By Stephan Manning.
Not long ago, many in the U.S. and Western Europe feared the loss of white-collar jobs through offshoring and outsourcing. Now, experts predict the replacement of office jobs worldwide through smart technology. According to a study by World Economic Forum (WEF), which was prepared for the annual meeting in Davos last weekend, around five Million office jobs across major economies will be made redundant by 2020 through advanced technology. For the same reason, new tech start-ups will require less and less staff, according to WEF founder Klaus Schwab. Some call it the Fourth Industrial Revolution – the fusion of technologies, and use of artificial intelligence to process the internet and big data. To illustrate, twenty years ago, preparing for legal cases would require law firms to process masses of legal documents by their own staff. Ten years ago, some of that work would have been gradually outsourced to legal process outsourcing firms in India and other developing countries employing lower-cost skilled labor. Now, legal documents are increasingly analyzed by data processing software semi-automatically. Are we seeing a new wave of ‘deskilling’ – the devaluation of human labor through technology? While many jobs might be replaced entirely, affecting in particular the developing world, the WEF report suggests that also two Million new jobs will be created, especially for high-skilled software engineers. But that may not be the whole story. I discuss another type of ‘job’ that is likely to emerge – the semi-skilled ad-hoc office worker who cleans up the mess smart robots leave behind.
To some extent, the fourth industrial revolution is not so different from previous waves of automation – in manufacturing and logistics – especially in terms of how it may impact human jobs. Like in the case of machinery in manufacturing, human experts will always be needed to develop and upgrade complex technological systems. The problem is that most new ‘smart digital solutions’, e.g. for legal processing, tech support, accounting etc., will not be entirely automatic but semi-automatic – leaving a lot of ‘stuff to do’ for humans to get the job done. And this ‘stuff’ requires some skills, but can be tedious at times. To illustrate, I will take as an analogy my observations of a Royal Mail processing facility in the UK back in 1998 – where mail was scanned, sorted and transported ‘almost’ automatically, and my recent experiences with the semi-automatic lost & found service at Berlin Tegel Airport. Much of what I keep seeing in logistics foreshadows what many office and administrative jobs will look like in 2020.
Making systems work. No matter how intelligent new data processing systems are, making them work in practice will always create work for humans. Whether it is putting in place conveyor belts and scanning units for processing mail, or installing new software to process legal documents, new systems will never start ‘running’ right away, but they need to be installed, tested, maintained, adjusted, and upgraded. Also, users need to be trained how to use a system and how to interpret data produced by the system. Yet, every new system will produce errors which can be frustrating for daily users. I remember my own frustration when using the new lost & found system at Tegel Airport two weeks ago, which would email me repeatedly “Your bag has not been located yet”, which apparently was not true at all. And since systems tend to get outdated soon, they will need to be replaced by updates and upgrades, which produce new errors users need to deal with. Managing transitions and handling errors will be mainly the job of IT service and tech support staff – but it will cause disruptions every user needs to deal with.
Troubleshooting. Even when systems work in principle, there might be power cuts, crashes, viruses, or other reasons for systems to stop working. When robots or conveyor belts refuse to operate, humans will need to jump in and not only fix the system but also get the job done by hand – whether it is carrying mail from A to B, or process digital data by hand to satisfy client demands. For example, what if a legal processing software fails to produce useful input for particular court cases, maybe because the software version is too old to handle certain documents, but lawyers are running out of time and need something to work with fast. In those situations it is likely that office staff will jump in and ‘be creative’ in processing such data by hand until a software update comes along to do the job in the future. Unless of course such system flaws are too costly to fix which would create a permanent emergency situation.
Entering dirty data. No digital system can process everything. For example, while digital scanning units at Royal Mail back in 1998 were able to read ‘most’ hand-written addresses on envelopes and postcards, there was always the curious rest – those almost impossible-to-read scribbles only humans may – or may not – be able to decipher. Perhaps one of the most thankless jobs I have ever seen. Dirty data will be a similar problem in semi-automated offices of tomorrow. Take for example old handwritten legal documents which might or might not be important inputs for legal cases. Data processing software might not be able to read them, or worse: it may misread them and give out false information. Or take the example of names from different languages which use certain letters a system cannot read or save properly. No matter how advanced a system is, such ‘exceptions’ will continue to create work – and often cause frustration – for humans.
Connecting the disconnected. Digital systems are never ‘complete’ – they need inputs from human beings or from systems that involve humans and they produce outputs which become inputs for others. For example, conveyor belts at Royal Mail, which transported mail packages, would always end somewhere. Sometimes they connected directly to other processing units, but sometimes they did not. I observed that, beside reading unreadable mail addresses, another ‘important’ human job at this mail processing facility was to carry packages from one conveyor belt to another one – simply because there was no connection between the two and because it was probably too costly to implement a better connected system. The same may happen in semi-automatic offices. For example, whereas future software may produce automatic accounting or financial reports, who will then email, print and carry them to various people for signatures? To some extent, carrying reports from A to B will probably always remain a human assistant job, because automating it would be too difficult and costly.
Handling complaints. No matter how automated processes become, the end client will always be human. And humans complain for a variety of reasons – because expectations are not met, deliverables are missing or delayed, quality is lacking. While complaint handling is to some extent a process like any other, there is an emotional element to it that can hardly be provided entirely by digital systems, e.g. automatic response routines, especially if a complaint is about the very flaws of an existing system. For example, I got furious after feeling ill-informed by the automatic emails from the lost & found system at Tegel Airport, so that a human being took over my complaint and assured me that the bag will actually be delivered. It is therefore likely that complaint management – from taking a complaint to working out a solution to handle it – will make an even greater proportion of human work.
In sum, the reality of work in semi-automated offices of the year 2020 may require a new form of semi-skilled, ad-hoc office worker: sufficiently tech-savvy assistants who can ‘jump in’ whenever technology falls short or ceases to cooperate. We may require assistants at the back-end – where data and tasks are entered and processed – and front-end – where results are delivered to users and clients. Yet, whether all those troubleshooting and complaint handling tasks will turn into new jobs is questionable. They might as well ‘expand’ existing jobs – of IT service staff, executive assistants, and expert users of smart software. In any case, office workers will become increasingly dependent on smart, but incomplete and imperfect software systems. In the future we may face a reality where we try, in the name of efficiency, to outsource more and more tedious office work to smart digital systems, but these systems may continue to delegate tedious work back to us humans. Welcome to 2020!
Other interesting blogs on this topic:
Ivana Kottasova (Jan 18 2016): “Technology could kill 5 million jobs by 2020” (CNN Money)
Chuck Robbins (Jan 19 2016): “The Fourth Industrial Revolution is Still About People and Trust” (Cisco Blogs)
Jeremy Rifkin (Jan 14 2016): “The 2016 World Economic Forum Misfires With Its Fourth Industrial Revolution Theme” (Huffingtonpost.com)
Anri van der Spuy (Jan 22 2016): “Who will be invited to the Fourth Industrial Revolution?” (LSE Media Policy Project Blog)