Inside China’s smart factories where workers teach AI to replace them

China’s race to dominate artificial intelligence-driven manufacturing is entering a controversial new phase, with factories increasingly using cameras, sensors and human-motion tracking systems to teach robots how to perform jobs currently done by workers.

Across industrial hubs in Shanghai, Shenzhen and Jiangsu, Chinese companies are deploying AI-powered robotics systems capable of learning from human behavior in real factory environments.

The shift is fueling fears that workers are effectively training the machines that could eventually replace them.

The debate intensified after viral videos circulating online showed factory workers wearing head-mounted cameras while carrying out repetitive tasks such as sewing, folding and assembly-line packaging.

There have been a speculation that the footage was being used to train AI systems and humanoid robots to mimic human movements.

Read also: Nigerian man faces 40 years in U.S. prison over romance scam, money laundering 

Although some of the viral footage originated outside China, the controversy has drawn renewed attention to Beijing’s aggressive push toward automated smart factories and dark factories which is manufacturing plants where production can continue with minimal or even zero human workers.

How the technology works

The technology works by collecting enormous amounts of visual and motion data from workers.

Cameras attached to helmets, factory ceilings or robotic systems record how humans grasp tools, move components, inspect products or navigate factory floors.

The AI models then analyse those movements to train machines through a process known as imitation learning or embodied AI.

Chinese robotics startup AgiBot is among the companies leading that transformation and reports show the Shanghai-based company uses human workers and teleoperation systems to train humanoid robots for manufacturing tasks.

The robots learn by observing and repeating human actions on electronics assembly lines.

The company’s technology is already being tested at factories producing smartphones, VR headsets and consumer electronics.

Developers say the goal is to create robots capable of adapting to changing factory environments rather than simply performing one repetitive motion.

China’s government backs automation drive

China’s government has heavily backed the automation drive as the country grapples with slowing economic growth, rising labor costs and a shrinking workforce.

Reuters reported that Beijing has poured billions of dollars into robotics and AI manufacturing programs, while local governments offer subsidies to companies building humanoid robots and automated production systems.

The scale of deployment is expanding rapidly as China installed nearly 300,000 industrial robots in 2024 which is more than the rest of the world combined, according to reports cited by The Wall Street Journal.

Chinese officials argue that automation is essential to maintaining the country’s manufacturing dominance. State media has framed AI integration as a way to improve efficiency, precision and industrial competitiveness.

However the transition could trigger widespread disruption for millions of low-skilled factory workers.

Technology researchers say embodied AI where machines learn physical tasks through real-world interaction represents the next major frontier in artificial intelligence.

Unlike traditional industrial robots programmed for fixed tasks, newer humanoid systems can adapt, self-correct and learn from human demonstrations.

China’s humanoid robotics sector is experiencing an unprecedented boom, with over 100 companies driving rapid development and commanding a massive share of global shipments.

Read also: Nigeria unifies literacy delivery under one national standard—Alausa

Robot schools and training centers are actively increasing this growth by training machines on human datasets, transforming how AI and physical automation integrate.

For factory workers, however, the technology presents a growing uncertainty that the same hands now teaching AI systems may eventually become unnecessary on the factory floor.