Clerical work is ubiquitous. It is carried out in practically any place where things need to be filed and processed. Due to its routine and repetitive nature, clerical work is often labeled as uninteresting or even unintellectual; boredom may be seen as inherent to this work . When it comes to artificial intelligence (AI) and automation, the kinds of tasks considered most likely to be displaced are clerical. Many suggest that in the future, working conditions can be enhanced by new technologies as people are freed of tedious tasks—or so the reports of global influencers in this realm would have us believe. Data-collection and data-processing tasks will be irrevocably automated, and clerical workers will retrain for higher-skill tasks—or disappear as a profession.
In HCI, we recognize that such a vision of the future is largely a repetition of 1980s visions of office automation. HCI, and related fields such as CSCW, have a strong tradition of critically examining the ways in which professional communities are interdependent . This includes insights into the specialization of the clerical realm, with concrete implications for the design of new technologies. One finding from the early studies of office automation was the importance of examining the silenced perspectives—also referred to as invisible work—that make things function yet tend to fade into the background . The strategies we rely on in HCI, and the established set of practices we use to think through the implications of invisible work, have expanded over the years to include various types of design interventions and redefinitions of conceptual issues. The unique way in which these research strategies are applied is linked to what we came to know as a feminist perspective on office automation [1,3].
Computerization—the process of introducing PCs, local networks, and prototypes for the optimization of work practices into clerical professions—was supposed to automate the tasks of entering the same data into different information systems, or even make these tasks disappear . Yet, decades later and despite office automation, clerical work is still here. However, as automation marches on, it is becoming increasingly precarious. Clerical work now supports AI rather than being seen as a legitimate occupation in itself . The question then arises: What role might we as academics and interaction designers have in shaping the imagined future of AI and automation, considering the real effects they have on clerical professions?
In hospitals, automation and AI are imagined as tools for maintaining order and making sense of patient data. Using AI, we can detect patterns in patient data and speed up decision making on diagnostics and treatment, or even prevent disease altogether. But detecting patterns and speeding up decision making make up only a portion of medical work overall. Different from work done by machines, human data work is critical to ensure the situated registration of data to augment patient trajectories as they unfold. Activities across hospital wards are aligned so that patient trajectories stay coordinated and responsibility is clear at all times. Patients are scheduled according to urgency and rescheduled if conditions change and require immediate attention. Patients' casework is checked to make sure that the data better reflects the vagaries of reality. Finally, patients are organized according to disease-classification systems, which tend to be culturally specific. This is not a complete list of clerical tasks in hospitals, but the measure of success in clerical work is when all these and many other tasks are carried out in such a seamless fashion that they become invisible .
There seems to be no limit to the amount of patient data that can be produced, so data itself has become a proxy for all kinds of advancements that will take place in future hospitals. Paradoxically, as with the imagined future of office automation in the 1980s, the future of AI and automation in hospitals seems to have little or no place for clerical work. Of course, tasks are divided differently in different countries depending on whether there is a single, shared clerical profession or several types of professions (e.g., transcriptionist, coders, clerks, medical secretaries). But, acknowledging that clerical work is data work, despite differences across contexts, it seems counterproductive to reduce or undermine these professions at a time when we are seeing a constant increase in data. In the minds of many managers, AI and automation will displace clerical workers, thus cutting costs significantly. However, as we have learned from research on "the hidden faces of automation" , it takes work to train and maintain AI and automation systems.
Obviously, as modern hospitals increasingly rely on data, a reasonable concern is how well data will be interpreted and put to use if we make clerical workers redundant. For example, by design, physicians will collect data on patients and process it directly through many of the new interfaces of medical information systems. However, in reality, the data work often simply shifts hands from clerical workers to the physicians. Rather than these data tasks disappearing, we now observe physicians' complaints of death by a thousand clicks.
The future of AI and automation in hospitals seems to have little or no place for clerical work.
The central point here relates to the overall organization of medical work and how clerical work practices have developed alongside the growing importance of data. The complex alignment of tasks that occur as part of the mundane data work in hospitals is one example of this development. A second point is that clerical work and medical work are highly intertwined. Thus, clerical workers are deeply involved in medical work through their handling of patient data, the use of which is shaping how work is accomplished in hospitals.
Along with the continued refinement of existing technology (i.e., medical information systems), the workplace has seen the dramatic growth of sensor technology to produce large datasets about hospital work . In tracking the movement of personnel within hospitals, it is the clerical work that (again) goes unnoticed in the overall picture. This is unsurprising, since most clerical work takes place at a desk. While tracking in the workplace has received increased attention in HCI, it is still an open question how exactly we adjust our research methods accordingly to understand the broader implications of sensor-based data production as a larger phenomenon affecting us all.
