Artificial Intelligence in the Capitalist University

Academic Labour, Commodification, and Value

John Preston

• Publisher: Routledge
• Year: 2021
• Print ISBN: 9781032123622
• eBook ISBN: 9781000471502
• Edition: 1
• Topic: Education
• Subtopic: Education Theory & Practice

Using Marxist critique, this book explores manifestations of Artificial Intelligence (AI) in Higher Education and demonstrates how it contributes to the functioning and existence of the capitalist university.

Challenging the idea that AI is a break from previous capitalist technologies, the book offers nuanced examination of the impacts of AI on the control and regulation of academic work and labour, on digital learning and remote teaching, and on the value of learning and knowledge. Applying a Marxist perspective, Preston argues that commodity fetishism, surveillance, and increasing productivity ushered in by the growth of AI, further alienates and exploits academic labour and commodifies learning and research. The text puts forward a solid theoretical framework and methodology for thinking about AI to inform critical and revolutionary pedagogies.

Offering an impactful and timely analysis, this book provides a critical engagement and application of key Marxist concepts in the study of AI's role in Higher Education. It will be of interest to those working or researching in Higher Education.

1

Introduction

DOI: 10.4324/9781003081654-1

What Is AI?

Artificial Intelligence (AI) is one of the most nebulous technologies in contemporary capitalism. It has come to stand for a wide variety of machines, automations, simulations, and speculations. Intelligent assistants such as Alexa and Cortana, Machine Learning (ML), cyborgs, robots, expert systems, downloaded minds, automata, self-driving cars, pattern recognition, transhumanists, post-humanists, semantic nets, neural nets, natural language processing, godlike consciousnesses, and future automated existential threats are all parts of what might be called AI. In many ways there is nothing new about AI. Humans have always imagined enchanted objects and mechanical devices that might approximate some form of consciousness. Cave, Dihal and Dillon (2020) span human history in bringing together fictional and non-fictional accounts, of ‘AI Narratives’. Several chapters in their collected volume identify pre-capitalist accounts of AI from Greek philosophy and the Middle Ages. Although these are not AI as we might understand it, these conceptions of intelligent, reasoning, or conscious ‘things’ are already embedded in a historical mode of production (slavery or feudalism). In contrast, contemporary AI is largely a capitalist technology and (as I argue in Chapter 6) we often look at advanced AI not as a slave or a serf but as an owned machine (as part of capital) and speculatively as a future conscious, or even ‘labouring’, entity.

The contemporary history of AI (Woolbridge, 2020) makes clear that it is an over-burdened and contested concept. From the work of Turing until the development of semantic AI in what Woolbridge (2020, p.47) refers to the ‘Golden Age’ (from 1956 to 1974), there was an emphasis on ‘General AI’ with the prioritisation of general skills across a wide range of domains including perception, problem-solving, planning, and understanding. The perceived failures of this ‘Golden Age’, as these higher-order problems proved to be extremely challenging, led to a narrowing of the focus of AI (‘narrow AI’). AI technologies such as expert systems (which combined AI with human reasoning to address problems in a specific domain), behavioural AI (that focussed on AI and robot behaviour, rather than reasoning), and AI assistants (that worked alongside humans to solve problems) were more successful but still not a step towards more general AI, AGI (Artificial General Intelligence).

Contemporary AI is often associated with ML that is based on statistical techniques, big data, and massively increased computing power to build ‘neural nets’ that can be taught to solve specific problems. The ‘neural nets’ involved are not synonymous with neurons in a human, or even a computational brain, but allow practical gains in terms of pattern recognition and strategy (particularly in terms of games such as Chess, Go, or video games). ‘Narrow AI’ and ML are particularly important in terms of military, business, and government (especially policing) applications. Woolbridge (2020, p.210) explains that there is now a schism between ML and AI so wide that some researchers in ML do not consider themselves to be AI researchers at all. Although AI was never a ‘pure’ discipline, the move from AGI to narrow, specific, applications resonate with commercial imperatives in terms of increasing business profitability, imperialist prowess, and state control and repression. This tendency reflects the development and appropriation of science and technology for capitalism.

