When robots didn’t take our jobs
The COVID-19 pandemic has accelerated automation and digitalisation. How does that affect workers, what new roles and demands does it bring forth, and what is the balance between job destruction and creation? The answers invite both pessimism and optimism. We are still in time to create a successful future for work.
Barcelona, 2030. The population is ageing, the workforce is shrinking and the promised robots have not arrived. Automation has not progressed as fast as some people believed it would. Countries have not invested enough therein and the capabilities of artificial intelligence (AI) have not been that advanced to perform tasks with empathy. Faced with the dystopian scenario of job destruction in the past decade, today a very different scenario is coming into sight: that of a world in which there aren’t too many robots, but too few of them.
Sound plausible? Something along these lines was reported by The Economist in its 2019 special scenarios supplement “The World If”, an exercise in imagining more or less immediate possible futures. In its essay, The Economist argued that we are concerned about something that is mistaken or, at the very least, we are exaggerating machines’ capacity to displace human jobs. The evidence of a coming jobs apocalypse today is “remarkably lacking”, the article claims, given that “employment in the rich world reached record-high levels in 2019, while productivity growth in many countries was weak”.
 “What if robots don’t take all the jobs? A different dystopia: July 2030”, The Economist, 2019.
Europe could face a high-skilled labour shortage due to an ageing population.
In the same vein, the MIT study “The Work of the Future: Shaping Technology and Institutions” (2019) points out that the likelihood of robots, automation and AI wiping out huge sectors of the workforce in the near future is exaggerated. Both The Economist and MIT point to an ageing population and the limited influx of immigrants as part of the explanation. “We anticipate that, due to slowing labor force growth rates, rising ratios of retirees to workers, and increasingly restrictive immigration policies, over the next two decades industrialized countries will be grappling with more job openings than able-bodies adults to fill them”, they claim.
These predictions appear to contradict one of the most widely-referenced figures on the relationship between automation and job destruction. It concerns a 2013 study by the University of Oxford according to which 47% of American jobs are at high risk of being automated by the mid-2030s. The study refers to the percentage of jobs most vulnerable to this process. This does not mean that they will necessarily be automated, as their authors have spelled out on numerous occasions, without much success.
The Organisation for Economic Cooperation and Development (OECD) estimates that 14% of existing jobs in developed countries could disappear as a result of automation in the next 15 to 20 years. Another 32% of jobs are also likely to change radically as certain tasks are automated. A more recent report published by the World Economic Forum (WEF) refers to a double-disruption scenario for workers due to automation in tandem with the pandemic-induced recession. The WEF predicts that, by 2025, the time spent on current tasks at work by humans and machines will be equal; by then 85 million jobs may have been displaced, while 97 million new roles may have been created. If so, the number of jobs created will surpass the number of jobs destroyed. But it also notes that the trend towards job creation is slowing down, while job destruction is accelerating.
An analysis by McKinsey reveals that the jobs jeopardised by the pandemic are particularly vulnerable to automation. Specifically, it estimates that 94% of jobs in food service and building occupations, and almost 70% of those in wholesale and retail sectors could disappear. Along the lines of the WEF, McKinsey predicts that new jobs will offset lost jobs, and it even believes that Europe could face a high-skilled labour shortage due to an ageing population.
Ghost workers are part of the so-called on-demand economy or the gig economy, based on small jobs.
Forced to co-work
The desire to eliminate human labour always generates new tasks for humans. It is what anthropologist Mary Gray defines as “the paradox of the last mile of automation”: the reason is that the algorithms that drive machines and artificial intelligence are sufficiently flawed so as not to be able to manage without humans, as is the case with robots and other types of software.
Most automated tasks require different people to control and take care of the processes 24 hours a day. Hundreds of millions of people already perform human computing tasks necessary to develop and operate the websites and applications that we all use. TripAdvisor, Match.com, Google, Twitter, Facebook and Microsoft are some of the companies that make use of workers online, who hire through platforms such as Amazon Mechanical Turk. Gray calls them “ghost workers”. They perform tagging tasks for artificial intelligence systems, they moderate content, they write false recommendations and solve captchas, those annoying little tests – for example, recognising and writing the numbers and letters that appear distorted on the screen – that are presented as a final step when executing transactions online.
Ghost workers are part of the so-called on-demand economy or the gig economy, based on small jobs. The procedure is simple: now I need it, now I order it, now I have it. And the work will be done by whoever is available at that time. It is another derivative of digitalisation. We go from lifelong jobs to recruiting on a project basis, often mediated by online platforms. This applies to home delivery couriers or private transport drivers.
Work for robots
Automation promised that robots would work for robots, but the opposite has occurred: it is us humans who often work for robots. Or for algorithms. If not “for”, at least “with”. As long as machines do not eradicate employment, we will have to work with them if we do not want to lose our jobs.
Collaboration between people and machines will be increasingly common despite the challenges, which are not so new. At all stages of technology adoption we have had to adapt. We first learned to use tools, then machines, then computers, and now, as computer systems pretend to be smart, we go from using them to working with them. In the connected world we are asked not only to be familiar with new technologies, but to collaborate with them. We are asked to change our knowledge and our dynamics, to adapt to tools devised to complement us, to augment us and, in many cases, to replace us.
Cobots are collaborative robots, designed to physically interact with humans in the same workspace.
The future of work has already arrived for a vast majority of the online workforce. The pandemic has forced many workers to digitise at top speed. Eighty-four per cent of employers claim to be ready to rapidly digitise work processes and move 44% of their employees to remote working, according to the aforementioned WEF report “The Future of Jobs Report 2020”. The challenges here are productivity and well-being: creating a sense of community, connection and belonging through digital tools, and providing new types of services and aids adapted to the needs of telecommuting.
