Shruti Suman
Google and Microsoft are placing their bets on a 27-year-old entrepreneur to make artificial intelligence (AI) work effectively in India. Preethi P., who resides in a small village called Agara, situated three hours southwest of Bangalore, is one of the 70 workers employed by a startup named Karya. These workers are tasked with collecting text, voice, and image data in various Indian regional languages. Unlike many other data workers, Preethi receives decent compensation for her efforts, earning significantly more than her usual income as a tailor.
In just three days of working with Karya, Preethi earned 4,500 rupees ($54), which is more than four times her monthly income as a tailor. This income has allowed her to make a monthly installment payment on a loan taken to repair her home’s crumbling mud walls. Karya, founded in 2021, aims to meet the growing demand for data, especially with the recent surge in generative AI technology.
India is expected to have nearly one million data annotation workers by 2030, and Karya stands out by offering its workforce, primarily composed of women from rural areas, wages that are up to 20 times the prevailing minimum wage. Tech giants have recognized the need for high-quality data in Indian languages to better serve non-English speaking users, and they are partnering with Karya to address this challenge.
Microsoft has utilized Karya to source local speech data for its AI products, while the Bill & Melinda Gates Foundation is collaborating with Karya to reduce gender biases in data used for large language models. Google is also relying on Karya and other local partners to collect speech data in various Indian districts, with plans to expand to cover more languages and dialects.
The initiative is critical because many AI services have been predominantly developed using English-language internet data, which doesn’t adequately represent the diversity of languages spoken by users in countries like India. Karya, a social impact startup supported by grants, targets workers in rural areas to expand the diversity of languages represented in AI training data.
Karya’s founder, Manu Chopra, aims to combat poverty while improving the quality of data. The startup has provided an opportunity for thousands of people, including women, to earn income while contributing to AI development. Karya’s model is being considered for expansion to other regions in Africa and South America.
The work done by Karya not only enriches AI training data but also empowers people in rural areas like Yelandur to earn income and support their families. The significance of AI might not be fully understood by these workers, but the impact on their lives is undeniable.