From the outset, it is important that we understand the multiplicity of AI technologies, and that AI is not synonymous with an electronic ‘mind’ or ‘brain’, although conceptions of these are also a part of capitalist expression and understanding of AI. Dyer-Whiteford, Kjøsen and Steinhoff (2019) argue that ‘…it is only through some familiarity with the science of technology of AI that an effective critique can be mounted’ (Dyer-Whiteford et al, 2019, p.9), but it is increasingly difficult to identify a consistent AI scientific project. AI is obviously distinct from forms of mechanisation or computation but might include many different techniques and approaches. Following AI theory more generally, Dyer-Whiteford et al (2019, p.10) distinguish between (narrow) AI, AGI, and Artificial Super Intelligence (ASI). Narrow AI would be able to undertake well-defined, bounded tasks and demonstrate intelligence in a constrained domain, AGI would be able to undertake a broad range of generalisable tasks across broad domains, and ASI would surpass human intelligence across many areas. There have been a number of fictional representations of ASI such as HAL 9000 (2001), Skynet (The Terminator), The Matrix, and San Junipero (Black Mirror). The ASI existential threat has been a persistent one in fiction whether it is from a psychotic ASI in space, one purposively starting a nuclear war, or an AI creating a simulated universe for human pain or pleasure. These threats seem far from our current experience of an automated utopia. We don’t look like cyborgs, even though our hands are glued to tiny cellphone screens at work or home. We are not going into cyberspace, even though we will spend most of our day on the internet. At home we do talk to an AI to find out the weather tomorrow, but nothing substantively changes. Things are always moving and shifting as technology constantly changes, but the repetition of work and consumption continues.

The impact of AI on our lives, and what AI ‘is’ are, of course, subjects of AI ethics (Coeckelbergh, 2020). Although AI ethics concerns itself with social impacts, redistribution, transhumanism and post-humanism, environmentalism, racism, gender, and discrimination, it often does so within frameworks where there is no alternative to capitalism. AI can only be understood as capitalist AI. In the workplace, particularly in the factory, AI is just another form of technology which eliminates workers and skills whilst not necessarily creating new jobs or alternative forms of vocation. It is often acting to privatise existing skills into expert systems. As we live in a capitalist society, it is impossible to even think about AI outside of capitalism. We conceive of the ultimate AI (ASI) as a cunning, learning superintelligence. As with human labour before it, the first task of capitalism appears to be primitive accumulation to attempt to contain AI and turn it into a commodity which can be sold. Our views of AI are shaped by this capitalist schema. This schema is not some idealist thought about AI that can be debated through counterargument but a real, material, circumstance of our lives. Again, there is little that is new about this. As labourers, we have always had to adopt prosthetics at work. We wear glasses and hearing aids. Our ability to work is gauged in terms of whether a prosthetic might help us to do so. At work we wear uniforms, suits, welding masks, exoskeletons, and carry phones. Our brains have developed neurochemical pathways that are triggered by the thought of money. We have a symbiotic relationship with knowledge and the cyber-physical. We are not becoming ‘cyborgian’, capitalism brought us into that tightening relationship with machinery since day one.

There have been recent attempts to consider AI within Marxist theories. Significantly, Dyer-Whiteford et al (2019, p.15) refer to the centrality of AI as machinery in Marx's work, in terms of a supplement to human labour and as fixed capital as compared to the variable capital of labour. Whilst it is human labour that creates value, machines are ultimately only a ‘…supplement or force-amplifier’ (p.16) transferring their own value to the product. Their ‘social function’ is to reduce Socially Necessary Labour Time (SNLT) to produce ‘…relative surplus value’ (p.16) but this leads to greater competition between capitals and an increase in the ‘organic composition’ of capital (p.17). Machinery confronts and absorbs the worker, bending the task to its requirements and drawing in labour. Dyer-Whiteford et al (2019, p.17) interpret the impact of machinery on crisis as being both through the imbalance between machine driven production, falling wages, and the falling rate of profit (p.17). Dyer-Whiteford et al (2019, p.19) also considers that capital will mobilise the ‘general intellect’ relegating human labour to the supervisory process but also ‘undermining value’ and ‘abolishing work’, thereby foreshadowing capitalist collapse. This is part of formal and real subsumption under capitalism in terms of the changing of labour's social form (under formal subsumption) to where the ‘…content of labour changes’ (p.20), moving from the extension of the working day and an increase in absolute surplus value to the production of relative surplus value. The dawning of AI-Capital brings about a further stage of ‘hyper-subsumption’ (p.21) where ‘capital's autonomizing force manifests as AI’ (p.21). This is not a process that continues indefinitely. Increases in relative surplus value over time results in a lowering of the total surplus value (and hence profit), unless there is constantly expanding accumulation. AI represents a way out for capitalism as it will try ‘…with all force to maintain the value of value’ (Kurz, 2014, p.54). However, to consider that machines can produce value is part of what Kurz (2014, p.41) calls an ‘immortalization of value’ a ‘failure to escape the value fetish’. AI is best seen, then, as the continuation of capitalism in decline rather than prefigurative of a new and expansive mode, and this perspective sets the tone for this book.

