Living in the Digital Miasma: AI, the Illusion of Neutral Science, and Organizing for Power

Living in the Digital Miasma: AI, the Illusion of Neutral Science, and Organizing for Power

A review of The Eye of the Master: A Social History of Artificial Intelligence by Matteo Pasquinelli

By Nuzrath Hussain

Volume 27, no. 2, Political Economy of Science 

To a wastewater epidemiologist like me, Jathan Sadowski’s description of technology and capitalism as “toxic miasma” has a particular bite.1

Courtesy of Verso Books.

Once a term for an entire (and now discredited) way of explaining disease, it captures the invisible haze we currently live in. My scientific work concerns pathogens flowing through wastewater, their genetic fragments amplified and sequenced, data streaming into R scripts and increasingly, AI pipelines. I can’t escape these tools; they structure my practice. But AI’s pervasiveness feels less like a tool and more like a smothering presence that subsumes labor, flattens thought, and tightens the grip of technological capitalism under the guise of efficiency. This contradiction lies at the heart of the political economy of AI: We cannot use it without being used by it. In this review of Matteo Pasquinelli’s The Eye of the Master: A Social History of Artificial Intelligence, I elaborate on this thesis and propose that to work with AI is also to work against it, because neutrality has never been an option in this miasma.

Algorithms as Social Tools

Algorithms, as Pasquinelli shows, are not mere abstractions. They are social tools forged through labor, ritual, and power. From Vedic fire altars and Babylonian counting tablets to mercantile bookkeeping and the work songs that gave rhythm to collective labor, each step is tied to knowledge—and thereby, labor as reflective activity—embedding mathematics into the management of bodies and resources in order to discipline and control. By tracing algorithmic culture across civilizations, Pasquinelli critiques the tendency to treat algorithmic reason as a uniquely Western, modern achievement. To frame computation as Silicon Valley’s discovery is, he argues, a form of epistemic colonialism that erases other traditions. What is now rebranded as “machine intelligence” rests on long histories of knowledge extractivism facilitated by empire and capital.

In the 1960–70s, Donald Knuth’s recovery of “ancient algorithms” played a crucial role in legitimizing computer science as a new discipline. By rooting it in long cultural traditions, programmers positioned their field as a science with pedigree, not merely an engineering trade. This canonization was political. It helped institutionalize power and professional authority for a new class of mental workers. Again, algorithms are not neutral as they inevitably become tools to consolidate authority within knowledge infrastructures.

Babbage and the Mechanization of Mental Labor

The story of Charles Babbage’s Difference Engine is often told as a tale of lone genius and the birth of modern computing. But Pasquinelli reframes it within the political economy of nineteenth-century capitalism. Conceived in 1822, at the height of steam power and factory expansion, the engine was a vast mechanical calculator designed to automate the production of navigation tables vital for trade and empire—an effort contemporary with Watt’s steam governor (1780s) and Jacquard’s punched-card loom (1801). Babbage’s real milieu was not the academy but England’s industrial workshops—places of intelligence—where machinists and craftsmen provided the mindful hands that made mechanical contrivances possible. The impetus for mechanizing mental labor came not from abstract speculation, but from the imperial need for error-free logarithmic tables to guide maritime trade and colonial expansion.

From this context emerged Babbage’s two great principles. First, the labor theory of the machine posits that new machines imitate and replace a pre-existing division of labor. Second, the Babbage principle, in which dividing labor into modular tasks allows capital to measure, price, and purchase only the exact amount of skill required, thus transforming the division of labor into a calculus of value. Taken together, these principles yield a techno-economic axiom: Computation itself emerges as both the automation of mental labor and the quantification of its costs. To compute, under capitalism, is to measure labor in units of time, energy, and capital, while erasing the human costs beneath the hierarchy of skilled and unskilled work.

This logic extended into Babbage’s vision of science and society. Machines were not only productive apparatuses but instruments for the surveillance and measurement of labor. His mechanical notation anticipated programming languages, while his faith in the exponential accumulation of knowledge foreshadowed the ideology of today’s knowledge economy. For Marx, who engaged with Babbage’s Economy of Machinery and Manufactures in both the Grundrisse and Capital, the point was clear: Machines arise not from pure science but from the analysis of the labor process.2 In historian William J. Ashworth’s phrase, Babbage’s project was nothing less than “the march of the material intellect set to the rhythm of the factory.”3 The Difference Engine thus stands not as the dawn of neutral computation but as a machine theory of value—an attempt to inscribe capitalist hierarchies of labor into the very concept of calculation.

