No, Automation Isn’t Going to Make Work Disappear

Juan Sebastian Carbonell

Talk of a great technological replacement suggests that automation is rendering most workers obsolete. But innovation isn’t simply replacing human workers — rather, it’s created a battle over whose interests the new technologies will serve.

The problem with the transformation of work today is less that new technologies could eventually replace workers but that they are used to degrade working conditions, keep wages stagnant, and flexibilize working time. (Lenny Kuhne / Unsplash)

Interview by
David Broder

The COVID-19 lockdowns spurred renewed discussion of which jobs are necessary — and how far our societies remain centered around work at all. With the so-called “Great Resignation,” this took the form of a growing refusal to accept working dull jobs for poverty wages. Yet, in many “techno-optimist” accounts, workers’ power to accept or refuse employment is in any case on the decline. They claim that artificial intelligence and automation are bringing an unprecedented wave of redundancy — demanding, in turn, that we find other ways to guarantee citizens a stable income.

But, French labor sociologist Juan Sebastian Carbonell insists, the claim that new technology is replacing the need for a human workforce is shortsighted — and, in fact, a myth as old as capitalism itself. In his new book Le futur du travail, he argues that work is not disappearing but being transformed, with the material consequences of new technology, outsourcing, and subcontracting shaped both by management plans and worker resistance.

Jacobin’s David Broder spoke to Carbonell about the myth of the “great technological replacement,” the resilience of the workforce at a global level, and the bases of class identity in postindustrial economies.


DB

Your work resists the idea that we are seeing a “great technological replacement” of human labor. Why is that — and what do self-checkout machines tell us about this?

JSC

First, because it’s false: the technological replacement isn’t happening. I take the example of self-checkouts, because it was controversial in France in the early 2000s. The CFDT (Confédération française démocratique du travail, French Democratic Confederation of Labor) campaigned against them, saying that they’d be replacing supermarket checkout workers. Trade unions are sometimes themselves the victims of this illusion: the capitalist mythology of a “great replacement” of labor by machines. Yet, twenty years after their introduction, self-checkouts are present in only 57 percent of supermarkets in France, and where they are present, they are in addition to, not instead of, conventional checkouts with human cashiers. They’re also not ever so automatic: there are still cashiers to supervise and help out the customers, though their tasks have changed.

So, the book tries to question this common sense. For me, the problem with the transformation of work today is less that new technologies could eventually replace workers but that they are used to degrade working conditions, keep wages stagnant, and mount a major flexibilization of working time.

DB

You explain that “labor-saving technology” doesn’t remove the need for workers in general but changes the way that work is organized.

JSC

When you look at the concrete effects of new technology in work, you immediately see consequences that cannot be reduced to “replacement” alone. There is a substitution of specific tasks, which does not get rid of the jobs entirely.

Management’s goal can also be to deskill the workforce, so one worker is more easily replaced with another. This is linked with Harry Braverman’s thesis on the degradation of work in the twentieth century, which continues in the twenty-first. It says that when management introduces new technology, it does so to increase its control over the work process. I saw this a lot in the auto industry when they introduced so-called “digital manufacturing” in assembly plants: as one technician put it, the aim is that anybody can do anything in the workplace. Another important — albeit not necessarily deliberate — result is that you also need a reskilling of workers: some new machines will need programmers and maintenance workers, and this also means something of a transformation of skills within the workforce.

For example, in the context of the so-called Fourth Industrial Revolution and digital manufacturing, we are also seeing the appearance of data scientists inside factories. It’s still difficult to know how many jobs this represents. But sometimes you do need to hire different kinds of workers to do different kinds of things in the changing labor process.

This deskilling also brings an intensification of the work process. This simply means more work for fewer people with the same skills, during the same work time — something that is very common in the dynamics of technological change.

A further important consequence — usually overlooked by researchers, journalists, or so-called experts on new technology — is increased control and surveillance in the workplace. Management can more easily, for example, follow workers with GPS and control their movements, as in the case of couriers and delivery personnel.

I use the example of the Daily Telegraph, where management introduced motion sensors so that they knew whether workers were in front of their computers. You see this also in industry, again with the introduction of digital manufacturing, for instance by registering the moment that the machine stops running, so management knows if workers took a longer break. More generally, the use of data in manufacturing has the consequence of rendering the work process more transparent to the eyes of management. It thus knows when machines are being used, to what intensity, when are they going to break, when or how are they being used, and so on. MES (manufacturing execution system) software allows it to centralize all this information and have a clearer view of the work process. The apologists for the new technologies present this as progress, but they don’t talk about its negative effects on workers.

