The digital labour of artificial intelligence in Latin America: a comparison of Argentina, Brazil, and Venezuela

TL;DR

This paper exposes the often overlooked human labor behind AI in Latin America, highlighting economic hardships and inequalities faced by data workers across Argentina, Brazil, and Venezuela.

Contribution

It provides original comparative data on data workers in Latin America, revealing their conditions and the socio-economic factors influencing AI labor practices.

Findings

Data work is linked to economic hardship and inequality.

Despite high education, workers face disadvantages and informality.

Cross-country differences highlight local socio-economic contexts.

Abstract

The current hype around artificial intelligence (AI) conceals the substantial human intervention underlying its development. This article lifts the veil on the precarious and low-paid 'data workers' who prepare data to train, test, check, and otherwise support models in the shadow of globalized AI production. We use original questionnaire and interview data collected from 220 workers in Argentina (2021-22), 477 in Brazil (2023), and 214 in Venezuela (2021-22). We compare them to detect common patterns and reveal the specificities of data work in Latin America, while disclosing its role in AI production.We show that data work is intertwined with economic hardship, inequalities, and informality. Despite workers' high educational attainment, disadvantage is widespread, though with cross-country disparities. By acknowledging the interconnections between AI development, data work, and globalized production, we provide insights for the regulation of AI and the future of work, aiming to achieve positive outcomes for all stakeholders.