Artificial Intelligence, Labor, and the Social Question of Technology: A Dialectical Political Economy Analysis

Artificial Intelligence, Labor, and the Social Question of Technology:
A Dialectical Political Economy Analysis

S.K. Das, A. Borhan

Centre for Social Research (CSR), Dhaka, Bangladesh

Abstract

Artificial Intelligence (AI) has rapidly emerged as one of the most transformative technological developments of the twenty-first century. A widely circulated narrative claims that AI will surpass human intelligence and render large sections of humanity economically obsolete. This paper critically examines this claim from the standpoint of political economy and dialectical analysis. It argues that AI itself is not responsible for the displacement of human labor; rather, the social relations governing its ownership and deployment determine its socio-economic consequences. Drawing upon the theoretical framework of classical political economy and Marxist analysis of technology, the paper demonstrates that AI represents a form of collectively produced intelligence whose benefits are currently appropriated privately by corporate capital. This contradiction between social production and private appropriation constitutes the central socio-economic tension of the AI age.

The paper further advances two interrelated arguments that deepen this critique. First, it exposes how the historical plunder of traditional and indigenous knowledge by corporate actors under the banner of intellectual property rights has direct parallels with the contemporary appropriation of collectively produced data for AI development. Second, it challenges the philosophical framing of the term ‘artificial intelligence’ itself, arguing that what is called artificial intelligence is, in reality, an extension and crystallisation of accumulated human intelligence — linguistic, cultural, scientific, and experiential. The ideological labelling of this intelligence as ‘machine-generated’ risks erasing human intellectual labour from historical and economic recognition. The resolution of these contradictions, the paper argues, requires the socialization of the benefits generated by AI so that technological progress can serve humanity as a whole rather than a narrow class interest.

1. Introduction

Artificial Intelligence has become one of the defining technological developments of contemporary society. From automated decision systems and large-scale data analysis to generative models and robotic automation, AI is transforming economic production, communication, and governance. A dominant narrative within technological discourse asserts that AI will soon surpass human cognitive capability and render human labor largely obsolete [1–3]. According to this view, technological progress will inevitably produce widespread unemployment as machines replace human workers.

However, this interpretation overlooks a crucial socio-economic fact: technology does not operate independently of social relations. The consequences of technological change depend fundamentally on the structures of ownership, control, and distribution within society. This paper advances the thesis that the central contradiction surrounding AI is not technological but political-economic. AI represents the cumulative product of collective human knowledge, public scientific research, and social data, yet its economic benefits are increasingly concentrated in the hands of a small number of corporate actors.

Beyond the question of distribution, this paper raises two further issues that have received insufficient scholarly attention. First, the current regime of intellectual property rights has historically served as a mechanism through which corporations appropriate traditional, indigenous, and communal knowledge without adequate recognition or compensation. This pattern is now being reproduced at a vastly larger scale through AI development, which feeds on the accumulated linguistic and cultural heritage of all humanity. Second, the very designation of these systems as ‘artificial intelligence’ performs an ideological function: by attributing cognitive capability to machines, it obscures the enormous reservoir of human intellectual labour upon which these systems are entirely dependent. Consequently, the social question raised by AI is fundamentally a question of distribution, ownership, and the proper recognition of collective human intelligence.

2. Technology and the Historical Dynamics of Productivity

Throughout human history, technological innovation has served as a powerful driver of productivity growth. Economic productivity may be expressed as:

P = Q / L

where P represents productivity, Q denotes total output, and L represents labor input.

Technological innovation increases productivity by enabling greater output with reduced labor time. The steam engine, electrification, and digital computing each dramatically increased productive capacity during their respective historical periods [4]. Artificial Intelligence represents the latest stage in this historical trajectory. Unlike earlier machines that primarily augmented physical labor, AI systems increasingly augment cognitive labor, including pattern recognition, language processing, and decision support. However, increases in productivity do not inherently produce unemployment. Instead, their social consequences depend upon the institutional structures governing production and distribution.

3. The Political Economy of Automation

Within capitalist economic systems, technological innovation is typically introduced within the framework of profit maximization. In classical Marxian analysis, capital investment is divided into two components:

C + V

where C denotes constant capital (machinery, infrastructure, technology) and V denotes variable capital (labor wages). Profit derives from surplus value, denoted by S. The rate of profit is expressed as:

r = S / (C + V)

Under conditions of competitive capitalism, firms attempt to increase profitability by reducing labor costs and increasing mechanization [5]. AI technologies thus become instruments for reducing the proportion of variable capital in production. This process can increase profits for individual firms, but it simultaneously creates systemic contradictions. Workers who lose employment also lose purchasing power, which can undermine aggregate demand within the broader economy. This contradiction between productivity expansion and purchasing power reduction has historically contributed to periodic economic crises [6].

