1
Automation:
Is It Really Different This Time?
A summary review1
Judy Wajcman
Martin Ford, The Rise of the Robots: Technology and the Threat of Mass Unemployment, London 2016.
Richard Susskind and Daniel Susskind, The Future of the Professions: How Technology Will Transform the Work of Human Experts, Oxford 2015.
Erik Brynjolfsson and Andrew McAfee, The Second Machine Age: Work, Progress, and Prosperity in a Time of Brilliant Technologies, New York 2014.
John Urry, What is the Future?, Cambridge 2016.
I have lost count of the number of conferences I have attended on Robots, Artificial Intelligence (AI) and the Future of Work. Predicting the future has once again become big business, a sure sign of which is the plethora of books appearing on this topic â those chosen above are but a tiny sample of the genre.
Such conferences have a common format. A few humanlike robotic heads, often with female nomenclature, are displayed and we are encouraged to interact with them for the wow factor. Then a panel of geeks tells us, the lay audience, about their amazing advances, and how close they are to passing the Turing test (making interaction with social robots indistinguishable from human interaction). This is followed by some economists estimating the dire consequences of advanced technology for job prospects. Finally, a few futurists are also included, some even from the so-called Singularity University.2 I naively asked one of them where this university was based and was told âit isnât really a universityâ! Itâs a state of mind, man.
So let me first sketch out the prevailing predictions about employment, then say something about the hyperbole on automation, robotics and AI, and finally why we need more books like Urryâs What is the Future? that provide some critical distance on this futurist discourse.
Letâs begin with Fordâs The Rise of the Robots, the Financial Times 2015 business book of the year. The book is laudable as a trade book, a pacey read about how an increasingly automated economy will affect modern workers. From manufacturing to services, from higher education to healthcare, myriad developments in AI are addressed that, according to Ford, will leave no occupation untouched. The scope of the book is impressive, not only in providing an accessible overview of the latest advances in automation, but also in comprehensively rehearsing the economic and policy debates about the future of work.
It is a thoughtful book and while history is not Fordâs longbow, he does acknowledge that fears of technological unemployment are not new. Even the Luddites get a mention. The crux of his argument, however, is clear. All the books reviewed here say it with one voice: âthis time it is differentâ. Yes, the masses that were thrown out of agriculture found jobs in factories; yes, there was the expansion of the service sector. But this time it really is different. A new future is on its way, and it is scary. Fordâs book is peppered with words and phrases like âfrighteningâ, âtipping pointâ and a âperfect stormâ.
According to Ford, information technology (IT) is the game changer, a uniquely disruptive force that has no historical precedent. This is because it is not only the low-skilled that will be displaced â highly skilled professionals are also at risk of being displaced by machines. Where previous waves of automation ultimately created wealth and new sectors of employment, we are now witnessing a fundamental shift in the relationship between workers and machines. Machines are no longer tools; they are turning into the workers themselves. âAll this progress is, of course,â Ford writes, âbeing driven by the relentless acceleration in computer technologyâ (p. xii). As usual, Mooreâs Law is invoked to prove the inexorability of accelerating technical progress.
The popular commentators and journalists, not to mention the business consultants, seem to devour this bleak picture with a Frankensteinian relish. It is what Urry calls in his book the ânew catastrophismâ: we stand in awe â and terrified expectation â of what we have created, awaiting the devastating consequences.
So, what is the empirical evidence for Fordâs thesis? Interestingly, Ford pauses halfway through Chapter 2 to eschew a too simple narrative that puts advancing technology âfront and centreâ as the explanation for the troubling economic trends he identifies, but then quickly reasserts that ITâs relentless acceleration sets it apart. Tellingly, he says, âIâm content to leave it to economic historians to delve into the data.â Evidence is largely presented in the form of vivid stories about the feats of Big Data and âdeepâ machine learning. Here pride of place is given to artificial neural networks â systems that are designed using the same fundamental operating principles as the human brain â that can be used to recognise images or spoken words, translate languages, etc. Such systems already power Appleâs Siri and, potentially, could transform the nature and number of knowledge-based jobs. If IBMâs Watson can win Jeopardy! and Googleâs AI can recognise catsâ faces based on millions of YouTube videos, then, Ford surmises, few jobs will remain.
