Send message to Author

Paper short abstract

AI governance often focuses on automation while overlooking the human labour that sustains AI systems. In this paper, I examine how job displacement and hidden AI work in the Global South are interconnected, and why examining this connection matters for how AI is governed.

Paper long abstract

Recent AI developments have reshaped work in two interconnected ways. One, automation has replaced/restructured jobs in sectors such as customer service, writing, and administration. Simultaneously, AI systems depend on extensive human labour for data annotation and content moderation to function (Muldoon et al., 2024). This labour is often precarious, outsourced, and concentrated in the Global South (Plantin, 2021).

I argue that contemporary AI governance in the Global South treats automation as the primary focus while making the human labour sustaining AI largely invisible. Most policy debates and regulatory frameworks emphasise deployment, ethical use, and efficiency, while paying limited attention to working conditions and social costs embedded in AI production.

By analysing documented cases of data annotators and content moderators in India, Kenya, and the Philippines, I show how AI systems rely on sustained human judgement to classify images, interpret language, and filter harmful content (Gray & Suri, 2019; Pogrebna, 2024). The very outputs of this labour, in turn, enable further automation, contributing to job displacement elsewhere (OECD, 2019; Frey & Osborne, 2017).

By holding these two sides together, the paper reframes AI governance as an epistemic and labour question rather than a purely technical or ethical one. I show how governance frameworks that ignore this interdependence risk legitimising automation while leaving the human work behind AI unaccounted for. Recognising this relationship is crucial for developing AI governance approaches that address not only the effects of AI, but also the labour relations through which AI is made possible to begin with.