Algorithmic Power and the Dialectics of Humanity : Artificial Intelligence, Capital, and the Reconfiguration of Social Totality

1. Introduction

Artificial intelligence has emerged as one of the most consequential arenas through which contemporary societies negotiate questions of power, value, governance, and human agency. Algorithmic systems now mediate labor markets, border regimes, policing practices, welfare distribution, financial speculation, and cultural production, increasingly shaping not only institutional decision-making but also everyday social life. From predictive analytics in public administration to generative systems in knowledge and culture, artificial intelligence has become deeply embedded in the reproduction of social order (Moleka 2024 ; 2025).

Mainstream accounts of these developments overwhelmingly frame artificial intelligence as a technical breakthrough driven by innovation, efficiency, and computational progress. Such narratives present algorithmic systems as neutral tools that optimize decision-making and rationalize social processes. In doing so, they obscure the historical, political, and economic conditions under which artificial intelligence is produced and deployed. This technocratic imaginary detaches AI from the social relations that give it form, rendering power asymmetries, labor exploitation, and colonial legacies largely invisible (Couldry and Mejias 2019; Zuboff 2019).

Dialectical anthropology offers a critical alternative to these dominant framings. Following Marx’s foundational insight that social forms appear natural precisely because their historical conditions of emergence are concealed (Marx 1976), this article argues that artificial intelligence must be understood as a material condensation of capitalist accumulation, colonial epistemologies, and political struggle. AI is neither autonomous nor neutral; it is embedded within relations of production, governance, and meaning that shape both its design and its effects. Algorithmic systems are produced through labor, energy extraction, data appropriation, and institutional power, and they in turn reorganize how labor, authority, and knowledge are constituted.

From a Marxian perspective, artificial intelligence represents an intensification of capitalism’s long-standing drive to abstract, formalize, and commodify human capacities. If industrial capitalism mechanized physical labor, contemporary digital capitalism increasingly targets cognition, affect, and judgment as sites of value extraction (Marx 1976; Postone 1993). Data-driven systems transform everyday social activity into raw material for accumulation, extending processes of enclosure into domains previously situated outside formal markets. This movement echoes earlier moments of primitive accumulation, including colonial extraction and the commodification of land, bodies, and knowledge (Harvey 2003).

At the same time, artificial intelligence must be situated within the historical dynamics of colonial power. Fanon’s analysis of colonial domination emphasized that modern systems of control operate not only through economic exploitation but through epistemic violence—the imposition of classificatory regimes that dehumanize and govern colonized populations (Fanon 1963). Contemporary algorithmic systems reproduce and intensify such logics through racialized surveillance, biometric border controls, and predictive policing technologies that disproportionately target marginalized communities (Benjamin 2019; Browne 2015). From this perspective, AI is not simply a technology of efficiency but a continuation of colonial techniques of ordering, classification, and control under new technical forms.

Gramsci’s theory of hegemony further illuminates how artificial intelligence operates as a cultural and ideological force. Algorithmic rationality is increasingly naturalized as objective, inevitable, and apolitical, securing consent for forms of governance that displace democratic deliberation with technical expertise (Gramsci 1971). Decisions once subject to political contestation are reframed as matters of optimization, risk management, or prediction. This hegemonic framing obscures the normative assumptions embedded in algorithms while legitimizing technocratic authority.

Polanyi’s analysis of market disembedding provides an additional lens for understanding the social consequences of artificial intelligence. As intelligence, judgment, and social relations are abstracted into data and subjected to market logic, social life becomes increasingly subordinated to algorithmic calculation (Polanyi 2001). Yet Polanyi also emphasized that such processes generate countermovements. The commodification of intelligence provokes resistance precisely because it threatens the moral and social foundations of collective life. Contemporary struggles over data sovereignty, algorithmic accountability, and digital labor rights can be understood as manifestations of such countermovements.

Finally, the Comaroffs’ work on late capitalism highlights how abstraction, speculation, and governance increasingly operate through legal-technical systems that blur the boundaries between economy, law, and culture (Comaroff and Comaroff 2020). Artificial intelligence exemplifies this conjuncture. Algorithmic systems govern through probabilistic futures rather than empirical pasts, reshaping how risk, responsibility, and authority are allocated. In this sense, AI is not merely a tool of governance but a reconfiguration of social temporality itself.

