Confronting the social consequences of automation


As the world of work changes beyond all recognition, three very different approaches to dealing with the social dislocations caused by automation have been touted - training and reskilling, a universal basic income, a job guarantee - but each comes with its own problems.


“I want to say, in all seriousness, that a great deal of harm is being done in the modern world by belief in the virtuousness of work, and that the road to happiness and prosperity lies in an organised diminution of work.” ― Bertrand Russell, In Praise of Idleness and Other Essays


Last month a report from the Office for National Statistics estimated that around 1.5 million jobs in England run a high risk of being partially or completely automated. The report is just one of the latest in a long line of educated guesses about how artificial intelligence and automation will impact employment in the decades ahead. The ONS’s methodology references four previous studies – from the OECD, Oxford University, PwC and McKinsey – which variously estimated that between 9% and 35% of jobs in the UK are at high risk of automation, depending on whether the authors looked at the likelihood of automation at the job or task level.

Automation will render obsolete millions of jobs that currently provide gainful employment

Ignoring the variation in headline-grabbing predictions, the experts are in agreement on two points: firstly, that automation will render obsolete millions of jobs that currently provide gainful employment; and secondly, that these job losses will not be distributed evenly throughout society. The NAO report notes that 70% of the jobs it assesses as high risk are currently occupied by women, and that the age group with the highest proportion of high risk jobs is 20 to 24 year olds.

Washington-DC think tank The Brookings Institution conducted an analysis of the impacts of automation earlier this year and concluded that around 25% of US workers are at high risk of being automated out of a job. Their data suggested that, while the previous wave of automation was responsible for a ‘hollowing out’ of the labour market through the computerisation of routine middle-skill jobs, the next wave would hit lower-wage jobs which require low education. The ONS makes a similar forecast for the UK, estimating that 99% of the jobs at greatest risk of automation are held by people with A-Level or lower education.

The disparities don’t stop at sex, age and education either. Geography has a role to play too. The Centre for Cities think tank estimates that jobs outside of the South of England are nearly 40% more likely to be at risk of automation than those in wealthier locations close to London. This should come as terrible news to anyone who’s looked on in horror at the widening gulf in wealth and opportunity between former industrial heartlands and service-oriented cities. This gulf has resulted in increasingly polarised politics, impossible to ignore since the 2016 US presidential election and the UK’s EU referendum of the same year. Immigrants may have borne the brunt of right-wing anger, but the robots are ready to take over from cheap overseas labour when it comes to putting downward pressure on wages, as Bank of England officials noted in a recent blog post.

Some suggest that in the long run artificial intelligence and automation will lead to a new renaissance, where humans freed from the necessity of mundane, repetitive jobs will be able to spend more of their time pursuing art, music and philosophy. But even optimists must admit that, in the short term, automation will continue to produce massive social upheaval, as the resultant unemployment afflicts some groups to a much greater extent than others. Modern societies recognise unemployment as a social ill – the UK alone spent £264 billion in 2017 supporting people with no or low income – but current policy measures are clearly inadequate for tackling the huge disruptions just over the horizon.

Celebrated economist John Maynard Keynes defined technological unemployment in his 1930 essay, Economic Possibilities for Our Grandchildren, as “unemployment due to our discovery  of means of economising the use of labour outrunning the pace at which we can find new uses for labour”. Since then, many proposals have been put forward for how to deal with this “temporary phase of maladjustment”. Which one you favour will depend not only on whether you are a right or left winger, a technological optimist or pessimist, a libertarian or authoritarian, but also on whether or not you believe that work itself is central to the human experience.

Holding out for better jobs through training and reskilling

The prevalent view among those touting the benefits of automation technology is that it will add as many, if not more, jobs to the economy than it will take away. The latest report making such predictions comes from the rarefied air of Davos in Switzerland, as the World Economic Forum’s Future of Jobs 2018 paper predicts that, globally, robots and software will displace 75 million jobs by 2022 but create 133 million new ones. Closer to home, Deloitte carried out a study based on 15 years of UK employment data and estimated that 800,000 jobs had been lost but nearly 3.5 million new jobs created through the adoption of new technologies.