The Origins of the General Intellect and the Abstraction of Labor

Building on Babbage’s philosophy of machinery, Marx, further in Grundrisse, reframed the machine as the embodiment of collective social knowledge, or the “general intellect,” subordinated to capital.4 Pasquinelli traces this lineage by showing how the steam engine and coal supplied not just power but abstraction, binding energy and labor into calculable, disciplined flows. Each transition in power source was built to serve specific interests. The shift from waterwheels to coal-fired steam engines was less about efficiency than about securing the perfect “abstract energy” to be easily standardized, measured, and relocated near urban labor pools. This marriage of fossil energy with mechanical precision produced what Andreas Malm calls “fossil capitalism,” where energy infrastructures and labor control fused into a single system of exploitation.5

Feedback devices (like Watt’s steam governor) and data carriers (like the Jacquard loom’s punched cards) reinforced this abstraction, embedding workers’ rhythms and knowledge into machines that could be quantified, optimized, and owned. This can be seen as embodying the first cybernetic logic of transforming motion into regulated energy and instructions into stored information.

The Collective Worker of the Information Age

Pasquinelli frames the rise of cybernetics and information theory as a continuation of the long abstraction of labor inaugurated in the industrial age. Drawing on the French philosopher of technology Gilbert Simondon, he identifies a decisive shift. Machines bifurcate labor into two streams: energy (the brute force powered by coal and steam) and information (the gestures, measurements, adjustments, and skills workers provided to keep machines running).6 These “micro-decisions”—the small acts of gesture, measurement, and adjustment—were absorbed into the machine, turned into instructions, and eventually made measurable. We can see this as an early step toward what we now recognize as data.

By the mid-twentieth century, with cybernetics and early computing, this process deepened to run production more efficiently. In the 1960s, Italian sociologist Raniero Alquati conducted workers’ inquiries at Olivetti and FIAT and found that information wasn’t just flowing top-down from management; it was constantly being produced by workers themselves and then siphoned off by the system, which he called “valorizing information.”7 In other words, the most valuable part of labor—knowledge—was taken by cybernetics, translated into numbers, and then used to control the very people who created it. Information, then, is not an external input; it is labor absorbed into machines and condensed into commodities.

Radical thinkers like William Thompson and Thomas Hodgskin had already argued that labor is always cognitive, that judgment and cooperation matter as much as muscle and sweat.8 They warned that when knowledge is captured by machines, it turns against the worker. Marx took this insight further and reframed cooperation not just as a property of individual workers or the genius of one, but as the power of the collective worker, gesamtarbeiter, a cyborgian social mechanism whose cooperation is both the source of machinic intelligence and the object of its measurement.9 In doing so, he also showed how knowledge, once alienated into machines, becomes capital’s property and is exerted as power over workers in the form of discipline, deskilling, and dispossession.

This insight links the abstraction of labor directly to the infrastructures of control: bureaucracies, feedback systems, and later digital networks that extend the “social factory” into society as a whole. The infrastructures that once tethered industrial growth to coal now tether “digital capitalism” to data centers whose vast energy demands scale the extraction of both information and natural resources. In both cases, what appears as progress is inseparable from regimes of labor discipline and ecological exhaustion.

Thus, the concept of “collective worker” is a reminder that any critique of AI must read models, datasets, and data centers as condensations of collective knowledge expropriated from living labor.

Cybernetic Mind, Contested Science, Dialectics 101

The story of “neural networks,” writes Pasquinelli, isn’t really about science uncovering the brain’s secrets. It is about how each historical moment projects its dominant technologies onto the mind: clocks in the age of mechanical time-discipline, telegraphs and relays in the age of industrial communication, computers in the age of information. When McCulloch and Pitts sketched the first artificial neurons in the 1940s, they were not simply imitating biology; they were translating the infrastructures around them—telegraph relays, switching circuits, control systems—into a model of thought. A dialectical perspective helps us see this not as a misstep but as a pattern. Scientific metaphors emerge from labor processes and material infrastructures then return as abstractions that reorganize those very processes. What appears as “brain science” is also a reflection of social hierarchies, labor relations, and the technical systems through which capital governs them.

This is where self-organization comes in. Engineers and cyberneticians claimed that machines could learn on their own, but what they were really doing was capturing the cooperative intelligence of workers and society, reorganizing it into circuits, and selling it back as “machine intelligence.” Neural networks didn’t triumph because they were better at mimicking brains; they triumphed because they were better at capturing and formalizing the messy, distributed cooperation that already holds social systems together. In this way, the abstraction of labor becomes the template for the cybernetic “mind.”

