Workers in Karur, India, are providing data to train artificial intelligence robots, a process termed ‘egocentric data’ collection. This involves individuals, such as textile workers like Shoba Kirubakaran and Kannagi Periyasaamy, recording their actions – including packaging, quality checks, and domestic tasks – using GoPros and smart glasses. This data is being utilized by companies like Objectways to create AI models capable of mimicking human movements for tasks like industrial automation and domestic chores. The practice is driven by the need for vast amounts of data to ‘teach’ robots, with workers receiving approximately $2.60 per hour for their contributions. Concerns are raised regarding the exploitation of labor and the potential displacement of workers, particularly given India’s large workforce. Experts suggest retraining initiatives and adaptation to new occupations may be necessary. The process involves replicating human actions, such as folding towels or chopping vegetables, to enable robots to learn through observation. Companies like Objectways are selling this data to AI developers globally, including to those creating domestic robots. The reliance on human data highlights the current limitations of AI, as robots require ‘motor skills’ to perform intricate tasks, and the cost and energy consumption of deploying robots at scale remain significant barriers in India.