Judging automation: Towards a normative critical theory

Introduction

Automation and the artificial intelligence (AI) industry are growing at an accelerating pace. There are now robots in almost all spheres of human life – from lawnmowing, floor cleaning and other domestic appliances to speech writing, chess playing, translation, facial recognition, and algorithmic model making. A vast amount of literature has also been produced on the subject over the past several years from a variety of points of view, ¹ including contributions from (neo-)Marxist scholars. ² Less intensively researched are reflections on automation from the points of view of human flourishing. ³ The question that needs to be posed is: What role does automation play in our lives and, given its current forms in today’s capitalism, what does it do to human flourishing? No doubt, this question cannot be answered in a single paper because of its scope and complexity. Nonetheless, it will be seen as a background question.

Any attempt to address the issue of assessing automation in the light of human flourishing depends on what we mean by ‘human flourishing’. Relying on my previous research (Bielskis 2017), this article will start with a brief account of ‘human flourishing’. Aristotle’s idea that the good human life is a life of activity will serve as a premise for our discussion. To claim that life consists of different activities is to recognise that life is always situated, historical, and is embodied in an existing institutional setting. The latter will be analysed from the point of view of historical materialism – that is, advancements in automation and AI will be discussed from the point of view of both the forces of production and the relations of production. It was Aristotle’s (2009: 1253b35–39) αὐτομάτους (automatous) in Politics that gave rise to the contemporary concept of automation. This article will trace how his notions of slavery, hard labour (douleia) and imagined self-moving machines were echoed in the classical political economy of David Ricardo and Karl Marx. Looking at automation from the point of view of historical materialism, it promises to free us from the slavery of hard work, yet the legal and political superstructure of capitalism means that the growth of automation in industry produces new forms of precarious labour. Finally, I will discuss Alasdair MacIntyre’s neo-Aristotelian notion of practice vis-à-vis Marx’s conception of alienated labour in capitalism. Given the conceptual structure – alienated labour (which prevents us from flourishing) versus non-alienated labour (as essential for human flourishing) – the question I pose is: Can we apply the normative-historical-philosophical schema of Aristotle-Marx-MacIntyre thus outlined to assess the current forms of automation?

A note of method: Good versus bad automation?

Assessing automation is an important theoretical and practical task. Technological advancement has been the primary feature of social life ever since the Industrial Revolution in the late 18th century. However, its benefits have not been equitably distributed within societies, while automation has meant diminishing opportunities for the middle classes (Frey 2019: 11). Automation in the domestic sphere has undoubtedly improved people’s lives (e.g. window-cleaning, floor-cleaning, and grass-cutting appliances), but automated customer services in the private or public sectors often contribute to frustration and social atomisation (e.g. consider airline and flight-ticket platforms such as eDreams: they have been fully automated since the COVID-19 pandemic, which comfortably allows them to avoid customers’ complaints when their flights are cancelled). In the sphere of policing, automation has also contributed to the endorsement of racist and sexist bias and has disproportionately targeted the poor and marginalised (Eubanks 2018: 11). Furthermore, scholars have shown that automation has been consistently employed in the process of the valorisation of capital at the expense of living labour, contributing to the deskilling of workers, especially during recessions (Blit 2020; Steinhoff 2021; Wendling 2009). Thus, the key question that needs to be posed is: How are we to assess the development of automation conceptually and theoretically? Are all forms of automation equally desirable? A task of theory is to provide a conceptual apparatus to evaluate social phenomena. What follows is an attempt to fashion a theoretical approach which would allow us to evaluate the current forms of automation.

