Rural India powers global AI models

TENDING crops by day and then logging on for a night shift of data labelling, 27-year-old Chandmani Kerketta is part of a rising rural Indian workforce helping power an artificial intelligence revolution.

- 

A winter view in Zadoi County of Yushu Tibetan Autonomous Prefecture, northwest China's Qinghai Province.

(Xinhua)

From her home in India’s eastern Jharkhand state, Kerketta is part of an Al-driven labour shift that the government hopes will transform lives, including by bringing more women into the workforce.

The work is basic but essential for machine learning: data labelling, annotation and quality checks.

It is the type of information key for driverless cars, for example.

“This job helped me finish my studies, and help at home on our farm,” Kerketta said as she tended tomatoes and peas.

Kerketta, from one of India’s constitutionally recognised tribal communities, was the first in her family to attend college.

She initially worked as an office assistant at a data-processing firm in Jharkhand’s capital Ranchi, where she watched employees working at computers.

But after a computer course at her village school, Kerketta joined an estimated workforce of at least 200 000 annotators in India’s villages and small towns - a growing figure, and roughly half of the world’s data-labelling workforce, according to US-based Scry AI.

Rural-based workers can label hundreds of images, videos and documents during eight-hour shifts, either from home or from modest internet-connected centres.