Before artificial intelligence (AI) was invented as a technology, capitalism in the West was driven by what could be called a fantasy of AI. This means that the most important economic processes – from decision making to production – were gradually delegated to some higher, magically autonomous intelligence, imagined as, for instance, an “invisible hand” steering the “self-regulating market.” In the neoliberal age, this fantasy of AI has paved the way for the rise of actual AI technology. Ultimately, something was born at the very intersection of AI as a fantasy of capitalism and AI as a technology of capital – and that something could be called AI-driven capitalism, in short: AI-capitalism.
Today, this far-reaching commingling of capitalist fantasy and capitalist technology is underlying processes of ‘market-friendly’ privatization and ‘market-transformative’ disruption—both coming to a head in ‘AI-powered geoengineering’ in the context of the climate crisis. Meanwhile, AI-capitalism’s consequences impact an increasing number of social fields (finance, logistics, etc.) and, last but not least, the design, valorization, and perception of human labor. At first glance, the most urgent problem is that under AI-capitalism human labor seems – across classes and contexts – to be gradually becoming extinct, although it is in fact undergoing deep transformations. Thus, at the end of the day the task is to debunk the extinction myth and to inquire how it conceals the far-reaching restructuring of work. In other words, rather than buying into the myth of human labor as a fading reality, it is necessary to tackle labor as a buried reality that needs to be excavated from beneath dominant narratives and power structures.
AI hypes and the rise of financialization
Of course, this story has many beginnings. One of them unfolds during the first big AI hype that affected wider audiences. Propelled in the 1960s by the writings of Norbert Wiener and J. C. R. Licklider, AI spread like wildfire at US universities, giving rise to the popular image of the ‘machinic superbrain’—even becoming a movie- star in Stanley Kubrick’s “2001: A Space Odyssey” (1968). Only a few years after the ‘AI winter,’ which had been declared in 1984, a new AI hype followed. This time managerial and banking circles embraced AI in a big way, which helped establish ‘neural networks as the almighty panacea of capitalism.’ Retrospectively, the two big AI hypes catalyzed a transition during which capitalism as a self-reproducing, self-running, and ultimately ‘smart’ system was reinvented. In this period capital made a leap into a new, immaterial era—trained on the all-encompassing instrumentalization of labor, while designed to become entirely unaccountable to labor as well as labor struggles e.g. in the form of worker (and student) revolts, but also resistance movements in (former) colonies. This attempt to fuse profitability with immunity in an expanded realm of capitalist magic is indicated by the rise of financialization.
The 1990s saw a large number of handbooks published on financial AI, for instance “Neural Networks in Finance and Investing: Using Artificial Intelligence to Improve Real-World Performance” (1992), “Trading on the Edge: Neural, Genetic, and Fuzzy Systems for Chaotic Financial Markets” (1994), “Neural Networks in the Capital Markets” (1994), “Artificial Intelligence in Finance & Security” (1995), and “Neural Networks for Financial Forecasting” (1996). Written for entrepreneurs, such books were suggesting that capitalism was an optimizable self-running system—‘you only need to correctly implement AI at capitalism’s crucial nodes... such as your company.’ One decade later, the financial crisis of 2007-2008 showed the eminent danger of this widespread belief. Causing immediate shocks all over the planet (banks collapsing, millions losing their homes, countries going bankrupt, etc.), the long-term devastations of the financial crisis are still unfolding and have yet to manifest themselves. This slow violence is comparable to what capitalism has left in the colonies as a trail of destruction—readable in terms of character and extent hundreds of years afterward.
One would assume that the financial crisis would have prompted a profound interrogation and theorization of financial AI based, e.g., on the aforementioned literature from the 1990s. Yet, most of the critical analysis avoids exploring AI’s role in finance and the financial crisis. If technology is discussed, then on general terms only. If fantasy is discussed, then it is not (explicitly) linked to AI. Thus, the deeper connection between AI and capitalism remains hidden in plain sight—a blind spot, that is. And the notion that ‘human labor being on the verge of extinction’ remains largely uncontested. In other words, the case of finance shows that AI-capitalism successfully mobilizes novel capacities to generate new margins of profitability and mechanisms of immunity against critique, crisis, and (worker) resistance.