There is a growing body of philosophical literature on technology and automation. Ontological accounts of technology have been developed from a Deleuzean point of view (Smith 2018), focusing on technological artefacts as externalised organs. Ever since Martin Heidegger’s (1977 (1954)) essay ‘The Question Concerning Technology’, a strong interest has been developed in phenomenological reflections on technology and its all-penetrating nature. Following Heidegger, most of these accounts have been predominantly negative interpretations of technology and its far-reaching social effects. ⁴ Herbert Marcuse, in part, took on board Heidegger’s negative account of technology yet developed a mixture of phenomenology and neo-Marxist critical theory in his critique of ideologically driven technological societies (Marcuse 2002). A contemporary follower of Marcuse, Andrew Feenberg (1999, 2010) has both criticised Marcuse’s dismissive attitude towards technology in capitalism and continued the theoretical critique of its colonising of aspects of the lifeworld. Marxist scholars have also contributed to the debates on the nature of technology in capitalism (e.g. Dyer-Witheford 1999; Steinhoff 2021; Wendling 2009). On the other hand, there is also a growing body of literature on technology and AI written from the broadly understood point of view of Aristotelian virtue ethics (Ratti and Stapleford 2021; Vallor 2016).

What has not been undertaken so far is an attempt to develop a theoretical intersection between Aristotelian and Marxist accounts of understanding and assessing the advancement of technology and automation. Both perspectives and their intersections are important because, on the one hand, Marx and Marxian critique allow us to analyse the ongoing process of technological advancement from the point of view of dominant power relations, while the Aristotelian emphasis on the good life and moral excellence enables us to judge concrete embodiments of technology and automation in the light of human flourishing. However, taken separately, they miss what is essential. The Marxian critique of the process of the valorisation of capital and its forms today as well as Marx’s conception of autonomy and emancipation need further spelling out. ⁵ The question that we need to pose is: What are the ethical conditions of emancipation in a society of non-alienated social relations? On the other hand, Aristotelian virtue ethics will be naïve without Marxian sociology, without historical materialism which situates contemporary advancements in AI and automation against the background of the dialectical relationship between the forces of production, of which the AI industry is an important part and the relations of production. To conceptualise these theoretical intersections between Aristotelianism and Marxism in order to evaluate automation will be the main task of this article. Hence, the question I am posing is not ontological. It is not what AI and automation are. Rather, the question is: What role does automation play in our lives and, given its current forms in today’s capitalism, what does it do to human flourishing?

Amy Wendling’s work has been a notable exception in this respect. It puts both of these perspectives together. In her outstanding book, Marx on Technology and Alienation, Wendling argues that the abundance of material wealth needed for creating a communist society, where technology is directed towards the creation of use values rather than being subordinated to exchange value, as well as free time for self-development are required. Although Wendling does not use Aristotelian vernacular, she asserts that Marx’s critique and account of alienation in capitalism provide us with the opportunity to move beyond technophobic approaches to technology in the vein of Heidegger. In line with Marx’s historical materialism, Wendling sees technology’s liberating aspect in a free society where the system of technology-mediated wealth creation aims at generating various use values to satisfy diverse needs and bring enjoyment. By tracing Marx’s intellectual journey and his oscillation between labour as a Hegelian form-giving activity and the thermodynamic notion of energy, Wendling argues that Marx’s insight should be geared towards emphasising freedom from labour and free time for enjoyment. According to her, Marx’s least theorised idea is that ‘human [higher] activity is not reducible to labour’, and she conceptualises enjoyment as the simultaneous cultivation of ‘mind, body, and emotions’ as well as ‘sociality and solitude’ (Wendling 2009: 87, 117). A part of such cultivation of mind is our ability to balance the public and private spheres and set time for the greater enjoyment of non-productive activities, such as caring for children and loved ones, nonreproductive sex as well as certain forms of religious life, and finally ‘the risibility of the practice of philosophy’ (Wendling 2009: 117).

The philosophical approach in this article is complimentary to Wendling’s work. However, by putting a stronger emphasis on the ethical nature of human flourishing, and life as activity, the theoretical articulation of Aristotle-Marx connection allows us to evaluate the existing life practices in capitalism from a normative point of view. Putting aside the philological issue of whether higher human activity is or is not reducible to non-alienated labour, ⁶ the key distinction drawn in this article is that between alienated labour as the instrument in the process of capital augmentation and meaningful non-alienated labour which serves in our pursuit of excellence and human flourishing.