DB

You tell us that some defenders of the “great technological replacement” thesis admit that previous waves of technological innovation reorganized rather than replaced the workforce, but insist that with the Fourth Industrial Revolution and artificial intelligence, this changes. To give a concrete example, I’ll save time on writing up this interview by using an AI-transcription service, but it’ll still take hours for me to edit the results. So, what actually is different?

JSC

Experts and futurologists are conscious that past waves of automation have not brought about the end of work. So, they try to see what is specific about the new wave of automation today. And they say that it’s going to bring a further, final rupture in capitalism, finally bringing work to an end.

The title of my book, Le futur du travail, has a double meaning in French: the debates on the future of work refer to the future, but for me they resemble the past. I emphasize the continuities between past waves of automation and the current one, saying that robotization and software in factories and offices had more or less the same consequences as the introduction, for example, of digital technologies in the factory or of AI today.

While AI is presented as the technology that is going to put an end to white-collar as well as blue-collar work, it is still very limited in its uses and potentials. The programmer still has a central role not only in the creation but also in the day-to-day running of the AI: they provide its architecture, its learning method, the data that feed it, and so on. So, we’re a long way from a technology that functions entirely autonomously. Some researchers talk about AI as a general-purpose technology, meaning that it’s a technology whose uses are becoming more widespread but whose consequences are still to be determined. But they will be so general that they’ll probably just create as many jobs as they’ll contribute to destroying. More specifically, artificial intelligence is not independent of workers, programmers, and developers working on this software.

So, behind the apparent novelty of these new technologies that are presented as a major disruption to capitalism, actually, there is the same continuity, the same logics of substitution, deskilling, and intensification of control. For example, translation software will replace some elements of translators’ work but could also make some specialties more valuable.

DB

I’d like to turn to the idea of “labor society.” As you tell us, overall employment levels haven’t dropped away, especially since the “full-employment” era included such a high nonactive population, for instance in domestic work. But nonetheless, it is widely claimed that the world of work is now so fragmented that it is impossible to speak of the common condition of “being a worker,” often characterized in terms of a now-bygone Fordist era.  Your work challenges this view. Why?

JSC

I try to historicize the discussion on the overwhelming precarity in today’s workforce, saying it is not necessarily so new but partly a product of the feminization of the workforce. If there has been a rise in unemployment in France (1 percent in 1968 to 10 percent in 2015), it is also because there is a drop in the “inactive” population, mainly women (who were domestic workers), who dropped from 27 percent in 1968 to 11.5 percent in 2015. If there was full employment during the “Trente Glorieuses” — the three decades after 1945 — it was also because the nonemployment rate (i.e., those neither employed nor seeking paid work) was high.

In a sense, precarity has been functional to capitalism since its beginning: what was new in the postwar era was the idea that capitalists would need, and work toward having, a more stable workforce. I try to develop this idea in the book. Even forms of work such as Uber and self-employment on digital platforms are not new. Self-employment mediated by other people has always existed in capitalism: I give the example of coal miners in turn-of-the-century France, who worked in a system like Uberization, where some were hired not by the bosses or directly by the mine but by middlemen, called “butties,” who paid them by the cartload — something Émile Zola describes in Germinal. The difference today is that this middleman is an impersonal digital platform.

So, I’m trying to say that this common experience of work never really existed. In the debate in France, there’s a phrase I like: “The working class is no longer what it never was.” It never was this person hired on a stable job, working 9-to-5 as Dolly Parton put it.

More generally, I tried to say that labor remains central, in two ways, despite the changes. The first is that it remains the main way in which society produces and reproduces itself. And it remains central, also, in a social way. It remains central in the sense that work is also a social order incarnated in the wage-labor form and also in social representations, which give work its centrality in the life of individuals. Today, despite claims of generalized precarity, permanent employment is the norm in France: 75.2 percent of the whole workforce is employed on a permanent contract.

What changes with COVID is that in these representations, where work remains central, work is no longer idealized. With the “Great Resignation,” we see that people are more disillusioned with working conditions, with wages, and with work flexibility today. But looking at the number of resignations, at least in France, they have risen only because they are catching up with resignations that otherwise would have happened during the lockdowns, and because when there is growth, when accumulation begins again, there is more internal job mobility. For example, in France in late 2021, there was more hiring than resignations, while hiring on permanent contracts currently exceeds pre-COVID levels in late 2019.