4. Artificial Intelligence as an Extension of Collective Human Intelligence

Contrary to popular narratives, AI does not emerge independently of human society. Rather, it is fundamentally built upon a vast foundation of collective human activity. AI capability may be conceptualized as:

AI = f(Kₕ, Dₕ, C, E)

where Kₕ represents accumulated human knowledge, Dₕ represents human-generated data, C denotes computational infrastructure, and E represents energy resources.

Modern AI systems depend extensively on human-generated linguistic, cultural, and scientific data. Large language models, for example, are trained on enormous corpora of human texts, scientific literature, and digital communication [7]. Furthermore, the scientific research underlying AI has been heavily supported by public institutions and publicly funded universities [8]. In this sense, AI represents a concentration of collective human intelligence, rather than an autonomous technological entity.

The very term ‘artificial intelligence’ is, therefore, philosophically misleading. What is designated as ‘artificial’ is, upon closer examination, nothing other than crystallised human intelligence — the accumulated linguistic patterns, logical structures, cultural meanings, and scientific reasoning of countless human generations, processed and recombined through computational systems. When a language model generates a coherent argument, it does so by drawing upon the cognitive labour of the philosophers, scientists, writers, and ordinary people whose words constitute its training data. The ‘intelligence’ exhibited is not created by the machine but summoned from the collective intellectual heritage of humanity.

This distinction carries profound political-economic consequences. If AI is recognised as an extension of human intelligence rather than a replacement for it, then the economic value it generates cannot be legitimately claimed as the exclusive property of those who control the computational infrastructure. The intellectual labour of all those who produced the underlying knowledge — including the vast, unnamed majority whose social and cultural output fed the training process — must be acknowledged as a constitutive force in AI capability.

Furthermore, the ideological framing of AI as ‘machine intelligence’ creates a dangerous discursive environment in which human cognitive labour is progressively devalued. If economic and policy discourse routinely describes AI as generating intelligence autonomously, the intellectual contributions of researchers, educators, writers, translators, and ordinary communicators are rendered invisible. This erasure serves the interests of those who own computational infrastructure by allowing them to present the value produced by collective human intellect as the product of their proprietary systems. The risk, therefore, is not only economic displacement but the systematic ideological discarding of human intellectual labour from the historical record of how knowledge and value are produced.

5. The Contradiction Between Social Production and Private Appropriation

The central socio-economic contradiction of AI can be expressed as:

Social Production ≠ Private Appropriation

AI systems are socially produced through the collective contributions of researchers, engineers, public institutions, and the broader public whose data forms the training foundation of modern algorithms. Yet the economic benefits generated by these systems are largely appropriated by private corporations controlling computational infrastructure and intellectual property. This contradiction closely resembles the structural dynamic identified in classical political economy, where the collective nature of production conflicts with the private appropriation of surplus value [5].

6. The Historical Plunder of Traditional and Indigenous Knowledge Under Intellectual Property Rights

The contemporary appropriation of collectively produced data and knowledge by AI corporations is not without historical precedent. The modern intellectual property rights (IPR) regime has, over several decades, functioned as a mechanism through which corporations have systematically extracted, commodified, and privatised knowledge that belonged to communities, peoples, and the common heritage of humanity. An examination of several well-documented cases reveals the structural continuity between earlier forms of knowledge plunder and the current accumulation of data by AI developers.

One of the most extensively documented examples concerns biodiversity and traditional medicinal knowledge. The case of the neem tree (Azadirachta indica), whose antifungal and pesticidal properties had been known and utilised by Indian and other South Asian communities for centuries, illustrates this dynamic clearly. In 1994, the European Patent Office granted a patent to the United States Department of Agriculture and the multinational corporation W. R. Grace for a method of producing a fungicidal product derived from neem. This patent was eventually revoked in 2000 following a sustained legal challenge led by the Research Foundation for Science, Technology and Ecology in India, which demonstrated that the claimed invention was based entirely on traditional knowledge already in the public domain [11]. The episode revealed how IPR frameworks, designed ostensibly to reward innovation, were in practice being used to enclose and commodify knowledge that communities had developed and freely shared across generations.

A parallel case concerns turmeric (Curcuma longa), another plant whose therapeutic uses — including wound healing and anti-inflammatory applications — have been extensively documented in traditional Indian medical texts for thousands of years. In 1995, two scientists at the University of Mississippi Medical Center were granted a United States patent on the use of turmeric for wound healing. The Indian Council of Scientific and Industrial Research (CSIR) successfully challenged this patent before the United States Patent and Trademark Office in 1997, providing documentary evidence from ancient Sanskrit texts and the Ayurvedic tradition to establish prior art [12]. This case prompted India to develop the Traditional Knowledge Digital Library (TKDL), a database of codified traditional knowledge specifically designed to prevent the future misappropriation of indigenous and traditional knowledge through the patent system.

