What is the relationship between logic and labour? Let us start with two propositions: first, digital computers are machines whose primary task is to perform logic, and second, automation is the process whereby mechanical performance substitutes for human labour. The conceit of this article is to square the circle between these two facts so that we can begin to have a working theory not of how digital computers, AI, etc. automate labour, but of what kind of automation is occurring.

Perhaps the most well-known contemporary reading of this relationship between logic and labour is to be found in Matteo Pasquinelli’s The Eye of the Master, where he claims: “ultimately, all labour is logic”.¹ My best interpretation of this claim, though it is not explicitly teased out in the book, is that all labour is reducible to its logical components because human labour necessitates some cognitive dimension and because the division of labour has a logical structure. The first problem with Pasquinelli’s argument is that it is manifestly untrue if we accept that labour is the human activity whereby nature is transformed for consumption or the furtherance of production.² If logic is present in this formulation, it is in an abstract and transcendental form, and the full activity of labour cannot be collapsed into logic. The second difficulty is this reading’s political consequence. Abstracting the social and material relations of labour away, as Pasquinelli does, leaves us with an idealist and ahistorical framework. Perhaps a clearer way to articulate the claim’s absurdity is to ask, “What kind of labour movement is possible if all labour is logic?” or “Does logic produce commodities?”

Pasquinelli is nevertheless correct to state that the relationship between logic and labour is crucial to the political economy of automation in the computer age. Where the combustion engine of a car automated the horsepower of the horse-drawn carriage, the digital computer automates computational tasks once performed by human beings. There is virtually no industrial production today which is not aided by some computational component. However, logic and labour remain discrete categories, whose relation is conjunctural and historically particular to the mechanical automation of work as aided by computation. It is in industrial capitalism, a mode of production that came into being only a few centuries ago and necessitated a complex division of labour, that computation became a relevant component of the labour process. Nowhere is the role of computation more evident today than in the organisation of the now-global division of labour, and in the circulation of raw materials and commodities, now described under the term logistics. Thus, it is clear that computation is not that which automates away logic or labour, but the technology capable of handling the vast complexity that sustains the daily operations of the production and reproduction of the capitalist world economy.

The advent of generative AI — a computational technology that bears notable aesthetic approximations to those intimate human activities that canonical computing could perform — has infected popular discourse with the notion that all human labour is replicable, if not replaceable, by machines. This stems from a kind of fetishistic attitude towards certain human activities over others, and the idea that the automation of factory work is somehow acceptable, but the automation of creative and intellectual work is a fundamental threat to our humanist self-understanding. Not only is this popular notion classist and philosophically dubious, but it also makes a greater analytical error: neither factory work nor intellectual labour is fully automatable. Moreover, a confusion is often made between the formal elimination of job titles through the reconstitution of the labour process under automation, and the real elimination of humans from the entire process. In Automation and the Future of Work, Aaron Benanav makes this distinction clear.³ While employers can justify lay-offs in the name of automation, such discourse distorts the underlying centrality of labour to production. Automation can reduce the total labour power needed, but it cannot eliminate labour entirely from the process. This is the insight wrested from preserving the distinction between mere logic and labour as a social relation, as metabolic activity between man and nature. Logic alone is not productive, and there can be no production without labour.

