There is much fear about Artificial Intelligence and automation in the workplace; popular books such as ‘Rise of the Robots’, ‘Inventing the Future’ and ‘The Economic Singularity’ are proliferating and all cover the forecast that robots might produce waves of unemployment in the already post-industrialized West.
Amidst this forecast, the idea of Universal Basic Income (UBI) has become popular amongst left and right thinkers as a potential solution to mass automated unemployment and help our society transition to a form of ‘post-work’ economy. Much of the rhetoric on this topic could be described as alarmist; the robots are coming very soon and we will all be left stranded in a radically different working landscape without suitable restructuring and policies (such as UBI) to deal with the consequences.
Technological unemployment does occur, but the prominence of the ‘coming of the robots’ thesis has once again ignited imaginations across the political spectrum, this fear may be overstated and there are a number of things to consider as to why it may be an economic panic rather than a certain or probable future.
This is not a new fear
The same fears, debates and rhetoric existed 40 years ago, if you watch BBC Horizon’s 1977 episode ‘When the Chips are down’ and updated the terms of reference contained within the episode, you could transpose the programme to 2017 and not notice. Fortunately we have 40 years of history to provide evidence that mass unemployment did not occur due to the rise of the microchip and machines. The economic landscape has radically changed since 1977 in the UK due to a variety of factors, but automated mass unemployment was not a significant factor at play. People were made unemployed, but mostly due to a change in political leadership and industries in the UK disappeared due to outsourcing, however new industries emerged and grew the workforce.
The economic reforms of the 1980s, outsourcing manufacturing and the microchip produced an increase in material wealth and opportunity in the UK. However, this was not evenly distributed across the geography of the country; many single industry towns experienced structural unemployment for many years, with some even suffering from this today. Though this occurred due to a political decision (or lack of) not to rectify or compensate for this change in the British economic landscape. The experience of localised (longer term) and national unemployment (shorter term) was not inevitable and nor was it the fault of technology, it was fundamentally a human decision.
The industrialisation of Britain in the 19th Century should also give us pause for thought. In 1750 the percentage of the population engaged in agriculture production has been estimated at 75%. Between 1750 and 1850 a combination of mechanisation, emerging agricultural capitalism, enclosures and new techniques in farming (such as the use of phosphate) increased yields on less land, reducing the need for as many humans to till the fields to feed the population[i].
By 1850, approximately 22% of the population was still employed in agriculture- a remarkable decline in employment in that sector- meaning many agricultural workers became unemployed from their traditional agricultural jobs and removed from their traditional lands. They did not starve on mass, they migrated to cities where new industries began to emerge; this is known as the industrial revolution[ii].
There was political turmoil associated with the process of industrialisation; the breakup of agricultural communities; the enclosure of the commons and the experience of new exploitative and dangerous industrial environments gave birth to the labour movement and resurrected proto-socialist and socialist ideas. However, despite being at the mercy of the newly emergent bourgeoisie, people did not experience mass unemployment for long due to the industrial revolution’s technological innovation. The post Napoleonic-war depression was arguably to do with post-war readjustment on the continent and an overproduction in Britain leading to a collapse in prices and economic crisis in the Western world. Economic historians may wish to challenge on this interpretation, but I do not think technological progress was the key cause of this dislocation.
The past does not indicate future performance, but the experience of the industrial revolution and subsequent effect of labour displacement does provide evidence and a useful historical case study to refer, to assuage fears of a similar revolutionary change in the economy today. Technological unemployment did occur but history has shown that is should not necessarily be considered a long term problem for capitalist economies. The resulting implementation of the ‘Corn Laws’ might provide an interesting historical parallel to the risk of falling prices as the result of automation and increases in productivity, but that arguably is a more classic economic conundrum rather than a unique experience that we might face in the 21st century.
Busts and mass unemployment have occurred at various times, but so far, technological unemployment has been a small and partial factor in those moments of crisis at best, never yet the key causation.
Businesses will take time to automate
Capital is expensive, do not be fooled that it isn’t because there is a cash glut in the global economy; it remains expensive because of opportunity cost, risk and probable returns on investment (ROI).
Capital has an opportunity cost for the lender/investor, as it can be used in a variety of ways with differing returns and risk. Stumping up money for a business comes with the risk that the business may go bankrupt and be unable to repay the capital, or produce a return that pales in comparison to other, safer options, that could have been alternatively be chosen. Even businesses that have raised capital, only to grow and become successful, can sometimes find that the equity they traded for initial investment is expensive on exit.
All this together (and a lot more that has been left out), means capital is expensive and this reduces the appetite to raise capital, lend capital, give away equity and replace labour (who can just be fired, as and when the need arises) with machinery (which is a sunk cost that depreciates with time). Businesses do raise capital and invest in new opportunities and technologies, but it is tempered by conventional business acumen and this will not disappear but be ever present in our economic landscape.
