In his latest book, Matteo Pasquinelli unfolds a history of artificial intelligence that arrives at an economy spearheaded by internet monopolies and digital capital, which divides and displaces labour to further exploit it. Following this line of reasoning, you could say that AI technology is, in fact, innovating exploitation and advancing precarisation.
The October 2023 publication ’The Eye of the Master: A Social History of Artificial Intelligence’ contributes to current disputes over artificial intelligence. The author Matteo Pasquinelli, Associate Professor in Philosophy of Science at the Department of Philosophy and Cultural Heritage of Ca’ Foscari University in Venice, theorises that algorithms mimic labour, not intelligence, and concludes:
“The debate on the fear that AI fully replaces jobs is misguided: in the so-called platform economy, in reality, algorithms replace management and multiply precarious jobs” (page 250).
The Division and Displacement of Labour
Taking a step back, algorithms can compute the cost of and help ascribe value to products, whether it be the products of labour or activities outside the workplace. In some cases, even, algorithms can compute labour and the activities themselves.
However, algorithmic computation warrants data inputs. On the topic of AI, the kind of technology many of us think of warrants numerous real-time inputs, which presupposes and perpetuates surveillance.
Therefore, more prominent AI developers trace our digital footprints by surveilling our every move online. As data, these footprints are fed to an algorithm, which clusters, classifies, and ultimately makes predictions about our future excursions. When these excursions involve commodity consumption, the predictions are sold to third parties.
Yet, Pasquinelli is not interested in this specific business model. Instead, he focuses on the algorithms that this industry relies on. Internally, these algorithms mimic the division of labour. Externally, they compute human activities and enable the valorization of them by mimicking the way these activities are organised in the workplace.
Pasquinelle writes: “the division of labour provides not only the design of the machinery but also of the business plan” (page 63).
The division of labour is the commercial dissection of human activity into measurable units. The division of labour is advantageous because it makes labour easier to surveil. However, surveilling labour does not always facilitate its automation, but it helps attuning the monetisation of labour.
When Pasquinelli writes that the division of labour provides the design of the machinery as well as the business plan, it means that machinery does not need to replace labour to be profitable. Machinery is a product of the division of labour, and a means to further dissect human activity into measurable units, from which company owners can profit.
This means that AI does not necessarily herald a radical break from the status quo. Rather, it is an exaggeration of the status quo. As a matter of fact, Pasquinelli explicates that AI is the latest product of the evolution of techno-economic principles that shaped the development of technology for hundreds of years because, historically, labour was never really replaced by machinery. Instead, machinery displaced labour when old crafts were dismembered by dissecting the individual tasks and mechanising them.
This is why Pasquinelli argues that AI is the continuation of the “systematic mechanisation and capitalisation of collective knowledge into new apparatuses” (page 94) that dismembers old crafts and advances the dissection of human activity into measurable units, from which company owners can profit. In other words, AI will not make humans obsolete. It is more likely that living labour will yet again be displaced, while AI helps attuning new ways of exploiting it.
The Pretense of Artificial Intelligence
AI embodies statistical derivation. The “characteristics of ‘intelligence’ that is anthropomorphised in AI systems is essentially the trick of projecting data on a multidimensional space in order to perform operations of clustering, classification, and prediction” (page 223).
What might look like independent, intelligent reasoning, is a type of statistical “intelligence” suited for the fixation of decision-making or bureaucratisation.
As a result, the prospect of AI replacing human intelligence is a mirage, which does not consider that “the error-correction techniques of deep learning have reached a computational limit and are unable to grow without paying exorbitant costs of energy and hardware resources which not even big corporations could soon afford” (page 249), he states referring to Neil Thompson.
Moreover, if you make the (illusory) distinction between mental and manual labour, it appears that many forms of manual labour are not, concurrently, automatable and profitable. Driving a car or sewing a piece of clothing are both examples of “manual” labour, which is extremely difficult to automate. Consequently, it makes more financial sense, as of now, to displace workers into low-wage service jobs like chauffeuring managed algorithmically [think: Uber/Wolt], than the actual business venture of self-driving cars.
This is why Pasquinelli reasons that AI will not substitute all labour, but, presumably, only the mental labour involved in managing it. In fact, AI technology has the potential to dismember the old craft of management by mechanizing and automating the surveillance and disciplining of manual labourers, and multiply precarious jobs, eventually instating new socioeconomic hierarchies based on the quantification of people’s productivity.
In short, internet monopolies inaugurated statistical valuation practices that could alter the capitalist economy by exaggerating preexisting techo-economic principles. These practices are the main outcome of a “monopolistic regime of knowledge extractivism” (page 236), as Pasquinelli calls it.
Nevertheless, ‘The Eye of the Master’ offers many relevant insights and perspectives on the history of AI and the current debate it sparked. One point of view, in particular, stands out:
We should not be fearful of AI. However, we should fear the underlying economic tendency of dissecting and displacing labour, which is accelerated by the digital extraction of data and the development of AI. Over time, this tendency ushers in new varieties of exploitation and economic precarity …
