ArticlePDF Available


This paper seeks to address emerging debates and controversies on the impact of robots and artificial intelligence on the world of work. Longer term discussions of technological ‘singularity’ are considered alongside the socio‐technical and economic constraints on the application of robotics and AI. Evidence of robot ‘take‐up’ is gathered from reports of the International Federation of Robotics and from case vignettes reported elsewhere. In assessing the contemporary relationship between singularity, robotics and AI, the article reflects briefly on the two ‘tests’ of artificial ‘intelligence’ proposed by the pioneer computer scientist Alan Turing, and comments on the efficacy of his ‘tests’ in contemporary applications. The paper continues by examining aspects of public policy and concludes that technological singularity is far from imminent.
Robots and AI at Work 205
New Technology, Work and Employment 33:3
ISSN 1468-005X
Robots and AI at work: the prospects for
Martin Upchurch
This paper seeks to address emerging debates and controversies
on the impact of robots and artificial intelligence on the world
of work. Longer term discussions of technological ‘singulari-
ty’ are considered alongside the socio- technical and economic
constraints on the application of robotics and AI. Evidence of
robot ‘take- up’ is gathered from reports of the International
Federation of Robotics and from case vignettes reported else-
where. In assessing the contemporary relationship between
singularity, robotics and AI, the article reflects briefly on the
two ‘tests’ of artificial ‘intelligence’ proposed by the pioneer
computer scientist Alan Turing, and comments on the efficacy
of his ‘tests’ in contemporary applications. The paper contin-
ues by examining aspects of public policy and concludes that
technological singularity is far from imminent.
Keywords: Alan Turing, artificial intelligence (AI),
computerisation, digitalisation, new technology, robots.
A recent report by the World Economic Forum (WEF, 2017) predicted a sea change in the
net effect on job totals as robotics and artificial intelligence (AI) systems are increasingly
introduced into the workplace. In the past, job displacement by new forms of computer-
ised technology was compensated for by job growth in other related areas. The WEF report
suggests that this trend may be reversing, so that the net effect of robotics and AI will be to
reduce rather than increase jobs. Moreover, there also appears a lack of governance of the
impact, in so far as ‘AI development has occurred in the absence of almost any regulatory
environment’ (WEF, 2017 p 50). The report follows academic commentary which also of-
fers a prognosis of net job loss. In terms of calculating risk, Frey and Osborne (2013) sug-
gest that almost half of all occupations and their related jobs in the United States may be
under threat of disappearance in the next two decades. Ford (2015: xii) predicts that work
will be transformed in a way ‘defined by a fundamental shift in the relationship between
workers and machines…machines themselves are turning into workers, and the line be-
tween capital and labour is blurring as never before’. More recent studies have been more
cautious in their predictions. Evidence assembled by the consultants PwC from a variety
of academic and government sources, suggests that whilst being disruptive to labour mar-
kets ‘any job losses from automation are likely to be broadly offset in the long run by new
jobs created as a result of the larger and wealthier economy made possible by these new
technologies. We do not believe, contrary to some predictions, that automation will lead to
mass technological unemployment by the 2030s any more than it has done in the decades
since the digital revolution began’. (PwC, 2018). In terms of occupation, the evidence
Martin Upchurch ( is professor of International Employment Relations, in the
Middlesex University, London, UK.
© 2018 Brian Towers (BRITOW) and
John Wiley & Sons Ltd.
206 New Technology, Work and Employment
© 2018 Brian Towers (BRITOW) and
John Wiley & Sons Ltd.
drawn together by PwC (2018 p3) identifies the likelihood of the most dramatic changes in
use in the transportation sector (especially if ‘driverless’ cars develop), and in financial and
legal services, as algorithms are increasingly applied to checking processes. There is gen-
eral consensus among case and predictive studies that less- skilled manual jobs are more at
risk from robotic displacement than high skill jobs. Person to person service jobs (e.g.
health and social care including social work) are the least likely to be at risk.
Apart from the effect on jobs, debate has focused on the disruptive and potentially
transformative effect of robotics and AI not only on the world of work but society more
generally. We have seen the introduction of new concepts fed by knowledge- based
digital work such as ‘immaterial’ (Hardt and Negri, 2000) or ‘free’ labour (Terranova,
2003), as well as a description of a new form of ‘technological singularity’. Singularity
refers to an end- point which, in the words of Good (1965) envisages a world where
everything is done and made by an ultra- intelligent machine able to ‘surpass all the
intellectual activities of any man however clever……..(so that) the intelligence of man
would be left far behind’. The debate has entered popular journalism, with a vision of
‘post- capitalism’ introduced by Paul Mason to include a vision ‘whereby human la-
bour becomes redundant, the long- term tendency of the rate of profit to fall is conse-
quently made obsolete, and where knowledge- driven production tends toward the
unlimited creation of wealth, independent of the labour expended’. (Mason, 2015: 136).
Public policy and political considerations have entered the fray. The UK Labour Party,
for example, with an eye on government has joined the debate with some vigour. Its
2018 report to Labour’s shadow chancellor John McDonnell on ‘Alternative Models of
Organisation’ warns of the dangers of job loss and suggests that ‘machine learning,
robotics, automation technology, artificial intelligence, the Internet of Things, digital
technologies – mean the coming wave of automation may well be different’. (Labour
Party, 2018 p 8). Part of the response, according to an earlier assessment by McDonnell,
should be to embrace ‘socialism with an iPad’ by investing much more in skills and
technology (McDonnell, 2015). The European Parliament of the EU has also voiced its
concern and is to consider the introduction of legislation granting self- learning robots
the status of ‘electronic personality’ which would allow them to be given legal respon-
sibility for their actions and subject to legal action against them should they go ‘rogue’.
(European Parliament, 2017). However, the supposed disruptive and transformative
effects of digitalisation and computerisation have been questioned. The latest major
study conducted by US- based economist Robert J. Gordon of technology in the United
States entitled The Rise and Fall of American Growth (2016) concludes that the IT revolu-
tion has led to less significant changes in productivity than a host of other technologies
including the telegraph, the electric light, or seemingly more mundane innovations
such as indoor plumbing and urban sanitation. The main reasons given for the caution
about the degree of disruption and transformation rest on the technical, legal and so-
cial constraints to the introduction and activation of robotics and AI.
There is clearly a debate to be had, not only on the interpretation of data, but also on
the balance between disruption of the labour market, and its potential transformation
into a world of work where human labour is redundant. There are concrete reasons
why catastrophic views for the future of human labour might be treated with caution.
As Nolan and Slater (2011) remark ‘The narratives are typically light on theory, resist-
ant to grounded historical and institutional interrogation, and commonly substitute
anecdote for searching empirical analysis’. A more grounded ‘socio- technical’ ap-
proach (MacKenzie and Wajcman, 1985) is needed to address these deficiencies of
analysis and to reveal technical, legislative, economic and societal constraints to the
process of change. The beneficial effects of robots and AI to the employer may also be
overstated, thus restraining adoption of such new technologies. Productivity improve-
ments, for example, may have only a short- term effect as gathering ‘capital- bias’ in the
production process restrict the opportunities for the extraction of surplus value from
the ‘living labour’ of human effort (Roberts, 2016).
This paper seeks to address the arguments, by contextualising the ‘debate’ on robots
and AI within longer term discussions of technological ‘singularity’, and further ex-
ploring the socio- technical and economic constraints on the application of robotics and
© 2018 Brian Towers (BRITOW) and Robots and AI at Work 207
John Wiley & Sons Ltd.
AI in the workplace. Evidence of robot ‘take- up’ is gathered from reports of the
International Federation of Robotics and from case vignettes reported elsewhere. In
assessing the contemporary relationship between singularity, robotics and AI, the arti-
cle reflects briefly on the two ‘tests’ of artificial ‘intelligence’ proposed by the pioneer
computer scientist Alan Turing, and comment on the efficacy of his ‘tests’ in contem-
porary applications. The paper continues by examining further social, technical and
economic constraints and concludes that technological singularity is far from
Accelerating change and singularity
Debates and on robotics and AI have a long history, focused on the prospect of techno-
logical ‘singularity’. The Hungarian mathematician Neumann János Lajos (John von
Neumann) first mooted the concept in the early 1950s as he reflected on the impact of
the next generation of computers. He referred to an ‘end- point’ in computerisation
where machines would be able to dominate production through processes of self-
improvement. AI by ‘deep’ or ‘machine’ learning, in this scenario, will allow for the
re- writing of AI’s own software to outstrip the functional capabilities of the human
brain, and would seemingly bring forward such an ‘end- point’. Raymond Kurzweil
(2005) has since revisited singularity and emphasises the nature of the contemporary
period of growth in technology (spurred by digitalisation). He concludes that by 2045,
the point of technological singularity will have been reached. Moore’s ‘Law’ is cited in
pursuit of this scenario (after Gordon Moore, the founder of Intel) which suggested
that the number of transistors per square inch on integrated circuits had doubled every
year since their invention (Moore, 1965). An adjunct to technological singularity can be
found in the concept of ‘economic singularity’ which articulates the jobless future as
human labour becomes redundant (Chase, 2016).
