Tilling the Data Farms of Guizhou
GUIZHOU, Southwest China — Every day, hundreds of vocational school students flock to a factory after class and sit down in front of rows of computers to label photos and analyze human speech. The data they generate is used in a variety of technology projects, from face and voice recognition to autonomous driving.
The factory is located in Bainiaohe Digital Town, a science and technology park around 50 kilometers from Guiyang, the capital of Guizhou province. Prior to an artificial intelligence conference held there in December of last year, Bainiaohe was virtually unknown.
With internet companies constantly improving their autonomous vehicles and computer programs besting elite players of the world’s most complex board game, China’s artificial intelligence boom is in full swing. But behind every big AI success story is big data: information that must be examined and processed by human beings. Today, many of these workers can be found in Guizhou.
Large corporations first started showing interest in the province — one of the poorest in China — in 2014. Because of its cheap labor and abundant electricity from hydropower stations, Guizhou was an appealing site for large data factories. In addition, the central and provincial governments began introducing preferential policies, such as science and technology parks, to help Guizhou outcompete its less-needy neighbors, attract tech companies, and improve provincial GDP figures. Today, China’s “big three” telecom operators have data centers in Guizhou, as do tech giants Apple and Foxconn.
Deng Xuechun, a 20-year-old student at Guizhou Forerunner College, has just started her shift at the Guizhou Mengdong Technology Co. Ltd. factory. Her job is to identify the vehicles in still frames from street camera footage, as well as any objects that may be obstructing the view. She is required to sit upright at all times and avoid speaking with her schoolmates sitting next to her.
Every day, Deng and her co-workers supply large artificial intelligence companies in China’s big cities with troves of data — but sometimes the sorting process can be difficult: Because she grew up in the mountains where cars are scarce, Deng can’t always recognize the manufacturer’s logo on each vehicle. She also has no idea whether driverless cars are safe, or whether she’ll even be able to get her own driver’s license in the future.
When asked why she came to work in Bainiaohe Digital Town, Deng tells Sixth Tone that the vaguely magical phrase “artificial intelligence” appealed to her. More importantly, however, she’s now able to earn money after class from her part-time job at the factory. Working seven to eight hours a day, five days a week, Deng is able to earn 2,000 yuan ($310) a month — a seemingly small sum that goes a long way toward supporting her rural family. Most of the other students working at the factory come from similarly humble backgrounds.
According to the National Bureau of Statistics, there were nearly 5 million impoverished people in Guizhou in 2015, accounting for roughly 9 percent of all impoverished people nationwide. To change the status quo, the government established vocational schools throughout the province with the aim of helping residents attain better jobs with higher incomes.
Data tagging doesn’t require a high level of education or complex skills, but it still takes time to get the hang of it. In her first month on the job, Deng earned just 800 yuan because she was slower and less accurate than her more experienced peers. The monotony of the work — and the toll it’s taking on her eyes — has prompted Deng to look elsewhere for employment after she graduates.
Others, however, have no intention of leaving the factory. Li Longguang, 23, started working as a data tagger in October 2015. He, too, struggled at first, earning just 500 or 600 yuan per month. Over the years, he’s seen the number of workers increase by leaps and bounds, from a few dozen to several hundred. After working a full year as a data tagger, Li was promoted to group leader of quality inspection, upping his monthly salary to 5,000 yuan. With his June graduation just around the corner, Li hopes to stay on at the factory and one day become a manager.
With increasing demand for data-processing services, the factory where Deng and Li work is constantly expanding its operations and hiring more staff. Recently, it rented a three-story building nearby. Li noticed that the sign on the outside of the building no longer said “data workshop,” but instead read “data factory.” He assumed it was just a typo.
He mentioned the apparent mistake to one of his co-workers, who offered a different take: “With 400 workers producing on a large scale, this is a factory,” said the colleague, without so much as turning away from his screen.
Contributions: Jacney Chan; translator: Lu Yunwen; editors: David Paulk and Lu Yunwen.
(Header image: Li Longguang walks between two rows of computers in Bainiaohe Digital Town, Guizhou province, March 25, 2018. He rose from his initial job as a data labeler to lead his own quality inspection team. Chen Jin for Sixth Tone)