China is paying thousands of people to fold laundry, open doors, and screw caps onto bottles. The goal is not to outsource housework but to train the next generation of humanoid robots.
Tech companies in China are hiring residents and factory workers to record themselves performing everyday manual tasks. These videos become the raw material that teaches robots how to move, grip, and interact with the physical world. The approach is a direct response to a global shortage of high quality training data for humanoid machines.
A nationwide video collection effort
Companies including the e commerce giant JD.com are leading the push. They have recruited people across the country to film themselves doing chores and repetitive factory work. One participant told reporters she spent hours recording herself folding shirts and sorting items at home. The videos are then fed into machine learning models that help robots learn by imitation.
The work is mundane but essential. Robots need thousands of examples of a single action to perform it reliably. By paying ordinary people to supply this footage, Chinese firms are building large datasets faster than competitors who rely on lab based demonstrations.
Factory floors become classrooms
In industrial settings, the data collection is even more structured. Factory workers are asked to perform their usual tasks while wearing motion capture suits or being filmed from multiple angles. These recordings capture not just the visual steps but also the force, angle, and timing required for each movement.
Local workers in manufacturing hubs have become part time robot trainers without leaving their jobs. For them, the extra income is a welcome addition. For the companies, it is a cheap and scalable way to generate the volume of data that humanoid robots need before they can be deployed in warehouses, kitchens, or homes.
Why this matters in China
China has set ambitious goals for robotics and artificial intelligence. Humanoid robots are seen as a strategic priority, with the government and private sector investing heavily in development. But training these machines has proven difficult. Robots that look human still struggle with basic physical tasks that humans find trivial.
By turning thousands of citizens into data collectors, Chinese tech firms are trying to close that gap. The strategy reflects a broader pattern in the country's tech industry: using large scale human labor to solve problems that other nations approach with smaller, more specialized teams.
For the people involved, the work is simple. They go about their daily routines or factory shifts while cameras roll. The footage they produce may one day help robots cook, clean, and assemble products. But for now, it is the humans who are doing the learning for the machines.