---

The question of logic and labour is often shoehorned into something much narrower. The most recent iteration of this, a debate in The Ideas Letter⁴ between Evgeny Morozov, Aaron Benanav and Leif Weatherby, makes useful advances toward what I will call a “generative AI socialist strategy”.⁵ Morozov argues we must overcome dominant Marxist paradigms which posit technology as constitutionally neutral or focus on deskilling and immiseration. Instead, he suggests: “[…] technology is not a surface onto which we project pre-existing values; it is one of the main sites where values are formed.”⁶ Benanav argues that capitalist firms are incapable of coordinating large-scale investment and data-infrastructuralisation.⁷ Weatherby claims that both Morozov and Benanav ignore the role of capitalist ideology and neoliberal rationality in the elaboration and implementation of generative AI.⁸ Though these contributions offer a sketch of a Marxist critique of generative AI, they start with too many assumptive priors. The most severe is the reduction of computation in capitalism to a system of organisation. Of course, this is important, but it misses something more general and pervasive: the role of technology in amplifying, augmenting, and rearticulating labour-power in the sphere of production. The production of goods and services in the economy is taken as a given, hidden behind the privileged analytical categories of planning, distribution, and value-formation. Morozov’s initial critique of Benanav disparagingly refers to the tendency that roots political and economic struggle in properly deliberative democratic discourse and governance — a kind of Webernianism. But as I intend to argue, neither Morozov’s insistence on radical experimentation nor Weatherby’s recentering on ideology goes far enough to avoid an all-too-common conceit of left technology critique: overinvesting in the cultural forms technology comes to represent, at the expense of the core class antagonism underlying it.

Indeed, all three approaches symptomatically tend towards foreclosing the possibility horizon of socialism and the radical potential of technology to increase the productive forces. It is a foreclosure in the sense that each of these critiques diverts our attention away from the sites of class struggle in the sphere of production. Whether it is to (re)distribution or to human ingenuity and creativity, the essential mediating role of labour in production gets diminished emphasis in the imaginative constellation of the struggle for a socialist future. I stress the increase in productive forces here as a dual rejoinder to Benanav’s distributionism and the more abstract critique Morozov offers of experimentation and socialist baroque, and as a step back from one of Benanav and Morozov’s core disputes. Benanav advocates for the formulation of democratic governance in the “Free Sector”, where the problem of profit-driven optimisation and the productivist technocracy of state-socialism can be circumvented. Morozov, weary of the “Free Sector” premise, argues against a kind of productivism to which he ascribes “top-down” “technocratic violence”. Neither Benanav’s multi-criterial model nor Morozov’s socialist baroque seems to appreciate the already tremendous power of technology to relieve labour of drudgery and to produce efficiently and at scale, assuming the crucial prerequisite that workers are in control of, and can reap the benefits of, technical productivity. Seen this way, technology can be returned to being an instrument — not a capitalist instrument, nor a socialist one — but a lever of production up for grabs in the class struggle. Weatherby appears more sensitive to this canonical value-neutral reading of techno-power, as his automation of the C-suite does not read technology as an agent in itself that comes to take (our) jobs. Rather, Weatherby claims that job titles are always already shaped by technical conditions whose non-agentic historical development creates redundancies and obsolescences. Where I disagree with Weatherby is his appraisal that some of the class struggle itself has been “taken over by machines”. The situation Weatherby describes appears to me to be much more narrowly framed within the familiar problematic of socialist discourse: questions of (more) just conditions for workers, the elimination of parasitic administrative bloat, and an expansive redistributive scheme.

The oft-referenced and oft-disputed structure and superstructure passage from the preface to Marx’s A Contribution to the Critique of Political Economy states:

“In the social production of their existence, men inevitably enter into definite relations, which are independent of their will, namely relations of production appropriate to a given stage in the development of their material forces of production. The totality of these relations of production constitutes the economic structure of society, the real foundation, on which arises a legal and political superstructure and to which correspond definite forms of social consciousness.”⁹

Technology, both as fixed capital and as an ideological apparatus, fits unevenly within this schematic. However, as Beverly Best reminds us, the social form Marx puts forth in this passage should occupy a central position in our interpretation of Marx’s total social theory. Best asks: “What sort of social dynamic does Marx articulate in this passage?” and answers: “Its animating force, according to Marx, is the social production of material and conscious life, a collective doing and making that inadvertently invents a world — a mode of sociality that excretes an assemblage of objectivities taken to be natural and intractable.”¹⁰ In light of this reading, it is perhaps untenable to maintain the thesis that technology is value-neutral. Nevertheless, the production of values still cannot simply be attributed to technology itself. Class society — in its totality — produces values; their substance is born out of relations of production, as Marx suggests. Those social relations entail tendencies that organise institutions in civil society, the state and the family. Technology mediates these values and gives form to their elaboration. Pace Weatherby, this is what Marxist ideology critique has to offer us: the proper placement of technology in a matrix of material processes, social relations, and ideas.