AI and Bias

As it is concerned with capitalism, this book takes an unusual angle in not being overly concerned with tracking the impact of specific aspects of AI on equity or social justice. Obviously, within capitalism, AI inevitably reproduces and produces massive inequalities, but it does not particularly matter to the argument presented here whether AI can, or should, be reformed to produce equitable outcomes, if AI is fair, or whether it is even accurate. By its nature capitalist work (whether it uses AI or not) results in inequalities between capitalists and labourers (and between groups of workers) and exploits and immiserates labour. This book is not primarily based on the moral case for addressing such inequalities as notions of injustice are primary to capitalism and obscured by principles such as ‘fair exchange’ (of course this does not prevent the use of moral language to describe the brutality, horror, and misery of capitalism). This is distinct from the approach taken in much of the critical work on AI, which considers the impacts of ML algorithms on social justice. In this work, social justice is taken to be a judgement on the equitable functioning of already existing capitalist processes such as the criminal justice system, autonomous technologies (such as driverless cars or autonomous weapons), or the allocation of benefit payments. AI is considered to reinforce existing biases, or to create new biases, but the capitalist system is not called into question. For example, Benjamin (2019a) shows how AI exacerbates existing racial biases in society in the criminal justice system, health, and education. Although this literature on bias is useful in mounting a critique of AI in Higher Education, the response that it elicits is largely affirmative in that it recommends better or different systems of AI, or the abolition of AI, inside an existing order (capitalism). The question then becomes how AI can act as a tool in social-democratic societies rather than addressing how AI supports the maintenance of systems (particularly capitalism). It also leads to critiques that sometimes involve simply listing the ‘bad things’ about AI in a particular area. For example, the argument concerning the decolonisation of Higher Education (HE) is an active debate at present on both sides of the Atlantic. Within this discourse, there is much current work considering the ways in which AI is in some sense racist, white, or colonialist in nature supported by work on ‘race’ and educational technology (Benjamin, 2019a, 2019b). Although there is a political and moral imperative to consider the social justice impacts of HE, sometimes arguments based on the socially just aspects of technology are prone to commodity fetishism because this work is based on the material features of the commodity form. This could be in terms of the appearance of AI (in terms of, perhaps, whiteness or its gendered nature), the nature of the algorithm, and the principles of AI in terms of logic which are thought somehow to be connected with racism or particular Eurocentric perspectives. Even in its own terms, this literature is sometimes based on Anglo-European and American perceptions and depictions of AI rather than considering Asian, Afrocentric, or Afrofuturist perspectives on AI (Benjamin's 2019b collection is an exception to this as it does consider non-Eurocentric contributions to the bias literature). In contrast, it is possible to take a Marxist view on humanism, in terms of the human species and species being, which is not necessarily based on essentialist conceptions of human nature, instead being grounded in a theorisation of human labour in capitalism. Through primitive accumulation, a labouring subject (collectively, the working class) is brutally separated from the property needed to produce commodities, from nature and from the means to sustain life. Only through capitalist labour (selling labour power as abstract labour) can the worker survive, and this is only through an inhuman system of really existing abstractions that validate abstract labour through commodity exchange (Pitts, 2018). This is not to deny the important work that is being done on race, gender, sexuality, and disability and AI in HE but to signal from the start the approach taken in this book is not particularly concerned with the accuracy (bias) or the equity of AI.

