Vineet Sawant has worked as a delivery driver in Mumbai for the past two years, facing unique challenges associated with language barriers. His primary language is Marathi, and his limited proficiency in English made it difficult for him to navigate job responsibilities, particularly when instructions were provided in English. Initially, he sought help from colleagues to understand his tasks, causing delays and potential errors in deliveries.
To address this issue, Zepto, his employer, partnered with Reverie Language Technologies a year ago to implement an AI translation service in its app. This integration allows drivers to select from six different languages, significantly easing communication. Sawant noted that he no longer struggles to interpret instructions, as messages are now provided in his preferred language, which enhances his efficiency.
Language challenges in India are widespread due to the country having 22 official languages and numerous dialects. Professor Pushpak Bhattacharyya from IIT Mumbai emphasizes that the lack of technology supporting these languages may exclude millions from accessing digital resources in sectors such as education and healthcare. While AI systems like ChatGPT are evolving, data availability for less common languages remains a significant hurdle. Many regional dialects lack sufficient digitized data necessary for creating robust AI language models.
Reverie Language Technologies is actively working to deploy its translation technology across various Indian companies, though there are concerns that focusing on more widely used languages could overshadow less common dialects. Meanwhile, Bhashini, a government initiative developed with Professor Bhattacharyya’s involvement, aims to create high-quality datasets for training AI in 22 languages. This project has already made strides, processing over a billion tasks and partnering with multiple government departments.
As the technology advances, initiatives like these could broaden accessibility for rural communities, facilitating essential services in native languages.
Source: https://www.bbc.com/news/articles/cn0qqzz1e4zo?at_medium=RSS&at_campaign=rss