Incorporating the use of automation computer technologies and systems into day-to-day work is a challenge that can be seen as a barrier. It calls for changing daily routines and deeply rooted practices, if not redefining roles. Companies estimate that approximately 40% of their workers will require retraining, according to WEF data. Retraining and redistributing tens of millions of middle-aged workers is a major challenge: 37% of European workers do not even have basic digital skills, as the Europe’s Digital Skills and Jobs Coalition points out. The challenge is heightened if we add a robot to the equation, be it a mechanical arm to assist in surgery or a cobot. Cobots (or collaborative robots) are designed to physically interact with humans in the same workspace. For example, in a factory, for the joint assembly of products. Their presence is growing in companies of all sizes, to the point that it is estimated that their market size will increase eightfold between now and 2026.
The dizzying growth of this industry will require people to build, train, and maintain cobots. These machines will replace the work of many people who, in turn, will have to learn to perform much more technical tasks. The collaboration between humans and machines promises to spare people from repetitive and alienating tasks, so that they can focus on other tasks with greater added value, whether they are technical, intellectual, creative, emotional or related to care and assistance. Automation also promises to increase human capabilities, something that to some extent those small devices that we all carry in our pockets already do.
 “Collaborative Robot (Cobot) Market by Payload, Component (End Effectors, Controllers), Application (Handling, Assembling & Disassembling, Dispensing, Processing), Industry (Electronics, Furniture & Equipment), and Geography – Global Forecast to 2026”. MarketsandMarkets, 2020.
Highly-skilled workers are not spared either from this dynamic, they are especially exposed to AI, compared to low and medium-skilled occupations, which are more susceptible to robots and software.
The truth, however, is that in the race between replacing and enabling technologies, the former are winning. This is the conclusion drawn by Carl Frey, one of the authors of the Oxford study “The Technology Trap: Capital, Labor, and Power in the Age of Automation” (2019), after reviewing a variety of recent technological developments, such as machine learning, machine vision, sensors, various subfields of AI and mobile robotics. While these technologies will generate new tasks that will require a workforce, replacing technologies still predominate. This trend, says Frey, is growing, which is not promising in terms of employment for an already battered middle class, whose jobs are the most susceptible to automation.
Highly-skilled workers are not spared either from this dynamic, they are especially exposed to AI – able to complete tasks that require planning, learning, reasoning, problem solving and prediction – compared to low and medium-skilled occupations, which are more susceptible to robots and software. In fact, information technology research and consulting firm Gartner predicts that 69% of routine work currently done by managers will be fully automated by 2024.
Whether these trends will be accentuated or mitigated remains to be determined. The balance between the elimination of employment, the replacement of tasks, the modification of roles or the creation of new jobs depends on multiple factors, on which it is still possible to take action.
All advanced economies have undergone major transformations in the workplace. The agricultural share of total employment in the United States declined from 60% in 1850 to less than 5% by 1970. In China, one third of its workforce moved out of agriculture between 1990 and 2015.
Automation has been displacing some jobs and creating others in new industries and businesses for a century and a half. As the MIT study I mentioned at the beginning of this article points out, the immediate historical background inspires hope: Between 1940 and 1980, rapid technological advances and institutions generated rising productivity and a rapid distribution of wages in a relatively uniform manner. The problem is that that dynamic was broken after 1980: only those workers with four-year university and post-graduate degrees were able to benefit from sustained income growth.
 Lund, S. and Manyika, J., “Five lessons from history on AI, automation, and employment November”. McKinsey, 2017.
Quality before quantity
This trend is avoidable, according to MIT, through strategies for productivity growth and income distribution with educational measures, labour regulations, public investments, tax policies, etc. that prevent the mistakes of the past. Now that technology has spared us from physical labour and computational work, and is beginning to make decisions for us, there is talk of investing in the quality of jobs, not the quantity of jobs.
Quality is precisely something that does not seem to be improving. The Economist was right: the global unemployment rate has dropped in high-income countries, and this may serve as a rebuttal to claims that technological change is resulting in massive job losses. However, as the International Labour Organization (ILO) points out in the report “World Employment and Social Outlook. Trends 2020”, job creation has mainly been concentrated in the service sector, where the average value added per worker is relatively low. The widespread precariousness of the on-demand economy’s small jobs and ghost jobs must be added to this.
Moreover, middle-income countries that have endured economic downturns in recent years still have high unemployment rates. The poor prospects for the world economy did not predict improvement before, and less now with COVID-19. With an unemployment rate above 16%, Spain does not have much to boast about when the average unemployment rate in the European Union is half that.
The pandemic has accelerated digitalisation, automation and robotisation. New and emerging technologies offer the potential to improve quality of life and working conditions. Whether opportunities are seized or not depends on the institutions, public and private leadership, claims MIT, “that transform aggregate wealth into greater shared prosperity instead of rising inequality”. If they succeed, and we succeed, when we look back at 2030 we won’t talk about when robots took our jobs, but about when automation and digitalisation brought forth a better world.
 “Encuesta de población activa (EPA). Serie histórica (datos en miles de personas)” [Economically Active Population Survey. Historical series (in thousands of people)]. Spanish National Statistics Institute (INE), 2021.
 “Unemployment statistics”. Eurostat, 2021. https://ec.europa.eu/eurostat/statistics-explained/index.php?title=Unemployment_statistics
Journalist and author specialised in science and technology. She contribues to El País, El Español (R+D), Xataka and Muy Interesante, among other media outlets. She is a professor in the Executive Master in Artificial Intelligence at the Institute of Artificial Intelligence.
From the issue
N118 - Apr 21 Index
Subscribe to our newsletter to keep up to date with Barcelona Metròpolis' new developments