Blackboxed circulation and the reinvention of ‘the factory’
The immaterial turn fostered by financialization privileges the circulation of derivatives and other immaterial products as the new profit paradigm. But circulation – as an ‘alternative to crisis-ridden production’ – is more than what just ‘disappears’ in digital networks. Intrinsically, it also involves material products, goods, and resources, and is supposed to take place practically everywhere. Cities are to be considered crucial nodes in this context. Here, most things are being moved along the ‘clandestine’ pathways of logistical infrastructure—and thus below the radar of perception and representation. In short, black box politics govern the infrastructure for both material and immaterial movements, that is, the basis of today’s life. This tendency is exacerbated as Alphabet, Uber, and Amazon expand their AI-powered logistical infrastructure, accelerating the privatization of formerly public services and the transformation of cities into ‘smart cities.’ In the course of this work processes – and their control and contestation – are increasingly withdrawn from public scrutiny.
Tellingly, heavyweights from the ‘old economy’ have also discovered the opportunities of this predicament. Revamping their business models, they are catching up and joining the trend of modularizing ‘the factory.’ A catalyst in this context is “Industrie 4.0”—or what could be called ‘the smart factory.’ Promoted by Germany after the 2008 financial crisis, CeBIT (at that time ‘the largest and most internationally representative computer expo’) was used as a platform. The vague vision of “Industrie 4.0” was quickly embraced by other countries such as the USA (“Industrial Internet”), France (“Usine du futur”), China (“China 2025”), and Japan (“Industrial Value Chain Initiative”). Unsurprisingly, “Labor Justice 4.0” is not exactly part of their agenda.
Taking the various threads of this development into account, one needs to acknowledge the following: logistical infrastructure is supposed to facilitate not only the circulation of products but also a circulatory mode of production—including the workers involved here. In other words, not only are the products of ‘the factory’ supposed to circulate, but also ‘the factory’ as such, less as a whole and more as a disassembled entity that gets partially reassembled over and over again, its
parts quickly moved along logistical infrastructure to places where they are projected to boost profits, e.g. due to a favorable regulatory framework. In this context it is crucial to remember the key characteristic of infrastructure: when it works without friction, it becomes imperceptible. In view of the above, one could say that the logistical infrastructure of ‘the factory’ and the labor it mobilizes are designed to be fluid so as to evade perception and representation as well as critique and resistance. Ultimately, this ‘design’ obfuscates how material and immaterial labor is being recalibrated under these conditions.
‘Generating something out of nothing’
Arguably, developing digital technologies is, at its core, about developing logistics-based technologies. Conceived thus, the Silicon Valley-led reintroduction of AI at the beginning of the 21st century was based on the requirements of logistics—and hence represents a radicalization of logistics. Thus, the task is to inquire what it means that companies like Alphabet, Uber, and Amazon are logistical enterprises at their heart. After all, their vision of frictionless logistics is boosted by ‘smart technologies.’
This means, for instance, that algorithms are deployed to write their own algorithms or to determine their variables. If this reflexive process is integrated into an algorithm, it becomes ‘self-learning.’ The programmers do not define the rules for its execution, but rather rules according to which the algorithm is intended to learn to reach a certain goal. In many cases, its solution strategies are so complex that they cannot even be comprehended afterwards. They can only be tested experimentally, no longer logically. “Such [self-learning] algorithms are basically black boxes, objects that can only be understood through their external behavior, but whose internal structure eludes recognition”, as Felix Stalder points out in his book “Kultur der Digitalität” (2016).
The blackboxed nature of these technologies contributes to the widespread belief that such self-learning algorithms work quasi-autonomously, as if no human input, training, and supervision was involved to actually get them to the point of operating quasi-autonomously. Ostensibly, they are capable of ‘generating something out of nothing.’ ‘No human labor needed!’ This glaring claim reveals the agenda of AI-capitalism as ultimately dedicated to reviving the goal of ‘frictionless capitalism.’ Here, value is supposed to be created in logistically optimized processes emancipated from human work. This implies not least a process emancipated from the trouble (read also: friction) that unruly workers have repeatedly caused the economy through strikes, revolts, and other forms of disobedience. Suspiciously convenient, isn’t it?
