Le capital algorithmique. Accumulation, pouvoir et résistance à l’ère de l’intelligence artificielle. Algorithmic Capital. Accumulation, Power and Resistance in the Era of Artificial Intelligence.

Jonathan Durand Folco & Jonathan Martineau, Montréal, Écosociété, 2023.

 

Le Capital Algorithmique is the sum of a multidisciplinary research project on the relation between technological developments and capitalism in the last two decades. The book develops an integrated theoretical framework to understand the social and technological changes that have altered our ways of living, working, being and interacting, and explores paths and solutions to change the course of these developments and collectively bring about a more just, democratic and ecological world.

The central assumption guiding the book is that we cannot understand the current waves of technological transformations without also addressing recent changes in capitalism, and conversely, we cannot understand the current stage of capitalism without an understanding of algorithmic technologies and AI. The aim is thus twofold: bring a holistic Marxian framework to AI studies, and bring critical theory and Marxist political economy up to date with the accelerated development of algorithms and AI.

The concept of “algorithmic capital” guides our inquiry into the rapid transformation of labor, temporalities, social relations, culture and politics we have witnessed in recent years. Building on the work of Nancy Fraser, we understand capitalism not only as a system of economic production, but as an “institutionalized social order” comprising the political sphere, social reproduction and a relationship to nature. As such, algorithmic capital is a multidimensional phenomenon: a logic of accumulation, a modality of social relation, a form of social power and an extractive industry, riddled with tensions and contradictions.

The book simultaneously builds on, synthesizes and bridges gaps in a rich, diversified and quantitatively important output of critical and mainstream literature on AI and current technological transformations. Contributions by Shoshana Zuboff, Kate Crawford and Ruha Benjamin, among other seminal non-Marxist critical studies, have been supplemented by more explicitly Marxist and socialist literature, for example on AI and production (Dyer-Witheford et. al.), on transformations of labor in the late digital age (A. Casilli, P. Jones, A. Benanav), on the political economy of platforms (N. Srnicek, C. Durand), on accelerationism and socialism (A. Bastani), on “technological solutionism” (E. Morozov), and on the social history of technology and everyday life experience (A. Greenfield, P. Marx, R. Seymour, M. Pasquinelli). Indeed, as much as critical contributions abound, the vast majority of the literature on AI and algorithmic technology, both critical and mainstream, focuses on a given sector, application, discipline or issue: AI and education, labor, business, inequalities, race, gender, security studies, law, transport, ethics, etc. Our contribution builds on such advances to construct a general critical theory of the current stage of capitalist development and the accelerated deployment of algorithmic technologies. This framework allows us to spread our analysis of capital’s new institutionalized social order in different spheres including social reproduction, nature-society relations, politics and power, culture, subjectivity and social relations, ethics, and practices of resistance.

One core thesis of the book is that capitalism has undergone a historical transformation in the last 15–20 years, bringing about a new stage of algorithmic capitalism, building on and then superseding the previous stage of neoliberal capitalism. This historical shift is occurring at the conjuncture of the emergence of big data and the rapid deployment of AI via advances in machine learning and deep learning; a new articulation of exploitation/extraction relations in capitalism based on the flourishing of a new business model of extraction and valorization of data; and the 2007–08 crisis of neoliberal capitalist accumulation and its aftermath. The book historicizes the post-2008 “austerity decade” as a transition period from neoliberal to algorithmic capitalism marked both by ruptures and continuities. Building up a theory of the new logic of algorithmic capital accumulation, the book critically engages with literature analyzing the current moment as “cognitive capitalism,” or more recently as “technofeudal” or “neofeudal.” Such contributions are helpful to grasp partial aspects of our new reality, but ultimately suffer from important limitations, and are best read as metaphors. Far from a neo-feudal “back to the future” moment, we have reached a new stage of capitalist development.

Over twenty chapters, organized and titled as straightforward theses, we make a series of arguments pertaining to the new institutionalized social order of algorithmic capitalism. On the topics of labor and social temporalities, we argue that algorithmic automation, far from leading to the “end of work,” or to “leisure society,” is rather pressuring labor and rendering it more precarious and insecure. Algorithmic capital comes with a new mode of exploitation/extraction that reconfigures global productive activities as well as labor markets, especially through extraction and valorization of data and the rise of “digital labor.” Ultimately, “algorithmic labor” is the sum of four types of productive activities reproducing algorithmic capital: digital, but also industrial/logistical, extractive and domestic. We mobilize social reproduction theory and critical feminist theories of affective labor to analyze the colonization of social reproductive space-times with algorithmic technologies and AI assistants. These processes rearticulate the relation between domestic labor and capital (a summary of this work has been published in English in Issue 8 of the journal Spectre). All in all, algorithms accelerate labor time, degrade leisure time, and increasingly connect social reproductive times to capitalist fluxes.