Structure of the Book

In this book I take a broad perspective on the different manifestations of AI in the capitalist university. To take a comprehensive view of approaches to AI, I consider AI in terms of hardware and software, ML, narrow AI, AGI, ASI, and optimisation. I examine the impact of AI on the control and regulation of academic work, its impact on digital learning, holography and remote teaching, the creation of digital commodities, its status as an academic field, ethical considerations, digital platforms, and existential threats to the university. The influence of AI on the capitalist university is not just in terms of its impact on replacing and subsuming academic labour but also in terms of commodity production, circulation, exchange, and the ‘abstract domination’ of capital. This includes the ways in which the social universe of capital acts back on each individual capitalist university through artefacts such as fatigue functions for workers and academic league tables that are the concrete forms of a ‘real abstraction’ that imposes increasingly pressurised work conditions on academic labourers. AI within the capitalist university is part of the totality of capital and capitalism. Despite this, what seem to be totalising capital relations are paradoxical, incomplete, and contested and in every moment that production and exploitation advance capitalism, it finds itself challenged, and it weakens, exposing alternatives that are prefigurative within itself. The critique presented here is, therefore, a negative critique, one that does not seek to reform the capitalist university, or reimpose a new model of political economy, but considers prefigurative possibilities beyond capitalism, including whether there are possibilities for AI in communism, beyond class and the ‘law of value’.

The empirical work on which the book is based emerged not from universities, but from research on the future of ‘digital manufacturing’ which involved examining the trajectories of AI, robotics, and advanced production through a variety of case studies. In this work, which involved ‘manufacturing as a service’ and digitisation the ways in which service industries were increasingly adopting the methods of ‘digital manufacturing’ became apparent. Digitisation, AI and ML, and optimisation are modular technologies that can be inserted into any business operation, and where differentiation between industry sectors is becoming less marked. Similar systems to those used by Uber, Amazon, and Netflix are also employed in manufacturing and universities, and models can be transferred between firms and industries. The university is increasingly concerned with commodity production in which digitisation and AI shape the form of the commodity produced and influence all aspects of activity (production, academic work, and time). Whatever was special about the university when compared to any industry or commercial sector is becoming a trace memory, erased by capitalisation, and dissolved by digitisation.

In Chapter 2, I argue that all global universities are capitalist universities. Whether the economic system is classified as a market, state, or hybrid form, the social relations and social forms in which universities exist are capitalist ones. Even those universities (which are often not formally recognised in national institutional systems) that are explicitly anti-capitalist are subject to the ‘vortex’ of the law of value (Neary and Winn, 2017). The capitalist nature and status of universities means that their attempts to visualise a more humanising, or even a more technical, vision of education, are consistently subject to the reduction of all entities and expressions to economic valuation and profit. Marx's work on value, its expression as ‘use value’ and ‘exchange value’, and the social forms of its existence (particularly in the commodity and in its objective form as money), provides the foundation for a consideration of the New Reading of Marx (Pitts, 2018) and value critique in the analysis of AI and HE. The New Reading of Marx (NRM) takes a historical perspective on work, labour (particularly abstract labour), and regards capitalism as (necessarily) a class system that imposes a mode of ‘abstract domination’, creating capitalist society that is purely concerned with commodity production and the accumulation of money (the objective form of value). Labour, as a specific historical form (our labour power, abstract labour, being a form that only exists in capitalism), is only socially validated (valued in capitalist society) if it is producing value i...

Table of contents

• Cover
• Half Title
• Series Page
• Title Page
• Copyright Page
• Contents
• Acknowledgements
• 1 Introduction
• 2 Capitalist Universities, AI, and Value
• 3 AI and the Subsumption of Academic Labour
• 4 The Pixarfication of Higher Education
• 5 Disaster Capitalism and Time in the Virtual University
• 6 AI, Existential Threat, and the Capitalist University
• 7 AI and the End of the Capitalist University
• Index