The fact that clerical workers in hospitals are typically assumed to be women also prompts the question of how this influences the promotion of AI and automation in technology design for this particular context. In her analysis of the industrial revolution in the home, Schwartz Cowan  points to the powerful narrative of displacement of women by new technologies. The narrative goes that "industrial technology has either eliminated or erased almost all their [women's] former functions." In fact, as with other types of work, when technology is supposed to revolutionize something, structural changes occur and new jobs are created, requiring a different set of skills .
Learning from the insights of early feminist work, research in HCI increasingly examines how new technologies transform professions and the conditions for work. In the U.S., hospital transcriptionists are increasingly independent contractors, which makes their work more precarious. As self-employed workers, they cannot make a living from only one job when paid 6 cents for each line they transcribe. On the other hand, if they decline the job they cannot collect unemployment insurance . Precariousness, as I use the concept here, is related to the conditions for work, but it is also related to the quality of work and the obligation of design to not let it go unnoticed.
A way forward is to turn to recent developments in HCI research that consider a feminist perspective, not only suggesting strategies that are applicable after the fact but also actively contributing to new design agendas . Focusing on invisible work is one such strategy, continuously refined to promote an appreciative perspective on different types of work. Take the example of clerical work in hospitals. In order to support critical decisions, there is a clerical task of interpreting how to enter data into one or more medical information systems. It may even be the clerical worker who initiates the appropriate action when errors are discovered. By identifying the potential of the clerical profession in the imagined future of AI and automation, we can promote core values of feminist HCI, including agency, identity, and social justice.
The future is not written in stone—I believe the strategies of feminist HCI can make a difference. In the FemTech.dk initiative at the University of Copenhagen, we are working toward increasing diversity within technology use and design. Here we have observed the powerful narrative about new industrial technology in hospitals and the consequences it has for clerical work. Though we are not yet seeing the negative consequence of precariousness already evident in the U.S., we argue that taking steps to counter the silencing of perspectives is still critical. This entails privileging a research perspective pointing to how data work is performed, and how it will be performed in the imagined future of AI and automation.
The increasing importance of data in hospitals is unmistakable. And, as data increasingly augments the complexity of hospital work, it is also clear that clerical work is not unintellectual or uninteresting. It is not clear that new AI and automation technologies will in fact replace clerical workers, although some jobs will require new skills. On the contrary, clerical work has developed to become quite specialized within the realm of data. Through the dissemination of our research, we have successfully added issues about technology and clerical work to the political agenda by pointing out how clerical work is not restricted to the filing and processing of data. Here it is important to note that having more voices strengthens the position of researchers. To reach a wider audience, we have to aim for changing both the technical attributes of design as well as the political attributes that govern how we think about work, data, and automation.
Early feminist HCI encouraged designers and academics to consider the politics of interventions in design . We as academics and interaction designers can work to broaden our understanding of how research into computerization and datafication can take a more active role in shaping society. We can rehearse arguments about socially just technology with researchers, designers, and policymakers. In this mission, we pursue a combined focus to understand how the conditions of precariousness and professionalizing are related to the quality of work. Maybe the future of clerical work is precarious. But we all have a shared responsibility across these different communities that we engage with to offer alternatives to the dominant predictions. Or at least to question them as we together shape the imagined future of AI and automation.
2. Møller, N.H, Shklovski, I., Silberman, S., Dombrowski, L., and Lampinen, A. Panel: A constructive-critical approach to the changing workplace and its technologies. Proc. of the 15th European Conference on Computer-Supported Cooperative Work – Panels, Posters and Demos. European Society for Socially Embedded Technologies, 2017.
5. Møller, N.H. and Dourish, P. Coordination by avoidance: Bringing things together and keeping them apart across hospital departments. Proc. of the International ACM Conference on Supporting Group Work. ACM, New York, 65–74, 2010.
7. Vinik, D. The real future of work. Politico Magazine. Jan/Feb 2018; https://www.politico.com/magazine/story/2018/01/04/future-work-independent-contractors-alternative-work-arrangements-216212
Naja L. Holten Møller is an assistant professor in the human-centered computing section in the Department of Computer Science at the University of Copenhagen, where she is part of the FemTech.dk research initiative. She is also a member of the newly established ACM Future of Computing Academy (ACM FCA). firstname.lastname@example.org
©2018 ACM 1072-5520/18/07 $15.00
Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, to republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee.
The Digital Library is published by the Association for Computing Machinery. Copyright © 2018 ACM, Inc.