Defining human flourishing

The importance of Aristotle, among other things, lies in what much later became known as his concept (and method) of teleology. Human life and its activities are directed to different goods. The first sentence of The Nicomachean Ethics captures his teleology most clearly: ‘[e]very art and every inquiry (methodos), and similarly every action (praxis) and choice (prohairesis), ⁷ is thought to aim at some good (agathon); and for this reason, the good has rightly been declared to be that at which all things aim’ (Aristotle 1980: 1094a1-2). The claim that some activities and their corresponding goods are more valuable than others and there are goods that we pursue for their own sake is concluded by Aristotle’s thesis that there must be the final good – the highest telos – for the sake of which we do everything else. This supreme end Aristotle calls eudaimonia – customarily translated as ‘happiness’, but a better rendering is ‘human flourishing’. ⁸ Martin Heidegger (2009) was right when he argued that ‘Aristotle had no “teleological” worldview’ and that ‘even a superficial understanding shows that τέλειον and τέλος do not mean “aim” or “purpose”’ (p. 57). Indeed, Aristotle’s so-called function (ergon) argument overrules any interpretation of telos in terms of the highest aim or purpose. Rather, eudaimonia as the highest good is activity understood as the exercise of our specifically human faculties while practicing (moral) excellence. This is what I take Aristotle’s formulation of the highest good as ‘activity of soul in the exercise of aretē’ (Aristotle 1980: 1098a17) to mean in modern vernacular. What distinguishes us from plants and animals and, therefore, is specific to humans, are our linguistic-rational powers (logos). To live a flourishing life is to actualise our rational-creative powers while exercising moral and intellectual excellences – or virtues – such as courage, moderation, justice, prudence, and philosophical wisdom. The good life is an activity corresponding to ergon – characteristic work and human ‘function’ – which consists of the variety of ways of using our rational powers. Finally, since human beings are political animals, individual human flourishing is impossible without collective human flourishing, without a collective ability to achieve the common good politically. ⁹

For the present purposes, such characterisation of human flourishing is important because it conceptualises life as an activity (in terms of energeia and praxis) and because there is already normativity inscribed in it. Contrary to other theories of normativity (such as deontology and utilitarianism), normativity here is inscribed in action itself – it is not external to it. Certain human activities are excellent insofar as they are characteristic of being human, thus acting them out or performing them contributes to human flourishing. So, creativity and our ability to reason well in achieving noble ends or exercising our rational powers in different games and meaningful practices (such as playing chess, reading and writing literature, philosophical reflections, creating art, and engaging in scientific enquiries) all contribute to human flourishing. Such conceptualisation of good human life therefore has important implications as far as the discussion on automation is concerned. Not all activities should be automated, but only those that infringe on our possibility to live creative and rational lives.

Aristotle’s ‘fragment on machines’

Aristotle’s account of human flourishing vis-à-vis discussion on automation is also important not only because it was Aristotle whose concept of automatous became the basis for the current use of ‘automation’, but also because he was the first to formulate, in a somewhat similar way, what Marx’s commentators called the ‘fragment on machines’ in his Grundrisse. It is important that the context of Aristotle’s claim on self-moving tools is his discussion on doulos and douleia:

We can imagine a situation in which each instrument could do its own work, at the word of command or by intelligent anticipation. A shuttle would then weave of itself, and a plectrum would do its own harp-playing. In this situation master builders (architectōn) would not need subordinates (hupētōn) and masters (despotais) would not need slaves (doulon). (Aristotle 2009: 1253b35–1254a1)

