Automation on our own terms
We need to rekindle that same spirit of pragmatic optimism which saw technology not as a threat, but as a force for human progress.
In 1921, Czech playwright Karel Čapek introduced the term ‘robot’ for the first time to the English language. His science fiction play, R.U.R., depicted a future where human clones would “do the work of two-and-a-half labourers”. Their purpose? To free people from oppressive toil and allow them to lead lives of leisure.
The story does not end well. Realising they are smarter than the humans who created them, Čapek’s robots overthrow their masters and, in typical cyborg fashion, begin to eradicate humans from the face of the Earth. Just one man is left standing.
Fast forward to 2018 and popular culture is again dominated by tales of machines gone rogue – from Ex Machina to the TV series Humans, and from Black Mirror to the film Automata.
But while the existential threat of robotics and AI remains firmly confined to science fiction, the prospect of new technologies changing the face of work appears real. PwC expects 7m UK jobs to be wiped out by 2040, whereas the Bank of England puts the figure at 15m by 2035. Whichever prediction you care to believe, the picture painted by thinktanks and consultancies is an alarming one.
Yet far from shying away from automation, we believe that the UK economy needs to accelerate its take up of technology if it is to move to a high skilled, high productivity and high pay paradigm. Automation must be pursued on our own terms, with good work guaranteed through a new economic settlement of mass ownership, a data commons and a reimagined social contract.
To anyone who doubts this mission can be achieved, look to the history of the RSA. Over the course of our 264 years, we have witnessed the birth of multiple industrial revolutions – from the advent of the first spinning frames to the birth of modern computing. And at every point we have sought to bend new innovations to the will of the many.
Our Premiums promoted technology for the “publick good”, including handmills that freed people to grind their own corn, and chimney sweeping inventions that did away with the need for child cleaners. Our Great Exhibition of 1851 raised the profile of the best industrial technology, from telescopes to early photography.
Minor as these activities may seem to us now, the note of hopefulness they struck was a radical departure from the despondent attitudes towards technology that plagued the early industrial revolutions. If we are to make the most of the fourth industrial revolution, we will need to rekindle that same spirit of pragmatic optimism which saw technology not as a threat, but as a force for human progress.
The myth of mass job losses
There are few better places to begin than to call out the myth of mass automation, which has for too long dominated the media and public’s attention.
New developments in fields such as deep learning, transfer learning and cloud robotics are indeed remarkable. Autonomous vehicles are now being tested in most developed countries, as are parcel delivery drones and cancer-detecting algorithms. Such feats would have seemed impossible just 15 years ago.
Yet for every jaw-droppingly impressive technology we hear of, there is another that silently falters without notice. IBM’s Watson computer has made several incorrect treatment recommendations for cancer diagnosis. Google Translate still struggles with translating large passages of text, despite years of tinkering. And Ocado’s complex warehouse robotics system continues to require end-to-end human involvement.
What is more, technology rarely automates whole jobs. More often it is designed to substitute for individual tasks. And because the vast majority of jobs contain tens, if not hundreds, of tasks, the removal of one or two by a machine is hardly terminal. No technology can fully substitute for a teacher, a carer, an architect or a construction worker. McKinsey estimates there are 2,000 different types of work tasks across all occupations.
Nor do technologies always substitute labour. Self-driving cars may replace taxi drivers and picking and packing machines may replace parts of a warehouse operative’s job. But CAD software extends designers’ abilities to create compelling visuals, just as robotic medical tools allow surgeons to make more precise incisions.
There are also cases where technology creates tasks that were never done by a human previously, or only by a fraction of the workforce. A prime example is the carebot ElliQ, which can remind people to take their medicine, set up video chats with family and friends, and recommend physical exercises. Given that none but the wealthiest of individuals have carers on hand 24/7, this device cannot be seen as encroaching on human turf.
On the occasions where automation does replace tasks and jobs, the savings to consumers and employers are not lost. In a process the RSA calls ‘recycled demand’, automation can lead to productivity gains and thereby cheaper goods for consumers. The money saved can be spent either on more of the same product or in another market, thereby reviving demand for labour.
One of the best examples of recycled demand can be found in the transformation of the 19th century garment industry. It is estimated that 98 percent of the labour required to weave a yard of cloth was automated as a result of new technologies, yet the number of textile weavers actually grew for a period because prices fell and demand was elastic.
The Great Exhibition of 1851, inspired by the RSA, raised the profile of the best industrial technology, from telescopes to early photography.
Quality over quantity
For all the talk of an impending labour market meltdown, joblessness in the UK has not been lower since 1975. More people want to cut their hours than work more, and involuntary redundancy rates have fallen steadily over the last decade.
Less certain, however, is how the quality of work will change as technology advances.
Many believe new machines will replace lousy jobs with better ones in emerging digital industries. New systems need to be designed and monitored, experts say, and their outputs explained. The number of programmers has grown by 40 percent since 2011, and IT directors have doubled over the same period.