As a mode of social power, algorithmic capital entrenches existing systems of oppression such as racism and patriarchy, most notably by automating the biases and discriminatory practices found in datasets used to train algorithms. Furthermore, algorithmic capital shapes new forms of power, domination and discrimination, including a concentration of socio-economic power in a few giant tech companies—which also levy important lobbying power; new power dynamics based on different levels of tech literacy and access to information in the population; and a consolidation of surveillance power in private firms, governments, police forces, and between individuals. Algorithmic capital also gives rise to a new form of what critical scholars such as Antoinette de Rouvroy and Karen Yeung call “algorithmic governmentality,” which is a power to direct, orient, nudge behaviors using access to information, algorithmic technologies and control over information environments. In our view, “algorithmic power” is rooted in and connected to capital accumulation, technological mediations of social life, alienation, and historical systems of oppression. Furthermore, the rapid dissemination of algorithmic technologies in state apparatuses, public administration, and judicial systems consolidates a commonality of interest between state and capital in the accumulation of algorithmic power. At the international level, tectonic shifts are manifested in tensions mounting between concurrent models of algorithmic capitalism deployed in the US and in China, while the international division of digital labor and novel forms of data colonialism configure exploitative relations between algorithmic powers such as the US and China and the Global South, and threaten to peripheralize regions, such as Europe, formerly at the core of global capitalism.

The new institutionalized order of algorithmic capitalism also finds expression at the level of culture, ideological forms and subjectivities. New pro-tech and pro-capital ideologies percolate from a strange partnership between Oxford philosophers and tech billionaires, such as long-termism, effective altruism and transhumanism. We engage these views building on Timnit Gebru and Emile Torres’s examination of the “TESCREAL” ideological nebulae. For us, TESCREAL is the esoteric core, while the “Californian” or “Silicon Valley ideology” of connectivity, technological solutionism and market libertarianism forms the mantle of an increasingly dominant ideology in algorithmic capitalism. New forms of subjectivities are also created by extraction and online socialization, and tensions arise between forms of “connected selves” and alienation, isolation and dispossession that come with the use of algorithmic technologies.

Building on the work of Kate Crawford and Andreas Malm, we inquire into the relationship between algorithmic capital and nature. Algorithms and AI are not simply flows of information in dematerialized clouds. They require a vast material infrastructure to function: billions of connected devices, servers, data centers, optic fiber, communication towers, electric energy, etc. Understood as an extractive industry, algorithmic capital comes with largely unsustainable ecological and energy footprints. Furthermore, building on Peter Dauvergne’s insights, the technological solutionism of AI environmental applications must be problematized. Algorithms will not save us from ecological disaster, but rather accelerate the ecological crisis.

As a critical theory of algorithmic capitalism, our contribution is grounded in an emancipation project. We explore paths of reform, revolution, and exit strategies from algorithmic capitalism. We develop a virtue ethics approach to favor collective resistance and emancipation from algorithmic capitalism. We analyze different scenarios of energy descent, democratization of technological development, and individual and collective technological sobriety. We revisit the age-old debate on socialist planning in light of the potentialities and limits of algorithmic economic planning, and present potential paths for a transition out of algorithmic capitalism. Ultimately, this historical stage of algorithmic capitalism has a beginning, a development, and will have an end. It is up to us to organize and federate the collective energies of resistance already flowing in the interstice of algorithmic power to bring about a postcapitalist, “technosober,” ecological, just and democratic world.

Le capital algorithmique is currently available in French from Écosociété. An English-language translation is forthcoming.

Jonathan Martineau is an Assistant Professor at the Liberal Arts College of Concordia University in Montreal. He is the author of Algorithmic Capital: Accumulation, Power, and Resistance in the Age of Artificial Intelligence (with Jonathan Durand Folco, Écosociété), and Time, Capitalism and Alienation (Brill).

Jonathan Durand Folco is an Associate Professor in the School of Social Innovation at Université Saint-Paul in Ottawa. He is the author of Algorithmic Capital: Accumulation, Power, and Resistance in the Age of Artificial Intelligence (with Jonathan Martineau, Écosociété) and À nous la ville ! Traité de municipalisme (The City is Ours! A Treaty on Municipalism) (Écosociété).

Image: Asian Development Bank, 37475-013: Madrasah Education Development Project in Indonesia, via Flickr, https://flic.kr/p/dQubTB