2013 TheFutureofEmploymentHowSuscept

Quotes

Abstract

The authors examine how susceptible jobs are to computerisation, by implementing a novel methodology to estimate the probability of computerisation for 702 detailed occupations, using a Gaussian process classifier. Based on these estimates, they examine expected impacts of future computerisation on US labour market outcomes, with the primary objective of analysing the number of jobs at risk and the relationship between an occupation’s probability of computerisation, wages and educational attainment. According to their estimates, about 47 per cent of total US employment is at risk. They further provide evidence that wages and educational attainment exhibit a strong negative relationship with an occupation’s probability of computerisation.

I. Introduction

In this paper, we address the question: how susceptible are jobs to computerisation? Doing so, we build on the existing literature in two ways. First, drawing upon recent advances in Machine Learning (ML) and Mobile Robotics (MR), we develop a novel methodology to categorise occupations according to their susceptibility to computerisation.[1] Second, we implement this methodology to estimate the probability of computerisation for 702 detailed occupations, and examine expected impacts of future computerisation on US labour market outcomes.

Our paper is motivated by John Maynard Keynes’s frequently cited prediction of widespread technological unemployment “due to our discovery of means of economising the use of labour outrunning the pace at which we can find new uses for labour” (Keynes, 1933, p. 3). Indeed, over the past decades, computers have substituted for a number of jobs, including the functions of bookkeepers, cashiers and telephone operators (Bresnahan, 1999; MGI, 2013). More recently, the poor performance of labour markets across advanced economies has intensified the debate about technological unemployment among economists. While there is ongoing disagreement about the driving forces behind the persistently high unemployment rates, a number of scholars have pointed at computercontrolled equipment as a possible explanation for recent jobless growth (see, for example, Brynjolfsson and McAfee, 2011).[2]

The impact of computerisation on labour market outcomes is well-established in the literature, documenting the decline of employment in routine intensive occupations – i.e. occupations mainly consisting of tasks following well-defined procedures that can easily be performed by sophisticated algorithms. For example, studies by Charles, et al. (2013) and Jaimovich and Siu (2012) emphasise that the ongoing decline in manufacturing employment and the disappearance of other routine jobs is causing the current low rates of employment.[3] In addition to the computerisation of routine manufacturing tasks, Autor and Dorn (2013) document a structural shift in the labour market, with workers reallocating their labour supply from middle-income manufacturing to low-income service occupations. Arguably, this is because the manual tasks of service occupations are less susceptible to computerisation, as they require a higher degree of flexibility and physical adaptability (Autor, et al., 2003; Goos and Manning, 2007; Autor and Dorn, 2013).

At the same time, with falling prices of computing, problem-solving skills are becoming relatively productive, explaining the substantial employment growth in occupations involving cognitive tasks where skilled labour has a comparative advantage, as well as the persistent increase in returns to education (Katz and Murphy, 1992; Acemoglu, 2002; Autor and Dorn, 2013). The title “Lousy and Lovely Jobs”, of recent work by Goos and Manning (2007), thus captures the essence of the current trend towards labour market polarization, with growing employment in high-income cognitive jobs and low-income manual occupations, accompanied by a hollowing-out of middle-income routine jobs.

According to Brynjolfsson and McAfee (2011), the pace of technological innovation is still increasing, with more sophisticated software technologies disrupting labour markets by making workers redundant. What is striking about the examples in their book is that computerisation is no longer confined to routine manufacturing tasks. The autonomous driverless cars, developed by Google, provide one example of how manual tasks in transport and logistics may soon be automated. In the section “In Domain After Domain, Computers Race Ahead”, they emphasise how fast moving these developments have been. Less than ten years ago, in the chapter “Why People Still Matter”, Levy and Murnane (2004) pointed at the difficulties of replicating human perception, asserting that driving in traffic is insusceptible to automation: “But executing a left turn against oncoming traffic involves so many factors that it is hard to imagine discovering the set of rules that can replicate a driver’s behaviour [. . . ]”. Six years later, in October 2010, Google announced that it had modified several Toyota Priuses to be fully autonomous (Brynjolfsson and McAfee, 2011).

To our knowledge, no study has yet quantified what recent technological progress is likely to mean for the future of employment. The present study intends to bridge this gap in the literature. Although there are indeed existing useful frameworks for examining the impact of computers on the occupational employment composition, they seem inadequate in explaining the impact of technological trends going beyond the computerisation of routine tasks. Seminal work by Autor, et al. (2003), for example, distinguishes between cognitive and manual tasks on the one hand, and routine and non-routine tasks on the other. While the computer substitution for both cognitive and manual routine tasks is evident, non-routine tasks involve everything from legal writing, truck driving and medical diagnoses, to persuading and selling. In the present study, we will argue that legal writing and truck driving will soon be automated, while persuading, for instance, will not. Drawing upon recent developments in Engineering Sciences, and in particular advances in the fields of ML, including Data Mining, Machine Vision, Computational Statistics and other sub-fields of Artificial Intelligence, as well as MR, we derive additional dimensions required to understand the susceptibility of jobs to computerisation. Needless to say, a number of factors are driving decisions to automate and we cannot capture these in full. Rather we aim, from a technological capabilities point of view, to determine which problems engineers need to solve for specific occupations to be automated. By highlighting these problems, their difficulty and to which occupations they relate, we categorise jobs according to their susceptibility to computerisation. The characteristics of these problems were matched to different occupational characteristics, using O*NET data, allowing us to examine the future direction of technological change in terms of its impact on the occupational composition of the labour market, but also the number of jobs at risk should these technologies materialise.

The present study relates to two literatures. First, our analysis builds on the labour economics literature on the task content of employment (Autor, et al., 2003; Goos and Manning, 2007; Autor and Dorn, 2013). Based on defined premises about what computers do, this literature examines the historical impact of computerisation on the occupational composition of the labour market. However, the scope of what computers do has recently expanded, and will inevitably continue to do so (Brynjolfsson and McAfee, 2011; MGI, 2013). Drawing upon recent progress in ML, we expand the premises about the tasks computers are and will be suited to accomplish. Doing so, we build on the task content literature in a forward-looking manner. Furthermore, whereas this literature has largely focused on task measures from the Dictionary of Occupational Titles (DOT), last revised in 1991, we rely on the 2010 version of the DOT successor O*NET – an online service developed for the US Department of Labor.[4] Accordingly, O*NET has the advantage of providing more recent information on occupational work activities.

Second, our study relates to the literature examining the offshoring of information- based tasks to foreign worksites (Jensen and Kletzer, 2005; Blinder, 2009; Jensen and Kletzer, 2010; Oldenski, 2012; Blinder and Krueger, 2013). This literature consists of different methodologies to rank and categorise occupations according to their susceptibility to offshoring. For example, using O*NET data on the nature of work done in different occupations, Blinder (2009) estimates that 22 to 29 percent of US jobs are or will be offshorable in the next decade or two. These estimates are based on two defining characteristics of jobs that cannot be offshored: (a) the job must be performed at a specific work location; and (b) the job requires face-to-face personal communication. Naturally, the characteristics of occupations that can be offshored are different from the characteristics of occupations that can be automated. For example, the work of cashiers, which has largely been substituted by self- service technology, must be performed at specific work location and requires face-to-face contact. The extent of computerisation is therefore likely to go beyond that of offshoring. Hence, while the implementation of our methodology is similar to that of Blinder (2009), we rely on different occupational characteristics.

The remainder of this paper is structured as follows. In Section II, we review the literature on the historical relationship between technological progress and employment. Section III describes recent and expected future technological developments. In Section IV, we describe our methodology, and in Section V, we examine the expected impact of these technological developments on labour market outcomes. Finally, in Section VI, we derive some conclusions.

II. A History of Technological Revolutions and Employment

The concern over technological unemployment is hardly a recent phenomenon. Throughout history, the process of creative destruction, following technological inventions, has created enormous wealth, but also undesired disruptions. As stressed by Schumpeter (1962), it was not the lack of inventive ideas that set the boundaries for economic development, but rather powerful social and economic interests promoting the technological status quo. This is nicely illustrated by the example of William Lee, inventing the stocking frame knitting machine in 1589, hoping that it would relieve workers of hand-knitting. Seeking patent protection for his invention, he travelled to London where he had rented a building for his machine to be viewed by Queen Elizabeth I. To his disappointment, the Queen was more concerned with the employment impact of his invention and refused to grant him a patent, claiming that: “Thou aimest high, Master Lee. Consider thou what the invention could do to my poor subjects. It would assuredly bring to them ruin by depriving them of employment, thus making them beggars” (cited in Acemoglu and Robinson, 2012, p. 182f). Most likely the Queen’s concern was a manifestation of the hosiers’ guilds fear that the invention would make the skills of its artisan members obsolete.[5] The guilds’ opposition was indeed so intense that William Lee had to leave Britain.

That guilds systematically tried to weaken market forces as aggregators to maintain the technological status quo is persuasively argued by Kellenbenz (1974, p. 243), stating that “guilds defended the interests of their members against outsiders, and these included the inventors who, with their new equipment and techniques, threatened to disturb their members’ economic status.”[6] As pointed out by Mokyr (1998, p. 11): “Unless all individuals accept the “verdict” of the market outcome, the decision whether to adopt an innovation is likely to be resisted by losers through non-market mechanism and political activism.” Workers can thus be expected to resist new technologies, insofar that they make their skills obsolete and irreversibly reduce their expected earnings. The balance between job conservation and technological progress therefore, to a large extent, reflects the balance of power in society, and how gains from technological progress are being distributed.

The British Industrial Revolution illustrates this point vividly. While still widely present on the Continent, the craft guild in Britain had, by the time of the Glorious Revolution of 1688, declined and lost most of its political clout (Nef, 1957, pp. 26 and 32). With Parliamentary supremacy established over the Crown, legislation was passed in 1769 making the destruction of machinery punishable by death (Mokyr, 1990, p. 257). To be sure, there was still resistance to mechanisation. The “Luddite” riots between 1811 and 1816 were partly a manifestation of the fear of technological change among workers as Parliament revoked a 1551 law prohibiting the use of gig mills in the wool-finishing trade. The British government however took an increasingly stern view on groups attempting to halt technological progress and deployed 12,000 men against the rioters (Mantoux, 2006, p. 403-8). The sentiment of the government towards the destruction of machinery was explained by a resolution passed after the Lancashire riots of 1779, stating that: “The sole cause of great riots was the new machines employed in cotton manufacture; the country notwithstanding has greatly benefited from their erection [and] destroying them in this country would only be the means of transferring them to another [. . .] to the detriment of the trade of Britain” (cited in Mantoux, 2006, p. 403).

