Startup Pays Indian Gig Workers to Train Robots

A new startup called Human Archive is turning India’s gig economy into a training ground for robots. Founded by researchers from UC Berkeley and Stanford, the company pays gig workers to wear camera-equipped caps and sensor-laden devices while going about their daily routines. The goal? To collect vast amounts of real-world physical data that can be used to train AI and robotics systems. As AI labs race to build robots that can navigate messy, unpredictable environments, they face a critical bottleneck: lack of high-quality training data. Unlike text or images, physical-world data—such as how a hand grasps a cup or how a person walks through a crowded room—is expensive and difficult to collect at scale. Human Archive’s approach leverages India’s large, tech-savvy gig workforce. Workers are paid to perform everyday tasks like cooking, cleaning, shopping, and commuting while wearing the recording equipment. The resulting data captures subtle movements, force interactions, and environmental variations that are nearly impossible to simulate. “We’re creating a dataset that mirrors the diversity of real human activity,” said a co-founder. “This is essential for building robots that can operate safely and effectively in homes, hospitals, and factories.” The startup has already attracted interest from major robotics companies and AI labs. However, the model also raises ethical questions about data privacy and fair compensation. Human Archive says it anonymizes all data and pays workers above local minimum wage. As the demand for physical AI training data explodes, Human Archive’s gig-based model could become a blueprint for how the industry scales—one camera cap at a time.