Sleepless labor in the ‘social factory’
Internet bots, also known as a ‘web robots,’ or simply ‘bots,’ are currently the most common form of AI. Over half of the movements on the Internet are produced by bots; they are at work virtually everywhere, from dating platforms like Wild to community spaces such as Wikipedia. As machines, bots are decoupled from hardware—they are machines literally made purely of text or code. As workers, bots are decoupled from conventional notions of labor—they can work day and night, as the US company BlackBeltHelp, for instance, promises: “Our AI-enabled Bot provides your end-users immediate access to support 24/7, anytime, anywhere.” Such a sleepless, 24/7 work cycle of machines has become the dominant mode of today’s digitally networked ‘social factory,’ where the whole of society has turned into a factory and the ‘production of the social’ has become one of the predominant goals of capitalism.
Today, the 24/7 mode of the ‘social factory’ subordinates labor in general to the logic of the autonomous machine. In the course of this, human labor is dehumanized, devaluated, and, consequently, appears obsolete—no longer capable of competing with machines since the worker’s body and brain depend on sleep and other forms of rest. But didn’t labor struggles in the 1970s already reveal that female cleaners in the UK, for instance, worked practically 24/7 like machines – cleaning offices at night and doing house and care work during the day, practically without any sleep? The existence of such labor struggles – as well as their echoes in the Global South – is rendered imperceptible by dominant narratives that subordinate labor to machine logic. Hence, exploring logistical infrastructure, one cannot but notice that the Amazon empire only seems to be completely robotized, and that the machine learning algorithm-powered processes are also carried out by human workers, who act here as hidden assistants to the machines. Moreover, when asking about how material and immaterial labor is being recalibrated today, it is crucial to challenge techno-fetishism that elevates the labor of calculating to ‘the ultimate and only valuable form of labor.’
Practically unquestioned, AI-capitalism is said to be centered on collecting, counting, and analyzing ‘the oil of the 21st century’—data. But isn’t this upvaluation of calculative labor how the diverse and interconnected facets of labor are suppressed? Web interfaces, for instance, only give the impression of being magically driven by bots because people perform the silent work of optimization in the background (‘web cleaning’ labor included). Also, the immaterial labor of looking at screens when, for instance, updating, sharing, and liking in social networks needs to be taken into account, since this very labor has become indispensable to the 24/7 production of capital in the cerebral cortex. Moreover, the fetishization of calculative labor neglects the material labor of manufacturing screens as well as assembling their components—and the fact that this labor has been relocated to sites outside the Global North where it is said that ‘capitalist extraction can still generate profits without being held to account.’
An automated world game?
Something further complicates the excavation of buried labor realities and struggles: the rapid scaling up of automated micro-structures of everyday life (for instance, AI is also presented as a presumably docile servant in everyday life, performing in the guise of ‘personal assistants’ such as Siri, Cortana, Bixby, and Alexa). This scaling up of automated micro-structures is particularly problematic because it appears complementary to the automation of macro-structures of global processes. At this juncture, the notion that ‘capitalism is a self-running system’ is gaining a total, all-encompassing quality—just like climate change. In fact, AI-capitalism has become interwoven with and inseparable from climate change.
Sea level rise, extreme weather events, and other environmental indicators always have a political, social, and, not least, economic dimension. In this sense, environmental issues are intrinsically linked to capitalism as an economic practice that hinges on the extraction of resources and labor on the one hand, and capitalism as a ‘global system’ that for centuries has been reconfiguring the planet on the other. In the wake of this, deliberate and large-scale interventions in the Earth’s climate system have created a new market. And, unsurprisingly, what this market is most fond of is nothing less than technological solutionism—with things like geoengineering powered by AI looming large on the horizon. At the same time, actors in mass media, politics and business tend to portray climate change as a ‘runaway monster’—a ‘social construction’ of ‘climate change as Frankenstein’ that clearly echoes the nature of AI-capitalism and only underlines the inevitability of measures that can meet today’s ‘most opaque autonomous non-human intelligence’ at eye level.
As two sides of one and the same coin, climate change and AI-capitalism evoke a world game running by itself—players (also read: workers) seem to be actively engaged, although most are unaware of the actual impact of their participation. AI seems left to its own devices. Unsurprisingly, the myth of the ‘automated world game’ obfuscates the havoc that the agenda of globally expanding AI as the fantasy and technology of capitalism has caused and is still causing. While capitalism’s devastating forces appear to go beyond anyone’s control – albeit primarily created by (white male) humans, they are often depicted as god-like forces – they affect everything/everybody on this planet, but not in the same way. Those who have contributed the least to the rise of these forces – e.g. in the form of global warming – are often those least prepared to deal with their consequences, and hence they are hit the hardest. This colossal injustice should urge us to debunk the myth of beyond-human-control forces and tackle the structures of power underlying the ‘automated world game.’