Building on these theoretical traditions, this article conceptualizes artificial intelligence as a contradictory social formation. While algorithmic systems intensify exploitation, surveillance, and epistemic domination, they simultaneously expose tensions within capitalism’s attempt to subordinate intelligence, creativity, and judgment to market logic. These contradictions destabilize established regimes of value and authority, opening spaces for contestation, resistance, and alternative imaginaries. A dialectical anthropology of artificial intelligence thus rejects both technological determinism and cultural pessimism, insisting instead on the historical openness of social futures.

2. Artificial Intelligence and Dialectical Method

Dialectical anthropology begins from the premise that social phenomena cannot be understood as isolated objects or autonomous systems but must be analyzed as historically produced relations embedded within broader structures of power, labor, and meaning. Artificial intelligence, despite its technical appearance, is no exception. To approach AI dialectically is to reject both technological determinism and cultural essentialism, insisting instead that algorithmic systems emerge from, reproduce, and transform existing social relations.

Marx’s critique of political economy provides the foundational orientation for such an analysis. For Marx, capitalism is defined not merely by markets or technologies but by a specific organization of social relations in which human capacities are abstracted, commodified, and subordinated to the logic of accumulation (Marx 1976). Technologies do not drive history independently; they materialize the imperatives of capital. Artificial intelligence thus cannot be understood apart from capitalism’s long-standing tendency to replace living labor with machinery in pursuit of surplus value. Yet, as Marx emphasized, this process is inherently contradictory. Machinery appears to liberate capital from dependence on labor while in fact deepening its reliance on human capacities in new and often obscured forms (Marx 1976; Marx 1993).

Recent scholarship has demonstrated that AI systems depend on vast infrastructures of human labor that remain systematically invisible. Data extraction, annotation, content moderation, and system maintenance rely on precarious workers dispersed across global labor markets, often under conditions of extreme asymmetry (Gray and Suri 2019 ; Tubaro, Casilli, and Coville 2020 ; Moleka 2025). From a dialectical perspective, artificial intelligence does not eliminate labor but reorganizes it spatially, temporally, and politically, intensifying exploitation while masking its social foundations. The apparent autonomy of algorithms thus functions ideologically, concealing the relations of production that sustain them.

Gramsci’s theory of hegemony deepens this analysis by illuminating how artificial intelligence operates not only through economic domination but through the production of consent (Gramsci 1971). Algorithmic systems increasingly present themselves as neutral, objective, and technically necessary, displacing political debate with claims of optimization and efficiency. Decisions regarding welfare eligibility, risk assessment, policing, and migration are reframed as technical problems rather than matters of social justice or democratic deliberation. This transformation exemplifies what Gramsci described as the naturalization of historically contingent power relations, whereby dominant groups secure legitimacy by presenting their worldview as common sense.

The hegemonic power of artificial intelligence is further reinforced through institutional expertise, corporate secrecy, and state reliance on technical authority. As scholars of science and technology studies have noted, algorithmic opacity is not merely a technical limitation but a political condition that shields systems from accountability (Burrell 2016; Pasquale 2015 ; Moleka 2025). Dialectical anthropology insists that such opacity be understood relationally: it is produced through legal regimes, intellectual property structures, and ideological claims about innovation that privilege capital over collective control.

Polanyi’s analysis of market disembedding offers another critical lens. Polanyi argued that capitalism’s attempt to subject land, labor, and money to market logic generated profound social dislocation, provoking countermovements aimed at re-embedding economic life within social and moral frameworks (Polanyi 2001). Artificial intelligence represents a further extension of disembedding, as intelligence, judgment, and social interaction are abstracted into data and subjected to algorithmic governance. Predictive systems reorder social life according to probabilistic futures, reshaping how risk, responsibility, and value are distributed.

Yet, as Polanyi’s framework suggests, this process is inherently unstable. The attempt to commodify intelligence generates resistance precisely because it threatens social reproduction and human dignity. Contemporary struggles over data protection, algorithmic bias, digital labor rights, and indigenous data sovereignty can be understood as countermovements against the unchecked expansion of algorithmic markets (Kukutai and Taylor 2016; Couldry and Mejias 2019).

