Robots Doing Our Dirty Work: Moralisation of health, routine care work, and machines


I am going to raise some questions about the replacement or supplementation of care work within a health care setting by robotics and possible impacts upon moralisation. Important questions concern not just the care work itself, but also information gathering and communication by humans or by machines, and status within a social system, and how these might be impacted by the replacement of supplementation of human labour by machines.

There will be no answers, since answers must await further empirical research: I hope merely to discuss how we might go about asking the questions.

The context for my initial thoughts on this were doing two diverse research projects at once: one on ethical questions in AI; the other, preliminary work in conjunction with a team of ethnographers examining routine care of dementia patients, including continence care. A hypothesis of this proposed work is that one of the functions of routine care work, perhaps especially for dementia patients who are on a trajectory of loss of some central markers of agency, is to maintain the person’s acceptable presence in the social and moral community. Thinking about this set me on the road to the thoughts in this paper.

In considering ethics in health care, the routine mundane aspects of life may be overlooked but are rich material for considering moralisation and value attributions. A focus on routine care work, often studied in ethnographies, is a rich source of value material; and because it is of low status in most social settings, including health care contexts, it may be the ‘canary in the cage’ for sniffing out the moral atmosphere.

First, preliminary brief remarks: about Individuals and social systems to stage set some of my concerns: In work on AI there are two approaches which can be usefully briefly contrasted. One, which focuses upon individual machines – which models AI upon the individual human agent, often indeed, the individual human brain. Ethical concerns may include, for instance, examining how to build ethics into a machine. Such an approach in AI tends to focus on getting the ethics right, and on the question of ‘what should be done?’

A second complementary approach in AI looks at autonomous systems. Such an approach is probably a richer model for thinking about ethics, and is certainly needed to complement and in some instances to correct the former model, which might skate over the wider socially embedded context of how decisions and behaviour are instigated, interpreted, and reacted to. Such an approach more readily asks questions concerning who is accountable, how do we relate to others in a system, and how is information distributed around a system.

I am going to be considering a broad approach to moralisation:

Moralisation in health can encompass many domains, and can involve both patients and staff, individuals and institutions or other social settings. In particular, I look at the nature of what I call the moral ecosystem – the local moral universe – formed by the surrounding social context and how it facilitates or blocks expressions of moralisation.

I consider also those judgements expressly labelled moral per se, or into any judgements or actions which have implications for differential or significant attributions of responsibility, praise and blame. Such practices can help or hinder the presentation of an individual as part of the moral community and the nature and extent of their inclusion in this community; this encompasses the question of whose viewpoints are taken as having moral weight, which is one of my major concerns.

So I am concerned with the content of moral judgements. Note that these can include attempts at moral rehabilitation of individuals (what’s in some contexts been called ‘sentimentalising work’, see later), or moral repair work, as well as negatively judgemental moralising. And in addition, the process(es) by which they are formed, including the (formal or informal) role of those individuals involved in forming moral judgements, and the organisational context within which moralising judgements are made. We are interested in how a decision has been arrived at, and who was involved, and communication is a key.

Note then that the flow of communication of moralising judgements and behaviour itself can affect actors within a system who are communicating moralising information, and this flow can be subject to systematic flows and blocks.

And note that thus, hierarchy and status are important variables.

Moral judgements, the attribution of blame, and inclusion within a moral community, occur within social, and, significantly, often hierarchical settings, and this is especially true of health care contexts. The ‘moral ecosystems’ that social settings create may be formalised, or semi-formalised. I stress ‘semi-formal’ for various reasons, including that much of the moralising and sentimentalising work which occurs within health care settings, vital as it may be, occurs almost incidentally to formal structures of role and obligation. There is a considerable body of work which demonstrates the importance of examining such social systems and organisational cultures for understanding the quality of care and my focus here is on moral aspects of this. Please note there’s not enough time to go into fine detail of all the points which I can only indicate – nonetheless, much of what I say has backing of empirical ethnographic findings.

