In homes, barbershops and kitchens across India, a new kind of digital work is quietly taking shape. According to trending reports, hundreds of thousands of workers are being paid to record everyday activities — cutting hair, cooking meals, cleaning rooms and folding clothes — on video. The footage is then sold to technology companies and research labs.
The goal is to train what is often called physical AI: robots and other machines that must move, grasp and react in the real world, not just answer questions on a screen. A chatbot can learn from text, but a robot that pours water, picks up a towel or opens a door needs to see how people actually do those things — the order of movements, the grip, and the small adjustments made when something slips.
That is why companies are turning to large amounts of real-world video. Simulated environments and laboratory demonstrations can teach only so much, and they often miss the messiness of ordinary life. Footage from many different homes, workplaces and streets can show variation in lighting, objects and body movements, helping models generalise from one setting to another rather than memorising a single perfect demonstration.
For workers, the reports describe modest pay — about $5 to $10 for eight to ten hours of recorded work — raising questions about fairness, consent and the value of the data being collected. Supporters say it offers flexible income and a chance to take part in a fast-growing technology, while critics ask whether people fully understand how the footage may be reused, and whether the pay reflects its long-term commercial value.
Whatever the answer, the trend shows how the next stage of artificial intelligence may depend less on words alone and more on watching human life, frame by frame. As physical AI moves from laboratories into homes and workplaces, the people whose movements teach it are becoming part of the story — and the debate about how they should be paid and protected is only beginning.
Source note: Based on trending news reports about workers in India recording everyday tasks for AI training; pay and scale are as reported and have not been independently verified here.