DB

You say that “deindustrialization is everywhere except the statistics,” especially outside of Western Europe. To take a major European industrial power like Italy, there is an experience of a couple of generations — roughly bounded by World War II and the 1980s, that saw the reemergence, rise, and fall of the labor movement — followed by large-scale deindustrialization and outsourcing, even if some important industries remain. Clearly, these combined factors provide a material basis for the narration of the “death of the working class.” But looking beyond the Global North, how real is this, and how far is this even a dominant discourse?

JSC

The “death of the working class” is an interesting question because one of the main reasons instigating the discourse on the end of work is deindustrialization and the end of blue-collar workers. What this misses is the globalization of value chains, the regionalization of industries, and the fact that — to take one prominent industry — there are today more auto workers worldwide than there were thirty years ago: far fewer in Italy, France, or the UK, but far more in China, India, and Latin America. Employment in the auto sector rose worldwide by 35 percent between 2007 and 2017. Take China, where employment in the sector rose 68 percent, to roughly 5 million workers in 2017, or Mexico, where employment doubled during the same period. At the same time, employment in the auto industry in France declined from 280,000 to 190,000 in the same period. That’s without taking into account the emergence of a battery value chain, whose effects on industrial employment are to be determined.

So, the “death of the working class” discourse is a Global North narrative, blind to the economic transformations of world capitalism. I use the theoretical framework of Beverly J. Silver, who says that capital faces two opposing forces. The first is the profitability crisis: capital searches for new countries where the labor force is cheaper, and new industries where it can invest, to counter the tendency of the rate of profit to fall. The second force is working-class organization. That’s why it always seeks out “disciplined” and “peaceful” working classes in Global South countries. But it also creates the same contradictions in these other countries. So, while it invests in creating new industries and new working classes in other countries, it also creates new labor conflicts and demands.

In this sense, alongside the deindustrialization of Northern countries, there is an industrialization of Southern and Eastern ones. Slovakia has more production of vehicles per person than any other country in Europe. Then, in northwestern Europe, you have a tendency also to create new hubs of industrial workers; in the book I give the example of logistics, which is one of the fastest-growing sectors of industrial work in the rich countries. There is a small boom in the number of workers in this sector, where jobs are usually manual, very unskilled, and very poorly paid. In France, you have now 800,000 blue-collar workers in logistical hubs in the peripheries of big cities. One can also think of the UPS Worldport in Louisville, Kentucky, with 20,000 employees. This again reflects the idea that where capital invests, labor conflicts emerge. You have seen this in France, and you have seen this in northern Italy, where there was a wave of strikes by migrant workers in logistical hubs. That is a direct consequence of this development of logistics as an industrial sector, as was the auto industry a few years ago.

Alessandro Delfanti said that Amazon is the new Fiat. I’m not sure if I entirely agree with that, because Fiat, now Stellantis, still exists. But the configuration of the workforce is somewhat similar. It means that there are young, unskilled, migrant, poorly paid, and highly concentrated workers in these new logistical hubs. And this is an explosive cocktail in a certain sense for the organization of the working class, and it could be a possible source of renewal also for the labor movement today.

DB

When we think of Fiat Mirafiori or Renault at Boulogne-Billancourt — historic “red fortresses” of organized labor — their significance wasn’t just the number of workers they employed or even the supply chains attached to them but also a certain symbolic importance as “national champions” and centers of industrial modernity that the workers battled to control. Today, there are bigger economic sectors that employ more people (e.g., tourism in Italy), but they don’t seem to have the same aggregating role, as a possible focus of class identity or vision of where power in society lies.

JSC

Yes, I see, and there are autonomists who say that “power is logistic, block everything!” and thus the logistics sector has a strategic importance in worker organization and overthrowing capitalism, if only we built stronger, non-bureaucratic, et cetera unions there. I think this is both true and untrue. Logistics has the particularity that you can’t easily do offshoring, simply because of how it works. At the same time, you have a real deindustrialization of Western Europe and the rich countries in general.