Within such prognoses of forthcoming singularity, the runaway nature of technolog-
ical advancement is a common theme. A runaway process is predicated on the notion
of ‘accelerating change’, whereby information technology has a special effect in induc-
ing an unstoppable and unquestionable transformation of work. It depends on a sup-
posed autonomy (Ellul, 1964: 14) in the application and effect of technology which then
produces its ‘runaway’ quality (Heidegger, 1977: 17). In such fashion, technological
singularity would be inevitable and simply a matter of time. Runaway and accelerat-
ing change have also been pertinent to longer term debates on the allegedly ‘special’
nature of information and communication technologies. Anthony Giddens (1999) has
been most prominent in promoting such a perspective. In his view, new information
technologies have been acting to engender the ‘runaway’ world in an unquestionable
and unstoppable fashion. However, we must consider the existence of human agency
in the process, both to design and to develop the technology, but also to temper and
shape its use. In this respect, the unquestioning nature of runaway change is highly ques-
tionable in itself. Judy Wajcman, for example, in introducing a socio- technical perspec-
tive, suggested of Giddens’ assumptions that ‘…he treats technology as an autonomous
force rather than as a sociomaterial ensemble of humans, machines, infrastructures
and everyday practices’. (Wajcman, 2016 p46). Labour is a significant agent in this
process, able to modify or resist technologies which appear to run counter to its inter-
est (Feenberg, 1991: 188). Rather than seeing runaway change, we find instead quan-
tum movements or spurts of change, or as Thomas Hughes suggests, a ‘momentum’
may be more apparent. ‘Momentum remains a more useful concept than autonomy…
it does not support the erroneous belief in technological determinism… (and) encom-
passes both structural factors and contingent events’ (Hughes, 1994: 80).
Over the last century, huge leaps of worker and organisational productivity have
also been predicted as time and space are condensed by the new information technol-
ogies, whether they be the telegraph or telephone or the computer and the internet of
things. John Maynard Keynes, reflecting on his contemporary experience with the
mass use of the telephone and telegraph, suggested a working week of no more than
208 New Technology, Work and Employment
© 2018 Brian Towers (BRITOW) and
John Wiley & Sons Ltd.
15 hours per week could be enjoyed as part of The Economic Possibilities for Our
Grandchildren (Keynes, 1930). He felt that the new communication technologies of the
time would massively increase productivity, and free up time for the worker. Following
the computer explosion based on silicon chips toward the end of the 1960s, the ‘futur-
ologist’ Alvin Toffler (1970) echoed Keynes with predications of a 4- hour working day.
Then as now, everyday political discourse became infected with excitement at the pos-
sibility of a new dawn for work and society. The technocratic dream of iPad socialism
described by John McDonnell was already anticipated in 1963 by the then leader of the
British Labour Party, Harold Wilson. In a major speech, he enthused over the ‘white
heat of technology’. Social democrats, he said, could ‘replace the cloth cap [with] the
white laboratory coat as the symbol of British labour’.1 As part of this zeitgeist, The
Collapse of Work (1979) and The Leisure Shock in 1981 were both written by the trade
union leaders Clive Jenkins and Barry Sherman of the UK white- collar union ASTMS.
The books predicted catastrophic levels of unemployment as a consequence of the de-
velopment of the microchip revolution, which could only be overcome by a policy of
work- sharing.
The predictions on working hours proved wrong. Average working hours in the
advanced industrial economies did fall in the immediate post- war period from 1950 to
1980. Much of this fall may have been associated with growing economic prosperity,
social democratic politics and trade union offensives in an era of relatively low unem-
ployment and expanding union memberships. Since the 1980s, the downward trend
has been reversed as many industrialised countries have now seen average working
hours increase (Lee et al., 2007: 32; Pradella, 2015). This era of mass computerisation
has, of course, coincided with the ravages of neoliberal capitalism, which may have
acted to negate any possible positive effects of computerisation on the length of the
working week. However, the optimistic predictions on productivity growth have also
proved false. This is important because the theses of Keynes, Toffler and other contem-
porary commentators are that increasing productivity will be the key to the leisure
society. There is a temporary boost to organisational and worker productivity from
one- off investments in new forms of information technology, but as Henwood (2003)
details such boosts appear unsustainable over time, for reasons explored later in the
paper. In aggregate form, the data from the last 30 years in advanced economies actu-
ally suggest a worsening, rather than improving trend. US Conference Board data an-
alysed by the economist Michael Roberts show us that between 1960 and 1980, average
productivity growth per year in advanced economies was slightly over 4 per cent, it
averaged approximately 2 per cent between 1980 and 2000 but fell further to 1 per cent
and less after 2000 (Roberts, 2014).
We need to explain such paradoxes if we are to fully assess the renewed challenges
of advancement in the technology of robotics and AI. Will such advances overcome
counter- tendencies toward stagnation, and what may be, if any, the technical, social
and economic limitations to world running toward the last ultra- intelligent machine?
If limitations and constraints are obsolete or absent, then what are the prospects for
singularity? To provide an answer to these questions, it is necessary first of all to re-
view the data on robot and AI usage, and to assess the technology as it pertains to the
world of work.
Robots and AI: some evidence
The International Federation of Robotics (IFR) recorded in 2016 a stock of 1.8 million
operational industrial robots worldwide. A ‘robot’ meets the ISO definition to be ‘an
automatically controlled, reprogrammable, multipurpose manipulator, programma-
ble in three or more axes, which can be either fixed in place or mobile for use in indus-
trial automation applications.2 This does not include the small number of professional
service robots or robots for domestic or personal use.3 On current trends, this would
mean the probability of over 3 million by 2020. The largest sector is the automotive,
with a share of 35 per cent of all purchases, closely followed by the electrical or
© 2018 Brian Towers (BRITOW) and Robots and AI at Work 209
John Wiley & Sons Ltd.
electronics sector with 31 per cent. Others showing increases include the chemicals,
rubber and plastics; metal; and food sectors. Both the auto and electrical sectors have
relatively complex manufacturing processes, entailing varying degrees of skill input
and supply chains. They represent industries with a history of mechanisation, sensi-
tive to consumer demand in a competitive product market. Demand for robots has
been increasing steadily since the 2008 financial crash, with Japan the biggest supplier
nation. The average annual increase in sales since 2008 has been in the order of 12 per
cent, but this hides considerable variance between sectors. Driving the increase has
been the electrical or electronics sector, with annual increase recorded in 2016 of 41 per
cent, while growth in the automotive sector appears to have peaked, recording only 6
per cent in 2016 (IFR 2017).
Three quarters of robots are sold in just five countries (IFR, 2017). These coun-
tries are China (with 30 per cent share), South Korea, Japan, USA and Germany. In
some other advanced economies, the patterns of demand have been variable. In the
UK, sales have been declining in the last decade but appeared to increase in 2016.
Sales are down in Italy but have increased in France. Outside of Europe, sales have
been decreasing in Brazil but increasing in Mexico. These patterns reflect the dom-
inance or absence of specific industrial sectors. Where strong growth is recorded in
those countries (in 2016) positively correlated with significant electronics or electri-
cal sectors.
The reality is that robot density remains tiny. The ILO estimates that the labour force
in China alone is 800,000,000. However, there remain only 68 robots in China installed
for every 10,000 employees in the manufacturing industry, and the number of new
robots installed in 2016 was 87,000 (IFR, 2017). In terms of robot density (outside auto
manufacture) ,South Korea topped the chart in 2014 with 475 robots installed per
10,000 employees. It was followed by Japan with 214 robots per 10,000 employees,
Germany with 181 and Sweden with 164 units (IFR, 2014).
