Rise of the machines

Why has the introduction of robots become such a contentious issue in the financial media and what are the negative charges laid against such machines, including their alleged role in increasing unemployment and widening inequality?

The Hollywood film Robot & Frank, which came out in the UK earlier this month, illustrates vividly how torn society is about the use of robots.

In the near future a son gives his ageing father, who happens to be an ex-convict, a robot companion to help him with tasks such as gardening. So to begin with the robot is playing a benign role by helping to improve an elderly person’s quality of life.

However, the father has other ideas. He teaches the robot to act as a high tech helper in a series of burglaries. The machine has defied the intentions of its makers by taking on malevolent tasks.

This quandary neatly sums up how the use of ever more robots to replace human labour is generally viewed. Although most people would acknowledge intelligent machines can play a positive economic role their application is also viewed with foreboding.

The main concern is not killer robots akin to the T-1000 or the T-X in the Terminator movies. It is instead directed at robots more akin to C3PO or R2-D2 from the Star Wars series: their designated role is to help people rather than to destroy them. Similarly the new generation of real robots, although not generally as advanced as those in science fiction, are not meant as threats to humans. Rather they are designed to enhance human capacities or to aid humans directly. Nevertheless in the eyes of many they are likely to cause economic and social dislocation.

This cover story will examine the implications of the introduction of ever more advanced robots into the economy. It will start by briefly outlining some of the positive tasks that modern robots are capable of performing. Next it will examine why the introduction of robots has recently become a contentious issue in the financial media. In conclusion, it will consider the negative charges laid against such machines including their alleged role in increasing unemployment and widening inequality.

Robotic advances

To gauge the spread of robots it is necessary to try to define more precisely what they are. There is no universally accepted definition but this is the entry from the Oxford English Dictionary: “Chiefly Science Fiction. An intelligent artificial being typically made of metal and resembling in some way a human or other animal.”

Although this is a useful starting point it has several elements that are increasingly open to question. For a start, robots are gradually becoming more contemporary fact than science fiction. It also looks likely that with the passage of time many robots will not resemble humans physically.

Take the driverless or autonomous car being developed by Google. These already exist although they have not yet been produced commercially. Such cars, essentially a form of robot, are capable of driving safely on busy normal roads without any human intervention. This task clearly involves an enormous amount of sophisticated perception and decision-making.

The economic and social impact of such driving robots could be enormous. It is claimed that if they were introduced on a large scale they could increase road capacity hugely. For example, cars could drive much closer together if the element of human error in driving was removed. It would also be possible for people to work or relax while their cars drove them to work or elsewhere.

The driverless car also point to the importance of “intelligence” in the dictionary definition. A car that simply enhanced the capacity of human drivers would not be a robot. Indeed modern cars already use a vast amount of computer technology. Nor would a vehicle that was driven by remote control qualify as robotic. But a car that has agency, that is it is capable of making complex decisions on its own, is a different matter.

“There should be more time to develop human knowledge in the broad-est sense of the word”

A drone or Unmanned Aerial Vehicle could also be robotic while at the same time looking nothing like a human. A drone operated by a remote human operator or a simple program would not be classified as a robot. But one that could make its own complex decisions, that had a form of agency, would.

Drones can have many applications beyond policing, military surveillance and even assassination. These can include monitoring livestock for farmers, exploring for resources, carrying payloads and scientific research.

Nor do drones have to be the size of conventional aircraft. Once the need for a pilot is removed it is possible to create mini-drones that are the size of an insect. Such miniturisation further increases the range of tasks of which they are capable.

However, this is far from an exhaustive list of the new applications for robots. There is room for them in education, healthcare as well as traditional manufacturing tasks. For example, robots are being developed that can perform complex surgery while others can do the difficult task of diagnosing sick patients. Agricultural robots, or agbots, are also increasingly being introduced to perform tasks on farms including intelligent fruit picking and weeding.

Nor should robots be seen as limited to the developed countries. Foxconn, the Taiwanese contract manufacturer that manufactures products for Apple and many others, announced back in 2011 it was introducing one million industrial robots to replace workers at its production lines in mainland China. Naturally such robot workers have no need for sleep, medical care or holiday pay.

Meanwhile, in India, Maruti Suzuki, has announced it is to introduce 800 robots at its new plant in Mehsana, Gujarat. Its aim is to make the plant 95 per cent automated.

Clearly this brief survey contains a wide range of different applications and levels of development. Some types of robots are already operating while others will take a while to be introduced on any scale.

Nevertheless they illustrate the tremendous potential of such technology. Machines are no longer just about replacing human muscle with mechanical power. Nor are they simply designed for automating routine tasks. Intelligent machines will increasingly be able to augment humans by performing difficult tasks better or doing things that are impossible for people.

Why the debate?

The voluminous current discussion about robots in the financial media raises the question of why they are attracting such interest now. Perhaps the obvious answer is that the world is on the cusp of a technological breakthrough. However, there are good reasons not to accept this claim at face value.