Furthermore, if we are about to experience an exponential increase in the use of artificial intelligence and robotics in business, then we should expect that new technologies employed in production processes will quickly become obsolete. Thereby increasing the risk of misallocated capital expenditure; managers and businesses pay a heavy price for getting this wrong. Any business that substitutes its workforce one day for an army of robots the next, might find the capital investment in that technology has quickly become substandard compared to the competition who held out a little long and has bought the ‘next generation’ of robots.
This will make it harder to calculate and be confident in forecasting a return on investment for CapEx; for the question of when to and what generation of robotics to invest in, will slow this process down. New technologies are first paid for by the richest people or companies, and usually are regarded as substandard after a short period of time. An example of this is the first iPhone’s; they were prohibitively expensive, difficult to use, lacking wider infrastructure to complement it and fundamentally they were a poor product.
The richest in society bought them because they could, which provided apple with immediate return on the high research and development costs of creating the first iPhone. Then the cost started to fall and the product improved in subsequent generations until it became affordable and user-friendly enough for it to be a genuinely mass consumed product.
The same will be true for industry, the large multinationals will and have been the first to use cutting edge robotics in production, but it will take time for the robots to improve their performance and become affordable for the majority of businesses.
Therefore it will likely be the economics of capital investment, coupled with the risk-averse nature of many businesses that limits the innovation and growth within commercial deployment of AI and robotics. There will be a ‘grace’ period, long enough for business, people and policy to change, plan and adapt to a more automated economy.
Furthermore, businesses are typically inefficient; their primary focus is on the next quarter and current financial year. Contrary to popular perception, most businesses do not turn a profit and find is difficult to do so. The likely experience of automation in the majority of businesses and sectors will probably reflect that of the IT revolution in the 1990s, which even now, has yet to be fully completed. All businesses now use computers, but tour most businesses and you might observe that their IT infrastructure and operational procedures haven’t progressed as much as expected.
Industries with more straightforward production processes have already automated production; the big monopoly businesses in various sectors that have the capital and incentive to automate in order to protect their leading position in the market. Small and Medium Enterprises simply do not have the strategic slack, capital or certainty to engage in a process of rapid automation; competing in the current business climate will take priority, leading to a much slower uptake of robotics in the workplace.
Economics is the study of how scarce resources are produced and distributed amongst a population, despite effectively producing an abundance of basic human needs, the UK continues to produce more goods and services that stretch beyond satisfying those basic needs.
One might point out that in a capitalist economy of private ownership, a person or group of people cannot really live a self-sufficient lifestyle outside of capitalism due to private ownership of land and the mean of production, therefore the majority of the population have to continue to engage in wage labour. This point cannot be overlooked, but it does not explain away our seeming desire for more goods and services beyond our basic human needs.
As traditional manufacturing shrank in the last 40 years, the service economy has grow in its place and it has found there is sufficient demand amongst the population to eat in restaurants, be entertained, drink in pubs and be served coffee in exchange from money. The material gains coming from cheaper production and goods due to outsourced manufacturing has produced a surplus of wealth which has been diverted into providing more services in the UK. A similar effect might occur from the productivity gains of a more automated working environment, thereby creating new industries and work opportunities for all.
It appears as if there is no limit to our demand, and as technology reduces costs we will likely see more opportunities to work will become viable and serve some previous unmet demand. The rise of services such as Uber -merely an aggregating app- has seen the emergence of a new supply of private car hire drivers, meeting a demand that always existed but was kept in check due to licensure and regulation of traditional taxi services. Companies such as Taskrabbit are capitalising on the market’s demand for ‘someone else’ to do everyday chores that previously people did themselves.
New industries and opportunities to work are emerging all the time, these may carry with them their own economic and political significance, such as the precariousness of employment within the so-called ‘gig economy’, but there appears ample evidence that there is much work to be done today and this will not decline in time for tomorrow.
Even if we take the belief that we will very shortly see a paradigm shift in AI and robotics that creates affordable and sophisticated robots that can do previously unmanageable tasks, there will remain gaps where it is either safer, easier, more acceptable or cheaper for human labour to do.
Previously unimaginable opportunities for work will quickly emerge to respond to people’s increasing material wealth surplus as a result of accelerating automation, perhaps in forms that make it impractical or expensive to automate in the near future. This may be a different type of work than we have previously seen, but there is no reason to assume there will be no ‘work’ to do that can be monetized in typical capitalist economic relationships.
It is our own desires and the structure of capitalism that can continually create the supply of work to fulfil unmet demand. Capitalism has and continues to be a dynamic and transformative force 200 years after it became the leading economic system in the world, it would take a brave forecaster to suggest that increasing automation in the 21st Century was the factor that tripped it up and finally dug its grave.
