Brown student maps the hidden global labor powering AI

Brown senior Cunyan Ma has created a free interactive map to make visible the people and companies behind AI systems. AI depends on human workers to label images and text, transcribe speech, assess chatbot responses, and perform other tasks needed to train and moderate models. Some must review disturbing material, including violent content and sexual abuse, so systems can identify and filter it. These jobs are often outsourced by major corporations to data-labeling firms that seek low-cost labor, with work concentrated in countries such as Kenya, Indonesia, and Colombia. Ma calls this arrangement “AI colonialism”: the extraction of difficult, low-paid labor in the Global South for profits largely accruing in the West.

The map connects data-labeling firms, their corporate customers, and locations where workers are based, drawing on existing databases, Ma’s own research, and news reports. Its current version covers 13 firms, 141 customers, and worker sites in 31 countries; Ma stresses that it is illustrative, not comprehensive. Her aim is to help the public understand the economic and political structures hidden beneath AI, rather than treating algorithms as autonomous creations. Brown professor Suresh Venkatasubramanian similarly emphasizes that AI is built from human knowledge and effort. Ma plans to expand the project, potentially mapping labor in the semiconductor industry as well. She also hopes the map could help dispersed workers find one another, strengthening their ability to organize and pursue a fairer share of the value their labor creates.