BENGALURU: A project in the state of Karnataka in India has utilized the AI technology to build the country’s first AI-based chatbot for Tuberculosis. Villagers in Karnataka read out sentences in their native Kannada language, which is one of India’s official languages, to help gather data for the AI chatbot.
India has over 40 million native Kannada speakers. It is one of India’s 22 official languages, and one of over 121 languages spoken by 10,000 people or more in the world.
Nonetheless, many of these languages are not covered by natural language processing (NLP), the branch of artificial intelligence which enables computers to understand text and spoken words. This means that hundreds of millions of Indians are excluded from useful information and economic opportunities.
Principal researcher at Microsoft Research India, Kalika Bali, stated, “For AI tools to work for everyone, they need to also cater to people who don’t speak English or French or Spanish.”
Karya, a tech firm, is working on building datasets for AI models for education, healthcare and other services in India by having thousands of speakers of different Indian languages generate speech data. The Indian government is also building language datasets through an AI-led language translation system called Bhashini.
This platform includes a crowdsourcing initiative for people to contribute sentences in various languages, validate audio or text transcribed by others, translate texts, and label images. Tens of thousands of Indians have contributed to Bhashini.
Researchers believe that this effort is essential as there are many challenges in collecting data for less common languages in India, including the fact that electronic records are not plentiful and there is a lot of code mixing present.
Economic value
Only a small fraction of the more than 7,000 living languages in the world are captured in major NLPs. English is the most advanced, and many AI systems and models are primarily trained on English, limiting access to vital information for those who don’t speak this language.
Grassroots organizations and startups are attempting to bridge this gap. Microsoft Research India and Google-funded Project Vaani are working on collecting and open-sourcing Indian language speech data to advance AI models for various applications.
Karya works with non-profit organizations to identify workers who are below the poverty line and pays them to generate data. By empowering them, there is a potential to build AI products for the community, such as in the areas of healthcare and farming.
Village voice
AI technologies are being used to provide outreach at the grassroots level in India. Google-funded Project Vaani is collecting speech data from about one million Indians and open-sourcing it for use in automatic speech recognition and speech-to-speech translation.
AI-based chatbots are also being used by social enterprises like Gram Vaani to respond to questions on welfare benefits, providing much-needed outreach and support to communities in need.
Swarnalata Nayak, a worker who contributes speech data in her native Odia language, has found additional income through her work for Karya. She states, “I do the work at night, when I am free. I can provide for my family through talking on the phone.”