Fanon’s analysis of colonial domination further radicalizes dialectical inquiry by foregrounding the racialized and epistemic dimensions of power. For Fanon, colonialism operated through the production of knowledge systems that classified, hierarchized, and dehumanized colonized populations (Fanon 1963). Artificial intelligence reproduces these logics through algorithmic classification systems that encode racial, gendered, and colonial assumptions into technical infrastructures. Predictive policing, facial recognition, and biometric border controls disproportionately target racialized populations, extending colonial forms of surveillance into digital space (Browne 2015; Benjamin 2019).

From a Fanonian perspective, AI is not merely biased; it is structurally implicated in regimes of domination that emerge from colonial histories. Dialectical anthropology therefore rejects reformist approaches that seek to “fix” algorithms without addressing the power relations that shape them. The question is not whether AI can be made fair in the abstract, but whose knowledge, values, and futures it is designed to serve.

The Comaroffs’ analysis of late capitalism further situates artificial intelligence within a broader transformation of governance and social temporality. In their account, contemporary capitalism increasingly operates through abstraction, speculation, and legal-technical regimes that blur the boundaries between economy, law, and culture (Comaroff and Comaroff 2020). Artificial intelligence exemplifies this conjuncture by governing through prediction rather than explanation, managing populations through risk scores, probabilities, and anticipatory logics. Social life is reordered around futures that have not yet occurred but nonetheless exert material force in the present.

Taken together, these theoretical traditions enable a dialectical understanding of artificial intelligence as a social totality. AI is not simply a technology applied to society; it is a historically specific configuration of power, labor, knowledge, and governance. It embodies capitalism’s attempt to resolve its contradictions through technical means, while simultaneously intensifying those contradictions. A dialectical anthropology of artificial intelligence therefore treats algorithms as sites of struggle, shaped by historical forces yet open to contestation and transformation.

3. The Commodification of Intelligence

The commodification of intelligence constitutes one of the most significant transformations of contemporary capitalism. If, as Marx argued, capitalism is defined by the systematic conversion of human capacities into commodities (Marx 1976), artificial intelligence marks a decisive extension of this process into the domain of cognition, judgment, and meaning itself. What is at stake is not merely the automation of tasks but the enclosure of intelligence as a productive force subject to ownership, extraction, and accumulation.

From a Marxian perspective, this development must be understood in relation to the historical logic of abstraction. Capital abstracts concrete labor into labor power, rendering qualitatively distinct human activities commensurable through exchange value. Artificial intelligence intensifies this abstraction by translating heterogeneous forms of social life—language, affect, movement, attention—into quantifiable data. These data are then mobilized as raw material for algorithmic systems that generate predictive value for capital. In this sense, AI does not simply process information; it reorganizes social reality according to the imperatives of accumulation (Marx 1976; Marx 1993).

This process echoes what Marx described as primitive accumulation, not as a singular historical event but as a recurring dynamic through which new domains of life are enclosed and commodified. Contemporary scholars have noted that data extraction functions as a form of digital enclosure, appropriating social activity that was previously embedded in non-market relations (Harvey 2003; Couldry and Mejias 2019). Everyday practices—communication, cultural expression, social interaction—are transformed into assets owned and controlled by corporate entities, often without meaningful consent or compensation.

Polanyi’s concept of fictitious commodities provides a crucial analytical lens for understanding this transformation. Polanyi argued that land, labor, and money are treated as commodities under capitalism despite not being produced for sale, a contradiction that generates profound social dislocation (Polanyi 2001). Intelligence, like labor, is not produced for the market as such; it is a constitutive feature of human social existence. The attempt to commodify intelligence through artificial intelligence thus represents a new form of fictitious commodification, one that threatens the social and moral foundations of collective life.

