Workers in China are teaching robots everyday tasks. Their movements become training data, helping machines learn household chores and respond to real surroundings.
The South China Morning Post reported on Liu Yun, who joined the programme after losing her job. The 40-year-old previously worked in the automotive industry. She struggled to find another job because of her age.
At a Beijing studio, Liu wore a sensor bracelet and chips on her fingers. Instructions appeared on a computer screen as she moved cups and scrubbed surfaces. She compared the experience to teaching a 2-year-old child.
Workers were closely monitored, with large screens displaying their productivity and accuracy rankings. The daily target was 160 samples, while leading workers completed over 260. Liu found the repeated tasks tiring. Washing dishes and wiping tables 200 times daily was already her limit.
Sun, a 22-year-old student from Shanxi, joined during his summer break while seeking an internship. He folded clothes, swept floors and recorded standard movements for training.
Day shifts paid 120 yuan ( ₹1,704), while night shifts paid 135 yuan ( ₹1,915). Sun felt the money did not justify the physical and mental strain. Many workers left after only a few weeks.
However, the work suits some people seeking flexible hours. Guo Hongjuan, from Jiangsu, collects data at home for around six hours daily, according to SCMP.
Her tasks include arranging books, wiping tables and moving objects. She earns about 4,000 yuan (nearly ₹57,000) monthly while caring for her child.
Robot training now extends beyond homes and studios. In August, Shenzhen company Physen AI partnered with a sanitation company.
Street cleaners wore devices that recorded their movements while sweeping roads. Their responses to passing electric bicycles helped capture changing conditions in real surroundings.
Online reactions have been mixed, with some users calling the work a cyber assembly line. Others compared it to home-based migrant labour. One worker said he had at least learned household chores.
“It is deeply ironic. Humans are teaching robots to work, yet those machines will one day replace the very workers who trained them,” wrote one user.
Reuters reported in August that human trainers in used headsets and controllers to guide simple tasks like sorting crates and making coffee.
Progress was slow, with beginners sometimes needing 300 attempts to produce a usable movement. Experienced trainers may need 50 attempts, according to Reuters.
Such robots could help address worker shortages as China’s population ages and shrinks. However, three staff members said the subsidised project currently lacks a clear route to profits. They cited high costs and low data prices.
Government and manufacturing investment are advancing faster than robots’ ability to work beyond carefully planned demonstrations, Reuters added.
