The Coloniality of Data Work in Latin America

Global inequality patterns are being reproduced in AI work: The research found that data labeling work—essential for training AI systems—follows centuries-old patterns where requesters are located in wealthy countries while workers are concentrated in the Global South, particularly Latin America. This perpetuates economic divisions rather than creating new opportunities for equitable development.

Workers have limited control over their own labor: The study reveals that platform design actively constrains workers' agency and decision-making when they're producing the annotations that train AI systems. Rather than empowering workers, the platforms' configuration restricts their ability to exercise judgment or influence how their work is used.

Northern ideologies shape the entire system: The research identifies how ideas and frameworks originating in wealthy, developed nations are embedded in these platforms to justify and normalize the current labor market structure. This ideological dimension means the inequality isn't accidental but systematized through how platforms are designed and justified.

Data work deserves scrutiny as a labor issue: By examining 93 platforms and conducting interviews with Venezuelan workers, this research demonstrates that the data annotation work powering AI systems is a legitimate labor concern worthy of the same scrutiny applied to other global supply chains, not merely a technical or efficiency question.