Others doubt a high-tech job revolution is around the corner. An investigation in 2013 by PwC found just 6 percent of all UK jobs that year were of a kind that did not exist in 1990. We may be creating jobs, the authors argue, but they are more or less the same as 30 years ago. Recent analysis by the IFS found that high-skilled jobs made up 46 percent of roles in 2016, barely higher than the 42 percent they occupied in 2005.
Pay is another area of contention. A study of 28 OECD countries by US economist David Autor found that, although technology has not been employment-displacing, it has reduced labour’s share in value added (with owners of capital – machines – gaining the rest). This does not necessarily mean wages have fallen for workers, but rather that they have missed out on the spoils of new wealth.
Again, these claims are contested. In 2015, Georg Graetz and Guy Michaels analysed industrial data for 17 countries from 1993-2007. Their results showed that industrial robots raised labour productivity, increased value-added, and augmented worker wages (although averages can hide wide variations in wage changes).
Technology’s impact on management practices is equally debateable. Biased algorithms used in recruitment could exclude minority groups from new job opportunities, surveillance software could erode the privacy of workers, and gig platforms – which would not exist without sophisticated algorithms – could atomise working partners, undermining job security in the process.
Alternatively, recruitment algorithms could remove bias from hiring decisions by focusing only on a candidate’s experience and qualifications, and surveillance software could prevent accidents and discourage workers from free-riding on the efforts of others. Microsoft, for example, has developed an AI-enabled ‘smart camera’ to detect unmanned tools, spillages and potential accidents in warehouses and factories.
Finally, it is impossible to predict how business models will evolve in response to technological disruption. It was not long ago that the music industry was thought to face decimation because of new streaming services. In the end, business models survived but flipped. Money was no longer to be made in album sales but live performances, with knock on consequences for all those working in the industry.
Too many robots? We don’t have enough
Different machines will have different effects on workers. Some will deskill jobs, reduce the bargaining power of employees, impinge on privacy, and put workers under greater scrutiny. Others will enliven and enlarge workers’ capabilities, help them to achieve more and better-quality work, and raise wages. Automation will create winners as well as losers.
Yet this debate is largely irrelevant if technology is not adopted, and herein lies the great irony of debates on technology and work. Despite the magnitude of commentary on automation, the RSA’s research shows our economy is automating relatively slowly and among only a narrow group of firms. A 2017 RSA/YouGov survey of UK business leaders found that just 14 percent of businesses are actively adopting AI and/or robotics, or soon plan to.
Other research comes to the same conclusion. The International Federation of Robotics finds the UK has just 71 robot units for every 10,000 employees, compared with 189 in the US and 303 in Japan. Overall business spending on ICT, machinery and other equipment has barely budged in real terms since the turn of the millennium.
Monday 19th November, 6pm GMT
Far from being a cause for celebration, low technology adoption rates could weaken the UK economy and our future prosperity
If automation is tough, running an automating business is even tougher. Earlier this year, cobot company Rethink Robotics went bust with a loss of 91 jobs after a major order did not materialise (cobot referring to machines that work collaboratively with workers rather than in isolation). Elsewhere, Johnson & Johnson had pulled the plug on its automated anaesthesiologist machine following disappointing sales.
Far from being a cause for celebration, low technology adoption rates could weaken the UK economy and our future prosperity. First, automation is a means to raise productivity, without which we are unlikely to see a return to meaningful wage growth. In terms of GDP per hour worked, UK workers are 26 percent less productive than their counterparts in Germany, and 30 percent less than US workers.
Second, without adopting technology our businesses cannot hope to be competitive internationally. Firms in tradeable sectors like finance and manufacturing are going head to head with rivals in China, India, Germany and elsewhere. If our businesses do not automate, they will struggle to cut costs and win clients, and jobs will be lost regardless. Automation in this sense can protect domestic work, not act as its adversary.
Third, widespread underinvestment in technology risks a small number of large, tech-led firms racing ahead of the competition and gobbling up market share in the process. Apple already shows signs of moving into healthcare, Facebook into banking and Amazon into bricks and mortar retail (see, for example, its recent purchase of Whole Foods). Concentrated markets are a threat to jobs and a risk to our democracy.
The value of automation is demonstrated by our European neighbours. Germany is one of the most automated economies, with more robots per worker than any other European country. But it also has one of the strongest manufacturing bases and has experienced real wage growth every year since 2014.
Sweden is another country that has embraced automation. According to an EU Commission Survey, 80 percent of Swedes have a positive view of AI and robotics, versus 60 percent of Brits. Why? Because they have created the mechanisms, such as Job Security Councils, to ensure the gains of automation are spread among the population. ‘The Robots are Coming, and Sweden is Fine’, ran the headline of a recent New York Times article.
Automation on our own terms
We need to accelerate the adoption of new technologies in a way that delivers automation on our own terms.
If technology is adopted without due care it will sharpen inequalities, deepen geographic divisions and entrench demographic biases within our workplaces. Educators, employers and policymakers need to be mindful stewards of technology, overseeing its creation and adoption, and establishing a new economic settlement for good work to prevail.