The jobs replaced by automation are very different in nature to those created by it

Those figures make for wonderfully optimistic headlines, but read beyond the first page of either report to find the major caveat: the jobs replaced by automation are very different in nature to those created by automation. The Deloitte report notes the ‘hollowing out’ effect, due to the fact that routine cognitive jobs that require precision, like bookkeeping, aren’t typically performed by low-skilled workers but are “increasingly automatable, in particular through rapid advances in technologies such as ‘big data’ and machine learning”. Regardless of the skill-level involved, it is jobs which involve routine tasks that provide the most fertile ground for automation. Hence the ONS report placing finance analysts at greater risk of automation than gardeners. Hence the sharp decline in employment for occupations like personal assistants, secretaries, clerks, cashiers and sales executives over the period between 2001 and 2015. The largest rises in employment over the same period, according the Deloitte report, have been among care workers, teachers, nurses, chefs, and project, sales and business development managers. In other words, jobs which require social and emotional intelligence – an area where automation, at least for now, lags behind.

If this trend continues, we can expect large numbers of people currently employed in medium-skilled yet routine jobs to be displaced by machines. Lots of those people will be able to find employment in lower-skilled (and lower-paid) jobs. For many that won’t be a practical or desirable option, but training and reskilling will be required to move into a higher-skilled role which is less susceptible to automation. The WEF report recognises this need:

“By 2022, no less than 54% of all employees will require significant re- and upskilling. Of these, about 35% are expected to require additional training of up to six months, 9% will require reskilling lasting six to 12 months, while 10% will require additional skills training of more than a year.”

But the authors also acknowledge that nearly a quarter of the companies they surveyed are ‘undecided or unlikely’ to pursue the retraining of existing employees. Another recent report, this time from the McKinsey Global Institute, paints an even starker picture:

“The shift could be on a scale not seen since the transition of the labor force out of agriculture in the early 1900s in the United States and Europe, and more recently in China. But unlike those earlier transitions, in which young people left farms and moved to cities for industrial jobs, the challenge, especially in advanced economies, will be to retrain midcareer workers. There are few precedents in which societies have successfully retrained such large numbers of people.”

Who will even take up the challenge? According to OECD data, public spending on training and skill support has been falling across most member countries. In the UK, government spending on employment policies, in real terms, has fallen from £5bn in 1994-95 to £2.6bn in 2017-2018. Even if government had the appetite for large-scale retraining and upskilling programmes, is it best placed to identify the skills shortages faced by businesses looking to realise the benefits of digitalisation, artificial intelligence and automation? The public sector’s track record with worker retraining programs would suggest not.

According to McKinsey, corporates have identified addressing the skills gap as an urgent business priority: sixty-two percent of executives at US and European companies with revenues over $100m believe they will need to retrain or replace more than a quarter of their workforce between now and 2023. But ‘retrain’ and ‘replace’ are two very different options. If the latter has a less negative impact on the bottom line, then the efforts of private enterprise will do little to halt the growth of an army of unemployed middle-aged, mid-career workers.

Despite the promise of new technologies to create novel kinds of economic activity, and new kinds of work, in the decades ahead, it seems that in the short term at least, technological unemployment is a certainty. Rather than trying to repurpose workers for the ever-shrinking number of middle-skill jobs available, it may be that we need to consider more radical policies for reducing, or mitigating the effects of, technological unemployment.

Dialling down on work with a Universal Basic Income

Among the most radical proposals for dealing with the social fallout from automation is a modified form of an idea that’s been around since the Renaissance. While Sir Thomas More, writing in Utopia, didn’t come up with Universal Basic Income, he did propose giving everyone the basic means of subsistence as a solution to the problem of thievery which would avoid using the gallows as a deterrent:

“Instead of inflicting these horrible punishments, it would be far more to the point to provide everyone with some means of livelihood, so that nobody’s under the frightful necessity of becoming, first a thief, and then a corpse.”