There are at least two possible explanations for the shift in attitudes towards technological progress. First, after Parliamentary supremacy was established over the Crown, the property owning classes became politically dominant in Britain (North andWeingast, 1989). Because the diffusion of various manufacturing technologies did not impose a risk to the value of their assets, and some property owners stood to benefit from the export of manufactured goods, the artisans simply did not have the political power to repress them. Second, inventors, consumers and unskilled factory workers largely benefited from mechanisation (Mokyr, 1990, p. 256 and 258). It has even been argued that, despite the employment concerns over mechanisation, unskilled workers have been the greatest beneficiaries of the Industrial Revolution (Clark, 2008).[7] While there is contradictory evidence suggesting that capital owners initially accumulated a growing share of national income (Allen, 2009a), there is equally evidence of growing real wages (Lindert and Williamson, 1983; Feinstein, 1998). This implies that although manufacturing technologies made the skills of artisans obsolete, gains from technological progress were distributed in a manner that gradually benefited a growing share of the labour force.[8]

An important feature of nineteenth century manufacturing technologies is that they were largely “deskilling” – i.e. they substituted for skills through the simplification of tasks (Braverman, 1974; Hounshell, 1985; James and Skinner, 1985; Goldin and Katz, 1998). The deskilling process occurred as the factory system began to displace the artisan shop, and it picked up pace as production increasingly mechanized with the adoption of steam power (Goldin and Sokoloff, 1982; Atack, et al., 2008a). Work that had previously been performed by artisans was now decomposed into smaller, highly specialised, sequences, requiring less skill, but more workers, to perform.[9] Some innovations were even designed to be deskilling. For example, Eli Whitney, a pioneer of interchangeable parts, described the objective of this technology as “to substitute correct and effective operations of machinery for the skill of the artist which is acquired only by long practice and experience; a species of skill which is not possessed in this country to any considerable extent” (Habakkuk, 1962, p. 22).

Together with developments in continuous-flow production, enabling workers to be stationary while different tasks weremoved to them, it was identical interchangeable parts that allowed complex products to be assembled from mass produced individual components by using highly specialised machine tools to a sequence of operations.[10] Yet while the first assembly-line was documented in 1804, it was not until the late nineteenth century that continuous-flow processes started to be adopted on a larger scale, which enabled corporations such as the Ford Motor Company to manufacture the T-Ford at a sufficiently low price for it to become the people’s vehicle (Mokyr, 1990, p. 137). Crucially, the new assembly line introduced by Ford in 1913 was specifically designed for machinery to be operated by unskilled workers (Hounshell, 1985, p. 239). Furthermore, what had previously been a one-man job was turned into a 29-man worker operation, reducing the overall work time by 34 percent (Bright, 1958). The example of the Ford Motor Company thus underlines the general pattern observed in the nineteenth century, with physical capital providing a relative complement to unskilled labour, while substituting for relatively skilled artisans (James and Skinner, 1985; Louis and Paterson, 1986; Brown and Philips, 1986; Atack, et al., 2004).[11] Hence, as pointed out by Acemoglu (2002, p. 7): “the idea that technological advances favor more skilled workers is a twentieth century phenomenon.” The conventional wisdom among economic historians, in other words, suggests a discontinuity between the nineteenth and twentieth century in the impact of capital deepening on the relative demand for skilled labour.

The modern pattern of capital-skill complementarity gradually emerged in the late nineteenth century, as manufacturing production shifted to increasingly mechanised assembly lines. This shift can be traced to the switch to electricity from steam and water-power which, in combination with continuous-process and batch production methods, reduced the demand for unskilled manual workers in many hauling, conveying, and assembly tasks, but increased the demand for skills (Goldin and Katz, 1998). In short, while factory assembly lines, with their extreme division of labour, had required vast quantities of human operatives, electrification allowed many stages of the production process to be automated, which in turn increased the demand for relatively skilled blue-collar production workers to operate the machinery. In addition, electrification contributed to a growing share of white-collar nonproduction workers (Goldin and Katz, 1998). Over the course of the nineteenth century, establishments became larger in size as steam and water power technologies improved, allowing them to adopt powered machinery to realize productivity gains through the combination of enhanced division of labour and higher capital intensity (Atack, et al., 2008a). Furthermore, the transport revolution lowered costs of shipping goods domestically and internationally as infrastructure spread and improved (Atack, et al., 2008b). The market for artisan goods early on had largely been confined to the immediate surrounding area because transport costs were high relative to the value of the goods produced. With the transport revolution, however, market size expanded, thereby eroding local monopoly power, which in turn increased competition and compelled firms to raise productivity through mechanisation. As establishments became larger and served geographically expended markets, managerial tasks increased in number and complexity, requiring more managerial and clerking employees (Chandler, 1977). This pattern was, by the turn of the twentieth century, reinforced by electrification, which not only contributed to a growing share of relatively skilled blue-collar labour, but also increased the demand for white-collar workers (Goldin and Katz, 1998), who tended to have higher educational attainment (Allen, 2001).[12]

Since electrification, the story of the twentieth century has been the race between education and technology (Goldin and Katz, 2009). The US high school movement coincided with the first industrial revolution of the office (Goldin and Katz, 1995). While the typewriter was invented in the 1860s, it was not introduced in the office until the early twentieth century, when it entered a wave of mechanisation, with dictaphones, calculators, mimeo machines, address machines, and the predecessor of the computer – the keypunch (Beniger, 1986; Cortada, 2000). Importantly, these office machines reduced the cost of information processing tasks and increased the demand for the complementary factor – i.e. educated office workers. Yet the increased supply of educated office workers, following the high school movement, was associated with a sharp decline in the wage premium of clerking occupations relative to production workers (Goldin and Katz, 1995). This was, however, not the result of deskilling technological change. Clerking workers were indeed relatively educated. Rather, it was the result of the supply of educated workers outpacing the demand for their skills, leading educational wage differentials to compress.

While educational wage differentials in the US narrowed from 1915 to 1980 (Goldin and Katz, 2009), both educational wage differentials and overall wage inequality have increased sharply since the 1980s in a number of countries (Krueger, 1993; Murphy, et al., 1998; Atkinson, 2008; Goldin and Katz, 2009). Although there are clearly several variables at work, consensus is broad that this can be ascribed to an acceleration in capital-skill complementarity, driven by the adoption of computers and information technology (Krueger, 1993; Autor, et al., 1998; Bresnahan, et al., 2002). What is commonly referred to as the Computer Revolution began with the first commercial uses of computers around 1960 and continued through the development of the Internet and e-commerce in the 1990s. As the cost per computation declined at an annual average of 37 percent between 1945 and 1980 (Nordhaus, 2007), telephone operators were made redundant, the first industrial robot was introduced by General Motors in the 1960s, and in the 1970s airline reservations systems led the way in selfservice technology (Gordon, 2012). During the 1980s and 1990s, computing costs declined even more rapidly, on average by 64 percent per year, accompanied by a surge in computational power (Nordhaus, 2007).[13] At the same time, bar-code scanners and cash machines were spreading across the retail and financial industries, and the first personal computers were introduced in the early 1980s, with their word processing and spreadsheet functions eliminating copy typist occupations and allowing repetitive calculations to be automated (Gordon, 2012). This substitution for labour marks a further important reversal. The early twentieth century office machines increased the demand for clerking workers (Chandler, 1977; Goldin and Katz, 1995). In a similar manner, computerisation augments demand for such tasks, but it also permits them to be automated (Autor, et al., 2003).

The Computer Revolution can go some way in explaining the growing wage inequality of the past decades. For example, Krueger (1993) finds that workers using a computer earn roughly earn 10 to 15 percent more than others, but also that computer use accounts for a substantial share of the increase in the rate of return to education. In addition, more recent studies find that computers have caused a shift in the occupational structure of the labour market. Autor and Dorn (2013), for example, show that as computerisation erodes wages for labour performing routine tasks, workers will reallocate their labour supply to relatively low-skill service occupations. More specifically, between 1980 and 2005, the share of US labour hours in service occupations grew by 30 percent after having been flat or declining in the three prior decades. Furthermore, net changes in US employment were U-shaped in skill level, meaning that the lowest and highest job-skill quartile expanded sharply with relative employment declines in the middle of the distribution.

The expansion in high-skill employment can be explained by the falling price of carrying out routine tasks by means of computers, which complements more abstract and creative services. Seen from a production function perspective, an outward shift in the supply of routine informational inputs increases the marginal productivity of workers they are demanded by. For example, text and data mining has improved the quality of legal research as constant access to market information has improved the efficiency of managerial decision-making – i.e. tasks performed by skilled workers at the higher end of the income distribution. The result has been an increasingly polarised labour market, with growing employment in high-income cognitive jobs and low-income manual occupations, accompanied by a hollowing-out of middle-income routine jobs. This is a pattern that is not unique to the US and equally applies to a number of developed economies (Goos, et al., 2009).[14]

How technological progress in the twenty-first century will impact on labour market outcomes remains to be seen. Throughout history, technological progress has vastly shifted the composition of employment, from agriculture and the artisan shop, to manufacturing and clerking, to service and management occupations. Yet the concern over technological unemployment has proven to be exaggerated. The obvious reason why this concern has not materialised relates to Ricardo’s famous chapter on machinery, which suggests that laboursaving technology reduces the demand for undifferentiated labour, thus leading to technological unemployment (Ricardo, 1819). As economists have long understood, however, an invention that replaces workers by machines will have effects on all product and factor markets. An increase in the efficiency of production which reduces the price of one good, will increase real income and thus increase demand for other goods. Hence, in short, technological progress has two competing effects on employment (Aghion and Howitt, 1994). First, as technology substitutes for labour, there is a destruction effect, requiring workers to reallocate their labour supply; and second, there is the capitalisation effect, as more companies enter industries where productivity is relatively high, leading employment in those industries to expand.

Although the capitalisation effect has been predominant historically, our discovery of means of economising the use of labour can outrun the pace at which we can find new uses for labour, as Keynes (1933) pointed out. The reason why human labour has prevailed relates to its ability to adopt and acquire new skills by means of education (Goldin and Katz, 2009). Yet as computerisation enters more cognitive domains this will become increasingly challenging (Brynjolfsson and McAfee, 2011). Recent empirical findings are therefore particularly concerning. For example, Beaudry, et al. (2013) document a decline in the demand for skill over the past decade, even as the supply of workers with higher education has continued to grow. They show that high-skilled workers have moved down the occupational ladder, taking on jobs traditionally performed by low-skilled workers, pushing low-skilled workers even further down the occupational ladder and, to some extent, even out of the labour force. This raises questions about: (a) the ability of human labour to win the race against technology by means of education; and (b) the potential extent of technological unemployment, as an increasing pace of technological progress will cause higher job turnover, resulting in a higher natural rate of unemployment (Lucas and Prescott, 1974; Davis and Haltiwanger, 1992; Pissarides, 2000). While the present study is limited to examining the destruction effect of technology, it nevertheless provides a useful indication of the job growth required to counterbalance the jobs at risk over the next decades.

III. The Technological Revolutions of the Twenty-First Century

The secular price decline in the real cost of computing has created vast economic incentives for employers to substitute labour for computer capital.15 Yet the tasks computers are able to perform ultimately depend upon the ability of a programmer to write a set of procedures or rules that appropriately direct the technology in each possible contingency. Computers will therefore be relatively productive to human labour when a problem can be specified – in the sense that the criteria for success are quantifiable and can readily be evaluated (Acemoglu and Autor, 2011). The extent of job computerisation will thus be determined by technological advances that allow engineering problems to be sufficiently specified, which sets the boundaries for the scope of computerisation. In this section, we examine the extent of tasks computer-controlled equipment can be expected to perform over the next decades. Doing so, we focus on advances in fields related to Machine Learning (ML), including Data Mining, Machine Vision, Computational Statistics and other sub-fields of Artificial Intelligence (AI), in which efforts are explicitly dedicated to the development of algorithms that allow cognitive tasks to be automated. In addition, we examine the application of ML technologies in Mobile Robotics (MR), and thus the extent of computerisation in manual tasks.

