Ebooka przeczytasz w aplikacjach Legimi na:
Odsłuch ebooka (TTS) dostępny w abonamencie „ebooki+audiobooki bez limitu” w aplikacji Legimi na:
STILL THINK ROBOTS CAN’T DO YOUR JOB?
Essays on Automation and Technological Unemployment
By Riccardo Campa
Copyright D Editore © 2018. All right reserved.
+39 320 8036613
1.2. Luddism: The First Reaction
1.3. Classical Political Economy: The First Denial
1.4. The Conversion of David Ricardo
1.5. Karl Marx: Beyond Economic Theory
1.6. The Marginalists: Mathematics versus Luddite Fallacy
1.7. The Keynesians: Technological Unemployment as a Fact
1.8. Reaganomics: The New Denial
1.9. Artificial Intelligence: The Specter of a Jobless Society
2.1. The McKinsey scenario
2.2. What is missing in this picture?
2.3. An alternative scenario
2.4. Education in a jobless society
3.1. Social robots and social work
3.2. Effectiveness and safety of human-robot interaction
3.3. Small-scale robots
3.4. From navigation and manipulation to interaction
3.5. Scenario and persona: the challenge of verbal interaction
4.1. Technology and Unemployment: The Vexata Quaestio
4.2. Some methodological tools for scenario analysis
4.3. The unplanned end of work scenario
4.4. The planned end of robots scenario
5.4. The unplanned end of robots scenario
4.6. The planned end of work scenario
4.7. An ethical judgment
5.1. Artificial Intelligence and Industrial Automation
5.2. Effects on the level of employment
5.3. Social stratification and the new generation robot
5.4. The need for a new socio-industrial policy
This is one of those books that one writes hoping to be wrong. The question with which I begin the book has recently been asked quite often. I ask it also to myself: Do I still think robots cannot do my job? My personal answer is simply “no”. Sooner or later, there will be robots that can teach and do science.
In spite of the fact that this is a collection of academic works, I ask my readers to allow me the indulgence of introducing the topic by offering a personal story.
I have always been fascinated by technologies, old and new, and especially by Artificial Intelligence and robotics. Not by chance, therefore, before turning into a social scientist I studied electronics. Still, I could never turn my back to the unwanted collateral effects of technological development.
When I was a teenager, I worked in a factory in summertime as a manual worker in order to pay for my studies. It was the 1980s, when the first wave of robotization was hitting Italian industries. I remember that every week a new machine was “hired” by my company, and a few fellow workers fired. Being seasonal workers we were not protected by long term contracts.
One day a computerized scale was introduced in my department. It was pretty obvious that it was there to do the job of my own team. I was at once fascinated and scared by that machine. On the one hand I was curious to see how it worked, on the other I knew it might lead to my firing. When the meal break started, by getting close to the machinery, I heard the boss saying that the hiring manager was looking but they had not yet found a worker who could supervise its functioning. So instead of joining my colleagues at the canteen, I started reading the instruction manual.
When the bell rang to signal the end of the meal break, I went to the boss and told him that I was a student of electronics and I knew how the scale worked. He was quite happy to have the machinery immediately in function, and I was happy to leave the physical work and turn into a supervisor. Even though I had to wait until late that evening to eat, I did not even feel hungry. I was proud of myself, and I thought my parents would be proud of me also, if they just could see me. I was just sixteen years old and it was only a modest seasonal job, but to me that “career advancement” meant a lot.
Still, what I predicted would happen, happened. My friends and colleagues were fired. I knew it was not their “fault”. Even if all of them did what I did—give up eating and study the instruction manual—only one supervisor was needed. The machine would have done the rest. I also know that some of those friends did not find a new job for long time.
This happened almost thirty-five years ago. It was my first experience with technological unemployment. By resorting to sociological jargon, I can say that my first knowledge in the sociology of work came from “participant observation.”
This probably explains why, once I became a professional sociologist, I focused so much on technology and future of work. I wrote much on these topics in both Italian and English. In this volume I present several of my works written in the English language. As often happens in a collection of essays published at different times, a few concepts and quotes are repeated. However, I wanted to leave the writings in their original form, as they were published by scientific journals. Here is a short description of the chapters.