The growth (from a low base) in robot and AI use can be attributed to four factors.
First, is the generalised improvement in the mobility of robots. Second, are the ad-
vances in ‘deep’ and ‘machine’ learning associated with improvements in the range
and scope of AI. AI application allows for greater levels of image and speech recog-
nition, while computer algorithms can store data of past behaviour and use it to
predict future behaviour, thus giving the robot the ability to ‘choose’ between ways
and means of performing a task with the appearance that they are ‘thinking’. The
‘thinking’ aspect depends on feedback loops or ‘backpropagation’ which allows al-
gorithms utilised by AI to attempt to operate as artificial neural ‘networks’.
Backpropagation was a method applied by Geoffrey Hinton in 1986 as a way of
mimicking the workings of the human brain and was lauded as the great leap for-
ward in AI capability (see James, 2017 for example). However, Hinton has since
expressed doubt about the method and urged the necessity to ‘throw it all away and
start again…. I don’t think it’s how the brain works. We clearly don’t need all the
labeled data’. (Hinton interviewed by James, 2017). Despite the emerging caveats
and caution, recent publicity given to robots ‘beating’ humans at Chess or Go! is a
testament to their ability to store and discriminate between data on past successful
moves (programmed in by humans) during the board games. Third, is the advances
produced from cloud computing, which allows the functions and tasks to be per-
formed by robots to be programmed remotely, rather than physically on the robot
itself while it is stationary. Ironically, this development means that the jobs of lo-
cally based humans whose task was to re- programme the robot will become de-
funct. Finally, the manufacture of robots, as with all machine- making, is subject to
economies of scale and competitive market pricing, meaning that the relative cost of
robots, when compared to that of human labour has been falling. This is apparent in
China where from IFR data, we find that while the numbers of workers replaced by
robots remained at 3 per robot between 2010 and 2016, the actual number of new
robots has soared by more than 10- fold between 2009 and 2016. This massive in-
crease was a product of the one- third drop in price of robots over the same period
compared to a doubling of workers’ wage rates (IFR, 2014).
210 New Technology, Work and Employment
© 2018 Brian Towers (BRITOW) and
John Wiley & Sons Ltd.
Beside cost, we must also note from a socio- technical perspective that other factors
are at play which constrain prospects for ‘runaway’ application of robotics and AI.
These other factors are summarised below.
Technical limitations
Progress toward a conscious and thinking robot is likely to be slow and ponderous.
Early assessments of the capability of AI referred to the work of the British computer
scientist Alan Turing. In his 1950 essay Computing Machinery and Intelligence, he began
with the statement “I propose to consider the question, “Can machines think?”” and
then presented two ‘tests’ necessary to judge whether an ‘intelligent’ robot could
‘think’ like a human. The first test is based on the proposition that a machine would be
able to ‘think’ if it could hold a conversation that was indistinguishable from one with
a human. Turing developed a metaphor of a party game (the ‘Imitation Game’) which
could be applied to the test, whereby a man and a woman go into separate rooms and
guests try to tell them apart by writing a series of questions and reading the typewrit-
ten answers sent back. He then asked if it would be possible for a computer to ‘fool’ the
guests into thinking a real man or woman was participating in the game rather than a
computer? (see Norvig and Russell, 2016 for a review of these questions). Subsequent
attempts over the following decades to develop a computer program that could pass
the test included the ELIZA program in 1966, and the PARRY a decade later. The latter
produced results which were no different from random application and failed to
convince. In more recent decades, the development of interactive ‘chatterbots’ (such as
Cortana and Siri) have advanced the process but have remained at the level of reactive
agents dependent on pre- programmed information and subsequent machine learning
rather than free- thinking intelligent machines with aesthetic and emotional as well as
linguistic intelligence (features Turing was also keen to explore). However, commen-
tators in the field of ‘android epistemology’ (e.g. Ford et al., 2006) have since ques-
tioned the usefulness of Turing’s ‘thinking’ test when applied to the utility of AI (see
also Whitby, 1988). There is no consensus on what ‘intelligence’ might be, and the
question may only be attempted by reference to philosophical as well as computa-
tional arguments. Copying human intelligence (however defined) may simply be a
distraction as it is often not what AI is designed to do (as it can perform functions
which complement rather than replace the human). Despite the caveats, the challenges
posed by Turing remain live, and serve to highlight the technical limitations on what
AI can and cannot do. This may be illustrated by his second test, that of a ‘halting prob-
lem’ whereby a computer using AI is constantly subject to feedback loops and may
never ‘know’ when it is ‘right’. The computer will continue to compute in an endless
cycle of feedback and re- calculation (Walsh, 2016: 34). The inevitability of continuous
feedback emerged from Turing’s calculations in 1936 whereby he ‘proved’ mathemat-
ically that a general algorithm to solve the halting problem for all possible programme-
input pairs cannot exist (see also Davis, 1958). A computer programme or algorithm is
thus faced with a ‘decision problem’, which can only be resolved by human interven-
tion, thus limiting the scope of applicability of AI.
The evidence on robot usage would suggest that for both tests considerable prob-
lems still exist which sit side- by- side with continuing problems of mobility and image
recognition. Robots also remain machines, and are subject to breakdown, necessitating
human minding and intervention which reduces their potential contribution to pro-
ductivity enhancement. The problem of mimicking human mobility is proving diffi-
cult to solve and efforts to create an affordable ‘plug and play’ robot have stalled. For
example, a leading company producing supposedly highly mobile robots, Rethink
Robots, announced redundancies of nearly a quarter of its staff in 2013 as technical
problems persisted (Tobe, 2013). To help understand such problems, one can imagine
a robot attempting to catch a tennis ball in flight. The velocity and trajectory need to be
finely calculated in a split second. Prior memory of the weight of the tennis ball is
needed to determine how firmly to grip the ball to avoid it bouncing back out of the
© 2018 Brian Towers (BRITOW) and Robots and AI at Work 211
John Wiley & Sons Ltd.
hand. Furthermore, the robot through prior learning must discriminate between a ten-
nis ball, a cricket ball and a table tennis ball in terms of flight, anticipated grip and
‘bounciness’. A human may remember such variables from previous experience but
for a robot this is a much harder task, requiring pre- programmed levels of image rec-
ognition, weight and geometry of flight paths. A human may also recognise immedi-
ately the intent of the thrower of the ball in terms of direction or speed (and even detect
‘fooling’ of intent such as that perceived by the goalkeeper facing the penalty taker), by
eye contact. All this is a logistical nightmare for a robot giving rise to what is known as
the ‘framing problem’, whereby a robot will not be able to make judgements beyond
the framework of the world that it has been programmed to understand (McCarthy
and Hayes, 1969). We can translate such dilemmas to manufacturing. The variables for
auto production, for example, are driven by consumer choice and can include scores of
final assembly options such as paint, trim, electric packs, adjustable seating etc. With
these problems in mind, Mercedes- Benz, a lead player in developing autonomous cars,
has now begun replacing its robots with humans in its factories due to this very lack of
flexibility in the robotic machine (Gibbs, 2016). Such conclusions are confirmed by
Sabine Pfeiffer in her study of German auto manufacturers:
During a normal and otherwise smooth shift, a worker responsible for the ballet of eight welding
and handling robots intervenes 20 to 30 times per shift—not because of technical incidents but in
order to prevent them. Although human work declined quantitatively over the years, its qualitative
role increased with automation
(Pfeiffer, 2016: 16).
As an alternative to full robotisation of the manufacturing process, the likelihood is
that auto producers will increasingly turn to the use of ‘cobots’ designed to undertake
less complex routine tasks alongside more flexible human labour.
Following the earlier work of Hinton et al., (Rumelhart et al., 1986), attempts to rep-
licate human ‘flexibility’ have been constructed through artificial neural networks.
Robotic ‘deep learning systems’ recognise a much greater range of image, colour and
speech patterns than they could in the past. Big data, transferred across platforms, can
also be used in algorithms to map employees’ behaviours, both physically and socially
(in terms of emails and correspondence etc.). However, they still depend on humans to
programme and code them. There is the potential for algorithmic bias reflecting past
biases (perhaps based on age, gender or race) acted upon by pre- algorithmic processes.