This threat becomes particularly visible when intelligence is detached from the social relations that give it meaning. Algorithmic systems reduce complex forms of judgment to statistical correlations, privileging calculability over interpretation. Such reduction does not merely distort intelligence; it reshapes it. As social actors adapt to algorithmic evaluation—optimizing behavior for visibility, ranking, or prediction—intelligence itself becomes reorganized around market and bureaucratic imperatives (Beer 2017; Fourcade and Healy 2017).

The commodification of intelligence is inseparable from global inequalities. Data extraction and AI development are embedded in transnational divisions of labor that mirror colonial and postcolonial hierarchies. While value is captured by corporations concentrated in the Global North, much of the labor that sustains AI systems—data labeling, content moderation, platform maintenance—is performed under precarious conditions in the Global South (Gray and Suri 2019; Tubaro, Casilli, and Coville 2020). This asymmetry reflects what the Comaroffs describe as the uneven geographies of late capitalism, in which abstraction and accumulation coexist with dispossession and informalization (Comaroff and Comaroff 2020).

Fanon’s analysis of colonial extraction deepens this critique by foregrounding the epistemic dimensions of commodification. Colonial regimes did not merely exploit labor and resources; they appropriated knowledge, categorization, and meaning, subordinating colonized populations through imposed systems of classification (Fanon 1963). Artificial intelligence reproduces these dynamics by embedding dominant epistemologies into technical systems that claim universality. The commodification of intelligence thus entails not only economic extraction but epistemic domination, as certain ways of knowing are rendered valuable while others are marginalized or erased.

At the same time, dialectical analysis insists that commodification is never total. Intelligence resists complete enclosure precisely because it is relational, situated, and socially produced. Meaning cannot be fully reduced to data without generating contradictions that destabilize algorithmic authority. Errors, biases, and failures within AI systems are not merely technical problems; they are symptoms of deeper tensions between lived social complexity and abstract market logic (Suchman 2020).

These contradictions give rise to resistance. Movements for data sovereignty, demands for algorithmic accountability, and critiques of extractive AI practices reflect what Polanyi described as countermovements—collective efforts to reassert social control over commodifying forces (Polanyi 2001). Indigenous data governance initiatives, for example, challenge the assumption that intelligence and knowledge can be freely appropriated, insisting instead on relational and collective ownership models that disrupt capitalist enclosure (Kukutai and Taylor 2016).

The commodification of intelligence, then, is not a completed process but a contested one. Artificial intelligence intensifies capitalism’s drive to abstract and monetize human capacities, yet in doing so it exposes the limits of market rationality. A dialectical anthropology of AI must attend to these tensions, recognizing intelligence as both a site of extraction and a locus of struggle over the future of social life.

4. Labor, Automation, and the Crisis of Value

Artificial intelligence fundamentally reconfigures labor under contemporary capitalism, not by eliminating work but by transforming its social, temporal, and epistemic conditions. In the dialectical tradition, labor cannot be reduced to a set of discrete tasks; it is a social relation and a site of value creation whose abstraction and commodification lie at the heart of capitalist accumulation (Marx 1976; Postone 1993). AI operates as both instrument and expression of this dynamic: it reorganizes labor, redefines expertise, and simultaneously threatens the very foundations of value.

Marx’s analysis of machinery illuminates this transformation. In Capital, machinery extends the productivity of labor while paradoxically deepening capital’s dependence on human capacities, even as it displaces certain forms of labor (Marx 1976). Contemporary AI functions analogously, automating cognitive, administrative, and even creative tasks while remaining fundamentally reliant on a vast, distributed, and often invisible labor force. Human labor persists in the generation of training data, annotation, content moderation, algorithmic auditing, and system maintenance, typically performed under precarious, low-paid, or outsourced conditions (Gray and Suri 2019; Tubaro, Casilli, and Coville 2020). This duality—displacement alongside indispensable human labor—reveals the contradictions of digital capitalism and challenges traditional theories of value based solely on measurable outputs.

Moishe Postone’s reinterpretation of Marx’s labor theory of value is particularly instructive. Postone emphasizes that the social form of labor, rather than labor as a quantity of effort, underpins capitalist value (Postone 1993). In the AI-driven economy, this distinction becomes critical. Algorithmic production abstracts labor into formalized inputs: data points, predictive models, and coded rules. Yet these abstractions rely on social, cognitive, and affective labor embedded in life-worlds. The commodification of intelligence produces a structural tension: labor is simultaneously rendered invisible and hyperexploited, destabilizing traditional measures of value and eroding the temporal and normative foundations of work.