So, how might the introduction of robotic machinery impact upon all this? I continue by discussing how moralising judgements, and attempts to combat negative moralisation, might occur within the social and hierarchical setting of a ward.

I am raising various abstract considerations, so let’s make it concrete. I will start with four examples that I observed while working as a nursing auxiliary on a long-stay ward in a central London hospital treating neurological conditions. These cases all show different aspects of the moralisation of health involving the lines of communication and importantly, hierarchy, within a particular ward and hospital culture.

One: a patient, Anna, had to take anti-TB medications, a prolonged regime, which can have the side effects of producing extreme nausea, and also depression. In her ward notes, it was noted that she had had several abortions, even though this did not seem relevant to her immediate medical needs, but was mentioned by staff on more than one occasion. She frequently complained of nausea, and frequently asked to see the doctor to get more anti-nausea medication. Nurses routinely took little note of her complaints, and dismissed her nausea as exaggerated. I attempted to intervene, since I myself had had to take a course of TB meds, and knew how sick and demoralised they make you feel. This took some struggle. I believe that moralising about her, delayed her access to anti-nausea medication.

Two: a patient, Brian, suffering from a condition which left him tube fed and unable to speak, communicating via a sign board. He used to wander up and down the ward, making distressing noises. This upset the routine of the ward and the other patients, although it also upset the staff. He was given sedation to discourage this. This worsened his ability to communicate. He was given further sedation, which left him bedridden. A nurse explained to me that the outcome would be that he’d get a chest infection from the tube feeding, coupled with the sedation, and die.

Three: a patient, Chris, with a late stage tumour, close to death. He wanted to stay on the ward to die, but the doctors wanted him to transfer to another hospital to have one last ditch treatment. He vanished from the ward one Saturday. I was sent out to scour the streets and the local pubs to find him. Eventually, the police found him half in and half out of a canal, considering suicide. The ward sister promised him he could stay on her ward. By first thing Monday morning, the doctors were back on the ward, had overruled the ward sister, and ‘persuaded’ him to move to the other hospital. We never saw him again.

Four, Dave, a wheelchair bound patient, much disliked on the ward for his loud discussions, judgemental views, and frequent demands on the staff. He needed treatment for severe constipation. One day, left on the commode behind a curtain, he called for the staff to attend to him. The nursing staff walked up and down the ward laughing and deliberately ignoring him. He eventually cottoned on and treated this as a joke. However, there was an element of malice in the nurses’ actions.

What is going on in these examples?

One: Anna. I was the lowest in the staff hierarchy. However, I possessed a particular type of knowledge, which other staff lacked – personal experience of medication side effects. But my low status meant this had little or no traction – what did I know about drugs? I did my best to counter the staff’s general view that the patient’s very complaints were symptomatic of her morally negative character.

Part of my motivation to do this was however, confidence from my knowledge that I was a graduate of the same Oxford college as one of the ward doctors (who routinely ignored my presence), hence encouraging an innate bolshy streak – this is one of the key moments when I first realised the institutional and social constraints on sharing relevant knowledge, and the contextual and differential operation of hierarchy; moreover, having more time to spend with patients, and hence a small amount of discretion in how I spent my time, I could also subvert the patient’s discredited status on the ward by singling her out for sympathy.

HCAs are low in the social hierarchy of hospital wards. Studies show that HCAs have distinct knowledge about patients yet there is a lack of any formal system for taking notice of this – they are a marginalised and low status group and consequently, form a group identity around this. This group identity however further marginalises them, and cements in-group behaviour. (Lloyd et al, 2011.)

Two: Brian disrupted the moral quiet of the ward, literally and figuratively. He disturbed others, but in so doing, he displayed his own distress, hence treatment could be presented as medical treatment for his own good. He was placed on a downward trajectory in an incremental way. One way of seeing this, indeed, is to consider him as ‘punished’ for these violations. His exclusion from the social sphere was worsened by the increased medication; this further escalated his exclusion. The nurses who witnessed this were distressed, but expressed powerlessness against medical prescribing.