In a notorious case, the algorithm- fed robot Beauty.AI only chose women of light skin
when asked to judge an international ‘beauty contest’ (Levin, 2016). Algorithms uti-
lised to predict who may perform best in a job (used by human resource departments
in recruitment and selection, for example) may be based on past results which ex-
cluded those people from non- standard backgrounds, or those in an anglo- saxon envi-
ronment with ‘foreign’ sounding names. Within the academic world, algorithm- based
plagiarism filters used in marking assessments may be biased against second language
students, who are less adept at reformulating words and phrases than native speakers.
Finally, there remains the problem of consciousness, which enables a human to reflect
and to understand context before deciding. Turing’s second test—the ability to know
when a ‘right’ decision has been made and to stop further computing, comes into play
here. A robot can within technical limits be programmed to perform new tasks.
However, expecting the robot to transfer knowledge gained in one task to another is a
much more technically difficult and expensive process. Daniel Dennett (1991: 431) sug-
gests that computers work very differently from the human mind—computers process
increasingly large numbers of information serially, while the mind involves the simul-
taneous interaction of different mechanisms and processes.
Even if such problems could be overcome further hurdles exist. The necessary fibre-
optic networks would need to be in place for high speed transmission of digitised in-
formation connected to AI to be useful. Such networks remain absent in most countries
outside the advanced industrial economies. There is a general absence of a common
platform and language (API- application programming interface) for computers to
212 New Technology, Work and Employment
© 2018 Brian Towers (BRITOW) and
John Wiley & Sons Ltd.
‘talk’ to each other, which restrict the development of ‘smart’ factories linked together
across global supply chains now integral to neoliberal capitalism. Finally, as Mahnkopf
(2017) reports ‘…the transition of companies to the digital landscape exposes them to
the dangers of cyber- attacks by individuals, inside or outside the firm, by computers,
social networks, by the cloud, nefarious organizations and governments’.
Social limitations
We must also consider social limitations, which receive little attention in more apo-
plectic accounts. It is in this realm that the ‘dialectic of technology’ becomes most ap-
parent. The limitations may entail legal constraints covering insurance liability or
personal privacy. With automated vehicles (or driverless/autonomous cars and air-
borne craft such as drones), the insurance liability in case of injury, death or damage is
likely to switch from individual (human) responsibility to the insurer. The insurer will
recover the cost of the claim from the manufacturer, placing an excess cost on manu-
facturing. In the UK, which seeks to place itself as a forerunner in insurance for auto-
mated vehicles, parliament is currently attempting to progress the Vehicle Technology
and Aviation Bill, on such a principle.4 Most ‘driverless’ cars will still have a driver
able to take over control, meaning a dual insurance policy will need to be constructed
when the driver is in control. Such difficulties surrounding insurance liability will be
confusing and likely to challenge not only the insurance industry, but also manufactur-
ers and governments. Ironically, a fast- growing area of AI application is in the insur-
ance and legal business, where cases are algorithmically assessed, and jobs are under
threat as a result. However, there are limits, as if any appeal against decision is made,
a human acting on behalf of an organisation must take responsibility. AI remains
computer- based, it is software linked to hardware and cannot be sued (hence the dis-
cussions within the European Parliament on the prospects of legal ‘rights’ for robots
and AI). The Japanese insurance company Fukoko Mutual laid off 34 employees and
replaced them with an AI system to calculate payouts to policyholders. The system is
based on IBM’s Watson Explorer, which, according to the tech firm, possesses ‘cogni-
tive technology that can think like a human’, enabling it to ‘analyse and interpret all of
your data, including unstructured text, images, audio and video’. (McCurry, 2017).
The Watson Explorer will be able to read tens of thousands of medical certificates and
factor in the length of hospital stays, medical histories and any surgical procedures
before calculating payouts. However, to legitimise the process in law, the sums will
not be paid until they have been approved by a member of staff.
Further socio- legal concerns include data privacy, not only of big data passed be-
tween organisations but also tracking data imposed by employers on individuals. Data
are central to algorithmic processes, but as Moore has commented in a review of legis-
lative developments:
Personal data is not static……..Some types of data can be traced back to an individual using a col-
lection of data points from public records which could be used to construct a picture of an individ-
ual, putting concepts of privacy under scrutiny. Indeed, previously anonymized medical records
have been excavated to gain information about people; school transcripts and church congregation
data resurrected. Not to mention the raft of data being collected by such behemoths as Facebook,
which introduces a new range of possibilities for the mix of personal and public identification, the
use of this data and legalities therein
(Moore, 2017)
Legislation concerning data privacy varies between countries. The European Union
is attempting to address the problem with its General Data Protection Regulation
(GDPR) which has the effect inter alia of preventing employers within the EU of
using tracking information on employees (the whereabouts and actions of ware-
house staff, for example) as a base from which to make HR decisions on employ-
ment, performance and disciplinary matters. These restrictions would drastically
affect many business and HR practices which now rely on algorithmic feedback.
© 2018 Brian Towers (BRITOW) and Robots and AI at Work 213
John Wiley & Sons Ltd.
Companies centred on the ‘gig economy’ would be most under threat. As Moore
remarks, the legislation ‘potentially fully disrupts the Uber business model and op-
erational practices’, simply because ‘Uber taxi drivers gain work through the use of
an app that directs customers purely based on algorithm…(whereby) movements
are entirely tracked and judgements about working practices made accordingly’.
(Moore, 2017).5
The most potent form of restraint on AI and robotics is likely to be worker resist-
ance. The relationship between technology and society not only reflects tensions
within the mode of production but also norms and expectations of the capital–labour
relationship. Historical accounts of automation point to many examples of resistance
to Fordist and Taylorist regimes of production by workers under the period of ‘mod-
ernism’. In the UK, for example, the 150,000 plus dockers in the port of east London
conducted a long, but ultimately unsuccessful fight against containerisation in the
decade from 1966 to 1976 (El- Sahli and Upward, 2015 p.2). Ten years later, 5,500
printworkers were sacked as they attempted to prevent the introduction of digital
journalism in Rupert Murdoch’s new printworks in the same east end of the city.
However, much contemporary theory forming the underlying base of shifts to the
‘knowledge economy’ or ‘immaterial labour’ appears to belie the material base of
digitalisation. This is despite the fact that cloud computers, for example, are rooted
in materiality, and consume huge amounts of energy both working in the cloud and
laying mile upon mile of fibre- optic cable to make the cloud operative. Similarly, the
post- industrial perspective of ‘new social movements’ (e.g. Gorz, 1999), or the ‘Third
Way’ of Giddens (1998) appear to obfuscate the material aspects of the mode and
social relations of production and consequentially downplay the centrality of class
struggle in shaping and re- shaping society (Upchurch and Mathers, 2011). In this
respect, the end of work and end of the working class theses rely on untested as-
sumptions of fed by the collapse of the certainties of social democratic institutions
such as collective bargaining and its associated strike activity. Within this realm of
thinking, the replacement society is the networked society, which reifies electronic
networking to a degree that organised labour is rendered passé, as trade unions find
difficulty in relating to the supposed reflexive and de- bureaucratised nature of the
internet (Castells, 1996).
Such prognoses of organised labour’s irrelevance in the face of the rise of informa-
tion technology may be misplaced. Recent examples of resistance to the new technolo-
gies of personal self- tracking or the vagaries of the ‘gig economy’ in employers such as
Uber or Deliveroo contradict these assumptions (Moore et al., 2017). Evidence of gath-
ering resistance has been assembled by Degryse (2016) for the European Trade Union
Institute which reports on a host of trade union responses including the Dutch unions’
campaign to Master the Robot (De robot de bass) and French trade unions ‘right to dis-
connect’ from electronic surveillance or emails beyond scheduled working time. So
there may be nothing ‘different’ about digitalisation, AI and robotics from other forms
of technology in its ability to crush the power of labour. Contestation will arise in from
labour in different forms which may act to temper, obstruct and even sometimes em-
brace the new wave of automation. Returning to Marx, we can note that he related the
formation and reformation of human society generally to the ‘…change and develop-
ment of the material means of production, of the forces of production…’ and lead to
the conclusion that ‘the mode of production of material life conditions the social, polit-
ical and intellectual life process in general’ (Marx, 1859). Marx here uses a dialectical
approach, relating technology, and its’ use, to the social relations observable within
society. This infers a natural contestation between classes over technological change
and indeed, resistance by the workers in the dying trades and occupations has often
defined industrial relations and the societal conditions of the age. Rather than new
technology heralding ‘the end of the working class’ we can observe that the composi-
tion of the working population continually shifts and changes with technical innova-
tions. Technology at work is thus mediated as part of its introduction, and is a product
of interaction and contestation between state, labour and capital. Resistance is not fu-
tile but inevitable.