The phenomenon extends beyond medicinal plants to encompass agricultural biodiversity and genetic resources. The patenting of basmati rice varieties by the American company RiceTec Inc. in 1997 provoked widespread outrage in India and Pakistan, as it claimed proprietary rights over strains that South Asian farmers had cultivated, selected, and refined across centuries of agricultural practice. Although several claims in the patent were eventually narrowed or cancelled following diplomatic and legal pressure, the episode demonstrated the willingness of corporate actors to appropriate collectively produced biological and cultural knowledge as private intellectual property [13].

These cases share a common structural logic. Communities invest centuries of observational, experimental, and experiential labour into the development of knowledge about the natural world. This knowledge is then extracted, nominally reformulated, and registered as private property under an intellectual property regime that does not recognise collective, intergenerational, or community-based modes of knowledge production. The original producers of this knowledge receive no royalties, recognition, or compensation. The entire accumulated value of their intellectual labour is transferred to corporate ownership through the administrative mechanism of patent law.

This historical pattern is now being reproduced at a qualitatively different scale through the development of AI. The training datasets of contemporary large language models encompass billions of texts, images, and other cultural artefacts produced by human beings across the globe. This includes the published works of authors, journalists, and academics; the creative output of artists and translators; the oral traditions and cultural expressions of communities that have been transcribed and digitised; and the everyday communicative labour of ordinary people expressed through digital platforms. All of this has been absorbed into proprietary AI systems without consent, compensation, or attribution. The intellectual labour of the world’s population has been privatised through the accumulation of data, just as the botanical knowledge of South Asian communities was privatised through the patent system. The mechanism differs in its technical character but is identical in its political-economic logic: the enclosure and private appropriation of collectively produced knowledge.

The legal and institutional frameworks governing this process remain deeply inadequate. Current copyright law was developed in an era of individual authorship and discrete creative works; it does not accommodate the collective, distributed, and often anonymous character of the knowledge upon which AI systems are trained. Calls for data sovereignty, algorithmic transparency, and the recognition of collective intellectual contributions are gaining momentum in policy discourse [14], but the structural power of the corporations controlling AI infrastructure continues to impede meaningful reform.

7. Historical Analogy: Machinery and the Industrial Revolution

The concerns surrounding AI strongly resemble earlier debates during the Industrial Revolution. In the early nineteenth century, mechanization provoked widespread fear that machines would permanently displace human labor. Movements such as the Luddites famously resisted industrial machinery. However, historical analysis shows that machines themselves were not the fundamental problem. Instead, the social consequences of mechanization were shaped by the economic structures within which they were deployed [9]. If the gains of mechanization had been socially distributed, they could have produced reduced working hours and improved living standards for all. Instead, the benefits were largely captured by industrial capital, producing both wealth concentration and worker displacement. The same contradiction now reappears in the context of artificial intelligence.

8. Socialization of Technological Gains

Alternatively, if the productivity gains generated by AI were socially distributed, technological progress could produce profoundly beneficial outcomes. For example, if AI were to double productivity across certain sectors, working time could theoretically be reduced:

T_new = T_old / α

where α represents the productivity increase. If productivity doubled (α = 2), an eight-hour workday could theoretically be reduced to four hours while maintaining equivalent output. Such outcomes would represent a transformation of social life rather than a crisis of employment.

9. Human Intelligence and Artificial Intelligence: A Necessary Clarification

The claim that AI will surpass human intelligence also requires careful philosophical and empirical scrutiny. Current AI systems operate primarily through statistical inference and pattern recognition. Mathematically, their operation can often be described through conditional probability models:

P(y | x)

where outputs are generated based on learned probability distributions derived from human-produced data. Human intelligence, however, encompasses dimensions that extend far beyond statistical computation, including:

• conscious intentionality
• ethical judgement
• historical reasoning
• creative imagination
• embodied social experience
• the capacity for genuine conceptual innovation

Consequently, AI should be understood not as a replacement for human intelligence but as a technological extension and recombination of recorded human cognitive output. The risk of designating AI as a superior or autonomous intelligence lies not merely in philosophical inaccuracy but in its political-economic consequences: it legitimises the displacement of human workers and the concentration of AI-derived value in the hands of those who control the infrastructure, by rendering invisible the collective human intellect that actually produces that value.