Our analysis builds on the task categorisation of Autor, et al. (2003), which distinguishes between workplace tasks using a two-by-two matrix, with routine versus non-routine tasks on one axis, and manual versus cognitive tasks on the other. In short, routine tasks are defined as tasks that follow explicit rules that can be accomplished by machines, while non-routine tasks are not sufficiently well understood to be specified in computer code. Each of these task categories can, in turn, be of either manual or cognitive nature – i.e. they relate to physical labour or knowledge work. Historically, computerisation has largely been confined to manual and cognitive routine tasks involving explicit rulebased activities (Autor and Dorn, 2013; Goos, et al., 2009). Following recent technological advances, however, computerisation is now spreading to domains commonly defined as non-routine. The rapid pace at which tasks that were defined as non-routine only a decade ago have now become computerisable is illustrated by Autor, et al. (2003), asserting that: “Navigating a car through city traffic or deciphering the scrawled handwriting on a personal check – minor undertakings for most adults – are not routine tasks by our definition.” Today, the problems of navigating a car and deciphering handwriting are sufficiently well understood that many related tasks can be specified in computer code and automated (Veres, et al., 2011; Plötz and Fink, 2009).

Recent technological breakthroughs are, in large part, due to efforts to turn non-routine tasks into well-defined problems. Defining such problems is helped by the provision of relevant data: this is highlighted in the case of handwriting recognition by Plötz and Fink (2009). The success of an algorithm for handwriting recognition is difficult to quantify without data to test on – in particular, determining whether an algorithm performs well for different styles of writing requires data containing a variety of such styles. That is, data is required to specify the many contingencies a technology must manage in order to form an adequate substitute for human labour. With data, objective and quantifiable measures of the success of an algorithm can be produced, which aid the continual improvement of its performance relative to humans.

As such, technological progress has been aided by the recent production of increasingly large and complex datasets, known as big data.[15] For instance, with a growing corpus of human-translated digitalised text, the success of a machine translator can now be judged by its accuracy in reproducing observed translations. Data from United Nations documents, which are translated by human experts into six languages, allow Google Translate to monitor and improve the performance of different machine translation algorithms (Tanner, 2007).

Further, ML algorithms can discover unexpected similarities between old and new data, aiding the computerisation of tasks for which big data has newly become available. As a result, computerisation is no longer confined to routine tasks that can be written as rule-based software queries, but is spreading to every non-routine task where big data becomes available (Brynjolfsson and McAfee, 2011). In this section, we examine the extent of future computerisation beyond routine tasks.

III.A. Computerisation in non-routine cognitive tasks

With the availability of big data, a wide range of non-routine cognitive tasks are becoming computerisable. That is, further to the general improvement in technological progress due to big data, algorithms for big data are rapidly entering domains reliant upon storing or accessing information. The use of big data is afforded by one of the chief comparative advantages of computers relative to human labor: scalability. Little evidence is required to demonstrate that, in performing the task of laborious computation, networks of machines scale better than human labour (Campbell-Kelly, 2009). As such, computers can better manage the large calculations required in using large datasets. ML algorithms running on computers are now, in many cases, better able to detect patterns in big data than humans.

Computerisation of cognitive tasks is also aided by another core comparative advantage of algorithms: their absence of some human biases. An algorithm can be designed to ruthlessly satisfy the small range of tasks it is given. Humans, in contrast, must fulfill a range of tasks unrelated to their occupation, such as sleeping, necessitating occasional sacrifices in their occupational performance (Kahneman, et al., 1982). The additional constraints under which humans must operate manifest themselves as biases. Consider an example of human bias: Danziger, et al. (2011) demonstrate that experienced Israeli judges are substantially more generous in their rulings following a lunch break. It can thus be argued that many roles involving decision-making will benefit from impartial algorithmic solutions.

Fraud detection is a task that requires both impartial decision making and the ability to detect trends in big data. As such, this task is now almost completely automated (Phua, et al., 2010). In a similar manner, the comparative advantages of computers are likely to change the nature of work across a wide range of industries and occupations.

In health care, diagnostics tasks are already being computerised. Oncologists at Memorial Sloan-Kettering Cancer Center are, for example, using IBM’s Watson computer to provide chronic care and cancer treatment diagnostics. Knowledge from 600,000 medical evidence reports, 1.5 million patient records and clinical trials, and two million pages of text from medical journals, are used for benchmarking and pattern recognition purposes. This allows the computer to compare each patient’s individual symptoms, genetics, family and medication history, etc., to diagnose and develop a treatment plan with the highest probability of success (Cohn, 2013).

In addition, computerisation is entering the domains of legal and financial services. Sophisticated algorithms are gradually taking on a number of tasks performed by paralegals, contract and patent lawyers (Markoff, 2011). More specifically, law firms now rely on computers that can scan thousands of legal briefs and precedents to assist in pre-trial research. A frequently cited example is Symantec’s Clearwell system, which uses language analysis to identify general concepts in documents, can present the results graphically, and proved capable of analysing and sorting more than 570,000 documents in two days (Markoff, 2011).

Furthermore, the improvement of sensing technology has made sensor data one of the most prominent sources of big data (Ackerman and Guizzo, 2011). Sensor data is often coupled with new ML fault- and anomaly-detection algorithms to render many tasks computerisable. A broad class of examples can be found in condition monitoring and novelty detection, with technology substituting for closed-circuit TV (CCTV) operators, workers examining equipment defects, and clinical staff responsible for monitoring the state of patients in intensive care. Here, the fact that computers lack human biases is of great value: algorithms are free of irrational bias, and their vigilance need not be interrupted by rest breaks or lapses of concentration. Following the declining costs of digital sensing and actuation, ML approaches have successfully addressed condition monitoring applications ranging from batteries (Saha, et al., 2007), to aircraft engines (King, et al., 2009), water quality (Osborne, et al., 2012) and intensive care units (ICUs) (Clifford and Clifton, 2012; Clifton, et al., 2012). Sensors can equally be placed on trucks and pallets to improve companies’ supply chain management, and used to measure the moisture in a field of crops to track the flow of water through utility pipes. This allows for automatic meter reading, eliminating the need for personnel to gather such information. For example, the cities of Doha, São Paulo, and Beijing use sensors on pipes, pumps, and other water infrastructure to monitor conditions and manage water loss, reducing leaks by 40 to 50 percent. In the near future, it will be possible to place inexpensive sensors on light poles, sidewalks, and other public property to capture sound and images, likely reducing the number of workers in law enforcement (MGI, 2013).

Advances in user interfaces also enable computers to respond directly to a wider range of human requests, thus augmenting the work of highly skilled labour, while allowing some types of jobs to become fully automated. For example, Apple’s Siri and Google Now rely on natural user interfaces to recognise spoken words, interpret their meanings, and act on them accordingly. Moreover, a company called SmartAction now provides call computerisation solutions that use ML technology and advanced speech recognition to improve upon conventional interactive voice response systems, realising cost savings of 60 to 80 percent over an outsourced call center consisting of human labour (CAA, 2012). Even education, one of the most labour intensive sectors, will most likely be significantly impacted by improved user interfaces and algorithms building upon big data. The recent growth in MOOCs (Massive Open Online Courses) has begun to generate large datasets detailing how students interact on forums, their diligence in completing assignments and viewing lectures, and their ultimate grades (Simonite, 2013; Breslow, et al., 2013). Such information, together with improved user interfaces, will allow for ML algorithms that serve as interactive tutors, with teaching and assessment strategies statistically calibrated to match individual student needs (Woolf, 2010). Big data analysis will also allow for more effective predictions of student performance, and for their suitability for post-graduation occupations. These technologies can equally be implemented in recruitment, most likely resulting in the streamlining of human resource (HR) departments.

Occupations that require subtle judgement are also increasingly susceptible to computerisation. To many such tasks, the unbiased decision making of an algorithm represents a comparative advantage over human operators. In the most challenging or critical applications, as in ICUs, algorithmic recommendations may serve as inputs to human operators; in other circumstances, algorithms will themselves be responsible for appropriate decision-making. In the financial sector, such automated decision-making has played a role for quite some time. AI algorithms are able to process a greater number of financial announcements, press releases, and other information than any human trader, and then act faster upon them (Mims, 2010). Services like Future Advisor similarly use AI to offer personalised financial advice at larger scale and lower cost. Even the work of software engineers may soon largely be computerisable. For example, advances in ML allow a programmer to leave complex parameter and design choices to be appropriately optimised by an algorithm (Hoos, 2012). Algorithms can further automatically detect bugs in software (Hangal and Lam, 2002; Livshits and Zimmermann, 2005; Kim, et al., 2008), with a reliability that humans are unlikely to match. Big databases of code also offer the eventual prospect of algorithms that learn how to write programs to satisfy specifications provided by a human. Such an approach is likely to eventually improve upon human programmers, in the same way that human-written compilers eventually proved inferior to automatically optimised compilers. An algorithm can better keep the whole of a program in working memory, and is not constrained to human-intelligible code, allowing for holistic solutions that might never occur to a human. Such algorithmic improvements over human judgement are likely to become increasingly common.

Although the extent of these developments remains to be seen, estimates by MGI (2013) suggests that sophisticated algorithms could substitute for approximately 140 million full-time knowledge workers worldwide. Hence, while technological progress throughout economic history has largely been confined to the mechanisation of manual tasks, requiring physical labour, technological progress in the twenty-first century can be expected to contribute to a wide range of cognitive tasks, which, until now, have largely remained a human domain. Of course, many occupations being affected by these developments are still far from fully computerisable, meaning that the computerisation of some tasks will simply free-up time for human labour to perform other tasks. Nonetheless, the trend is clear: computers increasingly challenge human labour in a wide range of cognitive tasks (Brynjolfsson and McAfee, 2011).

III.B. Computerisation in Non-Routine Manual Tasks

Mobile robotics provides a means of directly leveraging ML technologies to aid the computerisation of a growing scope of manual tasks. The continued technological development of robotic hardware is having notable impact upon employment: over the past decades, industrial robots have taken on the routine tasks of most operatives in manufacturing. Now, however, more advanced robots are gaining enhanced sensors and manipulators, allowing them to perform non-routine manual tasks. For example, General Electric has recently developed robots to climb and maintain wind turbines, and more flexible surgical robots with a greater range of motion will soon perform more types of operations (Robotics-VO, 2013). In a similar manner, the computerisation of logistics is being aided by the increasing cost-effectiveness of highly instrumented and computerised cars. Mass-production vehicles, such as the Nissan LEAF, contain on-board computers and advanced telecommunication equipment that render the car a potentially fly-by-wire robot.[16] Advances in sensor technology mean that vehicles are likely to soon be augmented with even more advanced suites of sensors. These will permit an algorithmic vehicle controller to monitor its environment to a degree that exceeds the capabilities of any human driver: they have the ability to simultaneously look both forwards and backwards, can natively integrate camera, GPS and LIDAR data, and are not subject to distraction. Algorithms are thus potentially safer and more effective drivers than humans.