214 New Technology, Work and Employment
© 2018 Brian Towers (BRITOW) and
John Wiley & Sons Ltd.
Orthodox economics rely on the balance between supply and demand of commodi-
ties as the key explanatory factor in determining use, exchange and price. More
trenchantly, there is an unquestionable faith that both demand and supply will be
sustained in the short to medium term. Indeed, such an assumption is implicit in the
‘runaway’ approach to new information technologies based on digitalisation and
computerisation. On inspection, we find that it is not the case that there is ever in-
creasing supply and demand for new information technologies. Demand and supply
are certainly tempered through the market, but the market can be saturated and
constrained by personal income restraints and inequalities. On the supply side, there
are concerns over the availability in the future of the rare earth metals necessary for
microchip and fibre- optic production such as indium, gallium, germanium and lith-
ium. The fact that such commodities are ‘rare’ makes them exhaustible and subject to
competition between capitals, further driving up prices and dampening demand and
any assumed ‘runaway’ effect. Congestion and stagnation will reduce demand and
consequently restrain not only the runaway prospects but also assumed gains in
productivity. Researchers at MIT have focused on this aspect and have highlighted
the reasons why ‘research productivity’ appears to stagnate rather than continue to
grow in discrete sectors such as information technology. Stagnation occurs ‘Because
it gets harder to find new ideas as research progresses, a sustained and massive ex-
pansion of research like we see in semiconductors (for example, because of the ‘gen-
eral purpose technology’ nature of information technology) may lead to a substantial
downward trend in research productivity’ (Bloom et al., 2017: 4). We may also study
the effect of ‘congestion’ by referring to the household and personal ‘adoption’ rate
of many new technological innovations aimed at consumers. There is usually a sharp
upward curve in adoption rates of new technologies, followed by a plateau effect as
demand is saturated and new products, sometimes but not always upgrading the
earlier ones, enter the market. The post- war boom in cars, refrigerators and landline
telephones flattened out in advanced western economies by the 1980s. Similarly, the
1970s boom in credit cards and colour televisions lasted just two decades. Is there
any reason to expect that consumer behaviour toward smartphones, robotics and
digital wearables will be any different?
The congestion and related effects have been econometrically tested. On productiv-
ity, key evidence published in 2015 from a data set of companies in 17 countries gath-
ered between 1993 and 2007, suggest that while productivity increases with robotic
innovation and some semi- skilled and lower skilled jobs are abandoned, ‘there is some
evidence of diminishing marginal returns to robot use —“congestion effects”—so they
are not a panacea for growth……this makes robots’ contribution to the aggregate econ-
omy roughly on a par with previous important technologies, such as the railroads in
the 19th century and the US highways in the 20th century’. (Michaels and Graetz,
2015). Computers, including robots, also represent a relatively small proportion of cap-
ital stock, and furthermore, investment has been declining since the height of the ‘IT
Revolution’ of the 1990s (Goodridge et al., 2012: 34). Evidence from the United States
suggest that investment in automation has even ‘decelerated’ in the last decade (Mishel
and Bivens, 2017). This is likely to be because while upgrades in software and hard-
ware are made (such as the constant revision of Microsoft Word, for example), the ag-
gregate effect of such upgrading is likely to be small compared to the initial investment
in the software. The management specialist Michael Porter (2001: 62) commented on
this likelihood and suggests that, ‘as all companies come to embrace internet technol-
ogy, the internet itself will be neutralised as a source of advantage’. Returning to Marx,
we must also note his description of a ‘lifespan of fixed capital’. An individual em-
ployer, and capital in aggregate, may delay purchasing of new technology until they
can be sure of sufficient rate of return on investment. Individual employers will thus
make plans to extend the physical life of pre- existing fixed capital (including both com-
puter hardware and software) as a way of reducing costs (see Weeks, 1981 p186 for a
full explanation).
© 2018 Brian Towers (BRITOW) and Robots and AI at Work 215
John Wiley & Sons Ltd.
Finally, on economics, we need to consider further the relationship between techno-
logical investment and the rate of return on such investment. Orthodox economics tend
to treat technology as a neutral factor in the social relations of production and in its eco-
nomic benefits, without consequences for the capital- labour dynamic. However, if we
draw from classical Marxist economics, we would deduce that the prime motive of in-
vestment in technology is to compete with other capitals by utilising technology to lower
unit labour costs and raise profitability. There is a tension between this need to compete
and the desire of the capitalist to recoup the investment made in new technology. This
can be achieved by increasing rates of exploitation of its workforce and/or by shedding
labour. In so doing, the phenomenon of capital- bias will emerge which serves to dampen
the rate of return on new investments. There occurs a rise in the organic composition of
capital measured by the ratio between constant or fixed capital, which Marx describes as
a product of past or ‘dead’ labour, and variable capital (capital invested in employing
labour- power), which activates the ‘living’ labour of workers in the production process.
Fixed capital, embodied in machinery and previously extracted raw materials, creates no
new value. It merely passes on its value in the process of becoming used by living labour.
As capital- bias takes effect, then the relative share of labour in any one production pro-
cess is reduced and hence the rate of return on capital investment (or rate of profit) falls
correspondingly. As already described, individual capitals must adopt technical innova-
tions to compete. In order to survive, they must match or undercut the generalised ‘so-
cially necessary labour time’ within the product’s sector, which determines the rate of
profit as it proceeds from the degree of exploitation of the workforce. By constant invest-
ment in machinery, they are sowing the seeds of stagnation and decline, by over- reliance
on fixed capital at the expense of variable. Contrary to the ‘post capitalist’ or immaterial
visions of the nature of production, there is no reason to suggest that investment in soft-
ware and the necessary computer hardware to envisage a networked production process
is any different from other forms of technological investment in this respect. Furthermore,
to overcome the deleterious effects of capital- bias, countervailing measures need to be
applied by capital, which involve getting ‘more for less’ from individual workers. Instead
of being a ‘neutral’ input, technology becomes instead a means by which to increase the
rate of exploitation of those workers left behind in the individual workplace (Adler,
1988; Hall, 2010). Either that, or the workforce is reduced even further, exacerbating the
negative effects of capital- bias by expanding ever more the organic composition of
(fixed) capital within the enterprise.
With all these caveats in mind, William Nordhaus (2015) seeks to draw conclusions
on the sum total of economic effects (both positive and negative) on the prospects for
technological singularity. Two ‘accelerationist’ mechanisms could develop to encour-
age singularity, either from accelerating supply or from accelerating demand. He then
applies a series of time- linked tests to both hypothetical scenarios, focusing on the key
input variables such as wages, productivity growth, prices, intellectual property prod-
ucts and R&D. Five of his seven tests for the likelihood of singularity proved negative.
These included the conditions for demand such as that for ‘accelerating productivity
growth’ and ‘rising wage growth’ (real time evidence, of course, shows a decline in
both factors in the last decades). The two that proved positive (including a ‘rising share
of capital’) indicated that singularity, if it did occur, would be at least 100 years away.
And as we have previously positioned, a rising share of capital may simultaneously
lead not only to decreasing rates of productivity growth, but also trigger a crisis of
profitability for capital in the long term. Returning to the vision of singularity which
has again become to obsess social commentators, we might even posit that the dream
of singularity, should it materialise, would thus be faced with a simultaneous collapse
of the underlying dynamic of capitalism.