First, we need a social contract fit for the modern labour market. If automation leads to even moderate job losses, inequalities of distribution, or puts downward pressure on wages, we will need a means of sustaining the living standards of people within work, not just outside of it. This provides one of the reasons for committing to Universal Basic Income (UBI) pilots (as the RSA is now supporting in Scotland); establishing a new welfare deal for the self-employed, with more rights in exchange for higher national insurance contributions; and creating Individual Training Accounts, which would give every worker an individual budget to finance lifelong learning.
But top down policy is not the only means of supporting workers. The RSA’s Future Work Awards will soon highlight inspiring examples of grassroots innovation that are reinforcing economic security from the bottom up. Among them are new insurance packages for gig workers, chatbots that can answer queries from workers about their rights, and recruitment algorithms that are purposely designed to boost diversity in hiring decisions. One Fellow, Stuart Field, is launching a Bread Funds scheme in the UK, which would provide the self-employed with a sick pay fund by pooling a small amount of money every month.
A new social contract should also feature a commitment to a shorter working week. As technology makes us wealthier, workers should in theory need fewer hours to maintain the same standard of living. Yet in the post-war period, we have managed to shave off just two and a half hours from the average working week. The solution will not come from policy change alone but from bottom-up experiments within workplaces, as we have seen recently in New Zealand and Sweden.
Second, we need to promote mass ownership and a stakeholder society. If automation means more income flowing to capital over labour, workers must have a stake in assets (the businesses and technology that are becoming ever more profitable) rather than simply relying on earnings. Labour’s recent announcement of a John Lewis-style share ownership scheme for employees could be a step in the right direction. The RSA has proposed a Universal Basic Opportunity Fund, which would be created through a government endowment, replenished annually with levies on wealth, profits and data transfers, to be invested in infrastructure and global equities to pay out periodic dividends to every citizen (workers and non-workers alike).
We should recognise, too, where stakes in ownership already exist and can be leaned upon. Millions of us already have stakes in businesses deploying technology through our pension schemes or other investments. These investments are often small individually, but campaigns like DivestInvest, which seeks to accelerate clean energy investment, show how collective power can shift business behaviour. The charity ShareAction has encouraged more than 100 investors to back the Workforce Disclosure Initiative, which puts pressure on companies to disclose information on how their workforce are treated.
But ownership should not stop at conventional shareholding. The RSA has promoted the community ownership of business as a way of giving people a stake in the services they rely upon and value. From pubs to shops to clean energy generation, this form of ownership primarily exists not to generate income but to ensure institutions serve local interests for the long haul. One example is South West Mutual, a customer-owned bank established by RSA Fellows that will work for the benefit of savers rather than distant shareholders.
Third, we need a new approach to data, which treats it less as an individual asset to be exploited and more as a common asset to support broader social goals. As pools of data expand to power new technologies like AI, we must ask how workers can have a greater say over how data is used and under what conditions. Should data be deployed to facilitate recruitment and interviews? Should employers collect more data to analyse worker performance?
Increasingly, there are calls for individuals to reclaim control over their own data, so that they can manage and monetise what they share. Jaron Lanier and Glen Weyl argue that data is a form of labour, and should be paid for as such. They imagine the rise of ‘data-labour unions’, organisations which can serve as gatekeepers of people’s data and negotiate rates like a traditional union would. John C Havens of the IEEE Global Initiative on Ethics of Autonomous and Intelligent Systems recently made the case that, in response to automation, workers can save their paychecks by becoming their own personal data brokers. However, the RSA and the Open Data Institute believe it is better to frame this challenge in terms of data rights that apply to all – not least because financial returns on data at an individual level are unlikely to reflect its real value. For example, if Facebook or Google were to pay individuals for their data, it would amount to less than $10 per year.
A new Rights Framework for Data, on the other hand, could help people exercise more power over how their data is used. A framework could, for example, limit workplace surveillance, which, according to a recent RSA/Populus survey, half of all workers fear.
Corporations, the state and public services must also develop transparent governance structures to demonstrate how data rights are safeguarded, while not shying away from using data to create better products and services. GDPR is a leap forward but more could do this voluntarily. For example, organisations could commit to disclosing which automated decision systems they use, for what purposes and with what safeguards.
Ending the digital dogma
In the heated debate that surrounds technology, it is easy to forget that we have choices.
Investors can choose which technologies to back. Tech companies can choose which projects to prioritise and which features to build into their products. Employers can choose which technologies to purchase and how to deploy them. Educators can choose which skills to equip young people with. And policymakers can choose the terms of our tax and welfare systems.
Just as the pioneers of the enlightenment struggled against the dogmas of church and state, so too must a 21st century enlightenment challenge the deeply embedded logic of scientific progress and the market. Rather than believe that if something can be automated then it must be, as a society we must continue to ask what technology is for. And more importantly, how will it help us to achieve the goal of good work for all?
Though they may have been painful in the short term, previous eras of technological progress were a tremendous force in making societies more free, humane and equal. It may not feel like it now, but in 30 years’ time we will undoubtedly be more prosperous. The question is whether everyone shares in the spoils.
How do we achieve 21st Century Enlightenment?
By Anthony Painter
It is movements and ideas combined that secure lasting change.
By Julian Astle and Laura Partridge
The teaching-to-the-test culture in our schools is failing to prepare pupils for the future they face.