Silicon Valley billionaires, perhaps feeling some responsibility, are flocking to the basic income banner

Since then, distinguished thinkers from Bertrand Russell to Martin Luther King Jr. to Sir Tim Berners-Lee have advocated government provision of a guaranteed income for all citizens to alleviate the poverty (all too ironically) caused by rising productivity. They’ve recently been joined by economist and Nobel laureate Angus Deaton, who sees what he calls Basic Income Grants as an effective way to combat the rising wealth and consumption inequality due to rapid technological change. Silicon Valley billionaires, perhaps feeling some responsibility, are also flocking to the basic income banner. Delivering the commencement speech at Harvard in 2017, Mark Zuckerberg told his audience that it was his family’s financial security that gave him the freedom to pursue a risky venture like Facebook:

“Now it’s our time to define a new social contract for our generation. We should explore ideas like universal basic income to give everyone a cushion to try new things.”

Tesla’s Elon Musk, Slack founder Stewart Butterfield and Y-Combinator President Sam Altman have also spoken in favour of Universal Basic Income as a way to unlock entrepreneurialism and put the brakes on growing inequality. Virgin founder Richard Branson has even suggested that a tax on the wealth generated by new technologies should fund cash handouts.

The idea of some form of basic income provision is gathering traction among both left- and right-wingers as a way to provide a social safety net without increasing the size of government. Small trials have taken place in Finland, California, Ontario and Italy. None have conclusively proven that a basic income can alleviate poverty or spur entrepreneurship, but then, none of the trials has been universal, since they involved cash stipends to only a small group of recipients, often with strings attached (citizens of Livorno in the Italian experiment, for example, had to be unemployed and prove that they’d been looking for work). Trials involving only a small portion of the population can’t be used to determine the effects on society as a whole.

Alaska has, since 1982, been sharing its oil revenue with citizens in universal annual payments ranging from around $800 to $2,000 per person, depending on oil prices. But that’s not enough for even the most frugal to live on, and so doesn’t really say much about what effect a universal basic income might have. Neither does it answer the question of how a genuine Universal Basic Income might be funded in a nation without such prodigious natural resource wealth. Some proponents, such as Bill Gates, have proposed a tax on robots to fund universal citizen’s dividends, arguing that it might help keep people in work:

“If you choose to buy a robot instead of employing humans, that is OK, you can do that, but the tax system will be pushing you to at least consider using humans more, unlike today’s tax system, which actually pushes in the opposite direction.”

But such approaches ignore the fact that much automation will be implemented through software at the level of the task, not the role, making it hard to definitively say whether an individual worker has been displaced by a robot or not.

Renegade economist and former Greek finance minister Yanis Varoufakis has proposed an alternative funding mechanism for what he calls a Universal Basic Dividend. Under this proposal society would become a literal shareholder in every corporation, with a fixed portion of all new equity issues (IPOs) placed in a public trust and the income distributed evenly to all citizens. Varoufakis argues that this is both simpler and fairer than a robot tax.

While disagreements over how to fund a universal basic income, and how much it would cost, account for much of the debate, other objections to a UBI centre around the value we place on work. Some critics of cash handouts for everyone argue that ‘free money’ would remove people’s incentive to work, thereby reducing productivity. But since we’re considering universal basic income as a response to lower demand for human labour, and skyrocketing productivity due to automation, we can set aside this complaint. (Also, the evidence from Finland’s admittedly limited trial indicated no impact on people’s likelihood to take up paid employment.)

A more radical critique, and one that is harder to dismiss, is built on the idea that work serves a much greater purpose in our lives than merely providing the means of subsistence. While endless leisure time to pursue our hobbies, interests and artistic yearnings may sound like the satisfying fruits of a nearly fully automated world, authors like Charles Hugh Smith, in Money and Work Unchained, suggest that those fruits may leave a bitter taste in the mouth, as they fail to provide the human capital (skills) and social capital (connections and social standing) that we build through work:

“When we deprive people of work, we deprive them of much more than their income, which could be replaced by Universal Basic Income (UBI). We deprive them of the wealth of social and personal value that work generates. In other words, we deprive them of the opportunity to build capital and wealth.”

Smith argues that work is central to our sense of purpose, meaning, dignity, self-worth and identity. But how can everyone have the opportunity to engage in meaningful work when the profit motive that drives automation results in an ever larger portion of activities being taken care of by machines?