Polanyi’s insight on fictitious commodities extends this argument into the social domain. Just as land and labor are “treated as commodities” despite not being inherently produced for the market (Polanyi 2001), AI abstracts human capacities—attention, cognition, creativity—into exchangeable units of value. This abstraction provokes social dislocations that manifest as labor precarity, algorithmic oversight, and intensified monitoring, echoing Polanyi’s thesis that disembedding generates profound social countermovements.

These dynamics are racialized and globalized, reflecting Fanon’s critique of colonial labor structures (Fanon 1963). The extraction of data and the provision of low-wage algorithmic labor are disproportionately located in the Global South, reproducing historical hierarchies of exploitation and epistemic domination. AI-driven labor regimes mirror colonial patterns, producing a dual economy of highly capitalized, automated outputs in the Global North and a globally distributed, precarious workforce in the South (Browne 2015; Benjamin 2019). This asymmetry underscores the continuing coloniality of labor under digital capitalism.

Simultaneously, automation destabilizes traditional forms of labor control. Algorithmic systems manage productivity, performance, and conduct through continuous surveillance, predictive assessment, and incentivized compliance, creating what scholars have termed “algorithmic Taylorism” (Mateescu, Nguyen, and Veale 2019). The reorganization of work along algorithmic imperatives transforms temporality: tasks are decomposed into micro-actions, performance is quantified in real time, and workers are evaluated by metrics designed to optimize efficiency rather than cultivate skill or meaning. Yet this very quantification exposes the limitations of formalized labor abstraction: errors, biases, and emergent behaviors challenge algorithmic governance, revealing the social and historical contingencies embedded in ostensibly technical systems.

These contradictions generate a profound crisis of value. Capital seeks to extract maximum surplus from labor through automation and datafication while remaining structurally dependent on human activity that cannot be fully captured or formalized. The social recognition, ethical legitimacy, and temporal coherence of labor are increasingly fragile, undermining the stability of capitalist value itself (Postone 1993; Fourcade 2018). Resistance emerges not merely as opposition to automation but as collective assertion of the irreducibility of human intelligence, relationality, and expertise. Labor movements, algorithmic audits, digital cooperatives, and data sovereignty initiatives exemplify these countermovements, contesting the alienation and abstraction imposed by AI (Kukutai and Taylor 2016; Couldry and Mejias 2019).

The crisis of value generated by AI thus has both structural and political dimensions. Structurally, it exposes the limits of abstraction: intelligence, judgment, and affect cannot be fully commodified without producing contradictions that destabilize accumulation. Politically, it creates the conditions for contestation, requiring workers, communities, and social movements to renegotiate the terms under which labor, knowledge, and intelligence are recognized and valorized. From a dialectical standpoint, AI’s automation and datafication are not endpoints but dynamic sites of struggle in which social reproduction, technological control, and human agency intersect.

In sum, the AI-driven reorganization of labor exemplifies capitalism’s intensification of abstraction while generating contradictions that destabilize value. It simultaneously obscures, reproduces, and transforms social hierarchies, revealing the inseparability of technical systems, labor relations, and historical power structures. Dialectical anthropology, attentive to these contradictions, interprets AI not as a neutral instrument of productivity but as a historically situated social formation in which exploitation, resistance, and reconfiguration of value are deeply intertwined.

5. Algorithmic Hegemony and Colonial Power

Artificial intelligence functions not only as a technology of labor and value extraction but as a vehicle for the reproduction of power through cultural, epistemic, and ideological means (Moleka 2026). Gramsci’s notion of hegemony provides the central analytic lens: power is secured not solely through coercion but through the consent of subordinate groups, mediated by institutions, ideologies, and everyday practices (Gramsci 1971). AI exemplifies contemporary hegemonic mechanisms by presenting itself as neutral, objective, and technically necessary, naturalizing specific forms of governance and social ordering.