The big data provided by these improved sensors are offering solutions to many of the engineering problems that had hindered robotic development in the past. In particular, the creation of detailed three dimensional maps of road networks has enabled autonomous vehicle navigation; most notably illustrated by Google’s use of large, specialised datasets collected by its driverless cars (Guizzo, 2011). It is now completely feasible to store representations of the entire road network on-board a car, dramatically simplifying the navigation problem. Algorithms that could perform navigation throughout the changing seasons, particularly after snowfall, have been viewed as a substantial challenge. However, the big data approach can answer this by storing records from the last time snow fell, against which the vehicle’s current environment can be compared (Churchill and Newman, 2012). ML approaches have also been developed to identify unprecedented changes to a particular piece of the road network, such as roadworks (Mathibela, et al., 2012). This emerging technology will affect a variety of logistics jobs. Agricultural vehicles, forklifts and cargo-handling vehicles are imminently automatable, and hospitals are already employing autonomous robots to transport food, prescriptions and samples (Bloss, 2011). The computerisation of mining vehicles is further being pursued by companies such as Rio Tinto, seeking to replace labour in Australian mine-sites.[17]

With improved sensors, robots are capable of producing goods with higher quality and reliability than human labour. For example, El Dulze, a Spanish food processor, now uses robotics to pick up heads of lettuce from a conveyor belt, rejecting heads that do not comply with company standards. This is achieved by measuring their density and replacing them on the belt (IFR, 2012a). Advanced sensors further allow robots to recognise patterns. Baxter, a 22,000 USD general-purpose robot, provides a well-known example. The robot features an LCD display screen displaying a pair of eyes that take on different expressions depending on the situation. When the robot is first installed or needs to learn a new pattern, no programming is required. A human worker simply guides the robot arms through the motions that will be needed for the task. Baxter then memorises these patterns and can communicate that it has understood its new instructions. While the physical flexibility of Baxter is limited to performing simple operations such as picking up objects and moving them, different standard attachments can be installed on its arms, allowing Baxter to perform a relatively broad scope of manual tasks at low cost (MGI, 2013).

Technological advances are contributing to declining costs in robotics. Over the past decades, robot prices have fallen about 10 percent annually and are expected to decline at an even faster pace in the near future (MGI, 2013). Industrial robots, with features enabled by machine vision and high-precision dexterity, which typically cost 100,000 to 150,000 USD, will be available for 50,000 to 75,000 USD in the next decade, with higher levels of intelligence and additional capabilities (IFR, 2012b). Declining robot prices will inevitably place them within reach of more users. For example, in China, employers are increasingly incentivised to substitute robots for labour, as wages and living standards are rising – Foxconn, a Chinese contract manufacturer that employs 1.2 million workers, is now investing in robots to assemble products such as the Apple iPhone (Markoff, 2012). According to the International Federation of Robotics, robot sales in China grew by more than 50 percent in 2011 and are expected to increase further. Globally, industrial robot sales reached a record 166,000 units in 2011, a 40 percent year-on-year increase (IFR, 2012b). Most likely, there will be even faster growth ahead as low-priced general-purpose models, such as Baxter, are adopted in simple manufacturing and service work.

in manufacturing, packing, construction, maintenance, and agriculture. In addition, robots are already performing many simple service tasks such as vacuuming, mopping, lawn mowing, and gutter cleaning – the market for personal and household service robots is growing by about 20 percent annually (MGI, 2013). Meanwhile, commercial service robots are now able to perform more complex tasks in food preparation, health care, commercial cleaning, and elderly care (Robotics-VO, 2013). As robot costs decline and technological capabilities expand, robots can thus be expected to gradually substitute for labour in a wide range of low-wage service occupations, where most US job growth has occurred over the past decades (Autor and Dorn, 2013). This means that many low-wage manual jobs that have been previously protected from computerisation could diminish over time.

The task model of Autor, et al. (2003) has delivered intuitive and accurate predictions in that: (a) computers are more substitutable for human labour in routine relative to non-routine tasks; and (b) a greater intensity of routine inputs increases the marginal productivity of non-routine inputs. Accordingly, computers have served as a substitute for labour for many routine tasks, while exhibiting strong complementarities with labour performing cognitive non-routine tasks.[18] Yet the premises about what computers do have recently expanded. Computer capital can now equally substitute for a wide range of tasks commonly defined as non-routine (Brynjolfsson and McAfee, 2011), meaning that the task model will not hold in predicting the impact of computerisation on the task content of employment in the twenty-first century. While focusing on the substitution effects of recent technological progress, we build on the task model by deriving several factors that we expect will determine the extent of computerisation in non-routine tasks.

The task model assumes for tractability an aggregate, constant-returns-toscale Cobb-Douglas production function of the form, :$(1) Q = (L_S+C)^{1-\beta} L^\beta_{NS}, \beta \in [0, 1],$ where $L_{S}$ and $L_{NS}$ are susceptible and non-susceptible labor inputs and C is computer capital. Computer capital is supplied perfectly elastically at market price per efficiency unit, where the market price is falling exogenously with time due to technological progress. It further assumes income-maximizing workers, with heterogeneous productivity endowments in both susceptible and non-susceptible tasks. Their task supply will respond elastically to relative wage levels, meaning that workers will reallocate their labour supply according to their comparative advantage as in Roy (1951). With expanding computational capabilities, resulting from technological advances, and a falling market price of computing, workers in susceptible tasks will thus reallocate to nonsusceptible tasks.

The above described simple model differs from the task model of Autor, et al. (2003), in that LNS is not confined to routine labour inputs. This is because recent developments in ML and MR, building upon big data, allow for pattern recognition, and thus enable computer capital to rapidly substitute for labour across a wide range of non-routine tasks. Yet some inhibiting engineering bottlenecks to computerisation persist. Beyond these bottlenecks, however, we argue that it is largely already technologically possible to automate almost any task, provided that sufficient amounts of data are gathered for pattern recognition. Our model thus predicts that the pace at which these bottlenecks can be overcome will determine the extent of computerisation in the twenty-first century.

Hence, in short, while the task model predicts that computers for labour substitution will be confined to routine tasks, our model predicts that computerisation can be extended to any non-routine task that is not subject to any engineering bottlenecks to computerisation. These bottlenecks thus set the boundaries for the computerisation of non-routine tasks. Drawing upon the ML and MR literature, and a workshop held at the Oxford University Engineering Sciences Department, we identify several engineering bottlenecks, corresponding to three task categories. According to these findings, non-susceptible labor inputs can be described as, :$(2) LNS = \Sum^n_{i=1} (L_{PM,i} + L_{C,i} + L_{SI,i}?$ where LPM, LC and LSI are labour inputs into perception and manipulation tasks, creative intelligence tasks, and and social intelligence tasks.

We note that some related engineering bottlenecks can be partially alleviated by the simplification of tasks. One generic way of achieving this is to reduce the variation between task iterations. As a prototypical example, consider the factory assembly line, turning the non-routine tasks of the artisan shop into repetitive routine tasks performed by unskilled factory workers. A more recent example is the computerisation of non-routine manual tasks in construction. On-site construction tasks typically demand a high degree of adaptability, so as to accommodate work environments that are typically irregularly laid out, and vary according to weather. Prefabrication, in which the construction object is partially assembled in a factory before being transported to the construction site, provides a way of largely removing the requirement for adaptability. It allows many construction tasks to be performed by robots under controlled conditions that eliminate task variability – a method that is becoming increasingly widespread, particularly in Japan (Barlow and Ozaki, 2005; Linner and Bock, 2012). The extent of computerisation in the twenty-first century will thus partly depend on innovative approaches to task restructuring. In the remainder of this section we examine the engineering bottlenecks related to the above mentioned task categories, each in turn.

Robots are still unable to match the depth and breadth of human perception. While basic geometric identification is reasonably mature, enabled by the rapid development of sophisticated sensors and lasers, significant challenges remain for more complex perception tasks, such as identifying objects and their properties in a cluttered field of view. As such, tasks that relate to an unstructured work environment can make jobs less susceptible to computerisation. For example, most homes are unstructured, requiring the identification of a plurality of irregular objects and containing many cluttered spaces which inhibit the mobility of wheeled objects. Conversely, supermarkets, factories, warehouses, airports and hospitals have been designed for large wheeled objects, making it easier for robots to navigate in performing non-routine manual tasks. Perception problems can, however, sometimes be sidestepped by clever task design. For example, Kiva Systems, acquired by Amazon.com in 2012, solved the problem of warehouse navigation by simply placing bar-code stickers on the floor, informing robots of their precise location (Guizzo, 2008).

The difficulty of perception has ramifications for manipulation tasks, and, in particular, the handling of irregular objects, for which robots are yet to reach human levels of aptitude. This has been evidenced in the development of robots that interact with human objects and environments. While advances have been made, solutions tend to be unreliable over the myriad small variations on a single task, repeated thousands of times a day, that many applications require. A related challenge is failure recovery – i.e. identifying and rectifying the mistakes of the robot when it has, for example, dropped an object. Manipulation is also limited by the difficulties of planning out the sequence of actions required to move an object from one place to another. There are yet further problems in designing manipulators that, like human limbs, are soft, have compliant dynamics and provide useful tactile feedback. Most industrial manipulation makes uses of workarounds to these challenges (Brown, et al., 2010), but these approaches are nonetheless limited to a narrow range of tasks. The main challenges to robotic computerisation, perception and manipulation, thus largely remain and are unlikely to be fully resolved in the next decade or two (Robotics-VO, 2013).

The psychological processes underlying human creativity are difficult to specify. According to Boden (2003), creativity is the ability to come up with ideas or artifacts that are novel and valuable. Ideas, in a broader sense, include concepts, poems, musical compositions, scientific theories, cooking recipes and jokes, whereas artifacts are objects such as paintings, sculptures, machinery, and pottery. One process of creating ideas (and similarly for artifacts) involves making unfamiliar combinations of familiar ideas, requiring a rich store of knowledge. The challenge here is to find some reliable means of arriving at combinations that “make sense.” For a computer to make a subtle joke, for example, would require a database with a richness of knowledge comparable to that of humans, and methods of benchmarking the algorithm’s subtlety.

In principle, such creativity is possible and some approaches to creativity already exist in the literature. Duvenaud, et al. (2013) provide an example of automating the core creative task required in order to perform statistics, that of designing models for data. As to artistic creativity, AARON, a drawingprogram, has generated thousands of stylistically-similar line-drawings, which have been exhibited in galleries worldwide. Furthermore, David Cope’s EMI software composes music in many different styles, reminiscent of specific human composers.

In these and many other applications, generating novelty is not particularly difficult. Instead, the principal obstacle to computerising creativity is stating our creative values sufficiently clearly that they can be encoded in an program (Boden, 2003). Moreover, human values change over time and vary across cultures. Because creativity, by definition, involves not only novelty but value, and because values are highly variable, it follows that many arguments about creativity are rooted in disagreements about value. Thus, even if we could identify and encode our creative values, to enable the computer to inform and monitor its own activities accordingly, there would still be disagreement about whether the computer appeared to be creative. In the absence of engineering solutions to overcome this problem, it seems unlikely that occupations requiring a high degree of creative intelligence will be automated in the next decades.

