The pace of change in technology shows no sign of slowing down, and as artificial intelligence develops, ever-more tasks that were previously carried out by humans could be done by machines.  We have seen technological change before, and history tells us that there will be winners and losers.  We must prepare economic and employment policies now, to protect those most likely to lose out, warns Diane Coyle.

• There are contradictory attitudes to the automation revolution– both enthusiastic and fearful
• So far, in the UK, it has not been putting people out of work – the employment rate is at a record high
• Machines do not always directly substitute for human workers and some may increase demand for human labour
• There are widely diverging estimates of how many existing jobs are vulnerable to automation
• There are some significant economic challenges to increased automation including the scale of this new wave of change and the fact that that in the past, the policy response to automation has been worse than inadequate
• There will be winners and losers and the time to talk is now

It is hard to miss the rise of the robots in the business news pages these days, but coverage of the automation revolution is conflicting. Sometimes it is all enthusiasm for the pace of technological change and the potential of ‘Industry 4.0’ or the ‘Fourth Industrial Revolution’. Robots will be able to care for the growing numbers of elderly people, cars will be self-driving and safer, drones will deliver our shopping. At other times, it is all fear about the robots taking all the jobs – and income – leaving mere humans redundant. The idea of a Universal Basic Income, which has come into fashion every time there is a visible wave of automation, is popular once again.

Robots – friends or rivals?

The first point to note is that whatever the future holds, there has not been enough automation so far – certainly in the UK – to put people out of work. The proportion of the workforce in employment is currently at the highest since the records began in 1971. Another way to make the same point is that the economy’s productivity (or output per hour worked) has been more or less flat for a decade. There is no evidence here of plentiful investment in automation that would have made those who are in work more productive. Some other rich OECD economies have high levels of unemployment but this is more because of either their post-financial crisis troubles (Greece, Portugal) or the strong incentives in their tax and regulatory systems against expanding employment (France, Spain).

The second, and most important point, is that capital equipment – even in humanoid electronic form – does not always directly substitute for human workers. There are many examples of new technologies increasing the demand for labour. One particularly striking one is the introduction of ATM machines during the 1980s. There were predictions of a job apocalypse for bank employees as the new machines rolled out. In fact, the number of employees in the banking sector increased. Released from the routine work of counting out cash and taking deposits, bank staff were able to take on new and more sophisticated tasks, such as savings advice and mortgage applications. The level of qualifications required increased, and so did their average pay.

This pattern of automation of the most routine work, and the upskilling of the work for humans alongside the new machines, is the story of the economy for at least the past 70 years and probably the entire period since the Industrial Revolution. It is not at all obvious, at least yet, that the new types of automation will be any different. Some technology experts argue that the advent of Artificial Intelligence means this time really is different, but the history of capitalism provides encouraging evidence of the adaptability of the economy in creating new kinds of goods and services, and therefore new kinds of work.

There are also widely diverging estimates of how many existing jobs are vulnerable to automation. One well-known prediction, much loved by journalists for its shock value, is nearly half of jobs could be replaced by machines within the next 10-20 years. A more detailed investigation of what kinds of work are sufficiently routine to be replaceable is 9% – assuming that businesses do want to go ahead and make those investments.

Past mistakes and poor policy responses

Even if the lower figure turns out to be closer to reality, there are some significant economic challenges arising from automation.

One is that the scale of this new wave of automation – even at the lower end of the estimates – will be significant. During the 1980s, for instance, in the UK manufacturing sector, employment declined by nearly 12% (employment in the whole economy fell by 4%) in a painful recession and leaving a legacy of deindustrialisation in many areas of the country. The second challenge is that the earlier experience shows the policy response to past automation has been worse than inadequate. The old industrial centres did get ‘left behind’, with many redundant workers unable to find new jobs, unsuited to the new kinds of skills needed in expanding businesses, and trapped in a spiral of unemployment, poverty and ill health that still characterises too many people in these parts of the country. Arguably, the geography of the Brexit vote last year reflects the inadequacy of the policy response to structural upheaval in employment a generation ago. That the political lesson has struck home is evident in the launch of a government Industrial Strategy  Green Paper in January.  The geography of the economy is also one of the key themes of the independent Industrial Strategy Commission.

For although the robot revolution, if that is what it turns out to be, might be good for humans in the long run – as past technologies have proven – it will also create winners and losers. There is a debate to be had about how to provide better for the losers next time around, and the time to have that discussion is right now, before the robots arrive, not after.