2 weeks ago
10
India’s linguistic landscape has long been characterised by remarkable diversity, accompanied at times by tension. This complexity is further highlighted during the ongoing debates and disputes over the use of particular languages in specific cities and states.
While the debate continues, AI could play a role in addressing it, or at least mitigate it as much as possible. Here is where initiatives like Bhashini, an Indian government project; AI4Bharat, a research lab at IIT Madras; People+ai, and Sarvam AI startup, come into the picture.
As Nandan Nilekani, co-founder of Infosys, said in a recent video featuring People+ai, “If you really want to make access to technology universal, we should be able to provide that access in a language of choice of the citizen. A big part of inclusion is access, a big part of access is having universal access in every language,” Nilekani said.
The most recent example of this approach was Sarvam AI translating Lex Fridman’s podcast episode with Prime Minister Narendra Modi into nine Indian languages, including Odia, Bengali, Tamil, Telugu and Malayalam, among others.
The podcast’s availability in multiple languages allowed the audience to view the podcast in different languages, making it more inclusive for those not fluent in English. This approach effectively addresses the linguistic diversity of the Indian audience.
Moreover, Sarvam AI recently announced its partnership with Unique Identification Authority of India (UIDAI) to improve the user experience for Aadhaar-related services.
Sarvam will deploy AI-powered voice-based interactions to enhance resident-centric use cases. The AI solution will support voice interactions and fraud detection in 10 languages, including Hindi, English, Telugu, Tamil, Marathi, Gujarati, Kannada, Odia, Punjabi, and Malayalam. More languages will be added in the coming months.
In the video, Vivek Raghavan, co-founder of Sarvam AI, explained that speaking in one’s native language and interacting with the digital world shouldn’t be a choice that a user should have to make. “You should be able to interact with this digital world in your own language and through voice, and, in many ways, the cultural imprint of the country is in the voice of the people.”
Commenting on this announcement, Pushpak Bhattacharyya, chairman of the National Committee on Indian Language Standards, told AIM that such AI initiatives backed by the government are very important for solving language problems in the country.
“The government plays a very crucial role there. The think tank is academia. The perspective, depth, and way forward come from academia. And the energy comes from industry and startups, who execute things on the ground,” Bhattacharyya said.
Similarly, Harneet Singh, founder of Rabbitt.ai, told AIM, “As a country, we seek a high level of personalisation…We’re not a self-centered economy, nor a self-driven one.” For instance, he explained that government agencies in India should guide people through the process of accessing their Aadhaar cards.
Multilingual and voice-based AI solutions can truly transform challenges related to access. “A lot of research is still needed to make these multilingual AI models sound natural and human-like so that people feel comfortable using them. We need more Indian voices and vernacular datasets to improve accuracy. Once that happens, it will be a game-changer for the country,” he added.
Bhashini, AI4Bharat Making Their Way
Launched by PM Modi in July 2022, Bhashini is a key component of India’s National Language Translation Mission (NLTM). It focuses on developing and providing tools and resources for language technologies, including speech recognition, translation, and natural language processing.
The goal is to enhance accessibility and inclusivity for users across different languages and regions in India. Bhashini chief Amitabh Nag, in a previous interaction with AIM, said that he believes his work belongs to everyone, and is not restricted to any political party.
“Throughout one’s life, regardless of daily endeavours, individuals should not feel pressured to learn another language,” Nag said, expressing a strong desire to eliminate the language barrier in this country.
The Lok Sabha secretariat and the electronics and information technology ministry have signed a memorandum of understanding (MoU) to develop Sansad Bhashini—AI-driven in-house solutions to streamline parliamentary operations in multilingual languages.
Sansad will provide its corpus of legislative records for training and fine-tuning these AI solutions, while Bhashini will contribute translation capabilities and technical expertise.
As a multicultural nation, one of the biggest challenges in India’s legal system has been language accessibility. A large part of Supreme Court judgments are delivered in English, making it difficult for non-English-speaking litigants to fully comprehend their cases.
To address this, the apex court, in collaboration with the National Informatics Centre (NIC), has deployed AI-powered translation tools that convert judgments into 18 Indian vernacular languages. These translated judgements are accessible through the eSCR portal.
“AI adoption in India will have much more variety, many more nuances, and a lot of diversity. That’s what makes it unique,” Bhattacharyya said, adding that parliamentary proceedings are publicly available, open documents. They should be accessible to as many people in India as possible, including students who are not proficient in English.
Notably, the introduction of the three-language formula under the National Education Policy (NEP) 2020, which recommends learning three languages with at least two being native to India, has further stirred discussions around language preference and regional identity.
AI4Bharat, which closely works with Bhashini to build these datasets and solutions, also plays a crucial role. Recently, the research team launched IndicTrans3, a multilingual translation model designed to support translations across 22 Indic languages.
Mitesh Khapra, head of AI4Bharat, wrote on LinkedIn, “Over the past four years, we have been on a mission to accelerate Indian language AI—building large-scale datasets, models, and tools, and releasing everything open source for the community. Now, all our contributions are available on Hugging Face!”
Why is This Important?
In a previous discussion with AIM, former head of People+ai, Tanuj Bhojwani, said that building AI in India is a game that’s very different from that in the West. “If you look at how much an AI solution could mean to a user, it’s much higher in India with a much larger volume,” he added. In the West, it’s about acquiring enterprise customers willing to spend millions of dollars.
In India, however, it’s a high-volume, low-value game, where the AI users typically do not pay so much. These users would feel more comfortable using AI in their native languages. This solution is at the population scale. For India to flourish in AI, models that understand India’s linguistic nuances and cultural complexities must be developed.
Nikhil Malhotra, chief innovation officer at Tech Mahindra, earlier told AIM that while building Project Indus for Tech Mahindra, he and his team visited different cities and towns in India to collect data for training models in Indic languages.
“This is important because for anyone in India, their first thought process triggers…in their own language—that’s AI for Bharat for me,” Malhotra explained, emphasising that this goes beyond people who think natively in English. According to him, this includes not just the rural population but also many homemakers in the country, who are first-time smartphone users.
For Ankush Sabharwal, founder and CEO of CoRover and BharatGPT, AI for Bharat means not just introducing AI to Bharat but also using AI to educate the Indian population about government policies, changes, and schemes.
“We are what we are,” Sabharwal said, emphasising that while educating the population in English and other skills is valuable, language should never be a barrier to access and opportunity. “Make everything natural and accessible [through AI].”
“Not just translation, but also allowing everyone to listen and speak, which is humane and natural, instead of reading and writing, will help bridge any technology gap, language gap and knowledge gap to a certain extent,” Sabharwal told AIM.
AI-driven language solutions aren’t just making things easier; they’re changing the game for accessibility in India with different languages. With Bhashini, AI4Bharat, People+ai, and Sarvam AI leading the way, we’re inching closer to a future where language is no longer a barrier.
