Abstract
The expansion of artificial intelligence (AI) has fueled debates on automation, job displacement, and ethical concerns, yet little attention is paid to the labor infrastructures supporting AI development. This paper critically examines the hidden labor behind AI training, focusing on Spanish companies outsourcing data processing tasks to Latin America. By drawing on labor geography, anthropology of labor, and social movements, the research highlights how AI labor dynamics perpetuate colonial economic structures in the digital age. AI training relies on human-annotated data, involving repetitive, low-paid tasks like image labeling and data classification. Despite their essential role, these workers remain invisible in discussions on AI innovation. European and North American companies systematically outsource these tasks to regions with weak labor protections and lower wages. While previous studies have analyzed outsourcing to Africa and Asia, this research explores the underexamined link between Spain and Latin America. Based on in-depth interviews with executives and outsourcing managers in Spanish firms, the study investigates how decision-makers justify AI outsourcing through narratives of efficiency, cost-effectiveness, and technological progress. These discourses obscure the precarious conditions of outsourced workers. Findings suggest that Spanish firms engage in “digital colonialism,” replicating historical patterns of labor exploitation within the digital economy. By identifying AI processing hubs in Latin America and analyzing outsourcing mechanisms, this study reveals the structural dynamics of “digital extractivism.” It calls for a critical examination of the geopolitical and historical contexts shaping AI labor, challenging simplistic narratives of digital disruption and advocating for ethical AI development.
Details
Presentation Type
Paper Presentation in a Themed Session
Theme
KEYWORDS
Artificial Intelligence, Labor Geography, Digital Colonialism, Outsourcing, Data Annotation, Spanish
