Report on Week 3 of Solidarity’s “Socialists in Labor” reading and discussion series, “Three, Four, Many Industrial Revolutions?”
BY IVAN DRURY ZARIN
At both my jobs, AI has shown up as a tool of surveillance and the degradation of worker’s control over processes of labor, not as the labor-saving device promised by its CEO high priests.
At my college job teaching international students, the main feature of AI is its degradation of teacher and student control over production. If a core element of production in a college are the artifacts of learning (essays, presentations, and other measurable outcomes of the learning process), then AI has degraded these products and shrunken the field of what teachers can ask and expect of students. Chat GPT, as a tool, has degraded the process of producing learning products on both sides: for teachers by interrupting the process of written interaction with students, and for students by degrading interactions with texts and with pedagogies in the classroom.
At my job as a bus driver, AI has arrived as a management tool for more perfect workplace surveillance and discipline, with the concentration of labour time as only a secondary aspiration. In April, Transdev, the French transit company that the British Columbian government sub-contracts to operate its paratransit services, rolled out new, AI equipped cameras in its busses. The old cameras would record if we hit a bump that rocked the bus, sending a 20 second clip to a unionized worker in our local head office. She would review these clips and send recordings of serious incidents to management. The new AI cameras record constantly and issue loud alarms and “real time feedback” on driving: mostly flagging that a driver has looked away from the road at the GPS map for too long, that you did not stop long enough at a stop sign, that you crossed a road line (or a tar line on the road that the camera mis-identified as a road line), and other pseudo infractions.
The new AI cameras record incidents at an order of magnitude greater than the old system. And instead of sending them to a unionized worker in our depot, it sends them to a AI assessment system operated by a third party surveillance company, which is overseen by non-union workers. This third party system then sends infractions it deems reportable, according to a black box rubric that our union does not have knowledge of, to management.
Transdev’s labor-saving uses of AI are more humble. Management has, at the same time as the AI cameras, replaced drivers’ quarterly in-person “refresher training” classes with videos we are forced to watch alone on our onboard tablets. I always found the most useful thing about refresher training was the opportunity to get together with a room full of other drivers, who we only ever see occasionally at the depot or on the road, and talk about our frustrations. These gripe sessions, which we’d squeeze in between the gaps of management’s educational program, or in response to these lectures, would be validating because I’d get to understand that something I find irritating or worrying was shared by other drivers, and they’d add their own insights to my complaints.
Replacing those classes with videos I have to watch alone on the bus means management is saving thousands of hours of paid time each quarter, because they no longer have to pay their hundreds of drivers to go back to the depot in the middle of a shift and sit for hours in a classroom together. Instead, they order drivers to pull over and complete a 40 minute video at a time in the day when there’s a gap in their schedule anyway.
In general, this week’s reading group on socialist responses to technology helped me understand these changes through a more class-struggle lens. As Ansar Fayyazuddin writes, in his review of some recent books about AI, the goal of capitalist technological innovation “is to reduce the allowed interventions of workers into the work process.” The most important insight from our readings and discussion, for me, was that I felt we were able to see through the mystique of evaluating the new technology of AI on its own supposed merits, and instead to see it as a tool of imminent capitalist workplace discipline that workers are fighting. In this light, the principle task for socialists regarding AI is clear: to intervene to support and sharpen workers’ class struggle response against AI.
Before this, I read a lot of socialists (online, anyway) defending AI through the hypotheticals of “how workers could make use of this technology in a socialist future.” But to cut through those abstractions and focus on the here-and-now is not really a novel interpretation of how socialists should respond to new technology in the workplace and society. In his classic Labor and Monopoly Capital, Harry Braverman argues that “Machinery comes into the world not as the servant of ‘humanity,’ but as the instrument of those to whom the accumulation of capital gives the ownership of the machines.” Sure, Braverman’s formulation leaves open the possibility of workers seizing and making different use of those machines, but in the meantime there is the more-pressing problem of how they are being used against workers in the capitalist present.
For this discussion, which had a feeling of urgency and also productive uncertainty, we were lucky to have two presenters, who took complementary approaches to the question of what technological development represents in class society, and how (or whether) we can apply those general guidelines to understanding and fighting AI. Jason Dawsey presented a history of Marxist critiques and responses to technological innovation and automation, focusing especially on the debates between the post-Trotskyist News and Letters organization and philosopher Herbert Marcuse. Ansar Fayyazuddin focused on the degradation of our creative, social worlds by the intrusion of homogenizing technology into fields of intellectual labor, which have, until AI, proven more resistant to the innovations of Fordism and Taylorism.
You can watch the video of Jason and Ansar’s presentations here, and listen to the audio here.
Problems of partial automation
With a case study of how socialist-humanists responded to the sudden explosion of automation in US factories in the 1970s, Jason explained that there has been a kind of a temporal split in Marxist thinking about technology. One side of the Marxist brain is committed to seizing capitalist innovation through embrace, emphasizing the positive future uses of technology that, in its capitalist container, is bad for workers. The other side, meanwhile, emphasizes the bad present.
Jason said that, in a special “automation” issue of News and Letters, the publication of the organization led by CLR James, Raya Dunayevskaya, James Boggs, and Grace Lee Boggs, workers responded to the automation of industry in the 1970s with fear. They were afraid that their jobs would being automated away, worried about greater management control, and resentful of labor leaders who were cooperating with management’s rollout of automation.