Workers are people and people adapt
The experience of this lag in automation in the majority of the business ecosystem will buy time for the current workforce to prepare for change. It is unlikely that many current workers will be doing the same job, for the same company, in ten years time. Therefore if a job is under threat of automation, with an expected lag on automation being implemented, then we should expect that the majority of current workers will have experienced one of the following: promoted into management; undergone a career change; re-skilled in anticipation of changes or have left the labour market completely, thereby reducing the risk of technological unemployment.
Young workers entering the labour market in the future will likely have different skills, knowledge and expectations compared to the current workforce. We can observe this change now, with near universal computer literacy among millennials. The next generation of the workforce will not be shocked by a changing labour market and will likely be expectant of what skills may be required within it.
The argument here is not to downplay the ethical or political objections to what the next (or current) generation of workers are facing, but to point out that they will have different interests compared to now, act in their own interests and may respond to the challenges in an unexpected form. The economy is never static and nor are workers, we will not know how the workforce will respond to increasing automation until it happens and we are limited in our knowledge and confidence to forecast how they might react.
Millennial’s so far have adapted to the current labour market, reducing their expectations of a ‘job for life’ and responding in their own interests to a labour market that requires constant updating of skills and job movement. This doesn’t rule out the possibility of a political reaction (no matter how small or great the chances of success), but often people normalise their experiences and acceptance and stability finds a new equilibrium.
Meet the real problem, same as the problem
The real cause for concern are issues that have already revealed itself to us; problems surrounding house prices, asset ownership, income inequality, the decline of the middle class and growing pension liabilities. Older generations of workers managed to secure homes, pensions and other assets, whatever their profession in the past, while younger workers (even those in traditional middle class professions) today are increasingly finding it difficult to save up for a home while rents are increasing due to a housing shortage.
The difference though, is that while older workers have homes and are asset rich and cash poor, younger workers are materially richer than their parents were at a similar age. The minimization of manufacturing and ‘jobs for life’ through outsourcing has led to a surplus that we now all enjoy. Hand-me-downs and little discretionary spending on consumer goods or toys was common place in the homes of the 1960s, 70s and 80s, even in wealthier middle class households. Today, even a minimum wage worker has at their disposal a powerful computer in the shape of a phone in their pocket and their own personal cinema through innovations such as online film streaming. This is a stark contrast in material wealth to the past and should not be easily forgotten.
This increased material wealth is perhaps the saving grace for our current economy; workers can still enjoy technology and increasing access to cheap entertainment despite their precarious position in the labour market and lack of access to assets. As these current young workers move into middle age we will see the next 20 years looking very different compared to the last 20 years, but the battlefronts are likely to be fought around house building, pension liabilities and government spending rather than a new age of mass unemployment.
The political reaction to this could just as easily take a right wing manifestation as a left wing one; people tend to become more fiscally conservative as they age due to changes in incentives and interests, large parts of which will remain despite the likelihood of reduced asset ownership in the future. We simply cannot forecast with any certainty how workers will react to a middle aged experience that very different from their parents’ generation.
This does not inevitably translate into a revolutionary moment; there will be consequences if these problems are not addressed but revolutionaries of left and right may be disappointed in how familiar the political alignment of the population will be in the future, there is a fair chance that the political landscape will look much like it does today and workers end up contented in adapting to this new economic landscape.
The challenge therefore is tackling existing issues such as the restriction on the housing supply and incentivising more geographically even economic development rather than the overstated coming of automation in the workplace creating sudden mass unemployment.
This is not to say that there will not be localised disruption and experiences of large unemployment in specific geographies, but this will likely be the exception, not the rule. Nor will there be a lack of a political reaction to navigate as class divisions and differing political preferences are an ever present feature of a democratic capitalist society. However, the 21st Century is a different country to the world of the late 1970s and early 1980s, workers and work, are more mobile which may mean these experiences are easier to resolve compared to the past.
In ten years time, the middle aged workforce will be made up of workers who entered the labour market with ICT skills and very different expectations of working life compared to workers 50 years of age and more. They will expect to be flexible and attuned to change as the norm in their career; they will be used to changing jobs frequently and respond to the conditions around them. People and the economy will adapt. Therefore this fear of automated mass unemployment is probably much overstated and so too, are the policy ‘solutions’ that are suggested by many thinkers as ‘necessary’.
Long term technological unemployment for the population is a possibility, nothing is certain in human affairs, but we must ask ourselves if it is more likely to occur in this wave of automation and associated productivity gains, compared to previous eras of technological revolution.
We may want to trial or implement ideas such as UBI for a variety of reasons or because we have certain political preferences; but this should not be offered up with such certainty as a solution to automated mass unemployment, as it is likely it is a ‘solution’ in search of a ‘problem’.