Human social intelligence is important in a wide range of work tasks, such as those involving negotiation, persuasion and care. To aid the computerisation of such tasks, active research is being undertaken within the fields of Affective Computing (Scherer, et al., 2010; Picard, 2010), and Social Robotics (Ge, 2007; Broekens, et al., 2009). While algorithms and robots can now reproduce some aspects of human social interaction, the real-time recognition of natural human emotion remains a challenging problem, and the ability to respond intelligently to such inputs is even more difficult. Even simplified versions of typical social tasks prove difficult for computers, as is the case in which social interaction is reduced to pure text. The social intelligence of algorithms is partly captured by the Turing test, examining the ability of a machine to communicate indistinguishably from an actual human. Since 1990, the Loebner Prize, an annual Turing test competition, awards prizes to textual chat programmes that are considered to be the most human-like. In each competition, a human judge simultaneously holds computer-based textual interactions with both an algorithm and a human. Based on the responses, the judge is to distinguish between the two. Sophisticated algorithms have so far failed to convince judges about their human resemblance. This is largely because there is much ‘common sense’ information possessed by humans, which is difficult to articulate, that would need to be provided to algorithms if they are to function in human social settings.

Whole brain emulation, the scanning, mapping and digitalising of a human brain, is one possible approach to achieving this, but is currently only a theoretical technology. For brain emulation to become operational, additional functional understanding is required to recognise what data is relevant, as well as a roadmap of technologies needed to implement it. While such roadmaps exist, present implementation estimates, under certain assumptions, suggest that whole brain emulation is unlikely to become operational within the next decade or two (Sandberg and Bostrom, 2008). When or if they do, however, the employment impact is likely to be vast (Hanson, 2001).

Hence, in short, while sophisticated algorithms and developments in MR, building upon with big data, now allow many non-routine tasks to be automated, occupations that involve complex perception and manipulation tasks, creative intelligence tasks, and social intelligence tasks are unlikely to be substituted by computer capital over the next decade or two. The probability of an occupation being automated can thus be described as a function of these task characteristics. As suggested by Figure I, the low degree of social intelligence required by a dishwasher makes this occupation more susceptible to computerisation than a public relation specialist, for example. We proceed to examining the susceptibility of jobs to computerisation as a function of the above described non-susceptible task characteristics.

...

...

...

VI. Conclusions

While computerisation has been historically confined to routine tasks involving explicit rule-based activities (Autor, et al., 2003; Goos, et al., 2009; Autor and Dorn, 2013), algorithms for big data are now rapidly entering domains reliant upon pattern recognition and can readily substitute for labour in a wide range of non-routine cognitive tasks (Brynjolfsson and McAfee, 2011; MGI, 2013). In addition, advanced robots are gaining enhanced senses and dexterity, allowing them to perform a broader scope of manual tasks (IFR, 2012b; Robotics-VO, 2013; MGI, 2013). This is likely to change the nature of work across industries and occupations.

In this paper, we ask the question: how susceptible are current jobs to these technological developments? To assess this, we implement a novel methodology to estimate the probability of computerisation for 702 detailed occupations. Based on these estimates, we examine expected impacts of future computerisation on labour market outcomes, with the primary objective of analysing the number of jobs at risk and the relationship between an occupation’s probability of computerisation, wages and educational attainment.

We distinguish between high, medium and low risk occupations, depending on their probability of computerisation. We make no attempt to estimate the number of jobs that will actually be automated, and focus on potential job automatability over some unspecified number of years. According to our estimates around 47 percent of total US employment is in the high risk category. We refer to these as jobs at risk – i.e. jobs we expect could be automated relatively soon, perhaps over the next decade or two.

Our model predicts that most workers in transportation and logistics occupations, together with the bulk of office and administrative support workers, and labour in production occupations, are at risk. These findings are consistent with recent technological developments documented in the literature. More surprisingly, we find that a substantial share of employment in service occupations, where most US job growth has occurred over the past decades (Autor and Dorn, 2013), are highly susceptible to computerisation. Additional support for this finding is provided by the recent growth in the market for service robots (MGI, 2013) and the gradually diminishment of the comparative advantage of human labour in tasks involving mobility and dexterity (Robotics-VO, 2013).

Finally, we provide evidence that wages and educational attainment exhibit a strong negative relationship with the probability of computerisation. We note that this finding implies a discontinuity between the nineteenth, twentieth and the twenty-first century, in the impact of capital deepening on the relative demand for skilled labour. While nineteenth century manufacturing technologies largely substituted for skilled labour through the simplification of tasks (Braverman, 1974; Hounshell, 1985; James and Skinner, 1985; Goldin and Katz, 1998), the Computer Revolution of the twentieth century caused a hollowing-out of middle-income jobs (Goos, et al., 2009; Autor and Dorn, 2013). Our model predicts a truncation in the current trend towards labour market polarisation, with computerisation being principally confined to low-skill and low-wage occupations. Our findings thus imply that as technology races ahead, low-skill workers will reallocate to tasks that are non-susceptible to computerisation – i.e., tasks requiring creative and social intelligence. For workers to win the race, however, they will have to acquire creative and social skills.

1. We refer to computerisation as job automation by means of computer-controlled equipment.
2. This view finds support in a recent survey by theMcKinsey Global Institute (MGI), showing that 44 percent of firms which reduced their headcount since the financial crisis of 2008 had done so by means of automation (MGI, 2011).
3. Because the core job tasks of manufacturing occupations follow well-defined repetitive procedures, they can easily be codified in computer software and thus performed by computers (Acemoglu and Autor, 2011).
4. 4An exception is Goos, et al. (2009).
5. The term artisan refers to a craftsman who engages in the entire production process of a good, containing almost no division of labour. By guild we mean an association of artisans that control the practice of their craft in a particular town.
6. There is an ongoing debate about the technological role of the guilds. Epstein (1998), for example, has argued that they fulfilled an important role in the intergenerational transmission of knowledge. Yet there is no immediate contradiction between such a role and their conservative stand on technological progress: there are clear examples of guilds restraining the diffusion of inventions (see, for example, Ogilvie, 2004).
7. 7Various estimations of the living standards of workers in Britain during the industrialisation exist in the literature. For example, Clark (2008) finds that real wages over the period 1760 to 1860 rose faster than GDP per capita. Further evidence provided by Lindert and Williamson (1983) even suggests that real wages nearly doubled between 1820 and 1850. Feinstein (1998), on the other hand, finds a much more moderate increase, with average working-class living standards improving by15 percent between 1770 and 1870. Finally, Allen (2009a) finds that over the first half of the nineteenth century, the real wage stagnated while output per worker expanded. After the mid nineteenth century, however, real wages began to grow in line with productivity. While this implies that capital owners were the greatest beneficiaries of the Industrial Revolution, there is at the same time consensus that average living standards largely improved.
8. The term skill is associated with higher levels of education, ability, or job training. Following Goldin and Katz (1998), we refer to technology-skill or capital-skill complementarity when a new technology or physical capital complements skilled labour relative to unskilled workers.
9. The production of plows nicely illustrates the differences between the artisan shop and the factory. In one artisan shop, two men spent 118 man-hours using hammers, anvils, chisels, hatchets, axes, mallets, shaves and augers in 11 distinct operations to produce a plow. By contrast, a mechanized plow factory employed 52 workers performing 97 distinct tasks, of which 72 were assisted by steam power, to produce a plow in just 3.75 man-hours. The degree of specialization was even greater in the production of men’s white muslin shirts. In the artisan shop, one worker spent 1439 hours performing 25 different tasks to produce 144 shirts. In the factory, it took 188 man-hours to produce the same quantity, engaging 230 different workers performing 39 different tasks, of which more than half required steam power. The workers involved included cutters, turners and trimmers, as well as foremen and forewomen, inspectors, errand boys, an engineer, a fireman, and a watchman (US Department of Labor, 1899).
10. These machines were sequentially implemented until the production process was completed. Over time, such machines became much cheaper relative to skilled labor. As a result, production became much more capital intensive (Hounshell, 1985).
11. Williamson and Lindert (1980), on the other hand, find a relative rise in wage premium of skilled labour over the period 1820 to 1860, which they partly attribute to capital deepening. Their claim of growing wage inequality over this period has, however, been challenged (Margo, 2000). Yet seen over the long-run, a more refined explanation is that the manufacturing share of the labour force in the nineteenth century hollowed out. This is suggested by recent findings, revealing a decline of middle-skill artisan jobs in favour of both high-skill white collar workers and low-skill operatives (Gray, 2013; Katz and Margo, 2013). Furthermore, even if the share of operatives was increasing due to organizational change within manufacturing and overall manufacturing growth, it does not follow that the share of unskilled labor was rising in the aggregate economy, because some of the growth in the share of operatives may have come at the expense of a decrease in the share of workers employed as low-skilled farm workers in agriculture (Katz and Margo, 2013). Nevertheless, this evidence is consistent with the literature showing that relatively skilled artisans were replaced by unskilled factory workers, suggesting that technological change in manufacturing was deskilling.
12. Most likely, the growing share of white-collar workers increased the element of human interaction in employment. Notably, Michaels, et al. (2013) find that the increase in the employment share of interactive occupations, going hand in hand with an increase in their relative wage bill share, was particularly strong between 1880 and 1930, which is a period of rapid change in communication and transport technology.
13. Computer power even increased 18 percent faster on annual basis than predicted by Moore’s Law, implying a doubling every two years (Nordhaus, 2007).
14. While there is broad consensus that computers substituting for workers in routine-intensive tasks has driven labour market polarisation over the past decades, there are, indeed, alternative explanations. For example, technological advances in computing have dramatically lowered the cost of leaving information-based tasks to foreign worksites (Jensen and Kletzer, 2005; Blinder, 2009; Jensen and Kletzer, 2010; Oldenski, 2012; Blinder and Krueger, 2013). The decline in the routine-intensity of employment is thus likely to result from a combination of offshoring and automation. Furthermore, there is evidence suggesting that improvements in transport and communication technology have augmented occupations involving human interaction, spanning across both cognitive and manual tasks (Michaels, et al., 2013). These explanations are nevertheless equally related to advance in computing and communications technology.
15. Predictions by Cisco Systems suggest that the Internet traffic in 2016 will be around 1 zettabyte (1×1021 bytes) (Cisco, 2012). In comparison, the information contained in all books worldwide is about 480 terabytes (5 × 1014 bytes), and a text transcript of all the words ever spoken by humans would represent about 5 exabytes (5 × 1018 bytes) (UC Berkeley School of Information, 2003).
16. A fly-by-wire robot is a robot that is controllable by a remote computer
17. Rio Tinto’s computerisation efforts are advertised at http://www.mineofthefuture.com.au.
18. The model does not predict any substantial substitution or complementarity with nonroutine manual tasks.

Appendix

The table below ranks occupations according to their probability of computerisation (from least- to most-computerisable). Those occupations used as training data are labelled as either ‘0’ (not computerisable) or ‘1’ (computerisable), respectively. There are 70 such occupations, 10 percent of the total number of occupations.