For movements inside science, teaching, and tech work, this realization matters. The metaphors of autonomy and “self-organizing systems” can sound emancipatory, but have historically been mobilized just as much for military planning, corporate management, and new forms of social control. From Cold War radar to today’s deep learning, self-organization has been a way of turning collective decision-making into data that can be governed from above.

Knowledge Economy and the Political Erasure of Mental Labor

In The Eye of the Master, we see how knowledge production has always been a site of struggle. Schools, workers’ institutes, and even political economy itself have been battlegrounds over who would direct the general intellect. The elites cultivated a labor aristocracy and celebrated craft in order to cement hierarchies, while some factions of the nineteenth century workers’ movement blurred mental into manual labor for tactical unity. Hence, both sides helped erase mental labor from view, resulting in political amnesia around mental labor.

That history repeats itself. Today, data centers, AI research platforms, and education technologies continue to capture the collective intelligence of researchers, teachers, and students, transforming our cooperation into proprietary infrastructure. AI’s progress is, in fact, a contested political-economic terrain regarding who supplies the general intellect, who owns its embodiments, and how its yields are circulated (or not) among those who produce it. For those of us working inside universities, labs, or classrooms, the lesson is sharp. The fight is not only about wages and working conditions but also about reclaiming the general intellect we collectively create and deciding whether it serves exploitation or emancipation.

Organizing Implications: How to Blow Up the Pipeline

AI is not thinking for itself. It is the latest machine for enclosing social cooperation—our collective problem-solving, teaching, and research labor—into statistical models. To resist it, we need to reassert that this intelligence comes from us, not from the machine, and organize around reclaiming how it is captured, owned, and used. But, like a blowfly carrying enteric pathogens, AI carries with it the shits of capitalist exploitation: data centers that guzzle water, burn fossil fuels, and seize land at the expense of marginalized communities. The pathogens here are the hierarchies—race, class, empire—that determine whose neighborhoods get poisoned, whose jobs get deskilled, and whose labor gets extracted. We’ve seen this play out already.

Across Indiana, where I live, community organizers at Citizen Action Coalition (CAC) like Ben Inskeep and Bryce Gustafson are already on the frontlines, knocking on doors, holding teach-ins, and mobilizing neighbors against Big Tech’s hyperscale data centers that drain aquifers, poison air, and hike utility bills.10 The CAC is forcing transparency where none exists by connecting residents with scientists to answer questions and waging legal fights against utilities that bend to corporate interests. Their work is a reminder that communities are already deep in this fight while science as an institution mostly stands aside, or worse, colludes with capital under the guise of neutrality to feed the very infrastructures that harm people. In Michigan, local anti-capitalist AI skeptics forced one data center out of the city of Ypsilanti only to see an alternative site proposed in a Black, working-class community nearby.11 The contradiction merely shifted. To truly resolve this, our fight must be dialectical: not just against AI as a technology, but against the infrastructures of energy, data, and militarized science that uphold it. That means tech workers must bargain over data, computing, and model outputs as products of collective labor. We must refuse metaphors that hide social origins, demand transparency on how “self-organizing” systems rewrite workflows, evaluation, and discipline, and ultimately build alternative, worker-governed infrastructures where the rules of learning—what is optimized and for whom—are set by those who supply the intelligence in the first place. These tools are already at hand and waiting to be wrested back for liberation.

We also need to push for collective control over the algorithmic pipeline, refuse militarized contracts, expose the surveillance logics, and fight for legal frameworks that actually bind capital instead of communities. Teachers have the urgent task of seeding a science for the people—one that draws openly from Marxist, eco-feminist, and other critical traditions. Breaking down the capitalist university, the defense lab, and the tech monopoly means putting today’s tools to use for emancipation, while remaking the conditions for entirely different tools tomorrow. Rewriting the rules of science requires remaking the rules of society. Only by changing society can we change science.12 This change will not be handed down. It will be built by an international working class that takes internationalism into its own hands.

Pasquinelli’s focus on Euroamerican intellectual history neglects the way AI has been embedded in the international division of labor and imperialism. Take the Tata Group, India’s largest multinational conglomerate, as an example. It is the backbone of the domestic tech economy and a collaborator in US-Israel’s military-industrial-geopolitical projects, extracting mental labor in Bangalore while profiting from genocide in Palestine.13 Meanwhile, in Tamil Nadu, tariff wars gut labor-intensive hubs like Tirupur’s textiles and gems, throwing thousands out of work.14 These aren’t tech jobs, but their loss swells the reserve army of precarious labor that global capital, including AI, feeds on. In nearby Salem, data-annotation shops turn workers into invisible ghost labor, poorly paid to label images, filter traumatic content, and train chatbots that power AI systems.