Like almost everyone else, he cites the Oxford Martin Schoolâs Frey and Osborne, whose line about half of US jobs being vulnerable to machine automation within the next two decades is endlessly repeated.3 This estimate, by the way, is based on an algorithm that predicts the susceptibility to automation of different occupations (rather than on the task content of individual jobs). That this methodology has been heavily critiqued has done nothing to halt its endless citation.4 They are both nice guys so good luck to them, but the uncritical proliferation of their findings is further proof of the pleasure â even pride â we take in the idea that a man-made, robot-worked utopia/dystopia is on its way.
The hyperbole about AI has reached such proportions that even New Scientist (16 July 2016) recently asked âWill AIâs bubble pop?â The author makes the point, familiar to sociologists of science, about the powerful role of metaphors in persuading us that these machines are acquiring human capacities. Yet artificial neural networks do not âlearnâ like we do, âcognitiveâ computing does not think, and âneuralâ networks are not neurons. The language is purposefully saturated with anthropomorphism. Rather than worry about the dreaded moment of Singularity, we should be concerned about the dominance of a small number of corporations who have this computing power and about the social consequences thereof. Such political questions are too often lost in our obsession with the robotic revolution we are set to witness.
In the crystal ball of Susskind and Susskind, this imminent revolution is seen to be even more dramatic than the forecast of Ford. While Ford believes that higher education and healthcare professionals are relatively immune from automation, the authors of The Future of the Professions specifically include them in their sweeping diagnosis about the end of the professions as we know them. In the internet society, they argue, we will neither need nor want doctors, teachers, accountants, architects, the clergy, consultants or lawyers to work in the way they did in the twentieth century. Although this will lead to massive job loss, this trend is a positive development as the internet will ultimately democratise expertise and empower people.
With a nod to Abbott,5 they begin by outlining the historical basis of professionalism as the main way expertise has been institutionalised in industrial societies. Until now there has been no alternative, as only human professionals have had the complex combination of formal knowledge, know-how, expertise, experience and skills they refer to as âpractical expertiseâ. But now, echoing the books above, we are on the brink of a period of fundamental and irreversible change, driven by technology. The authors envisage increasingly capable machines â from telepresence to AI â that will bring a fundamental change in the way that the âpractical expertiseâ of specialists is made available in society. These smart machines, operating autonomously or with non-specialist users, will perform many of the tasks that have been the preserve of the professions. The result will be the âroutinisation and commoditization of professional workâ, an argument much like Bravermanâs proletarianisation thesis but without the political economy. Here the only actors are the machines themselves.6
Richard Susskind has been a leading analyst of the impact of technology on the legal profession for several decades, and he is a firm believer in the positive opportunities for information sharing afforded by the internet. And the bookâs core moral argument is persuasive. Who would disagree that expensive and exclusive privileged elites need to be overhauled and instead we should promote the widespread distribution of expert knowledge? Indeed, the authors envision a model where most professional advice is delivered by automated IT systems, and is available free to users (just like Wikipedia). Once again, we are told about the unprecedented acceleration in the capabilities of IT, AI, Watson, machine learning, Big Data and affective computing. The nub of the matter here, though, is the premise that intelligent machines, drawing on vast amounts of data, will make better decisions than do mere flawed human experts. The archetypal example is the lack of sound sentencing by tired judges after lunch. Perhaps non-alcoholic lunches would be a simpler solution!