The first chapter traces a brief history of the concept of technological unemployment. The historical narration covers four centuries, since the beginning of the industrial revolution up to the present. As a consequence, it is highly selective, mainly based on sources in the English language and referring to only a few of the many social scientists involved in the debate. The scopes of the inquiry are essentially two. The first is to show that focusing on technological unemployment as an idea – and not simply as a phenomenon – is appropriate, because of the high level of controversy that still characterizing the debate. The second is to provide an understanding of critical societal changes occurring in the twenty-first century.
The second chapter proposes a short-term scenario analysis concerning the possible relations between automation, education, and unemployment. In my view, the scenario analysis elaborated by the McKinsey Global Institute in 2013 underestimates the problem of technological unemployment and proposes an education model which is inadequate for handling the challenges of twenty-first century disruptive technologies. New technological advances – such as the automation of knowledge work – will also affect the jobs of highly educated workers. Therefore, policy makers will not avert massive unemployment only by extending the study of math, science, and engineering. A better solution could be the establishment of a universal basic income, and the elaboration of an education model capable of stimulating creativity and the sense of belonging to a community.
In the third chapter I explore the most recent literature on social robotics and argue that the field of robotics is evolving in a direction that will soon require a systematic collaboration between engineers and sociologists. After discussing several problems relating to social robotics, I emphasize that two key concepts in this research area are scenario and persona. These are already popular as design tools in Human-Computer Interaction (HCI), and an approach based on them is now being adopted in Human-Robot Interaction (HRI). As robots become more and more sophisticated, engineers will need the help of trained sociologists and psychologists in order to create personas and scenarios and to “teach” humanoids how to behave in various circumstances.
The aim of the fourth chapter is to explore the possible futures generated by the development of artificial intelligence. The focus is on the social consequences of automation and robotization, with special attention on the problem of unemployment. To start, I make clear that the relation between technology and structural unemployment is still hypothetical and, therefore, controversial. Secondly, as proper scenario analysis requires, I do not limit myself to predicting a unique future; instead I extrapolate from present data four different possible scenarios: 1) unplanned end of work; 2) planned end of robots; 3) unplanned end of robots; and 4) planned end of work. Finally, I relate these possible developments not just to observed trends but also to social and industrial policies presently at work in our society which may change the course of these trends.
The aim of chapter five is to determine if there is a relation between automation and unemployment within the Italian socio-economic system. Italy is second in Europe and fourth in the world in terms of robot density, and among the G7 it is the nation with the highest rate of youth unemployment. Establishing the ultimate causes of unemployment is a very difficult task, and – as we said – the notion itself of ‘technological unemployment’ is controversial. Mainstream economics tends to correlate the high rate of unemployment in Italy with the low flexibility of the labor market and the high cost of manpower. Little attention is paid to the impact of artificial intelligence and robots on the level of employment. With reference to statistical data, we will show that automation can be seen at least as a contributory cause of unemployment in Italy. In addition, we will argue that both Luddism and anti-Luddism are two faces of the same coin both focusing on technology itself (the means of production) instead of on the system (the mode of production). Banning robots or ignoring the problems of robotization are not effective solutions. A better approach would consist in combining growing automation with a more rational redistribution of income.
The sixth chapter explores a more remote scenario, namely the hypothesis that machines could sooner or later “wake up”, become conscious, and have a role also in the pursuit of knowledge. It is a scenario analysis that often goes under the label of “transhumanism” and predicts the advent of the Singularity. Since the industrial revolution, humans have tended to reduce science to the ancillary role of an engine of technology. But the quest for knowledge using rational, scientific methods started at least two and a half millennia ago with the aim of setting humans free from ignorance. The first scientists and philosophers (at least that we know about because they wrote things down) saw knowledge as the goal, not as the means. The main goal was to understand the nature of matter, life, conscience, intelligence, our origin, and our destiny, not only to solve practical problems. Being skeptical of myths and religions, they gave themselves the goal to reach The Answer via rational and empirical inquiry. Transhumanism is a unique philosophy of technology because one of its goals is the creation of a posthuman intelligence. Several scientists share this hope: Making technology an ancillary of science, and not vice versa. By evolving and reaching the Singularity, the hope is that posthumans can achieve one the greatest dreams of sentient beings: finding The Answer.
