Student Maps AI's Hidden Global Data Labeling Chain
A Brown student researcher has created a map revealing the global supply chain of data labeling companies that train AI systems. Most major corporations outsource data labeling to specialized firms, which source inexpensive labor from the Global South—Kenya, Indonesia, Colombia, and other countries. Workers in these regions process AI-generated content, including child pornography, violent imagery, and bestiality, to train AI safety models. The current map documents 13 data labeling companies, 141 corporate customers, and 31 countries with identified data worker sites.
The research characterizes this arrangement as “AI colonialism”—extracting low-wage labor from the Global South. Companies intentionally maintain this supply chain in opacity to avoid accountability. The mapping combines existing company databases with original research on data labeling firm locations to create a public education tool that exposes these hidden structures. Project advisors emphasize that AI is fundamentally built on aggregated human knowledge and effort, not autonomous intelligence.
The researcher plans to expand the map to other industries such as semiconductors using power mapping techniques. A significant potential benefit is connecting distributed workers who currently lack awareness of each other, which could enable unionization. The ultimate objective is achieving more equitable distribution of AI labor and its economic benefits.
Read the full article
Continue reading at Hacker News →This is an AI-generated summary. Read the original for the full story.