Kim Moody, in a lecture in February 2026, bemoaned a similar tendency amongst major unions in the US, which have responded to AI by joining its boosterist chorus. Incredibly, he says, even the national teachers union, whose members’ interests seem obviously at odds with the incantations of AI hucksters, have been cheering rather than booing the introduction of AI to their workplaces.
Jason said that Herbert Marcuse, one of the best-known socialist humanist philosophers, responded to News and Letters by saying that they were getting the issue wrong. The issue, he said, was not technology itself, but technology within the capitalist mode of production.
It wasn’t until years later that Marcuse worked out his thinking on automation, in his major work, One Dimensional Man. There he said that a Marxist perspective that adopts a romantic view about technological development must be grounded in a particular historical moment in capitalist society.
Automation, he said, was a key centrifugal force in capitalist society. What it represented to workers was to make work more exhausting, more isolated, more deskilled. The issue he had was with partial automation: between parts of production that are highly automated and those that are not, it is human labour that fills the gaps.
Automated labour systems, unshackled from the rule of capitalist production, could reduce the work people must do, and free them. But, he says, that is not to say that a workers’ socialist society can simply inherit tech from capitalist society to apply it in new ways.
Technological innovation is not necessarily technological progress. Jason asked: “Who believes that a hydrogen bomb has any liberatory potential?” To use Marcuse’s formulation, a great refusal would be the most appropriate way to approach this sort of technology.
Categories of assessment
Ansar said that, in the past, before the industrial revolutions that introduced automation, deskilling, and homogenity to commodity production, the craft worker represented a threat to the capitalist because they possessed knowledge that the capitalist cannot possess. The only solution capitalists had to the problem posed by the craft worker was to replace them with another craft worker.
“This is why capital breaks down the work of the craft worker into discrete parts, repeatable and knowable at a glance,” Ansar said. A central goal of capitalism, then, is the deskilling of labour, whereby one worker can be swiftly replaced by another.
“When I was a grad student in Denmark,” Ansar said, “the government ran a commission to determine whether grad students and researchers were doing work worth what they were getting paid. If they had been able to conduct this study, they must have concluded that the work of researchers and artists refuses the universal measures required for commodified relations.”
The point is that intellectual labour is difficult to quantify. But capital’s “solution” to the craft worker’s irreplaceability is now being brought to research and creative work through the vehicle of AI, which is used to create junk that is leaving stains on our culture. Ansar said, “When the misnamed auto-correct interrupts our writing, it is hacking out originality and producing uniformity,” as capital has always done.
Ansar’s argument that AI is socially and creatively destructive is synthetic with Jason’s suggestion that AI technology may be productive only for capital, not for workers or social needs. Arguments to the contrary hinge on maybes and ifs, and depend a lot on the prognostics of companies that have invested billions of dollars in the promises of AI that have not, so far, materialized.
I think we need to be careful, when evaluating the prospects of AI’s productive usefulness as as a set of tools, to remember Braverman’s call to assess the machine not “purely in its internal relations, but in relation to the worker.” An assessment of the “internal relations” of the machine are the degree to which the designer was able to eliminate “any but the desired motion” in the name of mastering “the complete and continuous control of motion.” A social, rather than technical assessment of a machine considers “its effect upon the labor process.” (Braverman, 186-186)
I found Quinn Slobodian and Ben Tarnoff’s article “Muskism” to be useful in its review of the bygone era of Fordism. Beginning with Italian communist leader Antonio Gramsci’s social framing of Fordism as “a reconfiguration of social relations” under “the dictatorship of technical reason,” Gramsci argued that factory regimes overflow the point of production into society as a whole, bringing social relations into the service of commodity production. Slobodian and Tarnoff try to bring this unity of productive and social regulation to bear on the post post-Fordist regime we live under today, to limited effect, I think. But theirs is an important point that the effect of a machine upon the labor process should be analyzed social-totally, rather than within production-itself, as though the factory could occur in a vacuum.
Time to fight back not accommodate
One of Braverman’s most strategic arguments is that the capitalist introduction of new technology is itself a field of class struggle. Since the capitalist mode of production is always “being refined and perfected,” and because new generations of workers are “not formed within the matrix of work life, but are plunged into work from the outside,” … “the habituation of workers to the capitalist mode of production must be renewed with each generation.” Therefore, “the necessity for adjusting the worker to work in its capitalist form… becomes a permanent feature of capitalist society.” (Braverman, 139-140)
This is the process Braverman calls the “acclimatization of the worker to the new modes of production”; a capitalist victory that is only possible “as the capitalist mode of production conquers and destroys all other forms of the organization of labor, and with them, all alternatives for the working population.” This lesson applies to defensive fights as new labor-concentrating and labor surveillance technologies emerge. In his talk on AI hosted by the Havens Wright Center in February 2026, Kim Moody argued that the slow development of AI’s potentials affords us time to organize, oppose, and defeat its advance.
Braverman’s example of worker-resistance against new machines was Ford’s development of the assembly system for car making, which deskilled and degraded the conditions of work, and allowed for the generalization of uniform, low wages for workers. When it was only Ford employing such management techniques, workers would quit, to the point that, “toward the close of 1913 every time the company wanted to add 100 men to its factory personnel, it was necessary to hire 963.” But once every other car manufacturer had adopted the same factory regime, workers had no choice and were “habituated” to this new form of labor.
With AI a similar process is taking place, but socially rather than in any particular workplace. Capital is introducing AI measures at the edges of each of our lives and jobs, and it is seeping in. We can fight it and stop it, but not if we wait until it’s been absorbed into the body of our class through total immersion of our skins in its acerbic bath.