Studies show the importance of local ward culture in delivering care (Daykin & Clarke, 2000). My particular take on such findings is how fitting into a ward culture is a requirement for maintaining social i.e. moral standing. I shall discuss ward routines shortly.

Three: Chris was welcomed into the social world of the ward by the staff including crucially the ward sister; all expressed sympathy with his plight and his wishes. He wished to spend his last weeks on the ward where he’d spent months of his life. This would have represented a good outcome to his inevitable death; but for the overriding actions of the doctors, who turned up on the ward early, outside of their usual visiting times, and overruled the ward sister’s promise to the patient. The moralised acceptance of his wishes from a nursing perspective, was trumped by the dominance of the medical treatment paradigm of the higher status doctors. The ward staff were left aghast and demoralised.

Doctors have been found in studies to ignore HCAs although there are sometimes informal interactions with the doctors and culture varies widely (Jervis). Work has also contrasted different managerial styles – that of Pace – getting things done on time – versus Complexity – attending more closely to the complex needs of patients (Bridges, Williams.) Note that these styles will affect local cultures and hence the way in which the moral standing of the patients and the status of their views is presented locally, and will affect staff and their standing and how their views are heard.

Four: Dave was roundly disliked. He often expressed unpleasant views about others, and ordered the staff about aggressively, often instructing them how to do their job. At one point, I counted that he asked for something from the staff once every 90 seconds. By laughing at him when he was completely helpless, this seems like cruelty, but the staff kept this within limits claiming that they would see to him when they had time.

What’s going on? We can see a system of reinforcing social exclusion worsened by the stigma of ‘dirty’ work. Being associated with dehumanised individuals also leads to loss of one’s own social (and hence moral) status – courtesy stigma. A spiral of exclusion can occur, since to be excluded itself is a marker of dehumanisation. ‘Dirty work’ and association with ‘polluting substances’ exacerbates this. (Twigg, 2000; Bolton, 2005; Jervis, 2001, Stacey, 2005). Stress relieving humour in relation to pollution of feces is common in nursing situations to help cement group identity and combat low status (Jervis 2001, Lloyd).

In fact, my take on this was by releasing pent up frustrations with this patient, and finding a disruptive way of ‘getting their own back’, (indeed, their excuse for delay was that they were constrained by ward timetables) the nurses could be said to be acting to morally rehabilitate Dave to the ward: they made him pay for his ‘crimes’, and dissipated some of the underlying resentment towards him, in a spontaneous and creative way.

For now, because of time constraints, to summarise some important points:

Low position in the hierarchy may at once give particular knowledge, but paradoxically block its communication in the organisational culture. In some instances, such low status workers may be able to mitigate or disrupt these hierarchies.

The actions of health care workers may also serve to try to render a patient fit to participate in this moral community. This may also be by keeping a patient compliant with ward routines. Staff attempts to advocate for the patient and explain their needs as legitimate – sometimes against the dominance of a medical model of care (Lloyd).

From a systems point of view, the routine of the ward is also often a major factor in the moral world of the hospital; those who disrupt this may be in some peril of being downgraded in the moral community; witness patient B. Those who attempt to usurp this too are in peril – Witness Dave too, whose attempts to order the nurses about – i.e. to exert unauthorised control over the ward functioning – led to him being put ‘in his place’ in a particularly significant way.

The ways in which low status workers hence help to ensure the moral crediting of patients may take various forms, shaped by their very low status. Keeping the patient in line with ward routines, ensuring acceptable physical appearance, subverting where possible negative messages from others. Note their lessened ability to do this, and the informal ways in which such work is accomplished, and ways unseen to others of higher status.

There are findings of large differences in culture between settings. Note that the moralising work of health care staff may be accomplished via the help of high status.