But despite getting closer to the question of class formation, Weatherby ultimately halts his analysis at the prospect of the proletarianisation of executives, what he calls “the automation of the C-suite”.¹¹My argument is that technology is more fundamental to class differentiation than distributionist models suggest, but that the primary modes of struggle for workers are merely technologically inflected rather than technologically structured. Technological development is determined by and determinative of a qualitative change in the form of work, and a quantitative change in the total labour-power required to produce. Here, generative AI gets no special status over the hammer or the mill. The relationship between logic and labour is neither that the former subsumes the latter, nor that logic is some special subcategory of labour that generative AI encroaches on.

---

Questions of distribution and planning represent an ambitious agenda for the potential of computational infrastructure to help build a more just society. However, such discourses disregard more immediate political demands and tasks in the sphere of production, including but not limited to: the conventional fights for better wages and working conditions, the consolidation of class power, opposition to military-imperialism, ecological struggles, and mass-literacy campaigns. Such work is not mutually exclusive with that of social democracy, but it necessitates different tactics and involves different actors. The task of a left critique of technology is not merely to imagine a liberatory future of technology use, but to locate critical sites of struggle already within our present technological conditions, where pressure can be exercised. I want to argue that our analysis can neither relegate struggle to modes of distribution nor deflate analytical categories to the point of dissolution. If Benanav’s sophisticated vision of a multi-dimensional economy forecasts a new technological implementation of economic management, which at its current stage is distant from immediate concerns of technologically mediated precarity and exploitation, Pasquinelli sees this exploitation as existing in diffuse and microscopic domains where traditional modes of political struggle are not possible.¹²

Building on the operaist tradition — Romano Alquati in particular — Pasquinelli claims “still, the Marxian grammar of industrial machines appears insufficient to enter and frame the age of information machines.” Borrowing from Deleuze and Guattari, he suggests that a “new form of value accumulation driven by knowledge and by an active cognitive component that is part of the surplus labour produced by any subject.”¹³ The issue with these claims is that they offer very little in terms of politics. The cybernetic re-reading of Marx that Pasquinelli advocates only further abstracts and obscures the traditional modes of labour militancy —such as strike actions — and proposes nothing material or tangible in their place. The (over)investment in the “Fragment on Machines” in Marx’s Grundrisse in the case of Pasquinelli, and in “Credit and Fictitious Capital” in Volume III of Capital by the value-theoretical tradition, has yielded some important discoveries. Yet these have often been at the expense of the firm political-economic categories in Marx — namely, those of fixed capital and turnover time — that I believe give clarity to the crisis of modern technology and capital. My objection here is not to the development of new strategies or new forms of struggle, but to the implicit claim that traditional forms of struggle, of party organisation, and of labour organisation are outmoded tout court. The perceived failures of the industrial proletariat in the global North to constitute a revolutionary bloc, the changing landscape of labour, its shift toward service work and gig-ification, the production of surplus populations, etc. must necessarily augment our strategy, but cannot be made to displace the fact that the industrial proletariat maintains a particular and strategically intense relationship to the reproduction of society. The proliferation of advanced technologies, contrary to popular imagination, only further concentrates and centralises the power of an albeit shrinking industrial working class to disrupt and halt productive mechanisms. This is only further underscored by the reliance of the growing tech sector on extractive industries in the global South, which often employ casualised, but nonetheless more traditional forms of wage-labour.¹⁴

Neither social-democratic theories of distribution nor post-Marxist accounts of micro-labour sufficiently contend with our present crises. Disaggregating analytical categories like logic, labour, and automation is a necessary step towards building a critical theory of technology attuned to the primary contradiction between labour and capital. Labour — mental and physical — makes up a qualitatively heterogeneous and quantitatively homogenised resource in production. Our analysis of the political economy of computation must start from this germ and expand outward to the office, the factory, and the data centre. Only a robust account of labour’s capacities and technology’s unique mediating role offers the left a sufficient inventory from which to devise a winning strategy. Neither the collapse of head and hand nor the forced victory of one over the other will do. In short, not all labour is logic: this is a good place to start.