Algorithmic systems codify the implicit norms, assumptions, and epistemic hierarchies of their creators. This naturalization obscures the historically contingent nature of decisions that govern social life, from welfare eligibility to criminal justice risk assessments. What appears as impartial algorithmic judgment is, in reality, a sophisticated mechanism for producing consent to forms of control that would otherwise be politically contested (Burrell 2016; Benjamin 2019). In this sense, AI embodies Gramscian hegemony: it extends authority by shaping knowledge, shaping perceptions of fairness, and producing the legitimacy of technocratic governance.

Fanon’s analysis of colonial power amplifies this critique. Colonial domination was never purely material; it relied on epistemic hierarchies that defined who could speak, who could be recognized as knowing, and whose knowledge counted as valid (Fanon 1963). Algorithmic systems inherit and reproduce these hierarchies. Predictive policing, algorithmic sentencing, and surveillance technologies disproportionately target racially marginalized populations, reflecting patterns of historical subjugation (Browne 2015; Gangadharan 2017). AI thus represents not simply a tool of governance but a continuation of colonial epistemic power, reconfigured in the guise of neutrality and efficiency.

The Comaroffs’ analysis of late capitalism elucidates the intersection of abstraction, speculation, and governance in AI-driven systems. Algorithmic decision-making, by converting uncertainty into probabilistic forecasts, transforms social life into a domain of anticipatory governance (Comaroff and Comaroff 2020). Social actors are evaluated, disciplined, and governed according to models of risk rather than direct observation or historical causality. This restructuring produces both spatial and temporal reordering: populations are surveilled, monitored, and preemptively classified according to algorithmic metrics, reproducing inequalities under a veneer of technical rationality.

Polanyi’s concept of fictitious commodities resonates with these processes. By abstracting intelligence, attention, and social behavior into marketable data points, AI disembeds social life from collective norms, moral frameworks, and human judgment (Polanyi 2001). Yet, as Polanyi reminds us, disembedding provokes resistance. Indigenous data governance, algorithmic accountability movements, and activist interventions challenge the assumption that algorithmic knowledge is universal, highlighting the ongoing tension between technical authority and social legitimacy (Kukutai and Taylor 2016; Taylor et al. 2023).

Algorithmic hegemony is thus both material and symbolic. It operates through infrastructures, codes, and predictive models while simultaneously shaping what is recognized as knowledge, rationality, and legitimate authority. Critically, this power is not neutral: it is historically and geographically uneven, reinforcing global hierarchies rooted in colonial and postcolonial structures (Fanon 1963; Comaroff and Comaroff 2020). The automation of judgment extends the logic of colonial administration into the present, rendering historically marginalized populations subject to predictive oversight, risk scoring, and data extraction.

Recent ethnographic and critical STS research has documented these processes empirically. Studies of border management systems, algorithmic hiring platforms, and digital surveillance infrastructures reveal how AI produces both consent and coercion in complex, often invisible ways (González 2022; Choudhury et al. 2023). These works demonstrate that algorithmic governance is deeply embedded in social hierarchies: its effects are shaped by, and reinforce, pre-existing patterns of exclusion, dispossession, and epistemic marginalization.

Yet dialectical anthropology insists that such hegemony is neither total nor inevitable. Contradictions within algorithmic governance—errors, biases, opacity, and emergent social behaviors—expose the limits of technocratic control (Burrell 2016; Suchman 2020). These contradictions provide openings for resistance, contestation, and alternative forms of social knowledge. Movements advocating algorithmic transparency, community-centered AI design, and the decolonization of digital infrastructures exemplify countermovements to algorithmic hegemony, illustrating the dynamic interplay between power and resistance central to a dialectical understanding (Couldry and Mejias 2019; Gangadharan 2017).

In conclusion, artificial intelligence exemplifies the entanglement of technical systems with historical power, colonial legacies, and ideological consent. It functions simultaneously as a tool of labor reorganization, a mechanism for abstracting value, and an instrument of epistemic and cultural control. A dialectical approach situates AI within these interlocking dimensions, highlighting both its role in reproducing global hierarchies and the persistent potential for resistance and transformation.