I illustrate this via research study on the moral and sentimental work of the clinic shows how important informal work can be carried out in a dysmorphology clinic, where parents bring children for diagnosis. Dysmorphology is especially pertinent to discussion of moralisation, since it often produces visible abnormalities which in the case of facial abnormalities may be particularly discrediting. Parents routinely ruminate about whether they have done something to cause this. Genetic diagnosis may help to free parents from feelings of blame, and the clinic space functions as a ‘confessional space’, with assurances routinely given by the staff, often focusing around normalising the child, offering praise for the child’s behaviour, commenting favourably on the child’s appearance, using words like ‘adorable’, ‘handsome’, ‘pretty’. Such reassurance about development, and the placing of the child in a category of the normal, gains weight from the ‘objectivity’ with which this is performed, via measurements, comparisons over time, and with a large backing of professionally acquired expertise.

The moral inclusion of the child, and thus of their parents, into the accepting world of the clinic is often in stark contrast to the discrediting condition of the child in the wider community. Disruptive behaviour of the child which ordinarily serves to discredit, is viewed within the clinic as a sign of ‘liveliness’ or normality. This is in interesting contrast to the discrediting effect of disruptive behaviour of adults on a ward. The authors of the paper note varying practices in different clinics, but emphasise the importance of the moral and sentimental work of the clinic which extends beyond the ‘official’ work of the clinic.

I found this work of great interest in considering the possible impact of robotics, AI or other machinery in moralising in healthcare: the reassurances of the clinic staff gained weight from the conjunction of informalised, human judgement, and action but in conjunction with the ‘objectivity’ gained by ‘scientifically’ measured diagnosis. I ponder, then, about what interactions between human care, and robot or machine input, might most be fruitful. One question which occurs to me is how much the authority of the (medically qualified) clinic staff over this scientific knowledge might work in confluence with the ‘objectivity’ of the scientific diagnosis, bolstering the medical staff’s expertise. The high status of the medically qualified clinic staff at the dysmorphology clinic, within the health care moral ecosystem, is in these instances, of great value in combatting moralising judgements.

So, what conclusions can I draw?

The introduction of machines, of robots, might have multiple effects on the moral work of the hospital ward. This speculation suggests a very mixed picture.

Information/ communication:

Robotics or other machinery are likely to impact upon what data is recorded, and how it is shared. Information recorded by robotics might have a higher ‘objectified’ status than informalised information from front line care staff. However, in replacing care work by human workers, this might disrupt lines of communication of such information for moralising work including informal small talk care work which rescues patients as moral beings by presenting them as unique persons. Note this often includes work which is not any ‘official’ part of duties.

Here’s one thought. Note how discrediting work may impact upon social/moral status of both staff and patients. Research uncovers different paradigms for how ‘polluting’ matter is handled, which handling is part and parcel of maintaining moral and social status. These can include nursing (infection) and lay (disgust) paradigms, and humour. There is not sufficient time to explore this point here, but in other words, what is communicated, and how it is communicated, is not sheer raw ‘data’. How will the introduction of robotics into toileting, for instance, impact upon these complex dynamics?


How would the use of robotics in routine care work impact upon status, and hence upon the ecosystem of moralisation? We can perhaps find some clues in what has happened as nursing has become more professionalised. Research shows that nurses as take on more technical tasks, and more HCAs are used, HCAS are still seen as low status; their status may decrease, relative to the nurses. Indications are that status follows tech; low status follows routine care work (Lloyd). This ALSO suggests, that as more basic care work is shifted to lower status workers, then the knowledge gleaned at the bedside, MAY BE increasingly marginalised and locked out of the knowledge used to care for the patient and their roles in reconciling perspectives between different actors may be lost. I.e. it suggests that increasing tech (professionalization) may not only decrease opportunities for gathering such informal knowledge, it may further marginalise the kind of knowledge that might humanise patients in combating negative moralisation and in acts to incorporate them into the moral and social world of the ward.

