India’s “AI Trainers” Turn Everyday Household Chores Into Data for the Robot Future

In a modest kitchen in Chennai, Indian housewife Nagireddy Sriramyachandra straps a smartphone to her head and films herself slicing mangoes. The footage may look ordinary, but it is part of a fast-growing global effort to teach artificial intelligence systems how to understand and replicate human movement in real-world environments.

For each hour of recording, she earns just over two US dollars. The 25-year-old is among thousands of workers across India now employed in what has become an emerging segment of the digital economy: AI data training.

“Who else will give you 250 rupees an hour just for doing housework?” she said, adding that she views the work as both income and preparation for a future where robots may become common in households. “I may get a robot myself in the future.”

Unlike chatbots and image generators that learn from vast online text and images, robotics systems require first-person visual data to function in physical spaces. Companies are increasingly collecting what researchers call “egocentric data” — recordings captured from a human point of view — to help machines learn everyday tasks such as cooking, cleaning, and organizing objects.

Sriramyachandra sends her recordings through a dedicated application to Objectways, a data company with operations in India and the United States. The firm works with major technology clients and machine learning platforms, including systems used by large global corporations.

Across Tamil Nadu and other Indian states, workers wear head-mounted cameras and motion sensors while performing routine tasks. Some record themselves folding clothes or arranging objects, while others work in controlled studio environments designed to mimic homes.

At one such facility, rows of participants repeatedly perform simple actions in staged apartment rooms fitted with cameras and sensors. After thousands of hours of recordings, even the wallpaper is changed to diversify the data being collected.

Each video lasts only a few minutes, but workers may record dozens or even hundreds of clips in a single day. Some describe the work as repetitive yet manageable, though many say they feel constantly observed.

The industry is expanding as global investment in humanoid robotics accelerates. Analysts project that advanced robots capable of household and industrial tasks could number in the hundreds of millions within the next few decades.

Companies involved in data collection say the goal is to help machines eventually take over routine or physically demanding tasks, allowing humans to focus on higher-value work. Tasks being filmed range from folding laundry to preparing specific food items.

India has positioned itself as a major hub for this type of AI training work, employing both factory workers and informal laborers. Experts note that while the sector is growing, it also raises questions about the future of low-wage digital labor.

For many workers, however, the immediate reality is simple: everyday actions have become a source of income in a global industry preparing machines for a future that is still taking shape.