Why the Dominant Story around Technology Distorts the Truth

The most familiar element of the dominant story about the future is that of impending job losses. The stories dealing with new technology are dominated by messages that underline the speed and scale of change, and emphasise ‘disruption’. It is a story made up of different strands — for example some emphasise the potential for millions of job losses and the threat of what they call ‘societal disturbance’, while others hold out the dream of a future where robots do all the hard work. We are also urged to ‘innovate’ and change. Dystopia vies with utopia in these messages. The common thread is a tendency to view technology and technological change as a force of nature, something that is inevitable. Together these perspectives create a dominant narrative around technology and its impact on workers.

But new technology is not a force of nature. The impacts of technology are determined by policy choices, and at the end of the day, by the actions of people.

The problems with the dominant narrative are clearest on the issue of job losses. Analysts run into serious problems when they try to put a number on the jobs that may be lost and those that may be created.

Frey and Osborne published their famous paper on automation and jobs in 2013. Their assertion that 47% of jobs could be automated using currently available technology was, and remains, widely quoted.[1] Its basic method was replicated with amendments in other studies. But it wasn’t long before critics pointed out that it was not jobs that were automated but particular tasks. With this perspective new calculations were made by the OECD, which gave far lower figures of 9% in the OECD countries and 5% worldwide.[2] Then McKinsey weighed in on the debate using similar data, but different weightings.[3] They also looked at tasks not occupations as a whole. They concluded that less than 5% of US jobs could be fully automated, but that 60% of jobs could have about a third of their component tasks automated. Price Waterhouse Cooper then applied different weightings again and concluded that over 35% of jobs in the UK and US could be automated.[4]

There is no doubt that if automation was simultaneously applied to every possible task it would lead to widespread job losses. But there is a world of difference between what could be automated, and what will be.

The problem with these predictions is that the figures depend on the assumptions the researchers make. Change these assumptions and the figures change. Some experts have noted that these papers generally fail to take note of the complex web of issues that influence how and where technology is adapted.[5]

The problem is exacerbated by the fact that some of the experts and tech companies are using the more extreme forecasts of change as a marketing opportunity, while the media use them to sell papers or drive clicks. Nuance does not sell nearly as well.

Without a well-developed science or a widely-accepted methodology for forecasting the impacts of technological change we cannot be certain how realistic any of these predictions are.[6]

The most honest of the reports on automation are aware of this problem and apply important caveats like this one, “…it is important to keep in mind that these estimates refer to technological possibilities, abstracting from the speed of diffusion and likelihood of adoption […] Adoption, in particular, could be influenced by several factors, including regulations on workers dismissal, unit labour costs or social preferences […] In addition, technology will without doubt also bring about many new jobs.”[7] However, since few read the caveats, it is the numbers that remain in the popular conscience.

This leads us to believe that the figures for job losses are largely meaningless because the result will depend on a multitude of inter-related factors. Interestingly, The Economist, when analysing the ‘microelectronic revolution’ back in the early 1980s noted something similar:

“The most honest approach to estimating the jobs equation is that adopted by a study group set up by Britain’s department of employment. It concluded that assumptions had to be made about so many macroeconomic and other variables that constructing a forecasting model was pointless.”[8]

While it is clear that all new technology tends to eliminate existing tasks and therefore reduce the number of jobs, it also creates new tasks and new jobs. There is a lag between these processes of destruction and creation, and new jobs may not be in the same sectors, but nevertheless, some analysts state that there is more ‘work’ to be done today than ever before. We just have to look at the vast number of problems facing the world today to see that there is a lot of work that needs doing. The problem is that a lot of it isn’t valued by a system geared to exploitation and profit.

Furthermore, some experts argue that the relationship between automation and job losses is not linear. James Bessen’s research shows that automation can bring growing employment in occupations: by reducing the cost of a product it stimulates demand which causes more demand for labour in that occupation. And because automation can make labour more efficient it can increase demand for that occupation.[9] Bessen concludes that computer automation is associated with increased wage inequality within occupations, and with a ‘re-allocation’ of jobs that requires workers to learn new skills. This is a far cry from the loss of millions of jobs over a short period of time.