6. Contradictions and Countermovements

Artificial intelligence, as a social formation, is marked by a series of structural contradictions that are central to a dialectical understanding of contemporary capitalism. These contradictions emerge from the simultaneous expansion and abstraction of labor, the commodification of intelligence, and the consolidation of algorithmic hegemony. While AI amplifies exploitation, inequality, and epistemic domination, it also generates openings for resistance, contestation, and reconfiguration of social power—what dialectical anthropology identifies as countermovements (Polanyi 2001; Postone 1993).

Marxian analysis foregrounds the internal tensions of capitalist production: as AI abstracts and quantifies intelligence, it creates new forms of surplus value extraction, yet this process depends on human labor that remains socially embedded and historically contingent (Marx 1976; Marx 1993). The invisibility of digital labor, from content moderation to data annotation, contrasts with the immense value captured by AI-driven platforms, highlighting a contradiction between the formal abstraction of intelligence and the lived, concrete realities of labor (Gray and Suri 2019; Tubaro, Casilli, and Coville 2020). This tension destabilizes the coherence of capitalist value and exposes the limits of algorithmic governance.

Polanyi’s framework extends this insight into the social and moral domain. By treating intelligence and social behavior as fictitious commodities, AI disembeds human capacities from the relational and ethical frameworks that historically regulate social life (Polanyi 2001). The resulting dislocations—precarity, surveillance, and social stratification—provoke countermovements that challenge the legitimacy of abstraction. Indigenous data governance initiatives, labor activism, and civil society demands for algorithmic transparency exemplify these social forces, asserting the necessity of collective control, accountability, and relational knowledge (Kukutai and Taylor 2016; Taylor et al. 2023).

Fanon’s critique of colonial epistemic structures illuminates the racialized dimensions of these contradictions. AI reproduces patterns of surveillance, classification, and marginalization that echo colonial hierarchies, reinforcing epistemic domination while simultaneously producing points of resistance. Movements challenging biased predictive policing, algorithmic sentencing, and inequitable access to AI-mediated services exemplify the contestation of historically embedded power relations (Benjamin 2019; Browne 2015). Resistance is thus not merely oppositional but epistemically transformative, reclaiming knowledge, interpretation, and social meaning from algorithmic abstraction.

Gramsci’s notion of hegemony clarifies the mechanisms through which AI maintains authority despite these contradictions. Algorithmic governance secures consent by naturalizing technical rationality, rendering structural inequalities as neutral or inevitable (Gramsci 1971). Yet contradictions within AI—bias, error, opacity, and emergent behaviors—disrupt this authority, creating openings for countermovements. Public debates over data sovereignty, ethical AI design, and algorithmic accountability illustrate the ongoing negotiation of legitimacy and consent in algorithmic societies (Burrell 2016; González 2022; Choudhury et al. 2023).

The Comaroffs’ theorization of late capitalism emphasizes the temporal and spatial dimensions of these contradictions. AI’s anticipatory logic reorganizes social life around futures that are probabilistically modeled yet materially enacted in the present (Comaroff and Comaroff 2020). This predictive governance intensifies the abstraction of value and labor while producing social dislocations that fuel countermovements. Resistance is therefore not only a reaction to immediate exploitation but a challenge to the temporal logic through which capital seeks to preempt social and cognitive life.

Critically, these contradictions demonstrate that AI is neither neutral nor autonomous. It is a historically situated social formation in which exploitation, abstraction, and hegemony coexist with potentiality, agency, and transformation. Dialectical anthropology interprets these tensions not as anomalies to be corrected but as constitutive features of the system, revealing both the limits of capitalist abstraction and the possibilities for emancipatory interventions.

Countermovements are diverse in form and scope. They encompass grassroots digital cooperatives, algorithmic auditing initiatives, labor union interventions in tech platforms, policy advocacy, and epistemic decolonization projects. Collectively, they illustrate the capacity of social actors to assert relational control over intelligence, knowledge, and algorithmic infrastructures, challenging the dominance of capital and technical rationality (Couldry and Mejias 2019; Mateescu, Nguyen, and Veale 2019). In doing so, these countermovements reveal the dialectical interplay between domination and resistance, abstraction and materiality, automation and human agency.

