A growing number of gig workers in countries like Nigeria are finding a new income source: training humanoid robots by recording themselves performing everyday movements. Zeus, a medical student in central Nigeria, joins thousands of others recording hand gestures, body movements, and task demonstrations that feed machine learning models used to train physical robots. These workers strap smartphones to their foreheads and perform repetitive motions for hours, creating the dataset necessary for robots to learn human behavior. While the work provides supplementary income for workers in developing economies, the compensation remains minimal—often just a few dollars per session.
This emerging practice reflects a broader pattern in AI development: outsourcing labor-intensive work to low-cost regions while keeping the technology's profits concentrated in wealthy nations. The gig workers training humanoids represent an extension of existing crowdsourcing practices like data annotation and content moderation, which have long relied on underpaid workers in Asia, Africa, and Latin America. Unlike traditional manufacturing outsourcing, this work is largely invisible and unregulated, raising questions about worker protections, fair wages, and labor standards in the AI supply chain.
The practice highlights growing concerns about equity in AI development and the human cost hidden behind advanced technology. As humanoid robots and AI systems become increasingly sophisticated, the workers enabling that progress remain largely anonymous and uncompensated fairly for their contributions. Policymakers and tech companies now face pressure to establish ethical standards for AI training labor, ensuring workers receive adequate compensation and protections as the industry expands.