Computerisable
Rank Probability Label SOC code Occupation
1 0.0028 29-1125 Recreational Therapists
2 0.003 49-1011 First-Line Supervisors of Mechanics, Installers, and Repairers
3 0.003 11-9161 Emergency Management Directors
4 0.0031 21-1023 Mental Health and Substance Abuse Social Workers
5 0.0033 29-1181 Audiologists
6 0.0035 29-1122 Occupational Therapists
7 0.0035 29-2091 Orthotists and Prosthetists
8 0.0035 21-1022 Healthcare Social Workers
9 0.0036 29-1022 Oral and Maxillofacial Surgeons
10 0.0036 33-1021 First-Line Supervisors of Fire Fightingand Prevention Workers
11 0.0039 29-1031 Dietitians and Nutritionists
12 0.0039 11-9081 Lodging Managers
13 0.004 27-2032 Choreographers
14 0.0041 41-9031 Sales Engineers
15 0.0042 0 29-1060 Physicians and Surgeons
16 0.0042 25-9031 Instructional Coordinators
17 0.0043 19-3039 Psychologists, All Other
18 0.0044 33-1012 First-Line Supervisors of Police and Detectives
19 0.0044 0 29-1021 Dentists, General
20 0.0044 25-2021 Elementary School Teachers, Except Special Education
21 0.0045 19-1042 Medical Scientists, Except Epidemiologists
22 0.0046 11-9032 Education Administrators, Elementary andSecondary School
23 0.0046 29-1081 Podiatrists
24 0.0047 19-3031 Clinical, Counseling, and School Psychologists
25 0.0048 21-1014 Mental Health Counselors
26 0.0049 51-6092 Fabric and Apparel Patternmakers
27 0.0055 27-1027 Set and Exhibit Designers
28 0.0055 11-3121 Human Resources Managers
29 0.0061 39-9032 Recreation Workers
30 0.0063 11-3131 Training and Development Managers
31 0.0064 29-1127 Speech-Language Pathologists
32 0.0065 15-1121 Computer Systems Analysts
33 0.0067 0 11-9151 Social and Community Service Managers
34 0.0068 25-4012 Curators
35 0.0071 29-9091 Athletic Trainers
36 0.0073 11-9111 Medical and Health Services Managers
37 0.0074 0 25-2011 Preschool Teachers, Except Special Education
38 0.0075 25-9021 Farm and Home Management Advisors
39 0.0077 19-3091 Anthropologists and Archeologists
40 0.0077 25-2054 Special Education Teachers, Secondary School
41 0.0078 25-2031 Secondary School Teachers, Except Special and Career/Technical Edu-cation
42 0.0081 0 21-2011 Clergy
43 0.0081 19-1032 Foresters
44 0.0085 21-1012 Educational, Guidance, School, and Vocational Counselors
45 0.0088 25-2032 Career/Technical Education Teachers, Secondary School
46 0.009 0 29-1111 Registered Nurses
47 0.0094 21-1015 Rehabilitation Counselors
48 0.0095 25-3999 Teachers and Instructors, All Other
49 0.0095 19-4092 Forensic Science Technicians
50 0.01 39-5091 Makeup Artists, Theatrical and Performance
51 0.01 17-2121 Marine Engineers and Naval Architects
52 0.01 11-9033 Education Administrators, Postsecondary
53 0.011 17-2141 Mechanical Engineers
54 0.012 29-1051 Pharmacists
55 0.012 13-1081 Logisticians
56 0.012 19-1022 Microbiologists
57 0.012 19-3032 Industrial-Organizational Psychologists
58 0.013 27-2022 Coaches and Scouts
59 0.013 11-2022 Sales Managers
60 0.014 19-2043 Hydrologists
61 0.014 11-2021 Marketing Managers
62 0.014 0 21-1013 Marriage and Family Therapists
63 0.014 17-2199 Engineers, All Other
64 0.014 13-1151 Training and Development Specialists
65 0.014 43-1011 First-Line Supervisors of Office and Administrative Support Workers
66 0.015 19-1029 Biological Scientists, All Other
67 0.015 11-2031 Public Relations and Fundraising Managers
68 0.015 27-1014 Multimedia Artists and Animators
69 0.015 15-1111 Computer and Information Research Scientists
70 0.015 0 11-1011 Chief Executives
71 0.015 0 11-9031 Education Administrators, Preschool andChildcare Center/Program
72 0.015 27-2041 Music Directors and Composers
73 0.016 51-1011 First-Line Supervisors of Production andOperating Workers
74 0.016 41-3031 Securities, Commodities, and Financial Services Sales Agents
75 0.016 19-1031 Conservation Scientists
76 0.016 25-2053 Special Education Teachers, Middle School
77 0.017 17-2041 Chemical Engineers
78 0.017 11-9041 Architectural and Engineering Managers
79 0.017 17-2011 Aerospace Engineers
80 0.018 11-9121 Natural Sciences Managers
81 0.018 17-2081 Environmental Engineers
82 0.018 17-1011 Architects, Except Landscape and Naval
83 0.018 31-2021 Physical Therapist Assistants
84 0.019 0 17-2051 Civil Engineers
85 0.02 29-1199 Health Diagnosing and Treating Practitioners, All Other
86 0.021 19-1013 Soil and Plant Scientists
87 0.021 19-2032 Materials Scientists
596 0.94 49-3091 Bicycle Repairers
597 0.94 49-9091 Coin, Vending, and Amusement Machine Servicers and Repairers
598 0.94 51-4121 Welders, Cutters, Solderers, and Brazers
599 0.94 1 43-5021 Couriers and Messengers
600 0.94 43-4111 Interviewers, Except Eligibility and Loan
601 0.94 35-2015 Cooks, Short Order
603 0.94 47-3014 Helpers–Painters, Paperhangers, Plasterers, and Stucco Masons
604 0.94 43-4081 Hotel, Motel, and Resort Desk Clerks
605 0.94 51-9197 Tire Builders
606 0.94 41-9091 Door-to-Door Sales Workers, News and Street Vendors, and RelatedWorkers
607 0.94 37-1011 First-Line Supervisors of Housekeeping and Janitorial Workers
608 0.94 45-2011 Agricultural Inspectors
609 0.94 1 23-2011 Paralegals and Legal Assistants
610 0.95 39-5092 Manicurists and Pedicurists
611 0.95 43-5111 Weighers, Measurers, Checkers, and Samplers, Recordkeeping
612 0.95 51-6062 Textile Cutting Machine Setters, Operators, and Tenders
613 0.95 43-3011 Bill and Account Collectors
614 0.95 51-8011 Nuclear Power Reactor Operators
615 0.95 33-9031 Gaming Surveillance Officers and Gaming Investigators
616 0.95 43-4121 Library Assistants, Clerical
617 0.95 47-2073 Operating Engineers and Other Construction Equipment Operators
618 0.95 51-5113 Print Binding and Finishing Workers
619 0.95 45-2021 Animal Breeders
620 0.95 51-4072 Molding, Coremaking, and Casting Machine Setters, Operators, andTenders, Metal and Plastic
621 0.95 1 51-2022 Electrical and Electronic Equipment Assemblers
622 0.95 51-9191 Adhesive Bonding Machine Operators and Tenders
623 0.95 37-3011 Landscaping and Groundskeeping Workers
624 0.95 51-4033 Grinding, Lapping, Polishing, and BuffingMachine Tool Setters, Oper-ators, and Tenders, Metal and Plastic
625 0.95 43-5051 Postal Service Clerks
626 0.95 51-9071 Jewelers and Precious Stone and Metal Workers
627 0.96 43-5032 Dispatchers, Except Police, Fire, and Ambulance
628 0.96 43-4171 Receptionists and Information Clerks
629 0.96 43-9061 Office Clerks, General
630 0.96 11-3111 Compensation and Benefits Managers
631 0.96 1 43-2011 Switchboard Operators, Including Answering Service
632 0.96 35-3022 Counter Attendants, Cafeteria, Food Concession, and Coffee Shop
633 0.96 47-5051 Rock Splitters, Quarry
634 0.96 43-6014 Secretaries and Administrative Assistants, Except Legal, Medical, andExecutive
635 0.96 17-3031 Surveying and Mapping Technicians
636 0.96 51-7031 Model Makers, Wood
637 0.96 51-6064 Textile Winding, Twisting, and Drawing Out Machine Setters, Opera-tors, and Tenders
638 0.96 53-4011 Locomotive Engineers
639 0.96 1 39-3011 Gaming Dealers
640 0.96 49-9093 Fabric Menders, Except Garment
641 0.96 35-2014 Cooks, Restaurant
642 0.96 39-3031 Ushers, Lobby Attendants, and Ticket Takers
643 0.96 43-3021 Billing and Posting Clerks
644 0.97 53-6011 Bridge and Lock Tenders
645 0.97 51-7042 Woodworking Machine Setters, Operators, and Tenders, Except Sawing
646 0.97 51-2092 Team Assemblers
647 0.97 51-6042 Shoe Machine Operators and Tenders
648 0.97 51-2023 Electromechanical Equipment Assemblers
649 0.97 1 13-1074 Farm Labor Contractors
650 0.97 51-6061 Textile Bleaching and Dyeing Machine Operators and Tenders
651 0.97 51-9081 Dental Laboratory Technicians
652 0.97 51-9021 Crushing, Grinding, and Polishing Machine Setters, Operators, andTenders
653 0.97 51-9022 Grinding and Polishing Workers, Hand
654 0.97 37-3012 Pesticide Handlers, Sprayers, and Applicators, Vegetation
655 0.97 45-4023 Log Graders and Scalers
656 0.97 51-9083 Ophthalmic Laboratory Technicians
657 0.97 1 41-2011 Cashiers
658 0.97 49-9061 Camera and Photographic Equipment Repairers
659 0.97 39-3021 Motion Picture Projectionists
660 0.97 51-5111 Prepress Technicians and Workers
661 0.97 41-2021 Counter and Rental Clerks
662 0.97 1 43-4071 File Clerks
663 0.97 41-9021 Real Estate Brokers
664 0.97 43-2021 Telephone Operators
665 0.97 19-4011 Agricultural and Food Science Technicians
666 0.97 43-3051 Payroll and Timekeeping Clerks
667 0.97 1 43-4041 Credit Authorizers, Checkers, and Clerks
668 0.97 35-9031 Hosts and Hostesses, Restaurant, Lounge, and Coffee Shop
669 0.98 41-9012 Models
670 0.98 51-9061 Inspectors, Testers, Sorters, Samplers, and Weighers
671 0.98 43-3031 Bookkeeping, Accounting, and Auditing Clerks
672 0.98 43-6012 Legal Secretaries
674 0.98 53-3031 Driver/Sales Workers
675 0.98 1 13-1031 Claims Adjusters, Examiners, and Investigators
676 0.98 41-2022 Parts Salespersons
677 0.98 1 13-2041 Credit Analysts
678 0.98 51-4035 Milling and Planing Machine Setters, Operators, and Tenders, Metaland Plastic
679 0.98 43-5071 Shipping, Receiving, and Traffic Clerks
680 0.98 43-3061 Procurement Clerks
681 0.98 51-9111 Packaging and Filling Machine Operators and Tenders
682 0.98 51-9194 Etchers and Engravers
683 0.98 43-3071 Tellers
684 0.98 27-2023 Umpires, Referees, and Other Sports Officials
685 0.98 13-1032 Insurance Appraisers, Auto Damage
686 0.98 1 13-2072 Loan Officers
687 0.98 43-4151 Order Clerks
688 0.98 43-4011 Brokerage Clerks
689 0.98 43-9041 Insurance Claims and Policy Processing Clerks
690 0.98 51-2093 Timing Device Assemblers and Adjusters
691 0.99 1 43-9021 Data Entry Keyers
692 0.99 25-4031 Library Technicians
693 0.99 43-4141 New Accounts Clerks
694 0.99 51-9151 Photographic Process Workers and Processing Machine Operators
695 0.99 13-2082 Tax Preparers
696 0.99 43-5011 Cargo and Freight Agents
697 0.99 49-9064 Watch Repairers
698 0.99 1 13-2053 Insurance Underwriters
699 0.99 15-2091 Mathematical Technicians
700 0.99 51-6051 Sewers, Hand
701 0.99 23-2093 Title Examiners, Abstractors, and Searchers
702 0.99 41-9041 Telemarketers

