India’s Digital Infrastructure is Changing Super Fast with AI

India's-Digital-Infrastructure-is-Changing-Super-Fast-with-AIIndia's-Digital-Infrastructure-is-Changing-Super-Fast-with-AI

AI is transforming India’s e-governance landscape and reaching the farthest of underserved communities. From processing numerous queries round-the-clock to enabling bilingual support, the advancements in AI are making government initiatives more accessible and efficient than ever.

“Our AI systems now process between 5 to 7 lakh queries every month, operating 24/7, which is crucial in ensuring millions of citizens get timely assistance,” said Sharmishtha Dasgupta, deputy director general of the National Informatics Centre, during the Nvidia Summit held in Mumbai last week.

She goes on to mention that these queries range from enrollment and eligibility checks to updates and troubleshooting.

The system’s capacity to manage such high volumes highlights its scalability and efficiency, as well as its alignment with the goals of the Digital India campaign to make government services accessible to every citizen.

Dasgupta also noted that AI-powered bilingual support systems have been functional in bridging linguistic and digital divides.

“Bilingual AI-powered support systems are making interactions with government schemes like PM Kisan Samman Nidhi Yojana straightforward, reducing complexities and ensuring citizens can engage in their preferred language,” she added.

For instance, IRCTC, an extension of Indian Railways, is also using the conversational AI platform AskDISHA 2.0 chatbot to help customers book railway tickets through voice, chat, and click-based operations.

“In our constant pursuit to enhance the user experience, leveraging new age technologies, today we are taking a giant leap. Now passengers can book train tickets in a conversational manner, leveraging our AI Virtual Assistant, AskDISHA 2.0, powered by CoRover Conversational AI platform,” Rajini Hasija, CMD, IRCTC said.

Innovative Case Studies

Manohar Paluri, VP of AI at Meta, recently told AIM,“India is possibly among the top three in terms of Llama downloads and variants, and it’s also among the top two in terms of how many developers there are. The appetite for technology and the appetite of people adapting to new technology in this part of the world is amazing.”
Paluri has talked about one of the case studies, Pratham, a nonprofit for education.

“Pratham is an example of how this technology is being used to teach kids in an affordable way. So, you are scaling yourself where these chatbots actually can help you learn about a particular subject very quickly,” he added.

He explained how farmers can now use this technology in their native language and get the best information on agriculture and financial assistance, which was previously not feasible or accessible.

“So when you think about learning, it is going to be redefined now with technologies like Llama as well as MovieGen,” Paluri added.

He also discussed the benefits of an open ecosystem, where people can take the model and fine-tune it as per their needs.

“I was actually in Japan two weeks ago. There were Japanese versions of Llama, where they were able to tune it,” he said. Paluri also mentioned a Korean version and a Korean high school math version of Llama.

India’s AI Growth Trajectory

According to the principal scientific adviser’s (PSA) office, AI expenditure in India saw a 109.6% increase during the year 2018 to reach USD 665 million. Moreover, AI expenditure is projected to grow with a CAGR of 39% over the period 2019-2025, touching around USD 11,781 million by 2025. As per NITI Aayog, by 2035, AI has the potential to add USD 1 trillion to India’s economy.

Vernacular AI Models

Research and development firm Sarvam AI recently introduced its first version of OpenHathi, marking the release of India’s first Hindi LLM ((Large Language Model) in the OpenHathi series of models. Built on a cost-effective platform, this model, extending from Llama 2 7B, exhibits performance similar to that of GPT-3.5 for Indic languages.

Then comes Tamil Llama, developed by Kaggle Master Abhinand Balachandra. This Indic LLM is engineered to elevate the Tamil language domain and has been built on top of Meta’s Llama 2.

Ramsri Gourham Golla, another avid developer based out of Hyderabad, is working on building a Telugu Llama. Meticulously trained on Llama 2, his LLM showcases remarkable efficiency in the token count for Telugu text. As per Golla, the Telugu language consumes fewer tokens as compared to English.

A group of developers from Odisha are working on building Odia Llama, aka OdiaGenAI, a Llama 2 LLM fine-tuned for the low-resource Odia language using a prepared instruction set containing domain knowledge.

Not just in India. There has been a significant rise in the number of vernacular models globally built on top of Llama 2. Some of them include SeaLLMs, VinaLLaMA, LLaMAntino, and Dutch Llama, among others.

“We are trying to bring high-quality Indian tokens into Llama, so that it works for Indian languages,” said Paluri.

This means Lama, being the engine for Meta-AI, will work for Indian languages. This will help billions of people in India, who are using Meta-AI on WhatsApp and other Meta products.

The National Informatics Centre’s (NIC) AI platforms support both Hindi and English, making essential services more accessible for millions who may not be fluent in English.

Future Prospects

During the Nvidia Summit, Tanusree Barma, deputy director general of Unique Identification Authority of India (UIDAI), highlighted the transformative AI developments within UIDAI, as the organisation moves towards adopting and using indigenous AI capabilities.

Barma stated, “Within UIDAI, we have been silently having an AI revolution to enable indigenous implementation of LLMs as well as AI models to apply for biometric duplication and detection.”

Barma also talked about the critical role that AI now plays in the nation’s digital identity framework, emphasising how these AI models help strengthen security, improve accuracy and drive innovation for the UIDAI.

She further said that by focusing on homegrown AI solutions, UIDAI not only aims to reduce dependency on foreign AI technologies but also ensures data sovereignty and control, a major step in India’s broader AI ambitions.

Meanwhile, India is steadily moving towards its generative AI moment. Forging ahead is the initiative led by IIT Bombay for building foundational AI models called BharatGPT, a government-backed initiative, which aims to build multilingual and multimodal models through a public-private partnership (PPP). The consortium has received immense support from the community, with Jio’s chairman, Akash Ambani, also announcing his partnership for the mission.
To understand how BharatGPT aims to integrate generative AI with Bharatiya culture, AIM got in touch with Professor Ganesh Ramakrishnan from IIT Bombay, who is leading the initiative, and gained insights about what sets this mission apart from others in the field.

The post India’s Digital Infrastructure is Changing Super Fast with AI appeared first on AIM.

Follow us on Twitter, Facebook
0 0 votes
Article Rating
Subscribe
Notify of
guest
0 comments
Oldest
New Most Voted
Inline Feedbacks
View all comments

Latest stories

You might also like...