The topic of service automation is typically shrouded in both hype and fear. In Service Automation: Robots and the Future of Work, Leslie P. Willcocks and Mary C. Lacity present the results of a survey, client case studies and interviews with service automation clients, providers and advisers, engaging with both the benefits and possible challenges of present and future service automation technologies. While the authors could have been bolder in exploring some of the issues raised, Michael Veale recommends this book as a useful starting point for those looking to understand the growing implementation of automation in a variety of operational contexts.
If you are interested in this review, you may also like to listen to a recording of Professor Willcocks and Professor Lacity’s lecture, ‘Service Automation: Robots and the Future of Work’, recorded at LSE on 9 May 2016.
Service Automation: Robots and the Future of Work. Leslie P. Willcocks and Mary C. Lacity. Steve Brookes Publishing. 2016.
Booming headlines such as ‘Half of All British Jobs Could Be Replaced By Robots!’ and accompanying shelf-metres of popsci may be giving the particularly forward-looking a few sleepless nights. Yet these works rarely spur us to imagine the nitty-gritty details of a machine-staffed future. As it looks less and less likely that humanoid robots are on the immediate horizon, will boxy, distant descendants of automated checkouts replace our receptionists, lawyers and doctors? Where does this leave the day-to-day work of humans involved in commissioning and supervision? How will we train the developers, maintainers and managers, and how will we reward them? Will machines process their tasks on local hardware, or will that occur on a distant, centralised server — and, if so, where in the world will this server be? Those interested in the political economy of automation may see such questions as trivial, even boring. Robots are coming and the rest, my friends, is a footnote. Yet for management and business academics Leslie P. Willcocks and Mary C. Lacity, the immediate reality of these questions is both interesting and at the core of their new book, Service Automation: Robots and the Future of Work.
Personally, I believe the practice of automation will find it hard to transcend relatively dry commercial decisions involving people, institutions, rules, procedures and internal politics. It is, of course, much sexier to position automation as saviour or scourge: as a threat to work, a relief from drudgery or a catalyst for a new form of politics. But as Terry Gilliam noted in his 1985 film Brazil, the classic Orwellian dystopia seems significantly more realistic when you leave in the dysfunctional, clunky bureaucracies of our own world. In fact, there is a significant argument that it is analytically crucial to leave these in; that the dull drudgery of implementation is, in many ways, just as important as fanciful macro-level systems. Anticipating and understanding market structures and the roles, skills and values of decisionmakers and designers helps us to glimpse the realistic impacts of automation on labour and society in the short- and medium-term.
Willcocks and Lacity focus on robotic process automation — RPA for short — which is a type of automation especially useful for tasks with digital inputs and outputs. These tasks are usually manually undertaken by a human, and may include filtering, sorting, gatekeeping, inputting or otherwise transforming data. To give a few examples: text classification might be used to distinguish flavours of customer enquiry for the attention of different departments; automation may be used to link together otherwise incompatible systems; or to perform common, predictable data cleaning and imputation tasks. Some of these systems are relatively simple — flowchart-based tasks — while others rely on machine learning systems, the heuristic art of algorithmically training models from previously observed inputs and outcomes to try to make accurate predictions on new, unseen examples.
Image Credit: (Arthur Caranta CC 2.0)
Exploring the topic primarily through case studies and interviews, the authors touch on a range of interesting issues. Who programs or trains automatic routines? Usually this falls well within the responsibilities of IT departments or contractors, yet they point to many instances where it happens elsewhere in organisations. At times, these efforts are only noticed by IT when their own systems flag them as potential security risks. How has senior buy-in been achieved for these arcane projects with uncertain risks? Who are the major contractors in this space — are they specialised or generalised?
In many areas, Willcocks and Lacity could perhaps have been bolder. In the future, how many people do they envision needing IT skills — and is that congruent with current projections? Will cosy user interfaces substitute for computing knowledge? And if responsibility for automation is decentralised, how will security be achieved in this free-for-all world? Who will monitor the systems for clashes and feedback effects? These are larger questions, yet the present-focused case studies somewhat preclude a forward-looking, imaginative and normative vision of a future business environment. Indeed, when asked about job replacement, the case study interviewees eagerly report back that employees are merely redistributed among the company. This may be the case for pilot projects, yet the authors barely peek inside Pandora’s Box. Technologies of any sort rarely enter mainstream use quietly. Others must still envision what these systems might look like when this is more widely deployed, especially within companies that may have higher immediate cost pressures than early pioneers in comfortable market spaces.
Certain academic audiences may wriggle uncomfortably in their seats at parts of Service Automation. It is peppered with biz-speak, most harmlessly contained within industry block-quotes rearing every few paragraphs or in the (very welcome) ‘in their own words’ practitioner chapters that cap the volume. Still, the strange, hyping dialect sometimes makes a break for the main text — various businesses, for example, are described as being on their own ‘automation journeys’. Compounding this, an unclear research methodology and constant hum of positivity will likely leave critical readers feeling uneasy. The survey work has some, limited context — the respondent sample were participants at a major outsourcing conference — but the interviewees come from companies and case studies that pop up without explanation or selection logic, and the only context to most of the introduced ‘experts’ is that they work for some incumbent company in this space. This, coupled with the near-complete lack of any non-trivial or later cast-aside critical statements about the practices in the chosen case studies, leaves a sickly sweet, infomercial taste in a reader’s mouth. The ‘Lessons Learned’ sections are certainly useful — but to what extent can you be confident that they are really important without hearing about the failures or near-misses that led to their discovery?
Yet, despite these issues, I found the book useful. Operational questions at this level are often ignored. Even though Service Automation might not be a manifesto for future business nor truly address the challenge of its subtitle, ‘Robots and the Future of Work’, it opened my eyes to the first steps of practical automation in a wide variety of operational contexts. Practitioners could certainly do much worse than take this tome as their starting point should they wish to explore new technologies themselves. Yet it will be up to other scholars to take up the questions of who the real winners are in the implementation of light process automation — scholars who should not forget that there really are people who buy, bid for, manage, improve, interpret and retire these systems within the difficult and constrained contexts of their own organisations.
Michael Veale is a doctoral researcher sitting between the Department of Science, Technology, Engineering and Public Policy (STEaPP) and the Department of Computer Science at University College London. His research focuses on building new sociotechnical tools (such as methods, software and guidelines) to assess the impact of machine-learning algorithms that make societal decisions, particularly in the public sector. He has a BSc from LSE and an MSc from Maastricht University, tweets at @mikarv and can also be found at http://michae.lv. Read more reviews by Michael Veale.
Note: This review gives the views of the author, and not the position of the LSE Review of Books blog, or of the London School of Economics.