References

• Acemoglu, D. (2002). Technical change, inequality, and the labor market. Journal of Economic Literature, vol. 40, no. 1, pp. 7–72.
• Acemoglu, D. (2003). Labor- and capital-augmenting technical change. Journal of the European Economic Association, vol. 1, no. 1, pp. 1–37.
• Acemoglu, D. and Autor, D. (2011). Skills, tasks and technologies: Implications for employment and earnings. Handbook of labor economics, vol. 4, pp. 1043–1171.
• Acemoglu, D. and Robinson, J. (2012). Why nations fail: the origins of power, prosperity, and poverty. Random House Digital, Inc.
• Ackerman, E. and Guizzo, E. (2011). 5 technologies that will shape the web. Spectrum, IEEE, vol. 48, no. 6, pp. 40–45.
• Aghion, P. and Howitt, P. (1994). Growth and unemployment. The Review of Economic Studies, vol. 61, no. 3, pp. 477–494.
• Allen, R.C. (2009a). Engels’ pause: Technical change, capital accumulation, and inequality in the british industrial revolution. Explorations in Economic History, vol. 46, no. 4, pp. 418–435.
• Allen, R.C. (2009b). The industrial revolution in miniature: The spinning jenny in Britain, France, and India. The Journal of Economic History, vol. 69, no. 04, pp. 901–927.
• Allen, S.G. (2001). Technology and the wage structure. Journal of Labor Economics, vol. 19, no. 2, pp. 440–483.
• Armstrong, S. and Sotala, K. (2012). How we’re predicting AI – or failing to. Tech. Rep., Oxford: Future of Humanity Institute, Oxford University.
• Atack, J., Bateman, F. and Margo, R.A. (2004). Skill intensity and rising wage dispersion in nineteenth-century American manufacturing. The Journal of Economic History, vol. 64, no. 01, pp. 172–192.
• Atack, J., Bateman, F. and Margo, R.A. (2008a). Steam power, establishment size, and labor productivity growth in nineteenth century American manufacturing. Explorations in Economic History, vol. 45, no. 2, pp. 185–198.
• Atack, J., Haines, M.R. and Margo, R.A. (2008b). Railroads and the rise of the factory: Evidence for the united states, 1850-70. Tech. Rep., NBER Working Paper No. 14410, National Bureau of Economic Research.
• Atkinson, A.B. (2008). The changing distribution of earnings in OECD countries. Oxford University Press.
• Autor, D. and Dorn, D. (2013). The growth of low skill service jobs and the polarization of the US labor market. American Economic Review, vol. forthcoming.
• Autor, D., Katz, L.F. and Krueger, A.B. (1998). Computing inequality: have computers changed the labor market? The Quarterly Journal of Economics, vol. 113, no. 4, pp. 1169–1213.
• Autor, D., Levy, F. and Murnane, R.J. (2003). The skill content of recent technological change: An empirical exploration. The Quarterly Journal of Economics, vol. 118, no. 4, pp. 1279–1333.
• Barlow, J. and Ozaki, R. (2005). Building mass customised housing through innovation in the production system: lessons from Japan. Environment and Planning A, vol. 37, no. 1, pp. 9–20.
• Beaudry, P., Green, D.A. and Sand, B.M. (2013). The great reversal in the demand for skill and cognitive tasks. Tech. Rep., NBER Working Paper No. 18901, National Bureau of Economic Research.
• Beniger, J.R. (1986). The control revolution: Technological and economic origins of the information society. Harvard University Press.
• Blinder, A.S. (2009). How many US jobs might be offshorable? World Economics, vol. 10, no. 2, p. 41.
• Blinder, A.S. and Krueger, A.B. (2013). Alternative measures of offshorability: a survey approach. Journal of Labor Economics, vol. 31, no. 2, pp. S97– S128.
• Bloss, R. (2011). Mobile hospital robots cure numerous logistic needs. Industrial Robot: An International Journal, vol. 38, no. 6, pp. 567–571.
• Boden, M.A. (2003). The creative mind: Myths and mechanisms. Routledge. Braverman, H. (1974). Labor and monopoly capital: The degradation of work in the twentieth century. NYU Press.
• Breslow, L., Pritchard, D.E., DeBoer, J., Stump, G.S., Ho, A.D. and Seaton, D. (2013). Studying learning in the worldwide classroom: Research into edx’s first mooc. Research & Practice in Assessment, vol. 8, pp. 13–25.
• Bresnahan, T.F. (1999). Computerisation and wage dispersion: an analytical reinterpretation. The Economic Journal, vol. 109, no. 456, pp. 390–415.
• Bresnahan, T.F., Brynjolfsson, E. and Hitt, L.M. (2002). Information technology, workplace organization, and the demand for skilled labor: Firm-level evidence. The Quarterly Journal of Economics, vol. 117, no. 1, pp. 339–376.
• Bright, J.R. (1958). Automation and management. Division of Research, Graduate School of Business Administration, Harvard University Boston. Broekens, J., Heerink, M. and Rosendal, H. (2009). Assistive social robots in elderly care: a review. Gerontechnology, vol. 8, no. 2, pp. 94–103.
• Brown, E., Rodenberg, N., Amend, J., Mozeika, A., Steltz, E., Zakin, M.R., Lipson, H. and Jaeger, H.M. (2010). Universal robotic gripper based on the jamming of granular material. Proceedings of the National Academy of Sciences, vol. 107, no. 44, pp. 18809–18814.
• Brown, M. and Philips, P. (1986). Craft labor and mechanization in nineteenthcentury American canning. The Journal of Economic History, vol. 46, no. 3, pp. 743–56.
• Brynjolfsson, E. and McAfee, A. (2011). Race against the machine: How the digital revolution is accelerating innovation, driving productivity, and irreversibly transforming employment and the economy. Digital Frontier Press Lexington, MA.
• Campbell-Kelly, M. (2009). Origin of computing. Scientific American Magazine, vol. 301, no. 3, pp. 62–69.
• Chandler, A.D. (1977). The visible hand: The managerial revolution in American business. Harvard University Press.
• Charles, K.K., Hurst, E. and Notowidigdo,M.J. (2013). Manufacturing decline, housing booms, and non-employment. Tech. Rep., NBER Working Paper No. 18949, National Bureau of Economic Research.
• Churchill, W. and Newman, P. (2012). Practice makes perfect? managing and leveraging visual experiences for lifelong navigation. In: Robotics and Automation (ICRA), 2012 IEEE International Conference on, pp. 4525–4532.
• Cisco (2012). Cisco visual networking index. http://www.cisco.com/en/\us/ netsol/ns827/networking_solutions_sub_solution.html.
• Clark, G. (2008). A farewell to alms: a brief economic history of the world. Princeton University Press.
• Clifford, G.D. and Clifton, D. (2012). Wireless technology in disease management and medicine. Annual Review of Medicine, vol. 63, pp. 479–492.
• Clifton, L., Clifton, D., Pimentel, M., Watkinson, P. and Tarassenko, L. (2012). Gaussian processes for personalised e-health monitoring with wearable sensors. IEEE Transactions on Biomedical Engineering, vol. 60, pp. 193–197.
• Cohn, J. (2013). The robot will see you now. The Atlantic, February 20.
• Cortada, J.W. (2000). Before the Computer: IBM, NCR, Burroughs, and Remington Rand and the Industry They Created, 1865-1956. Princeton University Press.
• Danziger, S., Levav, J. and Avnaim-Pesso, L. (2011). Extraneous factors in judicial decisions. Proceedings of the National Academy of Sciences, vol. 108, no. 17, pp. 6889–6892.
• Davis, S.J. and Haltiwanger, J. (1992). Gross job creation, gross job destruction, and employment reallocation. The Quarterly Journal of Economics, vol. 107, no. 3, pp. 819–863.
• Duvenaud, D., Lloyd, J.R., Grosse, R., Tenenbaum, J.B. and Ghahramani, Z. (2013 June). Structure discovery in nonparametric regression through compositional kernel search. In: Proceedings of the 30th International Conference on Machine Learning, arXiv:1206.1846 [stat.ML]. Cambridge, UK.
• Epstein, S.R. (1998). Craft guilds, apprenticeship, and technological change in preindustrial europe. Journal of Economic History, vol. 58, pp. 684–713.
• Feinstein, C.H. (1998). Pessimism perpetuated: real wages and the standard of living in Britain during and after the industrial revolution. The Journal of Economic History, vol. 58, pp. 625–658.
• Ge, S.S. (2007). Social robotics: Integrating advances in engineering and computer science. In: The 4th annual International Conference organized by Electrical Engineering/Electronics, Computer, Telecommunication and Information Technology (ECTI) Association, Chiang Rai, Thailand, May 9â, vol. 12, p. 2007.
• Goldin, C. and Katz, L.F. (1995). The decline of non-competing groups: Changes in the premium to education, 1890 to 1940. Tech. Rep., NBER Working Paper No. 5202, National Bureau of Economic Research.
• Goldin, C. and Katz, L.F. (1998). The origins of technology-skill complementarity. The Quarterly Journal of Economics, vol. 113, no. 3, pp. 693–732.
• Goldin, C. and Katz, L.F. (2009). The race between education and technology. Harvard University Press.
• Goldin, C. and Sokoloff, K. (1982). Women, children, and industrialization in the early republic: Evidence from the manufacturing censuses. The Journal of Economic History, vol. 42, no. 4, pp. 741–774.
• Goos, M. and Manning, A. (2007). Lousy and lovely jobs: The rising polarization of work in Britain. The Review of Economics and Statistics, vol. 89, no. 1, pp. 118–133.
• Goos, M., Manning, A. and Salomons, A. (2009). Job polarization in europe. The American Economic Review, vol. 99, no. 2, pp. 58–63.
• Gordon, R.J. (2012). Is US economic growth over? faltering innovation confronts the six headwinds. Tech. Rep., NBER Working Paper No. 18315, National Bureau of Economic Research.
• Gray, R. (2013). Taking technology to task: The skill content of technological change in early twentieth century united states. Explorations in Economic History.
• Guizzo, E. (2008 July). Three engineers, hundreds of robots, one warehouse. IEEE Spectrum. http://spectrum.ieee.org/robotics/robotics-software/ three-engineers-hundreds-of-robots-one-warehouse.
• Guizzo, E. (2011). How google’s self-driving car works. IEEE Spectrum Online, October, vol. 18.
• Habakkuk, H.J. (1962). American and British technology in the nineteenth century: The search for labour-saving inventions. University Press.
• Hangal, S. and Lam, M.S. (2002). Tracking down software bugs using automatic anomaly detection. In: Proceedings of the 24th international conference on Software engineering, pp. 291–301. ACM.
• Hanson, R. (2001). Economic growth given machine intelligence. Technical Report, University of California, Berkeley.
• Hoos, H.H. (2012). Programming by optimization. Communications of the ACM, vol. 55, no. 2, pp. 70–80.
• Hornbeck, R. and Naidu, S. (2013). When the levee breaks: Labor mobility and economic development in the American south. The American Economic Review, vol. forthcoming.
• Hounshell, D. (1985). From the American system to mass production, 1800- 1932: The development of manufacturing technology in the United States, vol. 4. JHU Press.
• Jaimovich, N. and Siu, H.E. (2012). The trend is the cycle: Job polarization and jobless recoveries. Tech. Rep., NBER Working Paper No. 18334, National Bureau of Economic Research.
• James, J.A. and Skinner, J.S. (1985). The resolution of the labor-scarcity paradox. The Journal of Economic History, vol. 45, no. 3, pp. 513–540.
• Jensen, J.B. and Kletzer, L.G. (2005). Tradable services: Understanding the scope and impact of services offshoring [with comments and discussion]. In: Brookings trade forum, pp. 75–133. JSTOR.
• Jensen, J.B. and Kletzer, L.G. (2010). Measuring tradable services and the task content of offshorable services jobs. In: Labor in the new economy, pp. 309–335. University of Chicago Press.
• Kahneman, D., Slovic, P. and Tversky, A. (1982). Judgment under uncertainty. Cambridge University Press.
• Katz, L.F. and Margo, R.A. (2013). Technical change and the relative demand for skilled labor: The united states in historical perspective. Tech. Rep., NBER Working Paper No. 18752, National Bureau of Economic Research.
• Katz, L.F. and Murphy, K.M. (1992). Changes in relative wages, 1963–1987: supply and demand factors. The Quarterly Journal of Economics, vol. 107, no. 1, pp. 35–78.
• Kellenbenz, H. (1974). Technology in the age of the scientific revolution, 1500- 1700. The Fontana Economic History of Europe, vol. 2, pp. 177–272.
• Keynes, J.M. (1933). Economic possibilities for our grandchildren (1930). Essays in persuasion, pp. 358–73.
• Kim, S., Whitehead, E.J. and Zhang, Y. (2008). Classifying software changes: Clean or buggy? Software Engineering, IEEE Transactions on, vol. 34, no. 2, pp. 181–196.
• King, S., Bannister, P., Clifton, D. and Tarassenko, L. (2009). Probabilistic approach to the condition monitoring of aerospace engines. Proceedings of the Institution of Mechanical Engineers, Part G: Journal of Aerospace Engineering, vol. 223, no. 5, pp. 533–541.
• Krueger, A.B. (1993). How computers have changed the wage structure: evidence from microdata, 1984–1989. The Quarterly Journal of Economics, vol. 108, no. 1, pp. 33–60.
• Levy, F. and Murnane, R.J. (2004). The new division of labor: How computers are creating the next job market. Princeton University Press. Lindert, P.H. and Williamson, J.G. (1983). Reinterpreting Britain’s social tables, 1688–1913. Explorations in Economic History, vol. 20, no. 1, pp. 94– 109.
• Linner, T. and Bock, T. (2012). Evolution of large-scale industrialisation and service innovation in Japanese prefabrication industry. Construction Innovation: Information, Process, Management, vol. 12, no. 2, pp. 156–178.
• Livshits, B. and Zimmermann, T. (2005). Dynamine: finding common error patterns by mining software revision histories. In: ACM SIGSOFT Software Engineering Notes, vol. 30, pp. 296–305. ACM.
• Louis, P. and Paterson, G. (1986). Biased technical change, scale, and factor substitution in American industry, 1850-1919. The Journal of Economic History, vol. 46, no. 1.
• Lucas, R.E. and Prescott, E.C. (1974). Equilibrium search and unemployment. Journal of Economic Theory, vol. 7, no. 2, pp. 188–209.
• Mantoux, P. (2006). The industrial revolution in the eighteenth century: An outline of the beginnings of the modern factory system in England. Taylor & Francis US.
• Margo, R.A. (2000). Introduction to wages and labor markets in the United States, 1820-1860. In: Wages and Labor Markets in the United States, 1820- 1860, pp. 1–5. University of Chicago Press.
• Markoff, J. (2011). Armies of expensive lawyers replaced by cheaper software. Markoff, J. (2012 August). Skilled work, without the worker. The New York Times.
• Mathibela, B., Osborne, M.A., Posner, I. and Newman, P. (2012). Can priors be trusted? learning to anticipate roadworks. In: Intelligent Transportation Systems (ITSC), 2012 15th International IEEE Conference on, pp. 927–932.
 Michaels, G., Rauch, F. and Redding, S.J. (2013). Task specialization in US  cities from 1880-2000. Tech. Rep., NBER Working Paper No. 18715, National  Bureau of Economic Research.