Tamil Nadu’s state government, facing mass layoffs, has demanded resistance to US tariffs only to clash with the central government under Narendra Modi that aligns with Washington’s global value-chain strategy. Not long after the recent BRICS summit, where Modi posed alongside Putin and Xi, Washington followed with Trump’s announcement of a $100,000 H-1B visa fee, effectively shuttering the pipeline of Indian tech workers into Silicon Valley. India is still the single largest source of H-1Bs, with China second, meaning this chokehold doesn’t end exploitation but rather funnels skilled labor into more precarious offshore labor markets.15 This isn’t parochial politics. This is structural destabilization feeding labor into circuits of capital to power militarized innovation and push workers toward emigration and precarity overseas.16

Organizing in and beyond Indiana and India means recognizing this entanglement. To fight AI’s deskilling and data extraction in our labs, classrooms, and backyards is also to contest how these same infrastructures are weaponized abroad. And for us as knowledge producers, the responsibility is not optional. As scientists, engineers, and educators, we belong to the international working class. The task is militant and unavoidable: We must take ownership of the scientific pipeline itself, break with the bosses, join hands with our communities, and insist that knowledge, in all its forms, belongs to and benefits the people.

The Eye of the Master: A Social History of Artificial Intelligence
Matteo Pasquinelli
Verso Books
2023
272 pages

 

Nuzrath Hussain, MBBS MPH: Nuzrath is a medical doctor from the southern state of Tamil Nadu, India, interested in WaSH and climate justice, currently doing her PhD in IU School of Public Health. Her research here focuses on evaluating sanitation interventions in interrupting transmission of infectious diseases. She is currently a rank-and-file member of IGWC the grad workers union here @IU and ex-zonal organizer & executive committee member of Tamil Nadu Medical Officers Association. 

Notes

  1. Jathan Sadowski, The Mechanic and the Luddite: A Ruthless Criticism of Technology and Capitalism (University of California Press, 2025).
  2. Karl Marx, Capital: A Critique of Political Economy, Volume I, trans. Ben Fowkes (Penguin/New Left Review, 1976), 492–507; Karl Marx, Grundrisse: Foundations of the Critique of Political Economy (Rough Draft), trans. Martin Nicolaus (Penguin, 1973), 383–423; Karl Marx, Capital: A Critique of Political Economy, vol. 1, trans. Ben Fowkes (London: Penguin/New Left Review, 1976), 492–507.
  3. William J. Ashworth, “Charles Babbage and the Engines of Perfection,” Transactions of the Newcomen Society 73 (2002): 127–28; Marx, Grundrisse, 706.
  4. Ashworth, “Charles Babbage.”
  5. Andreas Malm, Fossil Capital: The Rise of Steam Power and the Roots of Global Warming (Verso, 2016).
  6. Gilbert Simondon, “Technical Mentality,” Parrhesia 7 (2009): 20, originally published as “Mentalité technique,” Revue Philosophique de la France et de l’Étranger 131, no. 3 (2006): 343–57.
  7. Raniero Alquati, Sulla FIAT e altri scritti (Feltrinelli, 1975).
  8. William Thompson, An Inquiry into the Principles of the Distribution of Wealth Most Conducive to Humane Happiness Applied to the Newly Proposed System of Voluntary Equality of Wealth (Wheatley and Adlard, 1824), 272; Thomas Hodgskin, Popular Political Economy: Four Lectures Delivered at the London Mechanics’ Institution (Tait, 1827), 97.
  9. Marx, Grundrisse, 694–712; Marx, Capital, 443–44.
  10. “Home,” Citizen Action Coalition (website), accessed September 24, 2025.
  11. “Township Urges Data Center Relocation,” Planet Detroit, August 2025.
  12. “Defending Science: SftP Statement,” Science for the People, April 7, 2025.
  13. “Diaspora Group Launches ‘Tata Bye Bye’ Campaign Against TCS Links to Israel Ahead of New York Marathon,” The Wire, October 21, 2024.
  14. Shilpa Ranipeta, “Tiruppur Exporters Fear 1.5 Lakh Job Losses, ₹12,000 Crore Revenue Hit as US Tariffs Loom,” CNBC TV18, August 26, 2025.
  15. “Trump’s $100,000 Fee on H-1B Visas,” American Immigration Council, accessed September 25, 2025.
  16. Rina Chandran et al., “AI Boom Is Dream and Nightmare for Workers in Global South,” Reuters, March 14, 2023; “Modi Should Strongly Oppose U.S. Imposing 50% Tariff: T.N. Chief Minister Stalin,” The Hindu, August 13, 2025.