In 2013, Frey and Osborne published The Future of Employment: How Susceptible are Jobs to Computerisation?, in which they explored the question of how many jobs were at risk of being computerised.
By combining their analysis of current machine learning technologies and US Labour Bureau data on 702 different occupations – breaking down occupations into a list of tasks involved – the authors were able to figure out the proportion of jobs in the US economy, as of 2010, which were in danger of being automated. They reported that
‘47 percent of total US employment is in the high risk category, meaning that associated occupations are potentially automatable over some unspecified number of years, perhaps a decade or two.’
While the authors stress against conflating automation of jobs and unemployment – they do not consider job creation and general equilibrium effects – the figure is testament to just how disruptive machine learning technologies are likely to be.
But there is more. Supposing that unemployment is seen on a sufficiently large scale, and as many lose their incomes, aggregate demand would decline, plunging the economy into a bout of demand-deficient decline. As the firms who laid off their workers see their sales fall, the blessing of the machine learning algorithms and sophisticated robotics might not make most a great deal better off.
The fourth industrial revolution could place policy makers in an unprecedented position, where conventional macroeconomic wisdom may not be sufficient. This seismic event has yet to occur; the UK’s participation rate is currently over 75%. But what should policymakers be doing now to evade this predicament?
In short, there are two options: give a man a fish, or teach a man to fish. A universal basic income, or raised guaranteed minimum income staves off deficient aggregate demand. Moreover, giving a man a fish today might offer him the time to train to become a computer programmer, journalist or any other profession unlikely to be automated soon. In fact, the benefits of a universal basic income might outweigh the costs, but this idea must be put to one side given there is no objective framework to calculate this.
The second option would be for policymakers to heavily invest in programmes which boost the human capital of workers across the economy – particularly those whose jobs are most at risk of being automated.
Going back to the fundamentals, mass automation is a great thing; it should be welcomed and encouraged. It will boost productivity, eliminate many laborious tasks and make everyone richer. But if a byproduct of this automation is large scale unemployment, the benefits will not only be unevenly distributed but would also be diminished on an aggregate level by there not being enough people to buy the newly automated goods.
By providing those whose jobs are most at risk of automation with access to training programs (be these vocational or higher education), policy makers would be able to fend off mass displacement of labour and enable more people to see their standard of living rise.