The march of the intelligent machines

A ‘second machine age’ is likely to test our ability to spread the rewards fairly, but is not something that will destroy jobs, as many fear

By Gavin Kelly  /  The Observer

Wed, Jan 08, 2014 - Page 9

Whether it is our humdrum reliance on supermarket self-service cash registers, Siri on our iPhones, the emergence of the drone as a weapon of choice or the impending arrival of the driverless car, intelligent machines are woven into our lives as never before.

It is increasingly common, a cliche even, for us to read about the inexorable rise of the robot as the fundamental shift in advanced economies that will transform the nature of work and opportunity within society. The robot is supposedly the specter threatening the economic security not just of the working poor, but also the middle class across mature societies. “Be afraid” is the message: The march of the machine is eating into our jobs, pay rises and children’s prospects. And, according to many experts, we have not seen anything yet.

This is because the power of intelligent machines is growing as their cost collapses. They are doing things reliably now that would have sounded implausible only a few years ago. By the end of the decade, Nissan pledges the driverless car, Amazon promises that electric drones will deliver us packages, Rolls-Royce says that unmanned roboships will sail our seas. The expected use of machines for everyday purposes is already giving rise to angst about the nascent problem of “robot smog,” as other people’s machines invade ever more aspects of our personal space.

As economically significant, perhaps, as the rise of supergadgetry is the growing power of software to accurately process and respond to data patterns. This raises the prospect of machines reaching deep into previously protected areas of professional work like translation, medical diagnostics, the law, accountancy, even surgery.

As yet, this techno-hype is not matched by much hard evidence. According to the International Federation for Robotics, the use of robotics in leading advanced economies has doubled in the last decade — significant, but less than you might expect.

However, the experience varies dramatically: Uptake exploded in China, while the UK lags far behind its competitors. The key question is whether the upward trend is about to take off, giving rise to sweeping changes in production that dislocate large tranches of the workforce.

That is certainly the view of several highly influential US economists, such as leading blogger Tyler Cowen from George Mason University, and Erik Brynjolfsson and Andrew McAfee from the Massachusetts Institute of Technology. In works with bracing titles such as Average is Over and Race Against the Machine, they have seized the public debate with their genre of arresting, unequivocal and futuristic argument that blends techno-optimism about the potential of machines with chilling generational pessimism about the divisive consequences for much of society.

Brynjolfsson and McAfee, whose new book The Second Machine Age is set to be one of the zeitgeist works of this year, argue that the digital revolution is about to crash into our jobs market. It has taken a while — Time magazine awarded the personal computer machine of the year in 1982 — but, they contend, the technology has now matured to a point where it will have the same scale of impact on production as the steam engine once did.

Similarly, Cowen speculates that the future belongs to a gilded 10 percent to 15 percent of workers whose skills will augment intelligent machines — the rest can look forward to long-term stagnation or worse. The harsh labor market experience of the young over recent years is a mere taster of what is in store. Growing numbers of low-skilled workers risk being unemployable: There will not be a wage at which it will be worth employing them. Swaths of the working poor will make ends meet only by migrating to areas offering very cheap housing, crumbling infrastructure and low taxes.

Welcome to the US future: burgeoning favelas leavened only by free Wi-Fi. Some of this has a dystopic, Blade Runner feel — it is striking how much of this economic futurology comes from the US. The more sober UK debate is concerned with deciphering the empirics of the recent past rather than conjecturing about the future.

At the London School of Economics, the respected economist Alan Manning, who has led work on the polarizing impact of technology on the UK job market, laments only half-jokingly that he would like to have the time to develop a new subdiscipline on “science-fiction economics.” It would bring rigor to our understanding of possible societies in which machines do radically more and humans less. For now, Britain looks overseas for visions of where the robots may lead the country.

As with all prophecies of doom, or indeed those of an impending economic boom, we should treat such visions with caution. Predictions about the uniquely transformational yet job-killing impact of technological change are as old as capitalism itself. There has never been an era without plausible experts warning the population that they are on the cusp of a new — usually scary — world resulting from technological breakthrough. Occasionally they are not wrong; mostly they are. Which is not to downplay technology as the motor of economic change. Time and again — from spinning wheel to steam engine — it has had disruptive implications for the workforce. However, labor displaced from field or factory eventually found new, more productive roles, demand expanded, living standards rose.

However, the lag can be a long one. Not long before his death in 1873, John Stuart Mill remarked that the industrial revolution had not yet had much impact. This seemed an extraordinary observation, but it captured at least a partial truth. As the economic historian Brad DeLong has shown, from 1800 to 1870 real working-class wages grew at just 0.4 percent a year, before tripling to 1.2 percent from 1870 to 1950 (reaching almost 2 percent in the golden postwar decades). Similarly, we are yet to experience the true gain, whatever it turns out to be, as well as the pain, of the robot era.

To get a better sense of the impact of technology on the labor market, we do not need to rely entirely on frothy speculation about the future. There is a decade or more of research to draw on. The rise of information and communications technology (ICT) is hardly new. The dominant view is that it has already been eroding a swath of jobs that involve repetitive tasks capable of being automated and digitized. This has disproportionately affected roles in the middle of the income distribution — such as manufacturing, warehousing and administrative roles.