In the seventh and last chapter I address the role of serendipity in the development of science and technology. It is a review of Robert K. Merton and Elinor Barber’s book The Travels and Adventures of Serendipity. A Study in Sociological Semantics and the Sociology of Science. Although this book does not treat the issue of technological unemployment directly, it critically discusses the orthodox Marxist theory on automation. According to this theory, scientific and technological discoveries are products of necessity. Industrial automation could not be developed in ancient times, because of the slavery mode of production. The cost of manpower was very low, so there was no need to produce machines. In the capitalistic mode of production characteristic of modern times, however, slaves are not available, so machines can fit the bill. While there is truth in this narrative, it is an oversimplification, because – as Merton and Barber convincingly argue – many scientific discoveries and technical inventions depend on chance and serendipity. Indeed, the fact that Heron and other Alexandrine engineers already projected and built automatons in ancient times does not fit Marxist theory.
If we ask common people if we need conscious computers and robots, the answer would probably be mixed, with – I guess – a majority against the idea. Personally, I am not prejudicially against the idea, but I think we should also take into account the possibility that conscious Artificial Intelligence may emerge from a serendipitous discovery, in an unplanned way and regardless of its social necessity.
In fact, as the development of Artificial Intelligence progresses, we may ask whether serendipity – intended as the capability of making fortuitous discoveries, or the ability to find something while we are looking for something else – will be a virtue we share with our mechanical children, or whether it will be a factor – maybe THE factor – that continues to differentiate humans from intelligent machines. In any case, we should consider the role of serendipity when we reflect and speculate about the future of work.
I am particularly indebted to Alan Sparks for his editorial contributions to various parts of this book. Besides being an award-winning non-fiction writer, Alan is also an accomplished computer scientist and astute social observer, and discussions with him have been very stimulating also with regard to content.
I am grateful also to Catarina Lamm for having translated from Italian into English chapters four and five of this book , and to Matt Hammond and Lucas Mazur for having proofread other fragments of the book. It goes without saying that any remaining inaccuracies in the facts or in the style are my own.
With regard to the title of this book, I have to credit Nikhil Sonnad, who published a press article in digital magazine Quartz entitled “Robot all too robot. Still think robots can’t do your job? This video may change your mind” (2014). After struggling to find a title, and after realizing that all the titles I was thinking of were already used for other books, I decided to borrow and reuse a fragment of that article title.
I also thank the readers of La società degli automi, a book written in my native language, partly covering the same topic but still more focused on Italian issues, that rapidly became a bestseller in Italy. Without the positive feedback of the public, my Italian publisher would have probably hesitated to print a second book on automation and technological unemployment in English. It will be a challenge also to D Editore to cross the borders and promote this book worldwide. So, the last thanks goes to Emmanuele Pilia for accepting the challenge.
The concept of technological unemployment is regaining momentum in the discourse of economists and economic sociologists. However, when analyzing the debate, what is most surprising is the substantial absence of agreement on the very existence of technological unemployment as a phenomenon. Some observers present technological unemployment as a sprawling monster that is completely subverting the global economy, while others conclude that this picture is just a mirage of doomsayers. Since reputable scholars are engaged in the debate, we cannot simply blame the polarization of narratives on the incompetence of one or the other school of thought. Even if the definitions of technological unemployment provided by different sources do not differ particularly, it has become evident that the terms contained in these definitions may assume different meanings depending on the theoretical perspective.
Unemployment is a phenomenon studied by both sociologists and economists. As Tony Elger (2006: 643) remarks, “[s]ociologists often focus on the experience and consequences of unemployment, leaving economists to analyze causes. […] However, consideration of the underlying processes that generate these patterns of unemployment exposes continuing controversy among economists, for example between neoliberal, neo-Keynesian, and neo-Marxist analyses of the political economy of contemporary capitalism. Thus, economic sociologists have to adjudicate between these different causal accounts [...]”
Unemployment is a complex phenomenon. “Economists distinguish between frictional unemployment, involving individual mobility of workers between jobs; structural unemployment, resulting from the decline of particular sectors or occupations; and cyclical unemployment, resulting from general but temporary falls in economic activity” (ibid.). To this list, one can add technological unemployment.
The Oxford Dictionary of Economics defines technological unemployment as follows: "Unemployment due to technical progress. This applies to particular types of workers whose skill is made redundant because of changes in methods of production, usually by substituting machines for their services. Technical progress does not necessarily lead to a rise in overall unemployment” (Black 2012: 405). As one can see, it is a concept that already includes a theory, since it puts into causal relationship two distinct phenomena: technological progress and unemployment. The disagreement between the different schools of thought mainly concerns the existence of this causal relationship.