Concluding remarks
Predictions of the end of the human job because of replacement by robots and AI are
lacking in sufficient analysis and evidence that cover the technical, social and
216 New Technology, Work and Employment
© 2018 Brian Towers (BRITOW) and
John Wiley & Sons Ltd.
economic effects. References to the 1920s/1930s, 1950s, 1970s and 1990s suggest that
predictions of emerging technological singularity proved to be false dawns. Many of
the ‘end of work’ scenarios, from J. M. Keynes, through Toffler, Gorz and Mason rest
their case on ever expanding productivity resulting from computerisation, informa-
tion technology, digitalisation or robotics/AI. Yet, aside from the ‘Golden Age’ of the
1950s and 1960s, we see declining rather than increasing productivity as the new
technologies become embedded. The reasons for the failure are complex and belie
any deterministic approach. First, while new technologies have the capacity to dis-
rupt modes of production, their potential to transform is constrained by the dialectic
of use and non- use conditioned by factors which go beyond mere technical innova-
tion. Second, we need to observe the continuing technical restraints on robots and AI
informed by the two Turing ‘tests’ constructed in the middle of the 20th century. The
validity of the tests has sometimes been questioned, but within the caveats the evi-
dence would suggest that AI’s capacity to overcome these tests, while considerably
enhanced in recent years, is still far from completion. This is more pertinent now,
precisely when the ‘rise of the robots’ is once again apparent in popular and some
academic commentary. This is not to say that the introduction of robots or AI into the
individual workplace will not raise worker productivity in the immediate. Job dis-
placement will always be a motive for the employer investment in robots/AI. The
point is that such gains remain short- lived as congestion and other effects subsume
the initial impact of investment. More importantly, robots remain as machines, they
are ‘dead’ labour which merely pass on existing value rather than create new value.
As such investment in the same will create a capital- bias effect within the factory or
office, leading in the medium term to a decline in the rate of return and a consequent
stifling effect on further investment.
1 content/uploads/2013/06/Labours- Plan- for- science.pdf
2 ISO 8373:2012
3 Some figures are here
4 2017/0143/cbill_2016- 20170143_en_2.htm
5 In 2016, Uber central computer was hacked revealing names and details of 57 million customers and
drivers (Lee, 2017).
Adler, P. (1988), ‘Automation Skill and the Future of Capitalism’, Berkeley Journal of Sociology 33,
Bloom, N., C. Jones, J. Van Reenen and M. Webb (2017), ‘Are Ideas Getting Harder To Find?’
Working Paper 23872, National Bureau of Economic Research, Cambridge, MA.
Castells, M. (1996), The Rise of the Network Society, The Information Age: Economy, Society and Culture
(London: Blackwell).
Chase, C. (2016), The Economic Singularity: Artificial Intelligence and the Death of Capitalism (London:
Three Cs).
Davis, M. (1958), Computability and Unsolvability (New York: McGraw-Hill).
Degryse, C. (2016), Digitalisation of the Economy and its Impact on Labour Markets (Brussels:
European Trade Union Institute).
Dennett, D. (1991), Consciousness Explained (London: Penguin).
Ellul, J. (1964), The Technological Society, trans. J. Wilkinson, (New York: Vintage).
El-Sahli, Z. and R. Upward (2015), ‘Off the Waterfront: The Long-Run Impact of Technological
Change on Dock Workers’, Working Paper 2015:11, Department of Economics, Lund
© 2018 Brian Towers (BRITOW) and Robots and AI at Work 217
John Wiley & Sons Ltd.
European Parliament (2017), Report with recommendations to the Commission on Civil Law
Rules on Robotics,
TEXT+REPORT+A8-2017-0005+0+DOC+XML+V0//EN (accessed 15 September 2018).
Feenberg, A. (1991), Critical Theory of Technology (Oxford: Oxford University Press).
Ford, M. (2015), The Rise of the Robots: Technology and the Threat of a Jobless Future (New York: Basic
Ford, K., C. Glymour and P. J. Hayes (2006), Thinking About Android Epistemology (London: MIT
Frey, C. and M. Osborne (2013), ‘The Future of Employment: How Susceptible Are Jobs to
Computerisation?’ Oxford University,
demic/The_Future_of_Employment.pd (accessed 15 September 2018).
Gibbs, S. (2016), ‘Mercedes-Benz Swaps Robots for People on its Assembly Lines’, The Guardian,
26th February,
people-assembly-lines?CMP=share_btn_fb (accessed 15 September 2018).
Giddens, A. (1998), The Third Way: The Renewal of Social Democracy (Cambridge: Polity).
Giddens, A. (1999), Runaway World (London: Profile Books).
Good, I.J. (1965), ‘Speculations Concerning the First Ultraintelligent Machine’, Advances in
Computers, 6, 31–83.
Goodridge, P., J. Haskel and G. Wallis (2012), ‘UK Innovation Index: Productivity and Growth in
UK Industries’, Nesta Working Paper No. 12/09.
Gordon, R. (2016), The Rise and Fall of American Growth (Princeton, NJ: Princeton University
Gorz, A. (1999), Reclaiming Work: Beyond the Wage-Based Society (London: Polity).
Hall, R. (2010), ‘Labour process theory and technology’, in P. Thompson, and C. Smith (eds),
Working Life: Renewing Labour Process Analysis (Basingstoke: Palgrave), pp. 159–81.
Hardt, M. and A. Negri (2000), Empire (Boston MA: Harvard University Press).
Heidegger, M. (1977) The Question Concerning Technology, trans. W. Lovitt, (New York: Harper
and Row).
Henwood, D. (2003) After the New Economy: The Binge….and the Hangover That Won’t Go Away
(New York: New Press).
Hughes, T. P. (1994), ‘Evolution of Large Systems’ in W. Bijker, T.P. Hughes, and T. Pinch (eds.)
The Social Construction of Technological Systems, 5th edn (first edition, 1987), (Cambridge, MA:
MIT Press), pp. 51–82.
IFR (2014), Executive Summary World Robotics 2014 Industrial Robots Report, (Frankfurt-am-Main:
International Federation of Robotics).
WorldRobotics_ExecSummary.pdf (accessed 15 September 2018).
IFR (2017), Executive Summary World Robotics 2017 Industrial Robots (Frankfurt-am-Main:
International Federation of Robotics) (accessed 15 September 2018).
James, M. (2017), ‘Geoffrey Hinton Says AI Needs To Start Over’ I Programmer. https://ww-
to-start-over.html (accessed 15 September 2018).
Jenkins, C. and B. Sharman (1981), The Leisure Shock (London: Eyre Methuen).
Jenkins, C. and B. Sherman (1979), The Collapse of Work (London: Eyre Methuen).
Keynes, J. M. (1963/1930), ‘Economic Possibilities for our Grandchildren’ in John Maynard
Keynes (ed), Essays in Persuasion (New York: W. W. Norton & Co), pp. 358–373. https://www. (accessed
15 September 2018).
Kurzweil, R. (2005), The Singularity Is Near (London: Viking).
Labour Party (2018), Alternative Models of Ownership (London: Labour Party)
Lee, D. (2017), ‘Uber Concealed Huge Data Breach’ BBC News, 22nd November, http://www. (accessed 15 September 2018).
Lee, S., D. McCann and J. C. Messenger (2007), Working Time Around the World: Trends in Working
Hours, Laws and Policies in a Global Comparative Perspective (Abingdon: ILO/Routledge).
Levin, S. (2016) ‘A Beauty Contest was Judged by AI and the Robots didn’t Like Dark Skin’
Guardian, 8th September,
intelligence-beauty-contest-doesnt-like-black-people (accessed 15 September 2018).
MacKenzie, D. and J. Wajcman (1985), The Social Shaping of Technology: How the Refrigerator Got Its
Hum (Milton Keynes: Oxford University Press).
Mahnkopf, B. (2017), ‘The (False) Promises of Digitalization’, Social Europe, 8 November
Marx, K. (1859), Preface to A Contribution to the Critique of Political Economy, https://www. (accessed
15 September 2018).
Mason, P. (2015), PostCapitalism: A Guide to Our Future (London: Allen Lane).
218 New Technology, Work and Employment
© 2018 Brian Towers (BRITOW) and
John Wiley & Sons Ltd.
McCarthy, J. and P. J. Hayes (1969), ‘Some Philosophical Problems from the Standpoint of
Artificial Intelligence’, Machine Intelligence 4, 463–502
McCurry, J. (2017) ‘Japanese Company Replaces Office Workers with Artificial Intelligence’,
Guardian, 5th January
McDonnell, J. (2015),‘How Labour will Secure the High-Wage, Hi-Tech Economy of the Future’,
The Guardian, 19th November
Michaels, G., G. Graetz (2015), Industrial Robots Have Boosted Productivity And Growth, But Their
Effect On Jobs Remains An Open Question,
Mishel, L. and J. Bivens (2017), ‘The zombie robot argument lurches on’ Report for Economic Policy
Institute, May 24th, Washington D.C.
