AI firms hire gig workers with wearable tech to train robots’ hands

AI firms are hiring thousands of gig workers across 50+ countries to train humanoid robots using wearable tech. Workers wear head-mounted cameras and motion sensors while folding laundry, washing dishes, assembling parts, and performing other physical tasks. Pay can be as low as $2.60 an hour, with rates cited around 250 rupees/hour in India. The data is used to teach robots embodied manipulation: real-world visual and spatial input for hand and object interaction that is hard to replicate via simulation alone. Companies such as Micro1 and Objectways run large-scale data-collection operations, and buyers reportedly include Tesla, Figure AI, and Scale AI. The sector spends $100 million+ annually on this kind of training data, alongside a broader investment surge in humanoid robotics (over $6 billion flowed into the space in 2025). Privacy and consent concerns are already emerging. Because devices capture personal environments (homes, family members, kitchens), questions remain about data ownership, how consent is obtained, retention periods, and whether footage could be reused beyond robotics training. AI firms’ push for “embodied AI data” is spreading globally, with similar efforts reported in China using VR headsets and exoskeletons.