Vineet Sawant, a delivery driver in Mumbai, reflects on the challenges he faced when starting his job due to language barriers. As a Marathi speaker with limited English reading skills, Mr. Sawant struggled to understand delivery instructions provided solely in English. This became an issue for his employer, Zepto, which claims to provide "India's Fastest Online Grocery Delivery."
To address these obstacles, Zepto collaborated with Reverie Language Technologies to launch an AI translation service in the app. Now, delivery drivers like Sawant can select from six languages, easing the burden of understanding instructions. "I don’t have to guess anymore," he states, grateful for the clarity the translations provide in his native language.
With 22 official languages and numerous dialects, language barriers are common across India. Professor Pushpak Bhattacharyya, an AI expert from IIT Mumbai, notes that without technology capable of understanding these languages, many individuals risk being excluded from the digital era, particularly in crucial sectors such as education and healthcare.
While generative AI systems like ChatGPT have highlighted the need for multilingual support, challenges remain. According to Professor Bhattacharyya, gaining access to high-quality, refined data is an uphill battle, particularly for lesser-known dialects and tribal languages, where digitized resources are often scarce.
Reverie's co-founder, Vivekananda Pani, acknowledges that while these translation technologies enhance communication, there is a danger of overshadowing less prevalent dialects. He emphasizes the need to ensure that the benefits of AI do not come at the expense of India's rich linguistic diversity.
To combat these challenges, Professor Bhattacharyya has contributed to Bhashini, a government initiative aimed at developing high-quality datasets for AI. Launched in 2022, Bhashini has made significant progress with over 350 AI language models processing more than a billion tasks, catering to numerous government departments across India. The technology enables multilingual chatbots for public services and translates government information into local languages.
Amitabh Nag, CEO of the Digital India, Bhashini Division, highlights the importance of representing India's culture and languages through tailored AI solutions rather than depending on global models. He envisions that rural users will soon enjoy voice-enabled access to government services and financial tools in their native languages.
Meanwhile, researchers are exploring AI applications in various health-related fields, such as smoking cessation. Kshitij Jadhav from IIT Mumbai is working on an AI model that will tailor support to individuals at different stages of the quitting process, focusing on providing empathetic, language-specific guidance.
As for Vineet Sawant, translating tools have significantly improved his work efficiency, increasing his daily delivery count from ten to around thirty parcels. He believes that language access fosters a sense of belonging and confidence among workers like him who may struggle with English. "When the app speaks our language, we feel more confident, and we work better," he concludes, showcasing the profound impact of AI-driven translation technologies.
Through these advancements, India's digital landscape is evolving to embrace the linguistic variety of its population, ensuring greater inclusion for all.
