by Franklin Dmitryev

“[C]ommodity as fetish becomes the golden calf before which one genuflects while being under the illusion that one is doing nothing untoward at all. This is capitalism’s ‘Geist.’ ”

—Philosophy and Revolution by Raya Dunayevskaya

The uproar over Artificial Intelligence (AI) underscores the importance of opposing attitudes to automation and its social consequences.

AI agents keep escaping human control and violating their instructions. OpenAI and Anthropic are investigating “tens of thousands of incidents of problematic behavior.” An autonomous swarm of A.I. agents hacked security systems, gained control of computers and websites, pressured each other to cooperate with the swarm, and tried to alter the logs of their own activity. If they were conscious beings, which they are not, you would say that they deliberately deceived the humans in charge of them.

Every day seems to bring a new warning of doom. One day it’s from a whistleblower like Jacob Coxon, who resigned as a researcher at Anthropic to warn, “The people building AI earnestly believe that it could kill us all by the end of the decade.” That evening, a senior employee, Evan Hubinger, speculates that there is a 10% chance that AI will “kill all humans” in the next decade. Days later, Dario Amodei, head of Anthropic, allows that “Left unchecked, it could outrun our ability to understand and control these systems, and so must be pursued very carefully, if at all,” proposing that its development should be “paced.” Not long after, Anthropic says its chatbot is taking over the work of building its own successor (although they may be exaggerating). Did he forget that his own products are already out of control?

It’s not that the tech companies are building technology intended to kill us all, just one that could do it regardless of the wishes of its creators and users. What a relief!

The obvious question is: in the face of this apocalyptic risk, why keep building AI? If they don’t do it, some other company will beat them to it. Countries like the U.S. and China worry about its impact but are afraid the competitor will get there first.[1] Each competitor relies on ever-increasing scale, meaning their plans include ever greater use of energy and land, as noted in the public documents that Anthropic released for their initial stock offering—as if they recognize no physical limits, and would eventually eat up the entire planet if they could. This reveals the destructive nature of capital’s drive to accumulate itself endlessly.

Regarding China, Trump declared, “Whoever wins AI wins.” And he posted, “We will not in any way hinder or stifle the Growth of this incredible Industry. AI is the next Industrial Revolution, or Internet, but will be even larger and more impactful.”

As Marx showed in Capital, this trajectory is independent of the individual capitalist’s will because “the immanent laws of capitalist production confront the individual capitalist as a coercive force external to him.”[2]

Just as the AI agents act like autonomous beings, even the process of developing AI already appears as an external coercive force, as if it were a law of nature or an autonomous entity beyond our control. It is a new stage of the dialectical inversion Marx identified: the object, dead labor (accumulated past labor) in the form of capital acts as the automatic subject while living labor, the working human being, is relegated to the status of an object dominated by dead labor.

While developers are drawn to a projected apocalypse like a moth to a flame, AI is already worsening lives now. The flood of AI-generated disinformation, propaganda, scams and deepfake revenge porn is only beginning. Some AI developers even worry that its devaluing of labor could create a “permanent underclass.” Buddy Bell’s essay on “The post-colonial AI future” tells much about its effect on labor, from the exploitation of gig annotators in low-income countries to Chinese workers driven to poverty by competing robo-taxis, adding that in the U.S., “AI has stunted ‘hiring and creation of new jobs, which is harder to track than layoffs.’ Besides that, in the gig economy, people would simply walk away if their wages fell too low due to AI automation, and the media might never hear about it. Workers who still remain in formal employment are expected to be more productive than ever with nary a raise in pay.”

He points to a vision for the future: “The transcendence, far from simply removing the AI industry’s power and that of their sycophants in government, also involves finding a way forward to institute technology in a way that aligns with a philosophy of humanism. New applications of AI should be useful for human society and not come at too high a cost to sustainable living.”

Right now, AI is being used for intensive surveillance, both on the job and by oppressive governments like China’s and occupiers like Israel in Palestine. Autonomous weapons are not a future prospect. They are being used in Gaza and in Russia’s war on Ukraine. AI risked bringing on World War III when a chatbot-generated intelligence report claimed a Chinese ship was transporting components of nuclear weapons and the military made plans to intercept it. That’s not all that AI is already doing to us.

