In the future, artificial intelligence might devalue, or even displace labour. However, we fail to see that the development of it has set in motion a cycle of labour displacement, especially in the Global South, making AI nothing more than a pretence. In fact, the controversy of AI is a reality shift that projects our fear of being replaced from the current to the future.
Gig work, also known as freelancing and online outsourcing, is thriving in South Asia. Globally one-third of online freelance workers are from India, according to the International Labour Organization, with two-thirds of remote workers located in the Global South. Coloniality is embedded in this upward trend, and the development of AI plays a huge role in revitalizing neocolonial rule.
Data annotation is essential to the development of AI. More specifically annotation “gigs” include labelling images and offering machine-learning algorithms contextual information about the data points they are trained to recognize and discern between.
While machine-learning algorithms themselves are “complicated outputs of intense human labor” (page 717), we tend to forget about the laborious task of sorting the data, which models are trained on, and the ongoing maintenance work, involving fixing errors and engaging with edge cases.
Annotation is an open-ended project, but the ongoing efforts of annotators are often forgotten. Ironically, the development of AI heralds the end of repetitive manual labour by completely overlooking that it, too, entails it for digital freelancers.
Annotators endure all the downsides of being self-employed. In some cases, the uncertainty and instability that follows self-employment are rectified with self-determination and flexibility. In most cases, though, the precarity of gig work captures annotators in a circuit of displacement.
Gigs are work “stripped of all its normal trappings: a schedule, colleagues”, and even “knowledge of what they [are] working on or whom they [are] working for”. Annotation is barely considered work and regularly referred to as ‘tasking’, reducing its agents to individualized taskers, who, due to the fluid nature of tasking where gigs appear and vanish without warning, are always alert and will work tirelessly whenever they get a chance to.
In the quoted article, the author also explains how one mistake results in taskers getting banned or booted, making the intensification of exploitation a dual matter. Due to the irregularities of the displacement of labour, and the displacement of annotation gigs as well, tasking permits an intensification of exploitation.
The global industry of annotation relies on taskers, and, in return, the industry presents the promise of flexibility. Nonetheless, what looks like easy money and labour autonomy, is a free-market trap where organizational costs are spared, and the accessibility of gigs lures the unskilled or unemployed into its circuit of irregular displacement.
Organizational shortcomings make tasking incredibly mobile. Gigs can be posted, cancelled, and reposted in accordance with global wage levels. Therefore, work is never guaranteed, and when tasking starts to provide people with a living wage, chances are it will disappear and reappear somewhere new where the demand for work is high, and requirements are low.
Cost savings are not necessarily the main driver for developers. Office pizza parties and pensions increase productivity, which is output measured against a fixed time unit. Instead, dispensing organizational costs is a means to an end. Developers are not looking to increase productivity. Rather, the desired outcome of dispensing organizational costs is mobility.
Overextending developers’ mobility is forming new labour hierarchies unequally allocated on a global scale, exacerbating subordinative exploitation and entrapping labour in a circuit of precarity, which secludes workers and strips them of their rights. In fact, the very concept of wages is under siege because, in the annotation industry, salaries are reduced to mere payments.
Labour is regenerative. The price of labour is thus the price of regenerating it, which is why wages or salaries, in this context, are terms used to describe the average cost of workers’ necessities of life. In annotation, on the other hand, there are no workers and employers, nor long-term commitments or relationships between the two. As a result, there is no labour, and developers only pay for the product of labour, not the regenerative labour itself. Purposefully or not, tasking is a venture of transfiguring variable capital and making it constant.
The premise of AI is lifting the burden of monotonous work off the shoulders of workers. Yet, it presupposes wearisome and repetitious tasking. Strikingly, AI is not liberating work from its mundanity. Quite the contrary, tasking is liberating work from workers’ necessities of life, i.e. work from workers. Additionally, tasking is cheapening labour.
Digitalization augments capitalist displacement, and the new technological endeavours of artificial intelligence will surely strengthen the tendency. It is not the final product of AI that we should fear. What is truly valuable, and, might I add scary, is the disintegration of jobs and the devaluation of labour, depreciating its regenerative qualities.
Tasking is already blurring the lines between jobs and gigs, essentially paving the way for a new economy. This means that debating whether AI will assist humans or take their place is a red herring, so to speak.
Tasking indirectly endorses developers’ free movement, which is ultimately liberating work from workers by overextending developers’ mobility. Liberating work from workers, as is the case with tasking, also means treating taskers like a finite resource, and their labour like a product out of tune with its special features.
In fact, the labour of taskers is treated like an energy stock, and appropriating variable capital allows developers to monitor the price fluxes of labour, like stocks. This possibility incentivizes displacement accordingly with the irregular rise and fall of prices globally, progressing precarization and the configuration of global labour hierarchies. Tasking is one way, out of many, to extract surplus value from the Global South.
The development of AI is causing a capitalist haze, wherein the buzz of AI’s potentiality leads to the misconstruction of Big Tech’s objectives.
Tasking is conjoined with the development of AI, hence, more broadly, the gig economy is intertwined with the operations of Big Tech. While the gig economy is driving down the global cost of labour, affecting all of us, taskers are effectively fuelling Big Tech’s imperialist mode of operation.
As a matter of fact, Big Tech is extracting labour and cheapening it, which leads to conceptually deteriorating the meaning of work and prompting the furtherance of precarization globally.
Photo: Nik from Unsplash.com
 Lehr & Ohm, David & Paul: ‘Playing with the Data: What Legal Scholars Should Learn About Machine Learning’ (2017)