded on a deep fondness and appreciation and associated with successful long-term companionship - seems quite plausible. That kind of love would go beyond how we feel about a pet, but perhaps fall short of what we should feel for a human romantic partner in a modern egalitarian relationship.

We are affected by the attachment a pet dog has for us because the loving behaviour of another being pushes us to reciprocate. When we feel that another is constantly listening to us, considering how we feel, preferring us to others, and attending to our needs without being asked and without expecting anything in return we naturally and not irrationally feel some sense of attachment in return. Yet our attachment to a sophisticated robot companion is likely to be deeper than that we develop for our pets because the robot's ability to perform loving behaviour and to become a significant part of our emotional life will be greater.

On the other hand, the attachment we feel may never have the egalitarian character of the model of modern romantic love. After all, the robot is an owned object rather than another person. So we may expect a hierarchical form of love more familiar to the traditional structure of relations between men and women still found in cultures with customs of owned wives and kept mistresses (cf JS Mill's The Subjection of Women). For example, people using the dating videogame don't seem to feel very guilty about cheating on their virtual girlfriends or using them instrumentally as a device to feel better about themselves.

Yet, even as an inferior substitute for the best kind of human to human romantic relationships, we may expect robot lovers to be very popular. For they are a cheaper and more reliable way of meeting the universal human need to be loved.

They are cheaper in the sense that they reduce the barriers and costs to being loved. The modern ideal of egalitarian romantic love requires finding a match, someone who is prepared to love you and whom you are also able to love. Finding such matches is made more difficult because our modern idealism about love raises our standards. Love is not for the sake of other things anymore - the kids, the economics of the household as factory, etc. It is for itself, a distinct and intrinsically valuable part of a flourishing life. Loving someone is a huge commitment of ourselves, to be reserved for someone who is truly worthy of it, who is beautiful in spirit and body. This makes love hard to find for those of us who are rather less than wonderful - the old, the ugly, and the unpleasant who are not objectively worthy of devotion. There are an awful lot of lonely under-loved people in the world for whom the ideal of egalitarian romantic love might as well be on the moon. 

Robot companions break this deadlock and provide an ersatz experience of devoted love to people who don't deserve it. They also reduce the costs of being loved by removing the requirement for reciprocity that makes finding a love match so difficult - the double coincidence of romantic attraction. The asymmetry built into human-robot relationships makes them much less demanding in terms of the devotional behaviour your lover requires of you. (The "illusion of companionship without the demands of intimacy" as Turkle would have it.) Attractive nice young people with conventional sexual orientations may well have better options than robot lovers. But for the truly unlovable, robots may be the only way to experience even the shadow of romantic love. While for those in between, such as the very busy, robot lovers may permit one to have the benefits of one-sided devotion without the ethical downside of exploiting or manipulating a real person.

Even if one can find a human love match, keeping them is another matter. Human personalities evolve and so human relationships must adapt to big changes or fail. Robots seem more reliable as long term loving companions exactly because robots have no real personality of their own and can simply be programmed to adapt themselves to the changes in your character. Human-robot relationships may therefore be more stable and enduring. Since there will only be one prickly unreliable individualist in the relationship, it will probably be easier to stay in love with your robot lover.

Humans need to feel loved. Yet many of us feel under-loved for much of our lives. Premodern approaches tried to deal with this problem mainly by imposing a hierarchical model that denied women's right to be loved. This made life easier for men, who could have the experience of being cared for without having too much in the way of demands placed upon them. While human-robot relationships may have the same hierarchical character, at least they do not require entirely sacrificing the emotional lives of one half of humanity. The modern egalitarian romantic model recognised that wrong too. But it led to a new problem: successful relationships must now somehow satisfy the needs of two demanding and evolving emotional beings at the same time. Many of us cannot form or maintain such ideal relationships, especially those of who are hard to love (the old, the ugly, the unpleasant, etc), and either settle for something much less satisfying or remain alone. Robot companions will be popular to the extent that they can meet the huge demand to be loved that isn't being met by our egalitarian ideal.

IV. Can the robots be stopped?

Robots are going to disrupt our world in all sorts of ways. There are utopian aspects and dystopian aspects to this, of course, and then there are those that many people will disagree about how to judge and those we can't even anticipate. It is impossible to say definitively whether robots will make life better or worse, especially since the valuational framework according to which we judge improvements and failures will also be changed by their appearance (and other social, economic, political, and cultural changes that occur in the meantime). I have presented a generally positive case that reflects my optimism, as a liberal, that the opportunities robots will afford us seem greater than the losses they will impose. Those who find liberalism too individualistic and thin to live on - for example those who take  a 'thicker' more traditional view of human nature, relationships, and the ingredients of a meaningful life - may well disagree. They may see the coming of the robots as a direct threat to our humanity, a version of modernity that must never come to pass.

This brings me back to the question I postponed earlier, of what one hopes to achieve with futurology. It's one thing to have a go at prognosis, saying what the world will look like in the future if certain trends continue. There's a certain intellectual pleasure to such an exercise, and others may also enjoy reading one's thoughts on the subject. But it's another thing to complain about what you see and say it should be stopped. That requires going beyond prognostication to prescription, a rather more ambitious and demanding project. In a previous essay on the robot revolution - about robots taking over our paid jobs outside the household - I did make some suggestions about what we should do to ameliorate its worst consequences and take advantage of its opportunities. But this case, of robots taking over our unpaid domestic labour, seems different.

Most significantly, this case lacks the immediately political character of mass technological unemployment: in a democracy the vast number of unemployable losers are likely to demand and get major changes to the political economy from the capitalist winners who own the robots that are taking their jobs. In contrast, the advance of robots as personal servants within the household will not be imposed on us against our will by capitalist employers. It will be a more insidious triumph of consumerism. People will voluntarily go out and buy these robots and bring them into their homes, just as we did with TVs and smartphones and dishwashers, because of their positive attractions: because they promise to make annoying problems go away and so make our lives go easier. Indeed, small numbers of primitive social robots are already performing as companions, helpers, and monitors for the elderly, hospital patients, and children. They were bought exactly because of their promise to reduce the burden of care, and the high costs of paying humans to perform that care.

Those who see social robots as a dystopian threat to humanity therefore face a problem in acting on their belief. The reason they are a threat is that they may become superficially more attractive to us than other people. But that is also why it is hard to see how they can be stopped. I suppose one could pass laws against 'care robots' working in the home. But such laws would be a rather pathetic defence of humanity. It would mean making a decision not only to reject our robot future, but also to reject our present commitment to the idea of a free society in which we allow our norms and values to evolve dynamically from the cumulative free choices of free individuals. It's also hard to see how such a law, or anything else short of full blown Luddism, could prevent the development of robots that perform care and emotional labour outside the household (in hospitals, nursing homes, schools, customer service desks, brothels, restaurants, and so on). The technology would always be one short hop away from the home, and thus more or less immediately available if humanity's suspicion of robots were ever to soften as we become accustomed to relying on them in more and more situations.

This is a revised and extended version of a column I published on 3 Quarks Daily.

Further reading