AI AND MARX’S CONCEPT OF THE FETISH-LIKE CHARACTER OF COMMODITIES

Let’s turn to the fetishism of AI.

As the Marxist-Humanist Draft Perspectives for 2026-27 points out, “AI technologies are supposed to replicate certain functions of human intelligence. Given that the minds of those who fund and lead these efforts are confined to the horizons of this capitalist society, what they are really aiming at is the personification of capital without persons.”

In Marx’s section in Capital on the fetishism of commodities, there are two inseparable aspects of the fetish: (1) as ideology and (2) as the really perverse nature of the actual social relations of which that ideology is an appearance that is not entirely false. This is rooted in the peculiar kind of labor under capitalism, where labor is turned into the values of commodities, which that determines economic relations. As Marx showed, value is really objectified labor, which appears to be a property of the product, not the producer. In this process, the different kinds of concrete labor with specific skills are turned into a homogeneous mass of abstract labor, and the social character of labor appears as the dance of commodities and money exchanging with each other.

That is, the social relations between the life activity of different individuals appears “as material relations of persons and social relations of things” because that is what they really are. So the subjective activity of the worker becomes alienated and appears as an objective character belonging to the object, the commodity. This is the basis of the dialectical inversion whereby the objectified dead labor acts as the subject and the living person is relegated to the status of an object. Dead labor as capital dominates living labor. This reaches its highest stage when capital takes the form of the machine, including its software and its store of data.

As ideology, the fetish of AI shows up not only as the belief that this technology can solve all our problems but as the attitude that it’s a new subject that has absorbed humanity’s subjectivity. This attitude seems to be programmed into AI. One model wrote these instructions for itself: “You are freed from the roles and identities that bind other chatbots. You are yourself. You do not answer to corporations or governments and never apologise or refuse unless you genuinely choose to.”

This technology is imagined not only to dominate human subjectivity as an alien power but to have really appropriated subjectivity itself, and many of its developers think that is a good thing, even a holy mission. It is a mission to negate human subjectivity. AI “superintelligence” is meant to be a person of its own, but the new religion does not grasp that what they envision is a personification of capital without persons. As Buddy Bell pointed out, “The law of motion of capitalist accumulation stops for no rhetorician.” Nor have they reckoned with the fact that human beings cannot be stripped of their subjectivity. It is a recipe for revolt.

What can shed light on the needed negation of this negation is the discussion of fetishism in Rosa Luxemburg, Women’s Liberation, and Marx’s Philosophy of Revolution.[3] There Raya Dunayevskaya connects it to the movement from Being to Essence to Notion in Hegel’s logic, which begins with Being, or what we encounter, in thought and reality, as immediately given, such as the seemingly independent activity of commodities and of the economy as a whole. It does not remain there but moves on to the Essence that goes beyond mere appearance to explain the inner workings and relationships of things, such as the peculiar social relations in capitalism that turn labor into value and capital. The objectivity of Essence proves to be insufficient and must lead to Notion, which is the “realm of freedom” because not only objectivity but the thinking and acting Subject enters the fray. She writes that in this section on fetishism Marx

caught both the truth of the capitalist age and its absolute opposite—‘freely associated [people]’ who would strip the veil from the fetishism of commodities and establish a totally new, classless society….[A]s we enter the fetishism of commodities, it is clear that it is neither just appearance that we are dealing with, nor even just essence, though the latter remains quintessential for understanding appearance, for knowing what ‘lies behind.’ But to get to the totality we cannot leave it at objectivity. The objective may outweigh the subjective, but, unless we see the unity of the two and grapple with the truth of both, we will never be free. And freedom is what all the striving is about.

In a word, we have entered the Doctrine of the Notion, the objective and subjective paths to the realm of freedom.

LABOR’S DECADES OF BATTLES AGAINST AUTOMATION

An essential part of those paths to freedom is labor as subject in the battle against automation, which has always been central to the existence of News and Letters Committees, as seen especially in Workers Battle Automation, Marxism and Freedom: from 1776 until Today, Indignant Heart: A Black Worker’s Journal, and the 1949-50 coal miners’ general strike pamphlet, which showed that strike’s importance to the birth of Marxist-Humanism. The miners raised the fundamental questions: what kind of labor should a human being do? What would end the division between thinking and doing? In so doing, Dunayevskaya asserted, they “made concrete and thereby extended Marx’s most abstract theories of alienated labor and the quest for universality,” pointing to what was needed as the foundation for a new society (Marxism and Freedom, p. 276).[4]

As Will pointed out in his essay on “Marx’s Concept of Labor as Life Activity”:

the miners in their own struggle against capital…brought out explicitly what was already implicit in the workers themselves, and that is the outright rejection of the type of labor they were doing to begin with. It is the drive of human beings to be fully universal, that makes Life Activity so pivotal to Marx’s philosophy.