Barriers to moral engagement:

Studies show that where nurses experience moral distress of not being able to deliver good care, they disengaged emotionally from patients (Bridges 2013); in so doing, the patients are ipso facto less integrated into the moral community, and the role of the nurses in acting as their moral advocates is lessened. This has implications for moral repair work. Hence, might the use of robotics help efficiency of wards hence lessening this cause of disengagement – counter to the concerns expressed above?


Ward routines – will robots help keep these going and hence help moral status of patients in not disrupting these, e.g. by lessening the disruption to ward routine of toilet visits – and/or help to solidify the importance of such routines, perhaps making violations of a rigid routine even less acceptable?

Work within wards shows how ward-level social and organisational conditions affects the delivery of individual care (Bridges et al, 2013, Patterson et al, 2011) If it’s ward level culture, then putting in machinery and expecting that this can be programmed in a general way might be problematic and miss opportunities to make the most of individual cultural dynamics of a ward. This has implications for how the implementation of robotics or other machinery supplementing or replacing human labour is done at a micro-level, ward by ward.


Much of the concerns discussed here amount to barriers to multidisciplinary team work – how might this be affected by the employment of machinery?

This will depend upon organisational culture and upon the background managerial styles. Hmmm. Might the use of robotics compound medical dominance? This might indeed be helpful in moral rehabilitation, in cases where the dominating medicos are sensitive and responsive – e.g. as in the dysmorphology clinic – but in cases where dominance of a medical model acts to block out the moralising beneficial influence of the lower grade staff, a problem.

My concern is that the very introduction of robotics might help to reinforce an ‘efficiency’ managerial style. To quote the authors of a metaethnography: “Nursing is then conceptualized as solely technical and physical work, while the more complex but less codifiable relational aspects of care are ignored or viewed as a ‘luxury’ by healthcare planners and managers” (Bridges); this might mean that opportunities for moral inclusion and moral rehabilitation are truncated.



Bolton, S.C. 2005. Women’s work, dirty work: The gynaecology nurse as ‘other’. Gender, Work and Organization, 12(2), 169–186

Bridges, J., et al., 2013. Capacity for care: meta‐ethnography of acute care nurses’ experiences of the nurse‐patient relationship. Journal of Advanced Nursing, 69(4): p. 760-772.

Daykin, N., Clarke, B., 2000. ‘They’ll still get the bodily care’. Discourses of care and relationships between nurses and health care assistants in the NHS. Sociology of Health & Illness, 22(3), 349-363.

Featherstone, K., Gregory, M. and Atkinson, P. A. 2006. The moral and sentimental work of the clinic: the case of genetic syndromes. In: Atkinson, P. A., Glasner, P. E. and Greenslade, H. T. eds. New Genetics, New Identities. Genetics and Society Abingdon:  Routledge, pp. 101-119

Jervis, L. L., 2001. The pollution of incontinence and the dirty work of caregiving in a US nursing home. Medical Anthropology Quarterly, 15(1), 84–99.

Lloyd, J., Schneider, J., Scales, K., Bailey, S., Jones, R., 2011. Ingroup identity as an obstacle to effective multiprofessional and interprofessional teamwork: findings from an ethnographic study of healthcare assistants in dementia care. Journal of interprofessional care, 25(5), 345-351.

Patterson, M., et al., 2011. From metrics to meaning: culture change and quality of acute hospital care for older people. NIHR SDO programme project, 3(1501):93

Twigg, J., 2000. Carework as a form of bodywork. Ageing and Society, 20(4):389-412.

Williams S, Nolan M, Keady J., 2009. Relational Practice as the Key to Ensuring Quality Care for Frail Older People: Discharge Planning as Case Example, Quality in Ageing, 10 3, 44 -55

Paula Boddington, Department of Computer Science, University of Oxford

We would like to thank the Future of Life Institute for generously funding the work of this project.