Addressing the need for purpose with a Job Guarantee

Hyman Minsky was a towering figure in the field of economics, one whose work on financial instability has seen a resurgence of interest in the aftermath of the global crisis of 2007-08. While the focus of his work was debt-fuelled cycles of boom and bust, he also dedicated much of his time to considering the problem of poverty caused by unemployment. Recognising that the welfare state was a good, but insufficient response (since it assumed that work was always available for anyone who wanted it), he proposed the idea of a government-backed job guarantee. Under this scheme, the state becomes an ‘employer of last resort’, providing meaningful work to anyone who can’t find employment elsewhere.

Championed by economists from the University of Missouri-Kansas City and the Levy Institute of Bard College, the Job Guarantee has picked up many notable converts in recent years. The UK’s Labour Party floated a watered-down version, applicable only to young people unemployed for 12 months or more, in their 2015 election campaign. Then in 2018, several leading figures in the US Democratic Party, including Senator Bernie Sanders, came out in favour.

How can the government guarantee meaningful work? And how would such a scheme be funded?

But while a Job Guarantee ostensibly addresses the need for purpose left unattended by universal basic income, actually implementing it throws up a whole host of questions and problems. How, for example, can the government guarantee meaningful work? And how would such a scheme be funded?

On the issue of costs, David Dayen, writing in The Intercept, argues that huge amounts of federal money in the US are already channelled into job creation efforts in the form of tax breaks and economic development subsidies for large employers, often without much in the way of results. With additional funding from higher taxes on the profits from automation, direct job creation by the government would be simpler, and possibly cheaper.

The question of what kind of work the government might be able to offer – considering that it would have to be low-skilled work to be a universal guarantee – might be answered by looking at existing, automation-driven trends. The rise in employment among care workers and teaching assistants reported by Deloitte over the past couple of decades could easily be augmented by additional government funding, while a rapidly ageing population and increasing class sizes assure demand for workers in areas where automation has yet to make major advances.

However, the highest barrier for the job guarantee to overcome is the lack of appetite on the part of public policymakers to take control of structural change in the economy. As a recent report from the IPPR Commission on Economic Justice on UK industrial strategy pointed out:

“The patterns of investment and production which have developed over the last 30 years have failed to generate a geographically balanced economy able to deliver sufficient numbers of well-paying jobs, to enable the UK to pay its way in the world or respond to future challenges. Policy must therefore go well beyond correcting ‘market failures’. Industrial strategy must be a more fundamental attempt to shift the volume and direction of private and public sector investment.”

Decades of over reliance on market mechanisms to deliver social good, and a rush to outsource government functions to the private sector have eroded the skills, capacities and ambitions of governments in many developed nations to the point where implementing industrial strategy would require a huge cultural shift in the corridors of power. After 30 years of ‘supply side’ economics, the countries of the western world have forgotten how to stimulate demand.

To work or not to work?

Which of the above-mooted solutions to technological unemployment you prefer will likely hinge on the importance you place on work as a generator of meaning in human life. Since political consensus on such a hotly-disputed issue is unlikely, the ultimate response to rapidly-declining demand for human labour may well involve a blend of basic income provision to alleviate poverty, a job guarantee for those who want to work, and training and education to ensure that the few remaining occupations are well-staffed. In the short term though, much might be accomplished by voluntary uptake of a shorter working week.

On the other hand, it could be that this discussion has been framed in terms that are already obsolete. Writing 56 years ago, cultural critic Marshall McLuhan predicted that the greater speeds and interconnectedness of electric technology would eventually dissolve the barriers between work and leisure:

“We have become like the most primitive paleolithic man, once more global wanderers, but information gatherers rather than food gatherers. From now on the source of food, wealth and life itself will be information. The transforming of this information into products is now a problem for the automation experts, no longer a matter for the utmost division of human labour and skill. Automation, as we all know, dispenses with personnel. This terrifies mechanical man because he does not know what to do about the transition, but it simply means that work is finished, over and done with. The concept of work is closely allied to that of specialization, of special functions and non-involvement; before specialization there was no work. Man in the future will not work, automation will work for him, but he may be totally involved as a painter is, or as a thinker is, or as a poet is. Man works when he is partially involved. When he is totally involved, he is at play or at leisure.”

But as the current pile up of political, economic and environmental crises make all too clear, before we can realise this future of meaningful, purpose-driven lives of dignity and total involvement in our intellectual and creative pursuits, we must first find a way to share the benefits of automation more equally.


Photo by Filippo Andolfatto.