This does not result in lower overall employment — for most economists the main change is to job quality, not quantity. There has been a rapid growth in demand for high-skill roles involving regular interaction with ICT, as well a rise in lower-paid work that is very hard to automate — from caring to hospitality. Consequently the balance of employment has shifted upwards and downwards with less in between; as Manning puts it, the labor market has been polarizing into “lovely and lousy jobs.” The impact of technology has been gradual, but inexorable — “it only goes one way,” he tells me. In some sectors the decline in employment and relative pay has been dramatic: The typical heavy goods driver receives less in real terms today than a generation ago.

Some of this is contested. Recent evidence suggests the extent of polarization may be overstated as it has not taken into account entirely new middle-income roles that replace old ones. Others point out that job-title inflation means that yesterday’s mid-level jobs are sometimes counted as today’s high-level ones. Some roles that are popularly assumed to have fallen prey to machines have adapted and survived — as US President Barack Obama realized to his cost when he asserted that ATMs have led to the demise of bank tellers (their numbers have risen). And it is important to keep a sense of proportion: Between 1990 and 2010 employment in hard-hit occupations in the UK like skilled trades fell by 25 percent and administrative jobs by 20 percent. Big losses, but they hardly represent the death of mid-level jobs.

A narrow focus on technology is also inadequate, as it fails to explain some of the big shifts of the past decade, like the explosion in rewards at the very top — 60 percent of the enormous increase in the slice of income flowing upwards to the richest 1 percent over the past decade went to those working in finance. To lay this at the door of the anonymous force called “technology” is to excuse way too much. Sure, developments in ICT were relevant, but they do not explain political choices over deregulation or account for rapacious rent-seeking by the financial elite. Wage inequality has many authors, from the demise of collective bargaining to the rise of globalization. As the influential Washington-based EPI think tank has argued: Do not make robots the fall guy.

Nor does an exclusive techno-focus illuminate the post-crisis polarization of the jobs market, which has seen recession-busting increases in high-paying jobs in sectors like business services alongside a big growth in low-paid work, with sharp falls in between in sectors like construction. Further signs of the impact of technology? Doubtful. This pattern has coincided with a demand-starved economy, an investment strike by business and plummeting wages. Indeed, recently the robots could be forgiven for worrying about their prospects given the falling cost of labor. It all adds up to a complex story. The hollowing out of the jobs market is real and important. However, its scale can be overstated and technology, though crucial, is by no means the only factor at work. None of this means we should be sanguine about the future.

Given the uncertainties and the capacity of market economies to adapt to shocks, many will assume that things will continue much as they have done. Perhaps. However, if the techno-enthusiasts are at least partly right, the consequences will be far-reaching.

Fortunately, perhaps, at least some of the issues that this would mean grappling with are more extreme versions of those we should be worrying about already. The rise of the robot is likely, for instance, to result in an increasing share of GDP flowing to the owners of capital at the expense of labor — something that has recently been occurring across many Organisation for Economic Co-operation and Development countries. An acceleration of this should rekindle interest in finding ways to distribute the ownership of assets more evenly, as well as finally prompting a serious discussion about shifting some of the burden of taxation from labor towards wealth.

Accelerating wage inequality, together with a rise in economic insecurity, would sharpen the need to bolster working-age welfare systems at a time when it is already creaking and has few political friends.

Whether the greater democratization of economic risk — if those in medicine, law and accountancy also feel the pressure — would shift the political dynamic remains to be seen.

We also need to focus on those occupations that are widely expected to grow in number and are dramatically less likely to be displaced by machines — such as care of the young and the old. They are heavily reliant on the state. Securing the fiscal basis and public consent to fund these crucial labor-absorbing industries over the next ageing generation, already an enormous issue, would become even more pivotal.

Historical weaknesses on education policy would cost us more dearly. The wage penalty arising from flaky sub-degree-level qualifications — a longstanding weakness — would rise, as would the premium for those who can combine rigorous analytical thinking with creativity. Massive wage returns are likely to flow to those with applied postgraduate degrees — ensuring fair access to them would become a more central feature of distributive politics.

Just as importantly, we need to prevent robot-fear being used as a force for fatalism. There are already voices arguing that the march of the machine means that a decent wage-floor is simply unaffordable. Yet the evidence that the minimum wage has worked well without costing jobs is vastly superior to that suggesting we are entering a new era of machine-peril. Let us not get too spooked.

J.M. Keynes, writing in 1930 as the Great Depression intensified, was prophetic about today’s public anxieties. “We are suffering from a bad attack of economic pessimism ... people say that the rapid improvement in the standard of life is now going to slow down.” He dismissed this sentiment, putting it down to the upheaval of rapid economic change, and argued that his generation’s grandchildren — today’s baby-boomers — would be better off, which of course they are.

We should be equally confident our own grandchildren will also grow up in a digital economy that is far richer than today’s, driven on in large part by further technological breakthroughs. It is harder than ever, though, to have the same confidence that this greater prosperity will be evenly shared out in the “age of the robot.”

Gavin Kelly is chief executive of the Resolution Foundation, an independent think tank aiming to improve living standards for low to middle-income families in the UK.