In conclusion, the contradictions generated by AI—between labor and automation, abstraction and lived experience, algorithmic authority and social legitimacy—constitute the central terrain of struggle in the contemporary digital economy. Dialectical anthropology provides a conceptual framework to understand these processes not as linear technological developments but as historically and socially embedded contradictions. Countermovements emerge from these contradictions, demonstrating the potential for reconfiguration, contestation, and the reassertion of human, social, and epistemic agency. AI, in this sense, is both a site of domination and a medium through which new forms of social, ethical, and political life can be envisioned.

7. Contradiction, Resistance, and Praxis

Dialectical anthropology emphasizes that domination is never total; it is inherently generative of resistance. The expansion of AI as a tool of labor abstraction, epistemic governance, and predictive control produces precisely such contradictions. Algorithmic systems, while operationalizing efficiency and profit, also create points of friction—labor disputes over AI-mediated work conditions, debates over algorithmic accountability, and movements for digital and epistemic sovereignty—that render domination contingent rather than absolute (Polanyi 2001; Postone 1993; Benjamin 2019).

Resistance to algorithmic domination manifests in multiple registers. At the level of labor, digital platform workers, content moderators, and AI annotators have organized to challenge exploitative conditions, asserting agency over the very infrastructure designed to discipline and abstract their labor (Gray and Suri 2019; Tubaro, Casilli, and Coville 2020). At the epistemic level, communities and scholars contest the racialized, gendered, and colonial biases embedded in algorithmic systems, advocating for transparency, fairness, and culturally situated knowledge frameworks (Browne 2015; Choudhury et al. 2023). These forms of resistance reflect Gramsci’s understanding of counter-hegemonic praxis: they do not merely oppose domination, but work to construct alternative structures of legitimacy, knowledge, and social authority (Gramsci 1971).

Fanon’s critique of coloniality is particularly instructive here. AI replicates historical patterns of epistemic hierarchy and control, yet resistance simultaneously enacts epistemic decolonization. Movements challenging predictive policing, discriminatory credit scoring, and the uneven global distribution of algorithmic labor represent collective refusals to accept the authority of purportedly neutral systems, reclaiming cognitive, social, and political agency from the machinery of domination (Fanon 1963; Comaroff and Comaroff 2020).

Polanyi’s insights into the social limits of market expansion illuminate why these countermovements are not anomalous but structurally necessary. Disembedding labor, intelligence, and social activity into algorithmic processes provokes social reaction precisely because society seeks to protect relational, ethical, and moral forms of life from complete commodification (Polanyi 2001). AI, therefore, is both a mechanism of control and a site of struggle—a terrain in which the dialectical interplay of domination and resistance unfolds.

Dialectical anthropology engages not merely in critique but in praxis: it situates AI within historical power relations, material conditions, and global hierarchies, rendering visible both forms of domination and the potentialities of resistance. This engagement allows scholars to imagine futures not predetermined by algorithmic authority but shaped by collective struggle over labor, knowledge, and social reproduction (Couldry and Mejias 2019; Taylor et al. 2023).

8. Conclusion

Artificial intelligence is frequently framed as a historical rupture or technological singularity, yet a dialectical analysis reframes it as a deepening of existing capitalist, colonial, and epistemic dynamics. AI intensifies labor abstraction, commodification of intelligence, and surveillance infrastructures, while simultaneously generating contradictions that destabilize regimes of value, authority, and knowledge. It is both an instrument of control and a medium through which human agency, resistance, and social innovation emerge.

A dialectical anthropology of AI treats it as a terrain of struggle rather than a completed system. Its trajectory will be shaped by ongoing conflicts over labor, epistemology, governance, and social reproduction. Recognizing these contradictions allows scholars to document emergent forms of resistance, expose inequities embedded in algorithmic authority, and imagine emancipatory alternatives.

Ultimately, the future of AI is inseparable from human action and social contestation. Anthropology’s role is not to predict technological inevitabilities but to elucidate the historical, material, and political conditions under which AI is embedded, contested, and potentially transformed. By rendering these struggles visible, dialectical anthropology affirms the enduring possibility of shaping a more equitable, accountable, and socially grounded relationship between humans, machines, and society.