Moore, G. E. (1965) ‘Cramming More Components onto Integrated Circuits’, Electronics 38, 8,
Moore, P. (2017), The GDPR, Algorithms and People Analytics, https://phoebevmoore.wordpress.
com/2017/11/07/the-gdpr-algorithms-and-people-analytics/ (accessed 10 August 2018).
Moore, P., P. Akhtar and M. Upchurch (2017), ‘Digitalisation of Work and Resistance’ in P.V.
Moore, M. Upchurch, and X. Whittaker (eds.) Humans and Machines at Work (London: Palgrave
Macmillan), pp. 17–44.
Nolan, P. and G. Slater (2011), ‘Visions of the Future, the Legacy of the Past: Demystifying the
Weightless Economy’, Labor History, 51, 1, 7–27
Nordhaus, W. (2015), ‘Are We Approaching an Economic Singularity? Information Technology
and the Future of Economic Growth’. Cowles Foundation Discussion Paper No. 2021, Yale
Norvig, P. and S.J. Russell (2016), Artificial Intelligence: A Modern Approach, 1st edn (London:
Pfeiffer, S. (2016), ‘Robots, Industry 4.0 and Humans, or Why Assembly Work Is More than
Routine Work’, Societies, 6, 2, 16
Porter, M. (2001) ‘Strategy and the Internet’, Harvard Business Review 79, 3, 62–78.
Pradella, L. (2015), ‘The Working Poor in Western Europe: Labour, Poverty and Global
Capitalism’, Comparative European Politics, 13, 596–613.
PwC (2018), Will Robots Really Steal Our Jobs?
sets/international-impact-of-automation-feb-2018.pdf (accessed 10 August 2018).
Roberts, M. (2014), Productivity, deflation and depression, https://thenextrecession.wordpress.
com/2014/01/20/productivity-deflation-and-depression/ (accessed 10 August 2018).
Roberts, M. (2016), ‘Can robots usher in a socialist utopia or only a capitalist dystopia?’, Socialist
Review, (July/August)
or-only-capitalist-dystopia (accessed 8 January 2018).
Rumelhart, D., G. Hinton and R. Williams (1986), Parallel Distributed Processing: Explorations in the
Microstructure of Cognition, Volume 1 (Cambridge, MA: MIT Press)
Terranova, T. (2003), ‘Free Labor: Producing Culture for the Digital Economy’, electronic book
review/technocapitalism, 2003-06-20 available at
thread/technocapitalism/voluntary (accessed 13 April 2018).
Tobe, F. (2013), “Rethink Robotics is Downsizing”,
Toffler, A. (1970), Future Shock (New York: Random House).
Turing, A. (1950), ‘Computing Machinery and Intelligence’, Mind, LIX, 236, 433–460.
Upchurch, M. and A. Mathers (2011), ‘Neo- liberal Globalization and Trade Unionism: Towards
Radical Political Unionism’, Critical Sociology, 38, 2, 265–280.
Wajcman, J. (2016), Pressed for Time: The Acceleration of Life In Digital Capitalism (Chicago:
University of Chicago Press).
Walsh, T. (2016), ‘What if We create human- level artificial intelligence?’, New Scientist, 19,
Weeks, J. (1981), Capital and Exploitation (Princeton, NJ: Princeton University Press).
WEF (2017), The Global Risks Report 2017 (Geneva: World Economic Forum)
Whitby, B. (1988), AI: A Handbook of Professionalism (Chichester: Ellis Horwood).
... It is noteworthy that chatbots do not act just as an employees' serving tool but enhance learning by changing the perspective from 'organising' knowledge and expertise of employees to 'servicing' them (Flanagan & Walker, 2021). Minimizing the cost and job displacement will always be a motive for the employer to invest in AI (Upchurch, 2018). However, there are still a lot of barriers to overcome, and the technological singularities of robots and chatbots are still far from imminent (Upchurch, 2018). ...
... Minimizing the cost and job displacement will always be a motive for the employer to invest in AI (Upchurch, 2018). However, there are still a lot of barriers to overcome, and the technological singularities of robots and chatbots are still far from imminent (Upchurch, 2018). In fact, despite multiple apparent advantages, surveys showed that CEOs and other senior executives are delaying in embedding AIpowered tools (as it could be chatbots) due to, among other reasons, an ambiguous vision of the usefulness of chatbots in recruiting and the reflection on employee's performance (Bughin et al., 2019). ...
Full-text available
Although artificial intelligence (AI) is broadly extended worldwide, full potential is not being reached in human resource management (HRM) so far. In fact, there is still a gap in analyzing the usefulness and limitations of chatbots in recruitment. This piece of research attempts to cover part of this gap paying particular attention to the Indian market. A qualitative analysis has been undertaken in selected Indian companies. One of the main conclusions drawn from the study is the need for mature chatbots in recruitment. Although chatbot has several features from HR professionals' perspectives, our paper reveals that chatbot has particular limitations in hiring mid-level and senior-level positions. Furthermore, there is a gap between what are Indian IT companies offering for recruiters and what are recruiters know about the latest development of the chatbots. The paper contributes as a food for thought for HR managers and academia by provides an extensive analysis of the usefulness and limitations of chatbots in the recruitment process in India.
... Spencer & Slater, 2020). On a similar notion, sticking to labour-intensive processes might in some cases simply be less cost-intensive and high investments into automation technology might also contradict with shareholder desires for short-term profits, while high wages can work as an incentive to automate (Lewis & Bell, 2019;Upchurch, 2018). Especially in highly industrialized countries, the limited availability of qualified workforce as well as the reduced potential for further productivity improvements might limit the potential returns for additional capital investment (Upchurch, 2018). ...
... On a similar notion, sticking to labour-intensive processes might in some cases simply be less cost-intensive and high investments into automation technology might also contradict with shareholder desires for short-term profits, while high wages can work as an incentive to automate (Lewis & Bell, 2019;Upchurch, 2018). Especially in highly industrialized countries, the limited availability of qualified workforce as well as the reduced potential for further productivity improvements might limit the potential returns for additional capital investment (Upchurch, 2018). ...
Full-text available
Companies across the globe have intensified the digital interconnectedness of their manufacturing processes. Much attention was devoted to how industrial employment will be affected in this new production paradigm. In this paper, we use survey data collected from German industrial workers in 2014 and 2020 to contribute to the literature on digitalisation and industrial employment. This is the first scientific study on Industry 4.0 that empirically deals with the development of key parameters of industrial employment over time. Our findings support the argument that whilst increased digital interconnectedness creates more opportunities for highly skilled workers, the extent to which manual workers will be substituted is often overestimated. Second, our data suggests that the operations of larger companies tend to be more highly digitally interconnected than those of smaller firms. We also provide evidence that German industrial workers are less likely to expect substantial job losses through digitalisation than in 2014.
... R.1 The integrator role of the biological and technological interaction (Matthew, 2020); R.2 The master's role in the interaction with the artificial intelligence (Upchurch, 2018); (Magee et al., 2011); R.4 The slave's role in the interaction with the artificial intelligence (Chitty, 2017;Turchin, 2018); R.5 The difficulty to identify a role (Goode, 2018). ...
... HR is familiar with digital tools facilitating daily work (Allal-Ch erif et al., 2021) but recently it is AI that transforms multiple activities of HRM (Vrontis et al., 2022). For example, scholars discuss an increase of productivity when employees cooperate with AI (Upchurch, 2018), control and evaluation of employees by algorithms (Kellogg et al., 2020), or consequences of AI implementation for various classes of workers (Frey and Osborne, 2017). Scholars use mostly conceptual theorising. ...
Full-text available
Purpose: The importance of artificial intelligence in human resource management has grown substantially. Previous literature discusses the advantages of AI implementation at a workplace and its various consequences, often hostile, for employees. However, there is little empirical research on the topic. The authors address this gap by studying if individuals oppose biased algorithm recommendations regarding disciplinary actions in an organisation. Design/methodology/approach: The authors conducted an exploratory experiment in which the authors evaluated 76 subjects over a set of 5 scenarios in which a biased algorithm gave strict recommendations regarding disciplinary actions at a workplace. Findings: The authors’ results suggest that biased suggestions from intelligent agents can influence individuals who make disciplinary decisions. Social implications: The authors’ results contribute to the ongoing debate on applying AI solutions to HR problems. The authors demonstrate that biased algorithms may substantially change how employees are treated and show that human conformity towards intelligent decision support systems is broader than expected. Originality/value: The authors’ paper is among the first to show that people may accept recommendations that provoke moral dilemmas, bring adverse outcomes, or harm employees. The authors introduce the problem of “algorithmic conformism” and discuss its consequences for HRM.