Given the technological advancement of the day, Aristotle’s notorious conception of ‘natural slaves’ as the animate tools of property for the performance of different tasks is treated here as an economic necessity. ¹⁰ His claim – that if we had tools moving ‘of their own motion (automatous)’, then we would have no need for slaves (douloi), slavish labour and servitude (douleia) – has a new meaning today in the age of automation. Without drawing the political conclusion that Aristotle wrongly drew, not only about the existence of ‘natural’ slaves but also about the exclusion of workers, women, artisans and everyone engaged in ‘slavish’ labour, we can nonetheless reasonably suggest that what Aristotle meant by doulos and douleia can be understood as alienated labour. Aristotle’s ideological conception of the human flourishing of slave-owners’ society suggests that there is no place for alienated labour in a good life. Following Aristotle’s distinction between the involuntary and the voluntary – a voluntary action is ‘that of which the moving principle is in the agent’ themselves, while involuntary actions are forced, compulsory or done out of ignorance ‘of which the moving principle is outside’ or ‘is acted upon’ by ‘men who [have] him in their power’ (Aristotle 2009: 1110a3-5) – we can provide a formal definition of alienated activity. Alienated labour is an estranged activity, the source of motion of which is outside the subject, who, out of external compulsion, is forced to act against their will or better judgement. Slaves were just such people in Aristotle’s Greece. Their labour was alienated: they were their masters’ tools, instruments without their own will. Marx’s notion of wage-labour as forced and alienated labour, as the ‘wage-slavery’ of 19th-century workers in factories, is conceptually close to that of Aristotle’s slaves. On the other hand, Marx’s conception of freedom, as we will see, is also, at least in part, akin to Aristotle’s notion of the flourishing of human agency. An essential part in a flourishing human life for Aristotle was the need to have free time for meaningful activities, which he conceptualised in terms of scholē – time set for engaging in the intellectual activities of self-cultivation, reflection and learning, without which eudaimonia is impossible. ¹¹

Aristotle’s legacy and classical political economy

Aristotle’s idea of self-moving tools and his notion of scholē, in their different forms, were continued by classical political economy. The tension between the free time of scholē and labour as douleia and the relationship between time and productive labour was locked in via classical political economy’s labour theory of value. It was Adam Smith who argued that the ‘value of any commodity (. . .) is equal to the quality of labour’ and that it ‘was not by gold or by silver, but by labour, that all the wealth of the world was originally purchased; and its value (. . .) is precisely equal to the quantity of labour’ (Smith 2007: 20). Smith argued that the use of land and capital also contributed to the value of a commodity and that the division of labour, trade and competition had produced conditions for wealth creation, at the core of which was competition and innovation, including investment into more sophisticated tools. It was David Ricardo, however, who realised the potential of machinery, speculating that if ‘machinery could do all the work that labour now does, there would be no demand for labour’ and ‘[n]obody would be entitled to consume anything who was not a capitalist, and who could not buy or hire a machine’ (Ricardo 1951–1973: 399–400). Much later, John Maynard Keynes, in his essay ‘Economic Possibilities for our Grandchildren’, originally published in 1930, argued that the development of technology and automation would shorten the working week to 15 hours in one hundred years’ time. Keynes’ naïve optimism was rooted not only in his belief in science, but also in the quasi-Aristotelian notion of freedom ‘from pressing economic cares’ and the ability to occupy our leisure in order to ‘live wisely and agreeably and well’ (Keynes 1963: 364). We are far less optimistic today: Keynes’ dictum that people (at least in most economically advanced societies) would work 15 hours per week in 2030 sounds as a utopia. Being a well-meaning liberal, Keynes never fully realised the nature of the self-valorisation process of capital and its exploitative nature. Hence, the renewed importance of Marx.

Marx’s ‘Aristotelianism’ and the notion of alienated labour

Marx was well-versed in Ancient Greek philosophy, including Aristotle. Calling Aristotle ‘the great investigator’ (Marx 1990: 151), Marx in part shared Aristotle’s conception of douleia and conceptualised his own notion of human ‘species being’ in a similar way to Aristotle’s conception of human flourishing. ¹² He saw Aristotle’s account of slavery as the consequence of a dominant ideology rooted in the slavery-based mode of production that did not allow Aristotle to theorise the concept of value rooted in labour. Marx’s claim, therefore, was that only when the universal equality of all human beings is fixed in a popular opinion can the value form become economic reality (Marx 1990: 152). Thus, abstract labour, as opposed to the concrete labour needed to produce a commodity with a particular use value, becomes the central concept of Marx’s political economy. As the human energy and skills needed to produce any commodity, abstract labour time becomes the source of the commodity’s value. Although the value form could only fully develop in capitalism, labour as such is ‘the condition of human existence (. . .) independent of all forms of society; it is an eternal human necessity which mediates the metabolism between man and nature’ (Marx 1990: 132). In capitalism, however, labour assumes its form of ‘wage-labour’ which, referring to Aristotle, Marx calls ‘wage-slavery’. Workers are free agents who sell their labour to anyone they like, while the buyers of labour do not own workers. However, the economic necessity of selling one’s labour in order to survive has a strong aspect of forced, involuntary activity. When at work, the worker is fully subordinated to the needs of capital, and the time spent working does not belong to the worker but to the buyer of labour. During this time, the worker does not functionally differ from the slave in the economy based on slave-labour: the worker is a tool in the service of capital production.

