Written by Dr Preeti Raghunath
Over the last few days, news of Kenyan workers calling for investigations into OpenAI’s exploitative work conditions has emerged, even as The Guardian called a recent ruling implicating Facebook to offer mental health support for its content moderators a “watershed moment”. A few months ago, Time broke a story on the low paid workers behind ChatGPT. The 2018 documentary film, The Cleaners had brought to light the difficult lives of the hidden workforce that keeps social media relatively less toxic. With AI and machine learning technologies rapidly changing the way we work and access resources, it is imperative that we bring focus to work/life in what I call, the data industries.
Defining data industries
Data industries are enterprise ventures of diverse proportions, from big technology corporations and platforms to small annotation firms located in the Global South that serve as conduits to a workforce that then work on outsourced data labelling jobs. Theyfurther include massive data centres built as part of cloud infrastructures as well as data analytics firms. The phrase ‘data industries’ allows us to trace, locate and bring the lens onto these varied enterprise ventures that have and continue to be deeply implicated in different ways in the business of datafication. These data industries exercise a huge influence, impacting lives and livelihoods, informing governance and framing regulation.
Towards this end, it is important to understand the modern genealogies of these corporations and enterprises. How can we trace some of the early manifestations of these industries, and where can we locate the logics and practices that drive them? In my recent writing, I suggest that, “datafication as a practice of making legible aspects of peoples’ being as quantified information is not new. From anthropometry to censuses to the adoption of technical standards for internet governance, the analogue-digital continuum of governing people and their activities as data is a particular practice of modernity, earlier forms of datafication across societies notwithstanding” (Raghunath, forthcoming in Technology and Regulation, 2023). By focusing on modern practices of datafication, one is able to historicise data and look for continuities and shifts in the lead up to its contemporary manifestations. In this endeavour, writings on the colonial topographies of internet infrastructure and data’s colonial political economy are very useful in enabling us to grapple with what historicity allows us to read in relation to data and technology infrastructures, and what it can be disparaging of.
Transcolonial railways as data industries
Against this backdrop, I trace the histories of modern data industries, for which I go back to an earlier revolutionary technology infrastructure, the Railways. Transcolonial railways, much like the early corporations and companies they had linkages with, were accompanied by massive amounts of analogue datafication through railway businesses and ventures that were contracted to build railways systems across empires.
In my current research, I study practices of datafication by colonial railway companies that were given contracts to build railway systems across Britain’s colonies. I recast these corporations and companies as early manifestations of the data industries, for they set in place practices of datafication and data-enabled colonial governance that accompanied their operations. I then explore transnational histories of labour in this study. I specifically look at the deployment of labour from the Indian subcontinent in the building of colonial railways across Britain’s colonies in present-day South and Southeast Asia, Africa and South America. Railway businesses like the East Indian Railway Company, established in 1845, were contracted to build the railways in British India and employed labour from various parts of the Indian subcontinent. Similarly, railway companies like the Great Indian Peninsula Railway, South Indian Railway and others drew on a vast labour force that was available as a result of dwindling agriculture and handloom sectors in British India, due to the colonial economy.
Further and importantly, labour from the subcontinent was also recruited as part of the system of indentureship which was actioned by the British colonial government after the transatlantic slave trade was abolished in 1834. They were transported to other colonies of Britain where they engaged in work on plantations, but also in the building of railway systems. For instance, the British East Africa Company built the “lunatic” railway line connecting Uganda and Kenya and deployed labour from Punjab. Labour from the subcontinent was deployed in large numbers in the building of the railways in Natal, in South Africa. In the 20th century, Tamils who had previously been taken to British Malaya were later recruited to build the railway line between Japanese Siam and Burma, remembered as “death railway”. Throughout my research, labour is not seen as a homogenous category but as one that is characterised by already existing hierarchies along class, caste, gender, religious, racial, regional and geopolitical lines. This becomes important in recognising forms of internal colonisation within societies.
Piecing together these histories of the data industries can help us further understand their contemporary workings across different sites today. It can help unearth logics and practices of datafication that serve as precursors to today’s digital data economy, replete with technocolonial logics, epistemic expropriation and extractive practices of engaging with human and more-than-human worlds.
Preeti’s current research is on the global data economy, exploring its historical making and contemporary manifestations.
Sociological Studies Research 