Data-labelling workshops once hailed as a solution in the country’s poverty alleviation campaign face a painful reckoning
7-MIN READ7-MIN ListenAlcott Weiin Tongren, GuizhouPublished: 6:00am, 5 Apr 2026Updated: 6:10am, 5 Apr 2026Before autonomous driving freed up the hands of Beijing’s middle class, thousands of workers some 1,500km (930 miles) away in China’s southwestern Guizhou province clicked away at computer screens to teach AI about navigating traffic.
In the mountainous city of Tongren, where incomes are less than half those in Beijing, the work of data labelling – marking residential buildings, pavements, roadways and traffic lights – shaped the artificial intelligence guiding those vehicles.
The job required little formal training and could be done almost anywhere, two factors that brought together the interests of tech companies seeking AI training data, the government aiming for job growth and workers needing jobs.
In China, AI-labelling workshops run by leading tech firms and supported by the state played a pivotal role in Beijing’s drive to alleviate absolute poverty in rural Guizhou, historically one of the country’s poorest provincial economies by GDP per capita.
One AI data-labelling poverty alleviation project created jobs for mothers with little education while enabling them to stay near home.
Success came early as the interests of three groups aligned.