• Mims, C. (2010 June). AI that picks stocks better than the pros. MIT Technology Review.
• Minka, T.P. (2001). Expectation propagation for approximate Bayesian inference. In: Proceedings of the Seventeenth conference on Uncertainty in artificial intelligence, pp. 362–369. Morgan Kaufmann Publishers Inc.
• Mokyr, J. (1990). The lever of riches: Technological creativity and economic progress. Oxford University Press.
• Mokyr, J. (1998). The political economy of technological change. Technological revolutions in Europe, pp. 39–64.
• Murphy, K.M., Riddell, W.C. and Romer, P.M. (1998). Wages, skills, and technology in the united states and canada. Tech. Rep., NBER Working Paper No. 6638, National Bureau of Economic Research.
• Murphy, K.P. (2012). Machine learning: a probabilistic perspective. The MIT Press.
• Nef, J.U. (1957). Coal mining and utilization. A history of technology, vol. 3, pp. 72–88.
• Nordhaus, W.D. (2007). Two centuries of productivity growth in computing. The Journal of Economic History, vol. 67, no. 1, p. 128.
• North, D.C. and Weingast, B.R. (1989). Constitutions and commitment: the evolution of institutions governing public choice in seventeenth-century england. The Journal of Economic History, vol. 49, no. 04, pp. 803–832.
• Ogilvie, S. (2004). Guilds, efficiency, and social capital: evidence fromGerman proto-industry. The Economic History Review, vol. 57, no. 2, pp. 286–333.
• Oldenski, L. (2012). Offshoring and the polarization of the US labor market. Tech. Rep., Technical report, Mimeo, Georgetown University.
• Osborne, M.A., Garnett, R., Swersky, K. and de Freitas, N. (2012). Prediction and fault detection of environmental signals with uncharacterised faults. In: Twenty-Sixth AAAI Conference on Artificial Intelligence (AAAI-12).
• Phua, C., Lee, V., Smith, K. and Gayler, R. (2010). A comprehensive survey of data mining-based fraud detection research. arXiv preprint arXiv:1009.6119.
• Picard, R.W. (2010). Affective computing: from laughter to IEEE. Affective Computing, IEEE Transactions on, vol. 1, no. 1, pp. 11–17.
• Pissarides, C.A. (2000). Equilibrium unemployment theory. MIT press.
• Plötz, T. and Fink, G.A. (2009). Markov models for offline handwriting recognition: a survey. International Journal on Document Analysis and Recognition (IJDAR), vol. 12, no. 4, pp. 269–298.
• Rasmussen, C.E. and Nickisch, H. (2010). Gaussian processes for machine learning (GPML) toolbox. The Journal of Machine Learning Research, vol. 11, pp. 3011–3015.
• Rasmussen, C.E. andWilliams, C.K.I. (2006). Gaussian Processes forMachine Learning. MIT Press.
• Ricardo, D. (1819). The principles of political economy and taxation. World Scientific.
• Robotics-VO (2013). A Roadmap for US Robotics. From Internet to Robotics. 2013 Edition. Robotics in the United States of America.
• Roy, A.D. (1951). Some thoughts on the distribution of earnings. Oxford Economic Papers, vol. 3, no. 2, pp. 135–146.
• Saha, B., Goebel, K., Poll, S. and Christophersen, J. (2007). An integrated approach to battery health monitoring using Bayesian regression and state estimation. In: Autotestcon, 2007 IEEE, pp. 646–653. IEEE.
• Sandberg, A. and Bostrom, N. (2008). Whole brain emulation: A roadmap, technical report 2008-3. Tech. Rep., Future of Humanity Institute, Oxford University.
• Scherer, K.R., Bänziger, T. and Roesch, E.B. (2010). Blueprint for Affective Computing: A Sourcebook and Manual. Oxford University Press. Schumpeter, J.A. (1962). Capitalism, socialism and democracy. Harper & Row New York.
• Simonite, T. (2013 June). As data floods in, massive open online courses evolve. MIT Technology Review. http://www.technologyreview.com/news/515396/ as-data-floods-in-massive-open-online-courses-evolve.
• Tanner, A. (2007 March). Google seeks world of instant translations. Reuters. http://www.reuters.com/article/2007/03/28/us-google-translate-id\ usN1921881520070328.
• CAA (2012). Roadside assistance customers benefit from smart support during peak and after hours. Tech. Rep., SmartAction case study of Canadian Automobile Association Saskatchewan.
• IFR (2012a). 68 robots perform farmer’s work. Tech. Rep., Case study of Fanuc Robotics Europe S.A., International Federation of Robotics, September 2012.
• IFR (2012b). World robotics 2012. Tech. Rep., International Federation of Robotics, August 30, 2012.
• MGI (2011). An economy that works: Job creation and America’s future. Tech. Rep., McKinsey Global Institute.
• MGI (2013). Disruptive technologies: Advances that will transform life, business, and the global economy. Tech. Rep., McKinsey Global Institute.
• UC Berkeley School of Information (2003). How much information? 2003.
• US Department of Labor (1899). Thirteenth Annual Report of the Commissioner of Labor. Two Volumes, Government Printing Office, Washington, DC.
• Veres, S.M., Molnar, L., Lincoln, N.K. and Morice, C.P. (2011). Autonomous vehicle control systems – a review of decision making. Proceedings of the Institution of Mechanical Engineers, Part I: Journal of Systems and Control Engineering, vol. 225, no. 2, pp. 155–195.
• Williamson, J.G. and Lindert, P.H. (1980). American inequality: A macroeconomic history, vol. 53. Academic Press New York.
• Woolf, B.P. (2010). Building intelligent interactive tutors: Student-centered strategies for revolutionizing e-learning. Morgan Kaufmann.,

volumeDate ValuetitletypejournaltitleUrldoinoteyear
2013 TheFutureofEmploymentHowSusceptThe Future of Employment: How Susceptible Are Jobs to Computerisation? Sep 2013?2013
 Author Carl Benedikt Frey + and Michael A Osborne + title The Future of Employment: How Susceptible Are Jobs to Computerisation? Sep 2013? + year 2013 +