HETEROMATING MACHINES, ISOLATION, INVISIBILITY AND THE USAGE OF TIME: A DISCUSSION ABOUT CHILEAN MICROWORKERS SUBJECTIVITY.

Artificial intelligence is often imagined as an abstract and autonomous technology: a set of systems capable of replacing, or eventually dispensing with, human labour. Yet empirical accounts of AI infrastructures suggest a more contradictory reality. Rather than eliminating work, contemporary AI systems depend on vast sociotechnical assemblages composed of minerals, cables, platforms, data centres, software architectures and, crucially, continuous human labour. Workers produce, classify and verify the data required to train models; they correct outputs; and, in many cases, they replace automated systems when these fail, remain incomplete or were never fully implemented.

This presentation examines AI not as an automating machine but as a heteromating assemblage: a system that redistributes tasks between humans and machines while creating new forms of dependence on dispersed and often invisible labour. Drawing on the Foucauldian notion of subjectivity and on the concept of the Marxist Social Factory, we argue that AI-related platform work produces specific forms of discipline and subjectivation. Workers are not merely inserted into pre-existing technical systems; they are shaped to fit the spatial, temporal and affective demands of remote platform labour. These processes also resonate with longer histories of informal and precarious work in the Global South.

Chile offers a particularly relevant site from which to examine these dynamics. Its labour market combines a long history of informality and neoliberal restructuring with a relatively strong currency position in relation to other countries in the region. This makes Chilean workers both differently positioned within global platform labour markets and analytically useful for understanding how AI production intersects with existing labour traditions, class trajectories and expectations of future work.

Based on ethnographic fieldwork, a questionnaire completed by more than 150 workers, and 30 semi-structured interviews, this presentation analyses how new forms of labour organization and algorithmic management delineate particular modes of subjectivation. The findings show that AI-related platform labour produces strong experiences of atomization, invisibility and isolation, making collective organization difficult. At the same time, as platforms specialize, workers develop differentiated strategies, identities and practices. These differences reveal not one homogeneous platform subject, but several ways of inhabiting and negotiating the unstable conditions of AI labour.

Nevertheless, common patterns emerge. Workers are required to capitalize fragmented time, remain available for irregular flows of tasks, and navigate a feast-or-famine model of income generation. They also become intermediaries between platforms, clients and recurrent technical failures, absorbing uncertainty as part of everyday work. In this sense, their labour is not external to AI systems but one of the conditions that allows these systems to function.

The presentation reflects on the methodological difficulty of studying a field that changed throughout the research process and will continue to change as the AI market expands, mutates or contracts. Even so, the Chilean case allows us to identify the contours of a specific worker profile: a temporary reserve labour force that often does not depend entirely on platform income but remains ready to give its free time to these systems while not seeing a future depending on it.