Technological unemployment can be studied at different levels of the economic system: at the level of individual actors, companies, productive sectors, countries, or global economy. That at least one individual has lost his job because the employer or the customer has purchased a machine that can accurately perform his/her duties is a fact that can hardly be denied. Similarly, it cannot be denied that entire companies have been automated and this process has resulted in a drastic reduction of employment inside the company. As well as it cannot be denied that, owing to technological innovation, entire economic sectors have been largely emptied of their workforce. The transition from traditional agriculture to intensive agriculture, through the use of agricultural machinery, herbicides, fertilizers, fungicides, etc., has led to demographic emptying of the countryside. The evaporation of jobs in the primary sector of the United States of America offers impressive numbers: in 1900 41% of the population was employed in agriculture, a century later, in 2000, only 2% of Americans still worked in same sector (Wladawsky-Berger 2015). A similar phenomenon was observed in the secondary sector, or manufacturing, at the turn of the twentieth and twenty-first century. In the United States, the ratio between employment in the factories decreased from 22.5% in 1980 to 10% today and is expected further decline to below 3% by 2030 (Carboni 2015). Similar situations can be observed in other industrialized countries, including Italy (Campa 2014a).
This emptying of whole sectors of the economy was accompanied by a migration of the workforce from one sector to another. A first migration was observed from agriculture to manufacturing, a visible phenomenon because it also led to a massive migration from rural to urban areas. A second migration of the labor force, less visible but equally significant, occurred from the manufacturing sector to the services sector (Campa 2007). Overall, at least so far, the increase in productivity in individual sectors has not resulted in the emergence of a permanent and chronic technological unemployment on a global level. This does not mean, however, that technological unemployment – at least as a temporary or local phenomenon – does not exist.
It should also be clear that the reabsorption of the unemployed into the economy has been possible thanks to two main levers: the first is free market, which enabled the birth and development of new sectors of the economy; the second is social and industrial public policies. The fact that both forces are at work is often obscured by the fact that observers are largely divided into two tribes: those who worship the Market as an almighty God, and those who attribute an analogous divine character to the State. Only those who do not profess either ‘religion’ can see that many factors have contributed to dampen the phenomenon of technological unemployment. Private entrepreneurs have created manufacturing industries and used the cheap labor flowing from countryside to city, in the nineteenth and early twentieth century. New enterprising capitalists have created service companies to redeploy manpower pouring out from factories, in the second half of the twentieth century. At the same time, trade unions and socialist political parties, through tough political and labor struggles, have succeeded in achieving steady reduction of working hours (even a halving of working hours, if we consider the period from the nineteenth century to the present), retirement and disability pensions, paid holidays, paid sickness, maternity leave, and other social rights, which on the whole have forced private employers to hire more workers than they would have hired in a laissez-faire capitalist regime.
Moreover, the idea that the equilibrium of a national economy is assured by the Invisible Hand is belied by the fact that employment crises have sometimes been resolved by the mass migration of workers from one country to another. This means that it is not written in the stars that capable private entrepreneurs and creative people who create new jobs, new companies, or even new economic sectors must continually rise. If they do not rise, if there are no social and cultural conditions that permit them to arise, the unemployment crisis generated by the introduction of new technologies can become chronic and irreversible in a specific geographical area. Finally, other forms of public intervention, such as industrial policies, have contributed to cushion the phenomenon. For instance, the creation of public manufactories, the nationalization of private companies, public contracts (just think of the incidence of military spending in the United States), wars, crime (the prison population in the US now exceeds two million individuals), as well as the creation of millions of jobs in the public service – jobs that are sometimes unnecessary and therefore constitute a permanent masked dole.
If you consider all these aspects, some of which are ignored by economic theory, it seems difficult to deny the existence of technological unemployment. Somewhat different is the question of whether it is a significant phenomenon on a global scale. From the psychological point of view, being replaced by a machine is certainly a big concern for those who lose their jobs, even temporarily. But the issue begins to acquire political relevance only if the proportion of individuals affected by the phenomenon is likely to disrupt an entire economic system. Throughout history, different moments when the phenomenon of technological unemployment has assumed critical proportions were observed. In these periods, the idea of technological unemployment has gained major relevance in the public debate.
Notoriously, a rather critical moment in European history was the transition from feudalism to capitalism, and not only for bloody political revolutions that accompanied the transformation. In the so-called feudal system, the creation of work did not constitute a problem, because social mobility was minimal. Children inherited the job of their fathers. The children of the farmers knew that they would be farmers themselves, or serfs. The children of the artisans learned their profession in the workshops of their fathers. The eldest son of an aristocratic family inherited the family estate, while his younger brothers were initiated in a military or ecclesiastical career. Daughters would be wives of men chosen by the father, or nuns. Beggars, robbers, vagabonds, prostitutes, and adventurers formed exceptions to the strict rule. In the Middle Ages, others were the economic concerns: wars, epidemics, famines. A serious problem that could arise was rather labor shortages as a result of these phenomena.
