People Are Getting Paid to Wear Cameras While Doing Chores. Here’s Why AI Companies Want the Footage

560

Thousands of people in India are taking part in a new type of job that sounds surprisingly simple: they wear a small camera while carrying out everyday household tasks. From washing dishes and making tea to folding laundry and organizing shelves, workers are earning around $3 an hour by recording their daily routines.

The footage isn’t being used for social media or reality TV. Instead, it’s helping train the next generation of artificial intelligence-powered robots.

Teaching robots how humans move

Unlike AI chatbots, which learn from billions of words found across the internet, robots need something entirely different. They must understand how people interact with the physical world—how they grip a mug, open a cupboard, peel vegetables, or carefully stack fragile objects.

To achieve that, AI developers are collecting enormous amounts of first-person video showing ordinary people completing routine tasks.

The recordings allow AI systems to observe tiny details that humans rarely think about, including hand movements, body positioning, object manipulation, and how people naturally adapt when something unexpected happens.

The rise of “Embodied AI”

Researchers call this technology Embodied AI—artificial intelligence designed not just to answer questions, but to physically interact with the real world through robots.

Companies developing humanoid robots believe these systems will eventually assist in homes, warehouses, hospitals, hotels, and factories. But before robots can perform even the simplest chores safely, they must first learn by watching humans.

Every recorded task becomes another lesson for the AI model.

Why companies need millions of hours of video

Training robots is far more complicated than training language models like ChatGPT.

A robot must learn:

• How tightly to grip different objects.
• How to avoid dropping or crushing fragile items.
• How to move efficiently around obstacles.
• How to react when objects are not where they’re expected.

Because every kitchen, home, and workspace is different, developers need massive amounts of diverse video data showing real people in real environments.

A new form of AI data work

The demand has created an entirely new category of employment. Instead of labeling photos or moderating online content, workers are now generating behavioral data by simply living their normal lives while wearing body cameras.

Although the pay—around $3 per hour—is modest by Western standards, it represents valuable income for many participants and supplies AI companies with one of the most important resources in robotics: authentic human behavior.

The future of home robots

Major technology companies are investing billions of dollars into robotics, believing that intelligent machines capable of household chores could become as common as smartphones in the coming decades.

For that future to become reality, however, AI first needs to learn from people. And for now, the best teachers aren’t engineers—they’re ordinary individuals making tea, washing dishes, folding clothes, and going about their daily routines while a small camera records every move.