The discussion of robots, both in fiction and reality, goes back a long way. Iconic robots in films include Gort in The Day the Earth Stood Still (1951), Robby the Robot in Forbidden Planet (1956), the first Star Wars (1977) and the first Terminator (1984).

In fact it can be traced even further than that. The term comes from a 1920 Czech play by Karel Čapek’s entitled R.U.R.: Rossum’s Universal Robots. The robots in that work were evidently assembled from artificially synthesised organic material. Not long after that, in 1927, the classic German science fiction film Metropolis, featured a robot as its central character. In 1942 Isaac Asimov, one of America’s greatest science fiction writers, compiled his three fundamental rules of robotics.

Even outside of science fiction there was talk of robot applications at least as far back as the 1970s. In 1979 a television advert for the Fiat Strada featured a high tech production line, an opera soundtrack and the sign-off line “handbuilt by robots”. No doubt the industrial “robots” were far less sophisticated than those of today but it does confirm the discussion is not itself new. Similarly London’s Victoria & Albert Museum had an exhibition on robots in 1984.

Several factors seem to lie behind the recent surge in interest beyond that of experts in the subject. First, it provides a way of attempting to explain the US’s slow economic recovery from the recession of 2007-08. For instance, unemployment peaked at 10.0 per cent in October 2009 according to official figures yet by January 2013 it had only fallen to 7.9 per cent (see graph one).

FS-11Mar-CS-graphs.jpg

Many commentators have previously attributed the slow recovery to failure to properly tackle a cyclical downturn or alternatively to an underlying crisis of innovation. But it has become increasingly popular to blame robots for increasingly replacing human labour.

The rise of robots is also being used for the related task of explaining rising economic inequality. According to this view business owners are increasingly taking the additional revenue provided by more automated production for themselves. Ordinary workers, on the other hand, are losing out. From this perspective, capitalists are using their monopoly power to benefit themselves at the expense of others.

This is a view which Paul Krugman, Nobel laureate and a columnist on the New York Times, has said has “echoes of old-fashioned Marxism” (“Rise of the robots”, 8 December 2012, New York Times blog). He has strong sympathy with this perspective despite describing himself as a liberal rather than a Marxist. In his view capitalist “robber barons” are harnessing robots to help them squeeze wages.

There is also another element to this discussion that is much less noticed or discussed. Many of the people whose jobs are apparently threatened by robots nowadays are professionals rather than manual workers. They are skilled medical practitioners, teachers or academics, and even journalists. The new generation of robots is not so much replacing muscle but supplanting brainpower, at least as it is traditionally harnessed.

As Walter Russell, one of America’s leading social commentators, has argued: “the same forces that attacked blue collar wages and employment in the last generation are now being felt by professionals; lawyers, journalists and professors now share some of the stress steelworkers have known for decades, and they don’t like it, and are very good at squealing.” (“Life After Blue: The Middle Class Will Beat The Seven Trolls”, The American Interest blog, 30 January 2013).

Blaming robots

The latter argument has not received nearly as much attention as it deserves but the contentions on high unemployment and widening inequality are widespread. It is becoming increasingly popular to blame these developments on the increased use of robots.

A 2011 book by two business professors from the Massachusetts Institute of Technology (MIT), Erik Brynjolfsson and Andrew McAfee, has become a key reference point for the debate. Race Against The Machine (RATM) claims to be upbeat about the new innovation and its impact on productivity. However, it also blames technological change for stagnating average (median) incomes in the US and the weakness of job creation. At the same time it points out that corporate profits have soared in the current recovery.

There are several clear points on which the book’s thesis can be challenged although there is substantial room for debate. The statistical data in particular is problematic.

In relation to job creation it should be clear that new technology can create as well as destroy jobs. The types of work done in contemporary America, for instance, are substantially different from those from, say, the 1970s, let alone the 1870s or the 1770s. There is no doubt that manufacturing, for example, employs far fewer it did a few decades ago even though total output is higher. Yet more people have found jobs in the service sector.

The authors of thecurrentmoment blog have pointed out that the labour participation rate – the percentage of the workforce that has a job – is broadly in line with its long-run average (see graph two) (“The persistence of work” 25 February 2013). There is no clear sign of a long-term reduction in the proportion of people employed.

Evidence of slowing productivity growth is more contested. It is easy enough to provide lists of exciting technological feats that robots are already performing or may be performing in the near future. But it should not be taken for granted that these are being implemented on a large scale or they are having a substantial impact on production. Over time the application of robots to the production process would be expected to show up in the productivity figures if it is having a significant economic effect (see graph three).

Robert Gordon, a leading expert on the subject, has made the opposite contention that the level of innovation is low. He recently wrote that the information technology’s “main impact on productivity has withered away in the past eight years”(Is US economic growth over? Faltering innovation confronts the six headwinds”, Centre for Economic Policy Research, Policy Insight No.63, September 2012).