In the name of ‘machine language’ and ‘machine learning,’ there is a growing tendency in economic and policy discourse to treat human labour and intelligence as merely transitional inputs that can be fully superseded once the technology matures. This framing must be challenged. Human intelligence is not a temporary scaffold that can be discarded once AI systems are built; it is the continuous, living source from which all further AI development must draw. Future AI systems will require new human-generated data, new human conceptual frameworks, and new human communicative practice. The notion that AI can achieve independence from human intellectual labour is not only empirically unfounded but politically dangerous, as it provides ideological cover for the erosion of labour rights, wages, and social recognition for cognitive workers across all sectors of the economy.

10. Toward a Dialectical Resolution

From a dialectical perspective, technological development generates contradictions between productive forces and social relations. The development of AI intensifies this contradiction by dramatically expanding productive capacity while existing economic structures concentrate the resulting benefits. The resolution of this contradiction requires institutional arrangements that align the social character of technological production with the distribution of its benefits. In practical terms, this may involve:

• public ownership of key AI infrastructure
• democratic governance of data resources
• social distribution of productivity gains
• reduction of working hours in proportion to technological productivity
• legal recognition of collective and indigenous knowledge as prior art in patent and data law
• equitable compensation frameworks for communities and individuals whose data and intellectual output constitute AI training sets

Such measures would allow AI to function as an instrument of collective human development rather than a mechanism of exclusion.

11. Conclusion

Artificial Intelligence represents one of the most powerful technological achievements of modern civilisation. Yet the central issue raised by AI is not technological but social. AI is fundamentally built upon the accumulated knowledge, creativity, and data of humanity as a whole. The contradiction emerges when this collectively produced intelligence becomes the private property of a narrow economic elite.

This paper has argued that the current moment in AI development is structurally continuous with earlier episodes of knowledge plunder: the patenting of traditional medicinal and agricultural knowledge by corporations in the late twentieth century was a rehearsal for the far larger enclosure of humanity’s collective intellectual heritage now occurring through the appropriation of data for AI training. In both cases, the intellectual labour of communities is absorbed without recognition or compensation, and its value is transferred to private ownership through the instruments of intellectual property law and capital accumulation.

The paper has also challenged the philosophical framing of AI as an autonomous or superior intelligence. What is called artificial intelligence is more accurately described as extended human intelligence — a sophisticated recombination of the cognitive labour of countless human beings, mediated by computational systems owned by a small number of corporations. The ideological risk of accepting the ‘artificial intelligence’ framing uncritically is that it renders human intellectual labour invisible, delegitimises demands for fair compensation and recognition, and provides cover for the progressive erosion of human workers’ economic position in an AI-saturated economy.

Consequently, the future of AI will depend not on the inherent capabilities of machines but on the social structures governing their use. If the benefits of AI are socially distributed and the collective character of its intellectual foundations is legally and politically recognised, the technology could enable unprecedented reductions in labor time, expanded educational opportunities, and greater cultural flourishing. If they remain privately concentrated, however, AI may deepen existing inequalities and social tensions in ways that recall, and vastly exceed, the dislocations of the industrial revolution. The question of artificial intelligence is therefore inseparable from the broader questions of how society organises the ownership and distribution of technological power, and how it recognises and rewards the collective human intelligence upon which all technological progress ultimately rests.

References
[1] Brynjolfsson, E., & McAfee, A. The Second Machine Age. W. W. Norton, 2014.
[2] Russell, S., & Norvig, P. Artificial Intelligence: A Modern Approach. Pearson, 2020.
[3] Bostrom, N. Superintelligence: Paths, Dangers, Strategies. Oxford University Press, 2014.
[4] Mokyr, J. The Lever of Riches: Technological Creativity and Economic Progress. Oxford University Press, 1990.
[5] Marx, K. Capital, Volume I. Progress Publishers.
[6] Mandel, E. Late Capitalism. Verso, 1975.
[7] Goodfellow, I., Bengio, Y., & Courville, A. Deep Learning. MIT Press, 2016.
[8] Mazzucato, M. The Entrepreneurial State. Anthem Press, 2013.
[9] Hobsbawm, E. Industry and Empire. Penguin Books, 1968.
[10] Varian, H. Artificial Intelligence, Economics, and Industrial Organization. NBER Working Paper, 2019.
[11] Shiva, V. Biopiracy: The Plunder of Nature and Knowledge. South End Press, 1997.
[12] Gupta, A. K. ‘WIPO-UNEP Study on the Role of Intellectual Property Rights in the Sharing of Benefits Arising from the Use of Biological Resources and Associated Traditional Knowledge.’ WIPO, 2004.
[13] Leskien, D., & Flitner, M. ‘Intellectual Property Rights and Plant Genetic Resources: Options for a Sui Generis System.’ Issues in Genetic Resources No. 6, IPGRI, 1997.
[14] Zuboff, S. The Age of Surveillance Capitalism. PublicAffairs, 2019.