Thus, Marx conceptualises alienated labour as an activity that produces not only a commodity but also ‘itself and the worker as a commodity’ (Marx 1988: 71). Instead of being the result of the free creative powers of the worker, commodity confronts the worker ‘as something alien, as the power independent of the producer’ (Marx 1990: 71). Even if it appears ‘as the result of free contractual agreement’, it nonetheless has a character of forced labour as it creates surplus labour which the capital owner receives without an equivalent or payment (Marx 1991: 958). Marx’s conclusion in the Manuscripts is that alienated labour has several interrelated aspects. Workers engaged in wage-labour are alienated: (1) from the products their labour produces; (2) vis-à-vis the organic creative process; (3) from fellow workers (given that labour is sold on the market and workers therefore compete among themselves, as all other homogeneous commodities do); and (4) from their ‘species being’, as they are forced to sell their labour in order to survive rather than to actualise their powers and flourish as human beings.

Marx on freedom, necessity and automation

Not everything Marx wrote on alienated labour is equally convincing and relevant today. However, before we assess these claims, it is important to look at one aspect of Marx’s account of freedom – namely, its relation to natural necessity. In Capital Volume III, Marx claims,

The realm of freedom really begins only where labour determined by necessity and external expediency ends; it lies by its very nature beyond the sphere of material production proper. (. . .) This realm of natural necessity expands with his development, because his needs do too; but the productive forces to satisfy these expand at the same time. Freedom, in this sphere, can consist only in this, that socialized man, the associated producers, govern the human metabolism with nature in a rational way, bringing it under their collective control instead of being dominated by it as a blind power; accomplishing it with the least expenditure of energy and in conditions most worthy and appropriate for their human nature. But this always remains a realm of necessity. The true realm of freedom, the development of human powers as an end in itself, begins beyond it, though it can only flourish with this realm of necessity as its basis. The reduction of the working day is the basic prerequisite. (Marx 1991: 958–959)

Here, Marx’s affinity with Aristotle is obvious: material production of external things (to use Aristotle’s words, pragmata – things – and ktēsis – property – as the sum of organa – tools, instruments), in as much as these things are means to something else, cannot be the end in itself. The production of things is a means to an end – human creativity and freedom. Marx phrases them in Aristotelian terms: the development of human powers which allow us to actualise our ‘species being’, our essential human nature. Yet freedom can be realised only within the realm of natural necessity. Cross-culturally, humans ‘must wrestle with Nature to satisfy [their] wants, to maintain and reproduce life’ (Marx 1991: 958–959). Production belongs to this sphere of necessity, the sphere which is historically mediated. Our needs historically expand, but so too do the forces of production and our ability to satisfy them. The growth of the productive forces in capitalism is enormous, yet the system of profit maximisation and the process of valorisation of capital subordinate human creative powers to the blind power of markets and competition. Thus, the expansion of machinery, as a consequence of blind competition, is, for Marx, the ‘capitalist material mode of existence’ (Marx 1990: 554). The subsumption of living labour to machines is the issue discussed in the fragment on machines in the Grundrisse (Marx 1993: 690–712).