Andrew Smithers, a prominent UK expert, concurs with Gordon’s broad conclusions. Despite a period of recovery in the 1990s he argues that the US’s labour productivity growth has steadily declined since the second world war (“US Productivity and Employment”, Smithers & Co, report No.409, 13 November 13 2012).

The authors of RATM, in contrast, argue that productivity growth is on an upward trend. This dispute can only be resolved through a detailed examination and even deconstruction of the productivity data that is beyond the scope of this feature.

As I have previously argued in Fund Strategy there is also a strong case that median income growth in the US, although poor, is not as bad as often assumed. America’s median income growth from 1979-2007, is normally estimated as about 3 per cent. But that is calculated on a particular set of assumptions. For example, it is pre-tax and before social transfers. But if it is calculated on an alternative, arguably more realistic, set of assumptions it could be nearly as high as 37 per cent (“Average incomes ‘did not stagnate’”, Fund Strategy, 30 April 2012. If the upper estimates are more accurate it would also call into question an element of the RATM theory.

The final element of a rebuttal to RATM’s claims is more conceptual but also more tricky. It contends that human ingenuity will find ways round the challenges posed by the introduction of robots. Humans invented the robots and they can also devise the best ways to harness them.

While this is true in the abstract it does not quite get to the nub of the RATM’s argument. The two authors claim to be optimists too. They argue that the challenge is to get humans to race with the machine rather than against it. In other words the claim that by improving “human capital” (skills) and organisational innovation it will be possible for humans to work better with machines. For the latter they give Apple’s App Store, Amazon marketplace and eBay as examples of new ways in which humans and machines can work together.

In the book’s conclusion they propose a 17-point “agenda for action” that includes measures in education, entrepreneurship, investment and regulation. Many of the proposals are widely accepted or even already being implemented. They include eliminating academic tenure, increasing the ratio of skilled migrants, teaching entrepreneurship and investing in infrastructure.

Despite the upbeat presentation this is a limited set of proposals. For example, much of the focus is within information technology itself – for instance in the greater use of Amazon or eBay – rather than the application of robotics to other areas such as the provision of energy or transport infrastructure. It is also largely focused on the distribution of goods or services rather than their production.

In their anxiety to manage the use of robots the authors devalue the meaning of education. For them it becomes synonymous with the improvement of “human capital”: practical skills that are suited to the modern labour market.

This is particularly ironic at a time when the application of technology should free humans from many practical chores. There should be more time to develop human knowledge in the broadest sense of the term; from basic science to visual arts to poetry.

The current discussion of robots is sorely lacking in imagination. Applying robots for the maximum benefit of humanity presents a formidable intellectual challenge. Harnessing human ingenuity is far trickier than simply asserting its importance.

_______

The end of work?

The debate about robots, particularly their purported negative effects, has close parallels with discussions of the end of work.

In itself the reduction in working hours or the elimination of work entirely can be seen as either welcome or as dangerous. It can be viewed as harnessing technology to set humanity free from toil or alternatively as machines, including robots, robbing us of work and even purpose.

Historically more optimistic social thinkers were generally positive about humans having to spend less time working and engaging in less physical toil. Nineteenth century economic writers such as David Ricardo and Karl Marx raised the spectre of machines replacing human labour.

Over the very long term the world has moved some way in the optimists’ direction. Robert Fogel, a Nobel prize-winning economist, has estimated that average working hours in the US fell from 182,100 per lifetime in 1880 to 122,400 in 1995 (The Escape from Hunger and Premature Death 1700-2100, p71).

Even the more recent figures indicate at improvement. According to the Bureau of Labor Statistics the average annual hours worked by employed person in the US fell from 1,829 in 1979 to 1,758 in 2011.

Of course such averages can mask divergent trends. In recent decades more women have entered the labour force and more people are employed in part-time work. The latter category includes many who are on poor pay overall. In any case the world is still a long way from the end of work despite the welcome long-term trend.

More pessimistic thinkers have usually taken a more negative view of such trends either raising the spectre of forced idleness or the decline of community. For example, Andre Gorz, a French green, wrote on the “end of the working class” in 1980 while Jeremy Rifkin, a high profile American green, had a book published on “the end of work” in 1995. More recently Ross Douthat, a New York Timescolumnist, wrote of the dangers of “a world without work” (New York Times, 23 February 2013).

However, the division between optimists and pessimists fails to capture the debate adequately. For example, Robert Skidelsky, a British economic historian, is among many green-tinged thinkers arguing that a reduction in working hours should be welcome. For him it is a question of sharing out the benefits of automation fairly between the rich and the rest of society (“The Rise of the Robots”. Project Syndicate, 19 February 2013).

The problem is that if the length of the working week falls but pay rates remain the same then incomes will also drop. Twenty hours’ work at a given rate will clearly pay much less than, say, 35 or 40 hours at the same rate.

The challenge is to decrease working hours while increasing pay at the same time. Such a goal can only be achieved through the imaginative application of technology in the productive process.