... Considering the singularity, such as the increase in the share of intellectual property goods in the capital stock, these tests eventually indicate that the singularity is still a long way off. Upchurch and Moore feel it is still uncertain if the singularity will occur, and they identify other criteria that restrict the singularity's occurrence (14). First, from a technological standpoint, it is extremely difficult to construct sentient, human-like robots. ...
Full-text available
Rapid advancements in artificial intelligence (AI) will have a dramatic impact on the global economy. This paper provides a systematic review of the economic impact of AI, focusing on the promotion of AI on productivity and economic growth; the impact of AI on labor employment; and the question of whether AI will increase income inequality. On this basis, a summary of how to implement public policies to reduce the potential negative impacts of AI on the employment structure and income inequality is provided. Finally, a summary and prospective research directions are provided.
Artificial Intelligence (AI) field of research started in the 1950s, for the purpose of understanding the nature in human intelligence (Jatobá et al. 2019). Artificial Intelligence accuracy and efficiency highly overreach the traditional management ability, AI highly beats human capabilities in accuracy and in processing data and storage, as human judgments in some cases are inaccurate and part from real situation; AI judgments are accurate and will help employees in making the right decisions and ongoing advancements can be reached (Wang and Li 2019). Human resource management (HRM) strategy is integrated with the organizations business strategy, HRM represents the organizations’ high level of decision making, HRM strategy focuses on employment policies and practices which consists of recruitment, selection, evaluation, development and retention of employees, and consultation and negotiation with individuals. Artificial intelligence is extremely important to be integrated in human resource management functions to support and run human resources functions efficiently (Jatobá et al. 2019). The research method will be systematic review aiming to find out the impact of artificial intelligence on enhancing human resource management functionality. It has been found that integrating artificial intelligence to the complex and various human resources tasks will reduce the large amount of time and efforts spent on performing those tasks, leading to efficiency and quality gains for the HRM functions (Arena et al. 2018).
Full-text available
This thesis explores how Uber reformats the urban and vice versa. Rather than taking for granted Uber’s success in remoulding the emerging ‘smart city’ in its own image, Uberising the Urban pays close attention to the contradictory, variegated and far from frictionless encounters between Uberisation and urbanisation. The thesis is particularly interested in those neuralgic points of contact where the abstract logics of Uber’s business model – its vectors of data extraction, labour exploitation and platform expansion – hit the urban ground of existing social and physical geographies. The Uberisation of the urban – such is this thesis’s main argument – does not take place in a material and social void; it unfolds in, with and against the dense social and material thickness of existing urban space. This argument is deepened in three case studies. Zooming in from different angles, these case studies show how the vectors of Uberisation have come up against the multiscalar and variously uneven urban grounds of the actually existing smart city of Toronto. While the first case study provides a detailed discussion of the conflictive processes leading up to the legalisation of Uber in Toronto and the parallel ‘regulated deregulation’ of the city’s taxi-cum-ridehail market, the second case study tackles the next subsequent ‘stage’ of Uberisation in Toronto: the proliferation of various public-private ridehail partnerships (PPRPs) between Uber and Lyft on the one hand and local and regional transit agencies in the GTA on the other. The third case study is concerned with Uber’s self-driving car programme and, in particular, the invasive practices of data extraction that Uber has implemented in Toronto – turning the city into a real-life urban data reservoir for the development of its self-driving software. A conclusion, shedding light on a potential reconfiguration of Uber towards more socially emancipatory ends, rounds out the dissertation.
Robots, artificial intelligence, and autonomous vehicles are associated with substantial narrative and image-related legacies that often place them in a negative light. This chapter outlines the basics of the “dramaturgical” and technosocial approaches that are used throughout this book to gain insights about how these emerging technologies are affecting deeply-seated social and psychological processes. The robot as an “other” in the workplace and community—an object of attention and discussion– has been a frequently-utilized theme of science fiction as well as a topic for research analysis, with many people “acting out” their anxieties and grievances. Human-AI contests and displays of robotic feats are often used to intimidate people and reinforce that individuals are not in control of their own destinies, which presents unsettling prospects for the future.
Motivated by the contradiction between a government's hard budget constraints and artificial intelligence, this study constructs a computable general equilibrium model embedded with various fiscal and tax policies to study the impact of artificial intelligence development on the Chinese economy under government budget constraints with different intensities. This paper seeks to find a reasonable policy that takes into account China's employment, income distribution, and budget constraints to achieve common prosperity. It finds that the softer the government's budget constraints, the smaller the negative impact of artificial intelligence on the economy. More specifically, allowing the government to increase its debts and spending is more effective than tax cuts. It is suggested that if the goal is to reconcile the contradiction between hard budget constraints and artificial intelligence, fiscal and tax policy combinations, together with an improvised soft budget constraint, are required to increase the taxation of capital to an appropriate degree. In order to resolve the contradiction between the government's hard budget constraints and the development of artificial intelligence in pursuit of common prosperity, a robot tax should be levied and automation capital taxed.
Industry 4.0 is fundamentally changing industrial employment around the globe. Yet, the burgeoning scientific and societal discourses both highlight the need for more empirical evidence of how Industry 4.0 affects industrial employment by taking into consideration varying circumstances and preconditions. Given the heterogeneity of industrial development and technology proliferation between countries, it is crucial to understand and compare future outlooks of industrial practitioners from different regions. This study analyses how Industry 4.0 will affect industrial employment comparing practitioners’ perceptions from Brazil, China, and Germany. Moreover, in our analysis we provide insights for different industrial sectors and company sizes. The study provides evidence that the effects of Industry 4.0 on staffing requirements differ between domains. Domains generally associated with larger shares of relatively low-skilled labour are expected to experience declines in staffing requirements, suggesting an increase in the polarisation between high- and low-skilled labour in terms of job opportunities. This effect is observable especially in larger companies. Moreover, an increase in the required employee qualifications is expected in all domains. Unlike in the other two countries, the inter-sectoral variability of perceptions is very small in China. There the company size seems to have the reverse effect on required qualifications compared to Brazilian and German companies, where less respondents expect higher qualifications for SMEs. Although we find sectoral differences both within and between countries, there is no clearly discernible trend allowing for generalizable sectoral conclusions, highlighting that impacts of Industry 4.0 on qualification should be further investigated under consideration of underlying contextual factors.
Long-run growth in many models is the product of two terms: the effective number of researchers and their research productivity. We present evidence from various industries, products, and firms showing that research effort is rising substantially while research productivity is declining sharply. A good example is Moore’s Law. The number of researchers required today to achieve the famous doubling of computer chip density is more than 18 times larger than the number required in the early 1970s. More generally, everywhere we look we find that ideas, and the exponential growth they imply, are getting harder to find. (JEL D24, E23, O31, O47)
Workplace tracking, monitoring and surveillance are in the ascendant. This chapter looks at everyday forms of resistance and evidence of dissatisfaction emerging in the context of digitalisation in workplaces via first-hand accounts from workers in warehouses, offices and gig economy work; and examples of trade union responses. Workers’ concrete labour is now increasingly subject to abstraction and extraction as new ways to measure our previously unseen labour become apparent and more subject to commodification in the process. In warehouse and factory work, accountability is transferred to workers but data not transparent. In professional workplaces, a new employer prescription reduces any possibility for negativity and workplace dissent by highlighting wellbeing. But the outcome is the same: intensified control mechanisms over workers.
What are the prospects for long-run economic growth?, the present study looks at a more recently launched hypothesis, which I label Singularity. The idea here is that rapid growth in computation and artificial intelligence will cross some boundary or Singularity after which economic growth will accelerate sharply as an ever-accelerating pace of improvements cascade through the economy. The paper develops a growth model that features Singularity and presents several tests of whether we are rapidly approaching Singularity. The key question for Singularity is the substitutability between information and conventional inputs. The tests suggest that the Singularity is not near.
We are suffering just now from a bad attack of economic pessimism. It is common to hear people say that the epoch of enormous economic progress which characterised the nineteenth century is over; that the rapid improvement in the standard of life is now going to slow down—at any rate in Great Britain; that a decline in prosperity is more likely than an improvement in the decade which lies ahead of us.