Although there is an ongoing scholarly debate on how to understand the fragment from Marx’s unpublished notes (e.g. Dyer-Witheford et al. 2019; Fuchs 2016; Hardt & Negri 2000; Steinhoff 2021; Virno 2006; Wendling 2009), I take its main message to be as follows. In the process of capitalist development, fixed capital as the means of labour expands to the extent that living labour becomes subsumed into the process of production guided not by the virtuosity of the worker but by the integrated system of machinery and science. In this process, ‘the means of labour passes through different metamorphoses, whose culmination is the machine, or rather, an automatic system of machinery’; the system of automation, in its most complete form, is ‘set in motion by an automaton, a moving power that moves itself; this automaton consisting of numerous mechanical and intellectual organs so that the workers themselves are cast merely as its conscious linkages’ (Marx 1993: 692). Marx’s claim that capital ‘thus works towards its own dissolution as the form dominating production’ (Marx 1993: 700) and his focus on the general intellect has been interpreted by post-operaismo theorists as a crack in the capitalist system, with the potential for an emancipated, utopian post-capitalist society (Bastani 2019; Hardt & Negri 2000, 2004, 2009). The general intellect, as general social knowledge, is essential not only for production (it is ‘the direct force of production’) but also conditions ‘the process of social life itself’, social practice, and ‘the real life process’ (Marx 1993: 706). The post-operaismo philosophers, and subsequent thinkers influenced by them (Mason 2015), argued that contemporary production based on the general intellect, information, digital code, and automation relies on immaterial labour which cannot be easily quantified in time, is affective and has the nature of the commons. Thus, production today relies on free-floating information, the nature of which is that of the public good, on the one hand, and automation, which makes human labour increasingly redundant, on the other. In as much as this is the case, then free-market capitalism, as a system of production based on the appropriation of surplus labour time and the trading of private goods, will cease to exist.

MacIntyre’s practice as non-alienated activity

Although the assessment of this speculative thesis goes beyond the scope of this article, ¹³ I want to suggest that we need to see it against the background of existing social relations. Digital capitalism is very much alive. ¹⁴ The social and economic power of platforms and other international corporations has grown rather than diminished, while citizens’ democratic control has lessened (Fuchs 2022). The exploitation of labour, especially that of precarious workers, has also increased (Bielskis 2015). Contemporary capitalism relies on labour, even if its forms have changed since the industrial era in the 19th and 20th centuries. The forms of alienation have also transformed and have been conceptualised differently (Honneth 2007), the matter to which I now return.

Alienated labour as conceptualised by Marx in the Economic and Philosophic Manuscripts is too abstract. At least by implication, it claims that all salaried and waged labour in capitalism is alienated. Clearly, this is not the case. The four characteristics of alienation spelled out above may not be present in most contemporary jobs dominated by services. While, granted, a great deal of office jobs are meaningless in the way described by David Graeber (2018), the question that still needs to be asked is: How can we conceptualise non-alienated, meaningful work which contributes to human flourishing and which, therefore, should be spared from being automated? Although this question cannot be comprehensively answered here, I want to suggest that Alasdair MacIntyre’s notion of practice can furnish us with a framework to conceptualise non-alienated labour. ¹⁵

By ‘practice’, MacIntyre means a socially established cooperative meaningful activity through which goods internal to it are realised while trying to achieve the activity’s specific standards of excellence, ‘with the result that human powers to achieve excellence, and human conceptions of the ends and goods involved, are systematically extended’ (MacIntyre 1985: 187). Examples of practices are (1) sports and games such as chess and football; (2) sciences such as philosophy, mathematics, physics and medicine; (3) arts such as literature, portrait painting, composition and the performance of music; (4) the activities of farming and fishing; (5) family life; and (6) politics in the Aristotelian sense of the word. Being a leading contemporary Aristotelian, MacIntyre’s notion of practice, and his philosophy in general, is a powerful contemporary rearticulation of Aristotle’s conception of human flourishing. Although practices cannot be straightforwardly seen as (meaningful) jobs – not all meaningful jobs are MacIntyrean practices – practices can also be interpreted in terms of the Marxian conception of non-alienated activity. Essential to practices is MacIntyre’s distinction between internal and external goods. Internal goods are the goods specific to concrete practices (e.g. poetry’s internal good is an excellent, thought-provoking poem and its aesthetic enjoyment by readers), while external goods are the same for all practices and include status, prestige, and money (MacIntyre 1985: 188). Our ability to value practices as meaningful activities while pursuing their (internal) goods as genuine human goods has an ethically transformative element. Practices educate us, enable us to exercise our rational and creative powers and give meaning to life. Pursuing only institutional power, money, and status alienates us from one another, as ‘characteristically they are such that the more someone has of them, the less there is for other people’ (MacIntyre 1985: 190). In short, when ‘a person focuses or is forced to concentrate only on activities aiming at external goods, he or she is alienated from (1) forms of work that are good, meaningful, important, and enjoyable as such; (2) common understanding, shared experiences, mutual recognition, and acting together with other people; (3) personal relationships with fellow human beings; and accordingly, (4) that which is essential and constitutive to human beings generally’ (Noponen 2011: 105).