The political potential of labor
Today, a delirious intensification of AI-capitalism is taking place: AI-powered empires run by global players such as Alphabet (formerly Google) on the one hand, AI fantasies projecting regions like Africa as testbeds for the ‘transformatory efficiency’ of machine intelligence on the other. Thus, rethinking the politics of labor has become an ethical and political necessity. This is also indicated by a quickly growing body of critical literature including “Cyber-Proletariat” (2015), “Automatic Society” (2016), and “Software, Infrastructure, Labor” (2016). Yet, lacking in this discourse is decolonization in general and the decolonization of labor in particular. So, where to begin? Cedric J. Robinson, for example, points in his book “Black Marxism” (1983) to “the conception of nature as constant capital and the fact that the organizers of the capitalist world system appropriated Black labor power as constant capital.” Françoise Vergès, referring to Robinson, suggests in her article “Racial Capitalocene” (2017) that the connection “between the Western conception of nature as ‘cheap’ and the global organization of a ‘cheap,’ racialized, disposable workforce” needs to be made and reflected upon. Making and reflecting on such connections is all the more important since AI-soaked beyond-human-control narratives distract not only from the modularization of ‘the factory’ but also from the reterritorialization of parts of ‘the factory’ to the Global South, and hence obscure the ‘cheap’ labor of workers there.
Thus, one of the major consequences of AI-capitalism is that labor struggles are forced to reinvent their bargaining power. The question is, how to resist the various strategies that aim to obscure the political realities of restructuring work and how to politicize the hidden labor in AI-capitalism. What if we began sharing experiences and practices of labor in everyday life around the world? And what if in the course of this we created a collective consciousness for our co-existence as workers? Could we then become capable of confounding, contesting and recoding the structures of power to emancipatory ends—and turning AI-capitalism against itself? Moreover, if AI-capitalism as an ‘automated world game’ generates and modulates climate change, and if the means of production (and circulation) have become means of climate production, then the question is what kind of impact does human labor have. Can human labor only support or also change the course of the game in question? The latter seems impossible. But what if one resisted submitting to the hopelessness of the climate crisis and the inevitability of AI-capitalism? What if workers seized the means of production (and circulation) as the means of climate production? What if they put those means in the service of environmental needs and justice? Human agency would no longer appear at ground zero—and the same goes for labor.
One thing seems certain: The more we become aware of how rendering labor imperceptible and meaningless consolidates structures of power that continuously aggravate inequality and injustice, the more we gain a perspective on how labor could be mobilized from within the workplace and against the very structures of power that are circumscribing it. This could activate the potential of the growing ‘reserve army’ of workers. After all, if human labor is indispensable but presented as disposable and even nonexistent, then capitalism’s dependency on labor has reached a critical limit. Labor gains a unique political quality at this limit.
Project website
Find all details and up to date information on the SILENT WORKS project here: silentworks.info
Timetable
1/31: Project Launch at transmediale; 5/29: Submission Deadline for Texts; 11/07: Exhibition Opening; 11/12-14: Conference (Workshops, Performances, Talks + City Tours); 11/20: Open House (Guided Exhibition Tour, Artist Talk + Performance); 11/28: Closing Event
Curators
Magdalena Taube is editor-in-chief of the internet newspaper Berliner Gazette, professor of Digital Media and Journalism at the Macromedia University of Applied Sciences in Berlin and guest lecturer at Leuphana University, Bard College and Humboldt University. She is the author of “Disruption des Journalismus” (2018) published by Institute of Network Cultures, Amsterdam and co-editor of numerous anthologies, most recently “A Field Guide to the Snowden Files” (2017). Her curatorial projects include “Signals. The Snowden Files in Media, Archives and Arts” (2017) and “BQV. Büro für Qualifikation und Vermögen” (2012).
Krystian Woznicki is a critic and the co-founder of Berliner Gazette. His recently published book “Fugitive Belonging” blends writing and photography. Other publications include “A Field Guide to the Snowden Files” (with Magdalena Taube), “After the Planes” (with Brian Massumi), “Wer hat Angst vor Gemeinschaft?” (with Jean-Luc Nancy). His new book “Undeclared Movements” – published by b_books – is forthcoming. His curatorial projects include “As Darkness Falls” (2014), “Temporary Embassies” (2008), and “Young Japanese Cinema” (1999).