Who trains the data for Artificial Intelligence in Brazil? New co-authored report

I am thrilled to announce the release of our newest report, Microwork in Brazil. Who are the workers behind artificial intelligence? which sheds light on the burgeoning landscape of remote, data-centered platform labor in the country.

🇬🇧Download the report in PDF

🇧🇷Fazer o download do relat贸rio em PDF

This comprehensive study is the result of a collaborative effort between the research center LATRAPS (Laborat贸rio de Trabalho, Sa煤de e Processos de Subjetiva莽茫o, Minas Gerais State University, Brazil), coordinated by Matheus Viana Braz, and the research program DiPLab (Digital Platform Labor, Polytechnic Institute of Paris, France), co-founded by Paola Tubaro and Antonio A. Casilli. This study draws on methods and results developed in various projects conducted between 2018 and 2023 in Europe and Latin America.

Viana-Braz-Tubaro-Casilli_Microwork-in-Brazil_EN-9
Main Highlights

HOW BRAZILIAN MICROWORKERS CONTRIBUTE TO AI

There are 54 microtask platforms operating in Brazil, belonging to different categories: data annotation and AI training, but also click farms that sell followers and likes on apps such as Instagram, YouTube, TikTok. Moderating violent and pornographic content on social media is the worst task, according to workers. There are some 鈥渨eird鈥 microtasks too, like taking photos of animal excrement in domestic settings to train vacuum cleaner robots to avoid it.

Read More

MICROWORKERS IN BRAZIL: YOUNG, WOMEN, WELL-EDUCATED


Brazilian microworkers are mostly young, between 18 and 35 years old (70.6%), women (63.9%), and married, living with a partner or living in common-law marriage (60.8%). The states with the highest number of workers are S茫o Paulo (28.8%), Rio de Janeiro (12.6%), and Minas Gerais (9.7%). Moreover, the workers鈥 education levels (Secondary education to Bachelor and Master) are above the national average.

Read More

MICROWORK IN BRAZIL IS LINKED TO UNEMPLOYMENT AND INFORMAL LABOR

33.5% of microworkers have no other paid work, and US$1.80 is the average hourly rate earned on microwork platforms. 38.8% of workers are either unemployed, without a professional activity, or working informally.

Read More

HOW BRAZILIAN MICROWORKERS CONTRIBUTE TO AI

There are 54 microtask platforms operating in Brazil, belonging to different categories: data annotation and AI training, but also click farms that sell followers and likes on apps such as Instagram, YouTube, TikTok. Moderating violent and pornographic content on social media is the worst task, according to workers. There are some 鈥渨eird鈥 microtasks too, like taking photos of animal excrement in domestic settings to train vacuum cleaner robots to avoid it.

Read More

MICROWORKERS IN BRAZIL: YOUNG, WOMEN, WELL-EDUCATED


Brazilian microworkers are mostly young, between 18 and 35 years old (70.6%), women (63.9%), and married, living with a partner or living in common-law marriage (60.8%). The states with the highest number of workers are S茫o Paulo (28.8%), Rio de Janeiro (12.6%), and Minas Gerais (9.7%). Moreover, the workers鈥 education levels (Secondary education to Bachelor and Master) are above the national average.

Read More

Our report delves into the unique characteristics of microworkers in Brazil, focusing on their demographics, work patterns, and earnings. It uncovers a unique trend where younger, highly educated women constitute the majority of this workforce, even earning more than their male counterparts. This intriguing gender dimension prompts us to explore whether this phenomenon is exclusive to Brazil or indicative of broader trends in the evolving market for online micro-tasks.

Furthermore, the study examines the intersection of microwork and global supply chains of data. It illuminates the intricate web that connects homes or cybercafes in Brazil to data centers in various countries, underscoring the significance of these global dependencies. By acknowledging and regulating these interconnections, we aim to contribute to the ongoing discussions surrounding the regulation of artificial intelligence (AI) and platform labor, ensuring fair and equitable outcomes for all stakeholders involved.

The Authors
Matheus Viana Braz

Assistant Professor in the Department of Psychology at the Minas Gerais State University (UEMG), Brazil, and a Professor for the Graduate Program in Psychology at the Maring谩 State University (UEM). Coordinates the LATRAPS.

Paola Tubaro
Paola Tubaro

Research professor (Directrice de Recherche) in sociology and technology at the National Centre for Scientific Research (CNRS) and member of Center for Research in Economics and Statistics (CREST) in France.

Antonio A. Casilli
Antonio A. Casilli

Professor of Sociology at the Polytechnic Institute of Paris 鈥 Telecom Paris. Co-director of DiPLab (Digital Platform Labor) and co-founder of the International Network on Digital Labor (INDL).

In the Media

The study 鈥淢icrotrabalho no Brasil鈥, reported by Folha de S.Paulo, illuminated the painstaking nature of AI training work. Workers invest hours without guaranteed compensation, spotlighting the precarious nature of such roles.

In another report by O Globo in August, the focus shifts to the meager remuneration received by Brazilian workers for AI training, raising questions about fair wages and labor standards.

Even more alarming is the revelations based on the report from Reporter Brasil that microtask platforms offer as little as R$ 20 for tasks, including impersonating medical professionals, underscoring the ethical dilemmas entwined with these platforms.

To delve deeper into the subject, you can listen to the interview with Matheus Viana Braz on Spotify (in Portuguese).