Since the 1960s we have seen the increasing use of automation in the workplace, but it seems clear that the best results are achieved when machines augment human labour, not when they replace it. In a recent example, Tesla has admitted that it’s production line was ‘over automated’ and this actually slowed production. Automation could not handle the problems and unexpected complexities of the production process.[10] In other cases, such as JD’s famous automated warehouse, the facility can only handle certain types of goods — another example of the way machines are not as flexible as human labour. Automaton can also reduce productivity by de-motivating workers, as occurred in some Soviet factories in the 1970s. We suspect that similar issues will occur across the transport and logistics sector as businesses try to implement new technologies. The point is that new technology is not a panacea — it will work in some areas, at some tasks, but not in others.

If we consider this alongside the fact that as societies we face enormous environmental, social, political and economic challenges, can we ever really run out of work? The answer is no. It’s just that much of this work is currently not valued or is drastically undervalued.

Sceptics on the employment impact of automation point to the following issues:

- In its normal operation an economy destroys and creates jobs constantly

- The job-loss figures the ‘automation and job losses’ papers produce are within the margins of this jobs ‘churn’

- The ‘long tail’ of the economy, the small and medium sized firms where most people work, does not have the capital to invest in new technology

- Some research shows that there are more jobs in areas that have adopted computers since 1980[11]

- Some research shows a small increase in jobs in sectors that have adapted ‘3rd and 4th industrial revolution’ technology.[12]

- There is no finite ‘lump of labour’ — there is always more work to do

- If technology were replacing jobs on a large scale, we would expect to see an increase in productivity. But the statistics show an overall decline in productivity growth across the OECD.[13]

- In neoliberal capitalism the tendency is towards short-termism and flexibility. Nothing is as short-termist and flexible as cheap, vulnerable labour. In many ITF sectors the cost of labour is still falling.

For most workers the main issue is that new technologies will change many aspects of their work, but not necessarily that their jobs will disappear. This perspective is reflected in the 2016 McKinsey report’s conclusion that 60% of jobs could (if automated today) lose up to a third of their component tasks.

This highlights that technology’s main impact will be to change jobs and work processes, rather than eliminate work altogether.

At the same time, there are some aspects of new technology that will be transformational — such as the collection and use of data from a world of connected sensors, including the widespread monitoring of the workforce. But the challenge here is more to working conditions and democracy rather than job elimination.

[1] Karl Frey and Michael Osbourne, ‘The Future of Employment: How Susceptible Are Jobs To Computerisation?’ September 2013.

[2] Arntz, M., T. Gregory and U. Zierahn, “The Risk of Automation for Jobs in OECD Countries: A Comparative Analysis”, ( May 2016), OECD Social, Employment and Migration Working Papers, №189, https://doi.org/10.1787/5jlz9h56dvq7-en.

[3] McKinsey Global Institute ‘A Future That Works: Automation, Employment, and Productivity’, January 2017.

[4] Price Waterhouse Coopers, ‘UK Economic Outlook’, March 2017. https://www.pwc.co.uk/economic-services/ukeo/pwc-uk-economic-outlook-full-report-march-2017-v2.pdf

[5] ‘Work in the Digital Economy: Sorting the Old from the New’, Valenduc and Vendremin, ETUI Working Paper, March 2016, p.16.

[6] Forecasting developments in a planned economy would be much easier. But in a market economy chaos dominates.

[7] Ljubica Nedelkoska and Glenda Quintini, ‘Automation, skills use and training’, OECD, March 2018

[8] ‘Microelectronics; All that is electronic does not glitter’, The Economist, 1 March 1980

[9] James Bessen ‘How Computer Automation Affects Occupations: Technology, Jobs and Skills’, Boston University School of Law & Economics Working Paper №15–49, November 2015.

[10] See ‘How Tesla “shot itself in the foot” by trying to hyper automate its factory, Quartz, 1 May 2018. https://qz.com/1261214/how-exactly-tesla-shot-itself-in-the-foot-by-trying-to-hyper-automate-its-factory/

[11] James Bessen ‘How Computer Automation Affects Occupations: Technology, Jobs and Skills’, Boston University School of Law & Economics Working Paper №15–49, November 2015.

[12] See ‘Digitalisierung und die Zukunft der Arbeit: Makroökonomische Auswirkungen auf Beschäftigung, Arbeitslosigkeit und Löhne von morgen’ ZEW, 2018.

[13] “The slowdown in productivity growth — already underway before the crisis — combined with sluggish investment, continued to undermine rises in economic output and material living standards in recent years in many of the world’s economies.” http://www.oecd.org/sdd/productivity-stats/ (retrieved on 26.04.2018)