But when you look back at the history of the workers’ movement, you see that some sectors that were at the forefront were not necessarily more concentrated or strategic. Erik Olin Wright distinguishes associational power — the power that comes from the collective organization of workers — and structural power, which comes from the location of workers in the capitalist economy. The leading role wasn’t always played by the ones able to stop the whole economy from running.

I take the example of mid-nineteenth-century France, where shoemakers were the most subversive workers: they had a strong associational power and were extremely organized. It is estimated that 4 percent of the people arrested for resisting Louis-Napoleon Bonaparte’s coup d’état in 1851 were shoemakers. They also created one of the first trade unions in Paris, in 1866. And many of the elected officials of the Paris Commune were also shoemakers, including Auguste Serraillier, who was informing Karl Marx about the events unfolding in Paris.

This leads me to the fact that, for example, platform workers don’t necessarily have the same structural power as, say, logistical workers, but there are similar dynamics of capital investment, work, and concentration. They have also been at the forefront of many struggles in France, in Italy, in Germany, and so on and the renewal of part of the workers’ movement in Western Europe. That is why I say that in these new sectors you find not only the same logics of capitalism in the organization of work but also the same logics of conflict.

DB

In your book, you relativize the real extent of Uberization but also cast the “platform economy” more as a pro-market fantasy than a viable economic model. Often, we see politicians dazzled by the idea that digital platforms are somehow not quite real and not bound by the “old” regulations, and in this sense they are a Trojan Horse for undermining labor conditions. But you also point to some material limits to this model’s expansion.

JSC

Yes, the first thing is to understand in relative terms the numerical importance of work for digital platforms. There are widely varying estimates: in France, between 1 and 6 percent of the population. There is a particular status called “microemployment” — not all the self-employed, and not including professionals who have their own firms like doctors and lawyers, but people that work, among other things, for digital platforms — and this only represents 2.8 percent of the total workforce. So, I wanted to first note the reality of these numbers.

Then, there are the material limits. The first is the problem capitalists have in establishing a reliable workforce. With this kind of work organization in digital platforms, sometimes working conditions are so bad that workers will not be loyal to the platform — for it also isn’t loyal to the workers. So, you have a lot of turnover, sometimes creating a problem in terms of the continuity of service.

The other problem is what Marx refers to as an anti-economic consumption of the workforce, from the point of view of capital: because working conditions are so bad, workers are extremely tired, and turnover is very high. For instance, some Uber drivers are doing up to sixty hours a week. This creates conditions in which the workforce is exhausted, again creating these problems of continuity of service.

The last problem is viability of this economic model. Take Uber: it’s not economically viable, except in certain very big cities like London, Paris, and New York, and in general, it loses a lot of money. This is also the case with Deliveroo.

This means some digital platforms in fact heavily depend on public subsidies. In France, there is a platform for domestic services like housecleaning, hairdressers, and sports coaches, called Wecasa. The state subsidizes clients for certain services, like housecleaning or childcare, to make it economically viable. If not, the platform couldn’t possibly pay workers sufficient incomes. So, they also can be attracted to it, in competition with other existing domestic services platforms, because it is artificially kept afloat by the state.

DB

It would seem, though, that while some of those workers could push back against these conditions by demanding recognition as employees — for instance, in Uber or Deliveroo — unionization would be a harder challenge when it comes to, say, Fiverr, when the job tasks are so divided up as to destroy the professional identity.

JSC

I see what you mean, though in the book I talk about the example of microworkers for Amazon Mechanical Turk (MTurk). This is the most fragmented workforce you can imagine: people working for a few minutes at a time for a digital platform, going “click, click,” in a completely anonymous way on the internet. But despite this extreme fragmentation, some have managed to organize not exactly in unions but in forms that resemble the “fraternal societies” that existed in nineteenth-century England. This was done also through the digital tools that capitalism has given us, turned against their masters in a certain way — using, for example, the forums that Amazon put online. They began discussing among themselves, saying we need to do this or that.

It’s funny because it really resembles the way in which workers used to think in the early nineteenth century, saying, for example, “We should write letters to Jeff Bezos saying, this is our situation, you need to change this situation. Could we be paid more? Could you make more transparent the way salaries are determined?” and so on. So, they created this association around their demands. Then Amazon closed down the forum, so they decided instead to organize outside of Amazon and create their own cooperative, which has a more transparent way of functioning, which was one of the main criticisms of MTurk.