The fundamental problem we have is that technologies are only as good as their makers. There is mounting evidence that machine-learning algorithms, like all previous technologies, bear the imprint of their designers and culture. Whether itâs Airbnb discriminating against guests with distinctively African-American names, Google showing advertisements for highly paid jobs primarily to men rather than women, or the use of data-driven risk-assessment tools in âpredictive policingâ, histories of discrimination live on in digital platforms and become part of the logic of everyday algorithmic systems.7 Even the much-lauded Wikipedia is skewed, in its representation of male to female scientists for instance. While the Susskinds are right to contest the power of the professions, they seem unconcerned with the rise of an even more powerful elite of male white Silicon Valley engineers whose values and biases will inevitably shape the technical systems they design. Making the politics of algorithms visible, explicit and accountable may turn out to be even more difficult than calling, say, lawyers to account.
I am with Brynjolfsson and McAfee who, in The Second Machine Age, argue that the most efficient future lies with machines and humans working together. Human beings will always have value to add as collaborators with machines. For a start, I do not believe that all the knowledge and experience, the âpractical expertiseâ of professionals, can be conveyed via online intelligent systems. Take the suggestion that even the problem of âempathyâ in delivering bad news in hospitals could be countered through an algorithm using consumersâ âpsychological and emotional profilesâ. Leaving aside the privacy issues this raises, the Susskinds do not grasp the nature of the âunrecognisedâ emotional work that is already delegated to largely female para-professionals such as nurses.
Indeed, the social character of skill and expertise, let alone the way that the professions have traditionally been structured around a gendered division of labour, gets no mention in this book (or in any of the others for that matter). We may be âsuckers for the wide eyes and endearing giggles of affective robotsâ, but to advocate the use of robots for empathetic care of the elderly mistakes the appearance of care with real empathy and genuine personal interaction. And anyway, as any roboticist will tell you, there is a huge chasm between the current claims about what these affective, sociable robots can technically feasibly do and what they really can do. Perhaps if eldercare was revalued and remunerated like, say, coding work, the putative labour shortages in this sector that robots are designed to alleviate would disappear. As they would if, more radically, housing and cities were redesigned so that the elderly were not relegated to separate places but were integrated into the wider civil society. But such thoughts are way beyond the scope of any of these books.
The Second Machine Age is the best of this bunch. While covering similar ground, Brynjolfsson and McAfee provide a much more balanced account of the pros and cons of automation on work. The book has been extremely influential, spawning a number of imitations (viz the Chair of DAVOS Klaus Schwabâs The Fourth Industrial Revolution). The titles of these books are themselves worthy of an article. Here, the history of technology starts with the Industrial Revolution (âthe first machine ageâ) and our interest in AI dates from the 1950s. If you want to remind yourself of how much older our obsession with the vitality of machines actually is, I suggest a quick visit to the webpage for the 2017 exhibition on Robots at Londonâs Science Museum.8
Brynjolfsson and McAfee are ultimately optimistic about the jobs that will be created as a result of the digital revolution. Although agreeing that many jobs will be swept away by innovations like the driverless car and 3D printers, they argue that, with the right policy levers, such advances can bring forth a bountiful future of less toil, more creative work and greater human freedom. Intervention is crucial given the worrying trends they identify: the polarisation of the labour market, the rise in income inequality and the âwinner-take-all economyâ. But, if we ârace with machines, instead of against themâ, we can take advantage of the uniquely human qualities of creativity, ideation and communication to create more high-quality jobs such as those of creative writers, digital scientists and entrepreneurs. While Brynjolfsson and McAfee also reify technology, treating it as a neutral inevitable force driving these changes, they are strong advocates of government investment in education and infrastructure to deal with its effects. For them, unlike Urry, the effects of technology are political but the causes are not.
Interestingly, like Ford, they propose a guaranteed basic income as one practical solution to the problem of technological unemployment. That this idea has once again become popular across the entire political spectrum makes me a little wary. It immediately conjures up a vision in which the Silicon Valley tech crowd continue to thrive on 24/7 working hours, while those left behind are paid to watch TV and sleep. This idea has a long and sound history and I am watching with interest the trials taking place in Finland and the Netherlands, for example. But in the current context, it is as well to foc...