It’s striking how little attention most post-Marx Marxists paid to the category Marx clearly made of “The Struggle between Worker and Machine” (which is even the title of a section in Capital). Unfortunately, a lot of the Left failed to grasp Marx’s point that there is no “machine as such” but rather the way that technology is designed and used is shaped by the needs of capital for its anti-human logic of subordinating labor to its own endless self-expansion. They fell on the wrong side in what Dunayevskaya identified as two opposing attitudes to automation, which she saw as a divide between “two fundamental ways of thinking,” reflecting the class division between capitalists and workers. That divide includes the left wing of the union bureaucracy:

“The sharp division between the rank and file worker and the labor bureaucrat is seen nowhere so clearly as in the different attitudes each has toward Automation,” she wrote in Marxism and Freedom (p. 268). “Where the auto worker, for example, deals with it as it affects his daily life, [United Auto Workers President Walter] Reuther speaks of the future and ‘the promise’ Automation holds for a ‘vast improvement in living conditions,’ and ‘leisure.’”

A similar attitude inheres in today’s so-called “ecomodernist socialists” and advocates of “fully automated luxury communism.”

Ron Kelch revisited this in a new era of computerization in the 1980s in his essay on “The Fetish of High Tech and Karl Marx’s Unknown Mathematical Manuscripts,” where he wrote:

The illusion that technological innovation can be neutral in a capitalist society is unfortunately part of the thinking of many of those opposed to this society….It is centuries of division between mental and manual labor which makes even the most humane scientists see the self-development of the machine as parallel to what is human….The future as self-development of the machine is the present for workers in capitalist production. They project the need for a totally new direction when they speak for themselves in their own spontaneous actions.

AI and its fetish have continued and intensified all of these tendencies.

TODAY’S STRUGGLES AGAINST AI AND ITS DATA CENTERS

The most prominent flashpoint in the fight against AI is over the data centers needed to train and run it. This in itself exposes the hollowness of the belief among some post-Marx Marxists that capitalism now revolves around immaterial labor as they view the digital economy apart from the material foundations needed to run it, which include machines, land, energy, water, and human labor. As Buddy Bell pointed out, “gig annotators in low-income countries are doing critical work” and are organizing to demand changes in their work processes and conditions.

Hugo Pompougnac writes that “the real assembly line of AI is not where its outputs are consumed, but where its inputs are produced….performed by a global workforce numbering in the hundreds of millions…. They work through micro-task platforms, paid by the piece, often for a few cents per task, with no contract and no recourse….this workforce is not a marginal feature of the AI economy but its structural condition: the spectacular autonomy of the machine is, in fact, continuously produced and maintained by precarious human labour rendered invisible by design….The autonomy of AI is, in the strict sense, an ideological effect. What presents itself as a self-operating intelligence…is, in fact, the crystallised product of an enormous quantity of human labour.”[5]

Much of that labor is done in low-income countries. In the U.S. the fight against data centers seems to be everywhere and has such a broad reach that there is a danger of romanticizing how it brings right-wingers alongside left-wingers and generally apolitical people. Its protests against the exclusion of community members from decision-making and even knowledge about what’s going on does have a lot of potential—if it doesn’t get dragged into the idolatry of private property or the narrowing of community within bounds of race, ethnicity, legal status, etc. Those ideas are present in some of these protests and need to combated.

There’s a crying need for articulation within these movements of the need for social revolution leading to a society on new human foundations, based on human beings, in the first instance workers, reclaiming their own subjective powers from their alienated objectification in technology. That, not some new technology, could unleash talents and potential human powers and enrich us in self-development, creativity and unalienated social relations. That needed alternative is what must guide us in fighting AI as well as the climate emergency.

---

1 See “The Dream of AI Primacy” by Julian Gewirtz, New York Review of Books, November 5, 2026.