Notes

1. Matteo Pasquinelli, The Eye of the Master: A Social History of Artificial Intelligence, 2023. [^]

2. As in Marx, this definition is provisionally borrowed from and developed out of a critique of the bourgeois economists, Adam Smith whose crude anthropology suggested “labour was the first price” and David Ricardo whose labour theory of value construed human labour as the necessary antecedent to production. See: Adam Smith, An Inquiry into the Nature and Causes of the Wealth of Nations, 1776; David Ricardo, On The Principles of Political Economy and Taxation, 1817; Karl Marx, Capital: A Critique of Political Economy, Volume 1, 1992. [^]

3. Aaron Benanav, Automation and the Future of Work, 2020. [^]

4. Evgeny Morozov, “Socialism After AI”, The Ideas Letter, 11 December 2025; Aaron Benanav, “A Real Political Economy of Technology: A Rebuttal to Morozov”, The Ideas Letter, 5 February, 2026; Leif Weatherby, “Automate the C-Suite: A Response to Morozov and Benanav”, The Ideas Letter, 5 February 2026. [^]

5. I am here referencing Laclau and Mouffe’s Hegemony and Socialist Strategy for two reasons. The first is a nod to a common gesture between the theorists I mention here (Pasquinelli, Morozov, Benanav and Weatherby) and Laclau and Mouffe towards a kind of post-Marxian theory. The association of the age of AI with a need to reevaluate or abandon the perceived rigidities of orthodox Marxism appears, to me, hasty. The second valence the Lacalu and Mouffe allusion seeks to invoke is the dimension of “strategy”. All three articles cited here emphasise a necessity for something of a socialist program for technology, or at least the preliminary theorisation to get there. Benanav stresses the need for democratic governance and is engaging in a highly practical application of political economy. Morozov writes of a somewhat more opaque but nonetheless motivated socialism-from-below, and Weatherby offers something of a heuristic in lieu of a program. In each of these approaches is an attempt to confront directly the imminent crisis of technology and economy. Though I am skeptical of where this lands — in Benanav’s case distributionism; Morozov’s socialist world-making; and Weatherby’s automation of the C-suite — I take well the shared ambition to theorise something beyond a purely negative critique, and to think a socialism beyond mere imagining socialism. [^]

6. Evgeny Morozov, “Socialism After AI”. [^]

7. Aaron Benanav, “A Real Political Economy of Technology”. [^]

8. Leif Weatherby, “Automate the C-Suite”. [^]

9. Karl Marx, trans. Ilinca Iurascu, A Contribution to the Critique of Political Economy, 1970, p. 20. [^]

10. Beverley Best, The Automatic Fetish: The Law of Value in Marx’s Capital, 2024, p. 2. [^]

11. Weatherby, “Automate the C-Suite”. [^]

12. Aaron Benanav, “Beyond Capitalism—1”, New Left Review, no. 153, July 2025; Aaron Benanav, “Beyond Capitalism—2”, New Left Review, no. 154, August 2025; Matteo Pasquinelli, The Eye of the Master; Matteo Pasquinelli, “On the Origins of Marx’s General Intellect”, Radical Philosophy, no. 206, 2019; Matteo Pasquinelli, “Artificial Intelligence or the Demonization of Mental Labor”, The Demon’s Brain, 2018; Matteo Pasquinelli, “Italian Operaismo and the Information Machine”, Theory, Culture & Society 32, no. 3, 2015. [^]

13. Pasquinelli, “Italian Operaismo and the Information Machine”, p. 5-9. [^]

14. See: Martín Arboleda, Planetary Mine: Territories of Extraction Under Late Capitalism, 2020. [^]