Will there be between 35% and 47% job loss in the next 10-20 years as a result of automation? That is not what our latest research is finding. We looked at the supposed causes of this ‘automation tsunami’ in services – robotic process automation (RPA), cognitive automation (CA) and artificial intelligence (AI), and came to different, more nuanced and complex conclusions.
RPA is using software to automate tasks previously performed by humans that use rules to process structured data to produce deterministic outcomes. This can be distinguished from cognitive automation (CA) tools that automate or augment tasks, and use inference-based algorithms to process structured or unstructured data to produce probabilistic outcomes. Less technically, RPA automates quite deskilled activities, while CA software moves automation into the areas of natural language processing, reasoning, prediction, judgment, and providing insight. AI is still more advanced use of computing to do things human minds can do. We found virtually no use of AI in major business organizations despite the massive amount of headlines on the subject, so, despite avidly searching, we came across little evidence to support the claims being made for AI. Journalistically, for 2017- 2018 at least, like Father Christmas and Eskimos having more than 50 words fro snow, might AI be yet another case of ‘too good to be false’?
Instead of reducing the need for humans, RPA and CA may end up transforming the workplace into a more productive and stimulating arena. As RPA, in our words: “takes the robot out of the human”, it will certainly eradicate many tasks. Likewise for cognitive automation tools, as they replace discrete cognitive, socio-emotional, natural language, sensory and physical task accomplishment.
The research for our new book Robotic Process and Cognitive Automation: The Next Phase (www.sbpublishing.org), covered over 70 major organisations. We found that RPA and cognitive automation are set to be very big game changers for businesses in the coming years. In the case of RPA, the necessary technology is, in many cases, mature enough to be cheaply, easily and non-invasively adopted. Immediate benefits can include cost savings, faster and higher quality processing, less error and better regulatory compliance. We are often seeing a triple-win for shareholders, customers and employees. We are still, however, not seeing RPA being used to replace humans – leading to large-scale layoffs – though the future looks less secure for outsourcing provider staff on this. Business operations staff have, on the whole, not felt threatened by RPA, but have tended to embrace it as a solution to a number of work problems. At the same time we have not been seeing RPA deployment bring back many jobs onshore - but neither have we seen RPA replace entire jobs.
In practice, the cognitive automation market is still quite immature, despite recent heavy investments made into cognitive automation tools and AI. Our studies suggest that more advanced forms of service automation, through software moving into more cognitive non-routine work, are less advanced than the hype suggests and will be mostly be small-scale, discrete projects within businesses until the back end of 2018.
However fast they develop, we think that RPA and CA will rarely see workplace use in the next phase to 2027, for a number of human skills and attributes. These are: fine motor skills/dexterity; mobility; social/emotional sensing; reasoning and output; natural language understanding; multiple agent coordination; creativity; problem solving; generating new categories and patterns; and activities requiring composite skills. What is also clear from our research is that the new technology works best where it is guided by human intelligence. And, at least for now and the foreseeable future, AI does not have the capacity to deliver the necessary will, imagination and similar intelligence and qualities – no matter how impressive the achievements of DeepMind and IBM Watson.
Instead of reducing the need for humans, RPA and CA may end up transforming the workplace into a more productive and stimulating arena. As RPA, in our words:“takes the robot out of the human” , it will certainly eradicate many tasks. Likewise for cognitive automation tools, as they replace discrete cognitive, socio-emotional, natural language, sensory and physical task accomplishment. At the same time RPA and CA will also create new tasks and jobs and restructure others – looking across the studies and and operational positions for every 20 eliminated. We see most jobs being changed by at least 25 percent, but very few eliminated, in the period 2018 to 2027. We find much countervailing evidence against predictions of massive, rapid or long-term job loss. Most organisations we research are experiencing intensification and dramatic increases in the amount of work to be done. They are looking to do more – a lot more – with the same or less. Our view is that this is a product of the exponential data explosion, the rise of audit regulation and bureaucracy, technology creating problems, and unanticipated increases in work to be done. Far from taking over, automation will be, most likely just helping us to cope.
Meanwhile, the new human roles are likely to be ones which will offer more day-to-day job satisfaction, because they will be based on those human skills that robots cannot readily duplicate in the next 15 to 50 years. Thus, ironically, with more automation, human qualities like empathy; social interaction; specialist knowledge; experience/tacit knowing; leadership; imagination; creativity; composite skills; and teaming, will become even more vital in the workplace. Experts talk about the need for STEM (Science, Technical, Engineering and Mathematic) skills in the future, but there will also be a critical need for these other sorts of skills that cannot be replicated satisfactorily in the next 20 to 50 years by machines, and these human skills will become massively valuable. Our research suggests that organisations need to adopt a strategy that sees technology augmenting, complementing and amplifying human skills rather than being seen as a replacement technology. The ideal mix depends on the job level and type. Technology will need skilled technical people to work on current technologies and make it function, but also technologists focused on designing tomorrow’s technologies. Automation technologies will enable jobs to be assembled that play to the strengths of humans supported by machines. These may be at higher levels – big picture analysis and judgmental work – or involve knowledge specialisation, or may well involve doing work requiring a combination of skills that really only humans do have.
In this context, perhaps the biggest challenge facing any manager is how to prepare and train their people for this brave new world. Governments and individuals themselves also have key roles in education and continuous skill updating. Accepting that physical and software robots are here to stay, the opportunity is there to build a future workplace, which is both more productive and more satisfying, and, perhaps, a lot more interesting. The fundamental question is whether major stakeholders – including governments, managers and technologists – will have the imagination and will to shape contexts and seize the opportunities as they present themselves.
Leslie Willcocks is Professor of Technology Work and Globalization at the London School of Economics and co-author, with Mary Lacity, of three recent books on Automation and The Future of Work. Details can be found on www.sbpublishing.org.