Since the launch of the Made in China 2025 initiative five years ago, China’s central government has issued a raft of policies and documents pushing companies to embrace intelligent manufacturing and industrial upgrading. Three decades of rapid development earned China the nickname “the world’s factory.” Now, faced with an aging population and dwindling pool of migrant labor, officials hope to keep the title via automation.

In theory, this process should create new knowledge-based and technical jobs for an increasingly skilled workforce. The Ministry of Human Resources and Social Security has claimed automation will lead to millions of manufacturing jobs for skilled workers. However, in the course of conducting fieldwork in Chinese factories, I saw firsthand the potential flipside of “industrial upgrading”: Labor is not necessarily upgraded along with technology. If anything, it is at risk of being downgraded.

According to data from the International Federation of Robotics, China’s robot density — the proportion of robots to employees — has risen faster than anywhere else in the world in recent years, from 25 units per 10,000 employees in 2013 to 140 units per 10,000 employees in 2018. Behind this increase is a significant reduction in employment, especially along the coast where the automation process is most advanced. According to media reports, the eastern Zhejiang province has lost 2 million workers in two years, and the number of workers in the southern manufacturing hub of Dongguan has dropped by 200,000 in the past three.

In the summer of 2018, I and one of my students, Xiaoxin, went undercover as ordinary workers at two factories in the southern Guangdong province. One of the two was a large manufacturer of air conditioners I had previously visited openly as a researcher. At the time, management claimed workers on the automated production line were hand-picked from the manual assembly lines for being “ambitious” and “fast learners.” I was told they were required to take part in a variety of training courses, needed to have secondary vocational education or higher, and received a salary twice that of those on the manual assembly line.

It was only after we entered the factories as workers that I discovered the truth. Some of the so-called hand-picked automated production line workers had been working there fewer than three months, while others were students on their summer vacations. Their duties were essentially the same as those on the manual assembly lines: simple and repetitive acts such as affixing foam or coils. They didn’t need additional education or skills, and they earned the same as those working on the manual assembly lines.

Indeed, we found many skilled workers in other factories we investigated had been replaced by robots, leaving them with no choice but to accept low-skill work. All their previous learning and experience was wasted, their pay downgraded. For example, welders in one auto parts factory used to receive a monthly subsidy of 350 yuan ($50) for performing a so-called core job. After the factory automated, their overtime pay and subsidies were both cut, and their monthly incomes slashed. In the words of one veteran worker: “When we’re completely replaced by robots, we’ll have to live on air.”

Ironically, even as automation devalues skilled labor, it does little to decrease workload intensity. At another factory we went to — a manufacturer of office desk chairs — I was assigned to the traditional manual assembly line, while Xiaoxin was put to work on the semi-automated packaging line.

The manual assembly line had more than a dozen workers, each performing a set of repetitive tasks. We worked nonstop whenever production needed to be ramped up, but could take short mental breaks when things slowed down. Workers on the manual assembly line still have some flexibility, as unfinished work can be temporarily set aside and other workers can lend a hand after they finish their own work. When production was higher or certain parts of the line couldn’t keep pace, the team leader allocated additional manpower to prevent parts piling up too much.

On the packaging line where Xiaoxin was stationed, however, the robots on the second half of the line constantly ran at high speed, leaving the human workers in the first half to find ways to keep up. On a day with lower production volumes, Xiaoxin might have 20 seconds to complete a task; the next day, when production quotas were higher, she might have just 15 seconds. If the workers in the first half of the line couldn’t keep up, all they could do was hit the pause button, package and process the goods, then let them flow into the machine end. But you can’t just keep hitting pause — the managers will come over to check on you. You just have to work like crazy.

After a few days, Xiaoxin had blisters on her hands and back pain. Essentially, automation allowed managers to use machines to fully maximize the labor of workers. But people can’t operate at full speed nonstop like machines can. It’s inhuman.

And when robots become the center of production, the space for worker resistance shrinks. One technician told us about a laborer on the packing line who deliberately sabotaged the machine, because it was going so fast that even a skilled worker like him couldn’t keep up. He damaged the outside of the packing boxes, forcing the machine to shut down when it couldn’t identify them. Management quickly caught on to what was happening, however, and the old workers were gradually replaced. Of the six workers on Xiaoxin’s automated packaging line, only the team leader and a forklift driver were longtime employees. The rest had all been there fewer than three months.

Few of them stood a chance of moving up the job ladder. Positions such as engineering technician account for only a fraction of the workforce. I met some young factory workers who thought they could get ahead by paying for training courses — a common tactic among previous cohorts of migrants. But their lack of formal education, among other difficulties, makes it hard for them to earn promotions by taking short training courses which don’t offer officially recognized certifications.

In short, as machines become the main locus of production, they, and not labor, determine its workflow and rhythm. Workers are reduced to a secondary role.

Yet the vast majority of workers remain unaware of the potential risks machines pose to their jobs. Manufacturing workers are still in short supply, and workers are still relatively able to find jobs. Some think machines will always need operators, and so don’t consider them a threat.

In the long run, rising unemployment and a weakening of workers’ influence over the production process are inevitable. In the words of one business owner, “What you don’t want are workers using their brains and doing what they think is best. Ideally, they just do whatever you want them to.”

This way of thinking is common among factory owners. They’re not just turning to automation to save on labor costs, but also to reduce their overall reliance on workers.

This will only become easier as robots become more affordable and technology continues to improve. At present, the price of a domestically made industrial robot has dropped from 100,000 yuan two or three years ago to around 30,000 to 40,000 yuan today. China’s hoped-for industrial upgrade is well underway, but it’s important to remember: When automation becomes the norm, it’s the capitalists who will benefit, not the workers.

Translator: David Ball; editors: Cai Yiwen, Kilian O’Donnell, and Shi Jiannan; portrait artist: Zhang Zeqin.

(Header image: A technical worker adjusts a robot arm at a workshop in Jixing, Zhejiang province, Dec. 3, 2018. Hu Lingxiang/People Visual)