There is, however, a conceptual rift between MacIntyre’s Aristotelian notion of practice and Marx’s conception of alienated labour in capitalism. That is, genuine practices exist and function in the economic system of capitalism as different forms of salaried or wage labour (e.g. practicing medicine in a public or private hospital or being an engineer in a private company). Of course, genuine practices face institutional pressures from the pervasive dictate of profit maximisation. However, they are and can be practices in as much as they withstand the corrupting nature of capitalist institutions. Thus, practices are small islands of human excellence and resistance against the economic pressures of institutional profit maximisation and the self-valorisation of capital.

Automation and digital capitalism today

It is at this point that we need to return to the issue of automation. The normative-conceptual principle for assessing the ongoing practices of automation should be the following. It is imperative to automate those forms of labour which are truly debilitating, forms which are alienating in character. However, we should not be so keen in advancing automation in the fields of work and labour which are creative, practice-based, and contribute to the development of our essential human faculties. This principle does not, however, mean that practice-based productive activities governed by the standards of excellence in pursuit of their internal goods, cannot apply different forms of automation in their own practices to improve them or make them more enjoyable. ¹⁶ As MacIntyre himself puts it, the ‘excellence [of a practice] (. . .) has to be understood historically. The sequences of development find their point and purpose in progress towards and beyond a variety of types and modes of excellence’; ‘[p]ractices (. . .) have a history: games, sciences and arts all have histories. Thus the standards are not themselves immune from criticism’ and improvement (MacIntyre 1985: 190). The key, therefore, is not whether some tasks of a practice as desirable work can or cannot be automated but what drives automation within practices: institutional pressure to generate more external goods such as money and profit (by, for example, getting rid of ordinary practitioners who are ‘too expensive’) or the desire to excel their standards and transcend old ones in pursuit of the genuine human goods (e.g. excelling diagnostic in medicine).

Automation, following Marx and Keynes, as we have seen, should free us from the necessity of hard labour so that we would have more time to engage in practice-based activities and, therefore, flourish. This is not, however, how the current forms of automation have developed in cybernetic capitalism. Marx’s historical materialism brings forth an important insight. Today, the development of the forces of production is so advanced that we could have a much shorter working day. Although working hours have decreased over the last 150 years, the precarity of employment has increased, while the line between free time and working time has become blurred both due to working from home and the constant availability of workers through a variety of communication channels and portable devices (Stone et al. 2017: ix). As Carl Benedict Frey (2019) has argued, well-educated professionals, especially in the United States, work longer today than low-skilled workers in manufacturing (p. 338). Insisting that we should understand the development of automation in terms of both labour-enabling and labour-replacing technologies, Frey (2019: 320) argues that approximately 47% of jobs are ‘susceptible to automation. Even if most of these jobs are low-income and low-skilled, the middle-class jobs of professionals are also vulnerable to automation. On the other hand, Philip Jones (2021) shows how capitalist platforms, while developing AI technologies, exploit data workers in the Global South as well as employ immigrants, prisoners, and other marginalised individuals. They comprise a growing, invisible cohort of ghost workers necessary for the AI industry. Thus, the nature of digital capitalism is highly monopolistic, exploitative and leaves the alienation of labour intact: social media platforms rely on data extractivism, surveillance and the invisible low-paid click work of ghost workers.

