Mumbai: On a humid afternoon in Mumbai, a delivery rider fumbled through an app that spoke only English, losing precious time and income with every misstep. It was a small, everyday frustration but one that illustrates a much larger problem: the digital world in India often speaks a language that millions of citizens cannot understand. Until recently, translation tools were clunky and unreliable. Then, with the advent of Indian-language AI, the app suddenly spoke his mother tongue Marathi and the barrier vanished. This seemingly small change is the tip of a transformative wave sweeping across India’s digital landscape, aimed at ensuring that no citizen is left behind due to language.
This effort is part of the Modi government’s next phase of digital public infrastructure, where technology is deliberately used to reduce inequality and widen access. Just as Aadhaar provided a digital identity and UPI revolutionized payments, initiatives like Bhashini and BharatGPT, cornerstone elements of the IndiaAI strategy, aim to create a unified digital voice for every Indian, no matter which language they speak.
At the foundation is Bhashini, launched in 2022, which seeks to make digital content and services accessible in all major Indian languages through advanced AI. Today, it processes over ten million translation requests daily. Importantly, it goes beyond dominant languages, leveraging the Bhasha Daan initiative to include historically neglected tongues such as Khasi, Manipuri, and Bodo. Its interfaces now power government portals from e-learning to telehealth, bringing critical services previously confined to English-speaking users into vernacular communities across the country.
Building on this is BharatGPT, India’s first sovereign large language model, developed with the collaboration of Indian institutes and startups. Unlike global models primarily trained on English, BharatGPT is tuned to Indian languages and cultural contexts. It is multimodal, supporting text, voice, and video interactions, and is designed for public services—helping citizens with tasks from ticket bookings to insurance inquiries. By focusing on context and cultural grounding, it avoids the generic, one-size-fits-all approach of many global AI systems.
The impact is already tangible. Rural women entrepreneurs can navigate government credit schemes independently, farmers can query mandi prices in their local dialect, and patients can access telemedicine advice without intermediaries. This is not just state-led; the ecosystem matters. AI4Bharat at IIT Madras builds open datasets for India’s less-spoken languages, while startups like Sarvam AI release conversational models aligned with the IndiaAI mission. International partnerships, governed by Indian data sovereignty rules, provide computational support while keeping the control and innovation domestic.
Yet challenges remain. Data coverage across dialects must expand, computing resources should be accessible to smaller institutions, and AI must reach the front lines in schools, clinics, and rural offices. Equally critical is investment in human capital: India’s AI talent remains concentrated and small in number, with top-tier researchers often migrating abroad. Without nurturing and retaining experts, India risks being a consumer of AI rather than a pioneer. Complementing this, ordinary citizens must gain AI literacy in their native languages to fully benefit from these platforms.
Globally, India’s approach is unique. In the U.S., AI is largely corporate-driven, focusing on model size for competitive advantage, with multilingual adaptation an afterthought. China’s state-led models prioritize control rather than linguistic diversity. The European Union has fragmented efforts lacking a public-first LLM, slowing real-world deployment. India, in contrast, is pragmatic and action-oriented: it steers AI to solve a social problem, giving citizens a voice online without chasing sheer scale or centralization.
The implications extend far beyond India. Much of the Global South faces a similar tension between linguistic diversity and digital access. If India’s approach succeeds, it could offer a blueprint for integrating billions into the digital economy while preserving linguistic identities a model that could transform both development and democracy across the world.