DB

As you refer to there, a theme of the book is that technology doesn’t simply produce negative outcomes — there’s a struggle over the effects, in which the technology is itself a tool. But what good examples can we point to where unions or workers’ organizations have been proactive in outlining how technological advance might be used in a socially useful way?

JSC

I’d struggle to think of any example of a workers’ organization that has put forward an original idea regarding technology, except the ones I mentioned of unions being skeptical about these new technologies. Indeed, workers are usually right to be skeptical about them, because when they’re implemented they have the negative consequences I described.

Thirty years before the CFDT ran this campaign against self-checkout machines, they did an inquiry with sociologists and published a book, very important at the time, called Les Dégâts du progrès (The Damage Progress Brings). It was a very interesting reflection on the use of technology in the workplace, basically saying that, of course, technologies are bad for workers when they are in the hands of the bosses, but they could be something different.

This is the idea I try to defend in the book. Technologies are not themselves emancipatory, as when they are in the bosses’ hands, they are part of the subordination of workers. But if technologies are accompanied by an emancipatory political project, they can have the exact reverse effect. This is why I ask why unions don’t demand public investment in bettering working conditions through new technologies in the workplace. When I talk to workers in the car factories where I’ve done research, they said of course they’re not fundamentally against new technologies if they free them from this or that physical task.

But the problem is, historically, the Left and the workers’ movement have demanded that automation mean the reduction of working hours. This demand is no longer enough: for reducing working hours has been systematically accompanied by the flexibilization of working time. In France, the thirty-five-hour week was seen positively by unions, but the scholarly literature argues that bosses used it to make jobs more flexible and intensify the workload, since it was calculated as an average over the course of a year, meaning that sometimes workers were doing six or seven days a week and furloughed at other times, creating an impossible situation for their families. People who call for a thirty-two-hour week haven’t properly taken stock of what happened with the thirty-five-hour week.

DB

One proposed solution to the effects of automation is Universal Basic Income (UBI). You reject the claim — the supposed disappearance of work — that often serves as the justification for the demand. But I’d be interested to dig down deeper into why you think the Left should reject the demand. Your book cites Daniel Zamora, who argues that UBI fits the logic of poverty reduction rather than the kind of equality and democratic control exerted by organized workers. But why isn’t a minimum income at least a foundation to build on?

JSC

The problem with UBI is that it rests on a premise that is normally false. Usually, the defenders of UBI do so either because they believe there is a precariat replacing the proletariat, or because they think that automation will eventually lead to a workless future and so we need a solution for all these people that will be eventually jobless.

But for me, the problems with UBI are more political. The main reason that I draw a lot on Zamora is that this demand narrows the horizons of the Left and the workers’ movement. If it used to demand the abolition of wage work, the abolition of the state, a classless society, the socialization of the means of production, UBI is just a redistributive measure binding the Left within a budgetary calculus.

Then the other problem is that it replaces the collective force of workers, with a personal and individualized — but also anonymous — relationship with the state. That’s why for me UBI is not only a problem but also could be dangerous for the Left, as a solution to the so-called crisis of work today.

DB

Many of your lines of argument seem to locate your work as a critique of autonomism, but these talking points are hardly unique to that milieu. For instance, the idea of the death of the working class also informs a certain left-populism that seeks to regroup the precarious and atomized under the banner of “the people.” Why do you think these ideas are so common?

JSC

Yes, the autonomist left shares a lot of misconceptions on labor with a certain “common sense” on the replacement of workers by machines or about the emergence of a “precariat.” This is also why I included a discussion of Marx’s famous “Fragment on Machines” from the Grundrisse and of the strategic power of workers in the logistics sector, on which autonomists often focus.

In my book I wanted to give some tools to the Left, especially in Western Europe, to think about the realities of work today and the subversive potential it still has. Some comrades in France said my book is about the working class, but it isn’t: I try to defend E.P. Thompson’s idea that the working class is actually the result of class struggle rather than the other way around. For me, class is a political, symbolic, and cultural issue — it never exists “in itself.” That is why it is necessary to start from work as it is today, by detaching ourselves from all the misconceptions about it — ones sometimes perpetuated by the Left — and instead think about class and class politics on a scientific basis.

Perhaps the missing chapter of this book is one about the struggles in the social reproduction sector, another sector in which work is undergoing major developments. Capital is massively hiring precarious women to work in commodified domestic labor. And this is a sector which, for me, can also be at the forefront of the class struggle and the workers’ movement. But others talk about it better than I would have done.