With the transition to capitalism, previously unknown problems arise: in particular, overproduction and unemployment. The introduction of machines in the production system and social mobility disrupt the traditional conception of work and life. To many, it appears inconceivable that someone willing to work cannot find a job. So much so that the first reaction of the political authorities is to limit the use of the machines where cause unemployment. Even mercantilist Jean-Baptiste Colbert, who gave great impulse to the industrialization of France by the creation of so-called Manufactures nationales, passed measures to restrict the use of machines in private companies.
Where the authorities do not intervene, the workers themselves may make a fierce and desperate struggle against the machine, of which we find a detailed account in Capital by Karl Marx (1976: 554-555): “In the seventeenth century nearly all Europe experienced workers' revolts against the ribbon-loom, a machine for weaving ribbons and lace trimmings called in Germany Bandmühle, Schnurmühle, or Mühlenstuhl. In the 1630s, a wind-driven sawmill, erected near London by a Dutchman, succumbed to the rage of the mob. Even as late as the beginning of the eighteenth century, saw-mills driven by water overcame the opposition of the people only with great difficulty, supported as this opposition was by Parliament. No sooner had Everett constructed the first woolshearing machine to be driven by water-power (1758) than it was set on fire by 100,000 people who had been thrown out of work. Fifty thousand workers, who had previously lived by carding wool, petitioned Parliament against Arkwright's scribbling mills and carding engines. The large-scale destruction of machinery which occurred in the English manufacturing districts during the first fifteen years of the nineteenth century, largely as a result of the employment of the power-loom; and known as the Luddite movement, gave the anti-Jacobin government, composed of such people as Sidmouth and Castlereagh, a pretext for the most violent and reactionary measures. It took both time and experience before the workers learnt to distinguish between machinery and its employment by capital, and therefore to transfer their attacks from the material instruments of production to the form of society which utilizes those instruments.”
David F. Noble (1995: 3-23) maintains that the Luddites are not to be considered technophobic. When the machinery was introduced in manufactures, the workers destroyed it because of necessity, not because of technophobia. Their choice was limited to three options: 1) starvation for them and their families; 2) violence against the uncompassionate owners of the means of production; 3) destruction of the means of production. Choosing the third option was the mildest way to communicate their discomfort as regards unemployment.
The reaction of the political authorities was clearly less mild. Such was the incidence of the phenomenon that the English government implemented the death penalty for Luddites. The ‘assassination’ of a machine was put on a par with the assassination of a human being.
In spite of the fact that the appearance of machinery produces worrisome social disorders, economists are reluctant to modify their theories in order to make place for technological unemployment. There are just a few exceptions. For instance, an attempt at conceptualization is found in James Steuart’s book An Inquiry into the Principles of Political Economy (1767), and precisely in chapter XIX (“Is the Introduction of Machines into Manufactures prejudicial to the Interest of a State, or hurtful to Population?”) . Steuart admits that the sudden mechanization of a segment of the production can produce temporary unemployment and, therefore, public policies are needed to facilitate the absorption of the labor force into other tasks. He is still persuaded that the advantages of mechanization outweigh negative side effects, but is also convinced that problems do not solve themselves. However, that of Steuart is an isolated voice.
Classical economics is dominated by Adam Smith’s optimistic perspective, which emphasizes the positive effects of mechanization and the self-regulating nature of market economies. In his masterpiece An Inquiry into the Nature and Causes of the Wealth of Nations, he provides evidence of a causal connection between high taxation and unemployment (Smith 1998: 1104), or excessive prodigality of the landlords and unemployment (Smith 1998: 448-449), rather than between the use of machinery and unemployment. Machinery is mainly seen as a means to increase the productivity of laborers: “The annual produce of the land and labour of any nation can be increased in its value by no other means but by increasing either the number of its productive labourers, or the productive powers of those labourers who had before been employed. […] The productive powers of the same number of labourers cannot be increased, but in consequence either of some addition and improvement to those machines and instruments which facilitate and abridge labour; or of a more proper division and distribution of employment” (Smith 1998: 455-456).