Evgeny Morozov has convincingly argued that the current AI industry, the lion’s share of which is controlled by no more than ten firms, bears too high a cost for its contributions to discovery and innovation (Morozov 2019: 67). Market competition forces each of these companies to invest billions of dollars in research and development in order to advance the same set of skills (Morozov 2020). Thus, the capitalist model of market competition is inefficient and wasteful. Billions are spent on the same skills developed by each company so that they can sell their automated tools as private goods in the future. The AI industry, therefore, should be understood as an example of a public good: these skills need to be developed once and then shared and made accessible to others (Morozov 2020). In this way, the costs of AI development will be greatly reduced, which will also increase quality as it will enable interested parties ‘to take advantage of network effects’ (Morozov 2020). Morozov therefore proposes that AI should be decommodified and made into publicly created and controlled digital infrastructure:

What is needed instead is a centralized approach, where artificial intelligence is conceived as an infrastructure with a political economy behind it. You engage in a well-planned way of funding and developing it, and you find a way to make it accessible to different players in the economy – perhaps on different conditions. Big companies may have to pay a higher fee, smaller companies a lower fee and NGOs, activists and startups nothing at all. All of that suddenly becomes possible when we manage to take that one big step towards legal, political and financial institutionalization. (Morozov 2020)

The argument for the decommodification of AI technology is indeed convincing. To continue with the alienated forms of digital capitalism a la ‘click and scroll, but we will extract your data in order to bombard you with advertising later’ is unsustainable. This is especially true as far as children and young adults – who have grown up with smartphones and social media and have been socialised by and through virtual capitalist digital platforms – are concerned. These platforms have glued children to their phones and computers, devices which may now potentially have more influence on the socialisation of children than parents or guardians. ¹⁷ If the commodification of the AI industry is not challenged, then the negative social effects of automation on human flourishing will continue.

Conclusion

Although the AI industry is relatively new and its long-term effects on human lives are yet to be determined, the current forms of automation and their political economy demonstrate high levels of capital concentration and monopolisation, social inequality, surveillance, data extractivism and the exploitation of an invisible army of low-paid workers, mainly from the Global South, performing the micro-work of preparing data. A few corporate giants based in Silicon Valley are enjoying the overwhelming market share and skyrocketing profits ‘not seen since the robber barons in the 19th century’ (c.f. Steinhoff 2021: 139). Labour in the AI industry is also highly segregated when machine learning scientists, a fraction of the highly professional ‘labour aristocracy’, can earn up to seven-digit annual salaries, while data analysists’ salaries are far more modest – not to mention ghost workers who, being exposed to the offensive and psychologically debilitating work of labelling photos, earn, at most, 40 dollars per ten-hour-working day (Gray & Suri 2019: 8; Steinhoff 2021: 152–153). Born out of the military industry after the Second World War and funded predominantly by public money, research in AI has become commodified. Even if or when they are used in the service of public goods, AI tools are acquired by state agencies through schemes of public and private partnership. Being commodified, automation serves the process of the self-valorisation of capital: the logic of market competition forces companies to use automation as a means of replacing labour, including that which is high-skilled.

Given the paradox that the AI industry faces – ‘it is hard for computers to do many tasks that are easy for humans, and vice versa’ (Frey 2019: 310) – it seems that the process of automation moves in the opposite direction to the Aristotelian-Marxian insight developed in this article. The high cognitive tasks of humans are being automated, while the human functions of ordinary perception and bodily movement are not (or at least not yet). IBM’s AI programme beat Gary Kasparov at chess in 1997. On its own, this fact does not make human engagement in the practice of chess obsolete, yet the question that needs to be raised is: If the capitalist practices of automation remain unchecked, and if Marx’s insight on the subsumption of labour by the system of self-moving machinery is correct, will humans – at least those of us who do not own automated machinery – not become the slaves of machines? The employment of AI by capital forces will likely produce a deeper class divide between the few owners of automated machinery and the rest of us – people who have relied on their labour to survive. It is imperative, therefore, to aim to put AI under democratic control – not only because of the wastefulness of market competition underlying the current AI industry, but also because AI as a public good in the form of shared digital infrastructure would allow greater democratic scrutiny and control. Too much is at stake for democratic societies to leave the logic of private capital – whose essence is best expressed by Marx’s prose ‘no admittance except on business!’ – to be the sole driver of AI and automation.