India, home to one of the world’s largest information technology engineers and scientist pools, saw an interesting debate over the course of direction the country should follow for its artificial intelligence (AI) industry which will reshape human life, business, and economy in future. While tech industry titan Nandan Nilekani, co-founder of software giant Infosys, advised against building the country's own large language models (LLMs), some other experts such as Google’s India Research Director Manish Gupta advocate building the country’s own large language model.
In recent times, two completely different views emerged about India’s future AI strategy. While the creator of India’s unique biometric-based identification system - Aadhaar, Nandan Nilekani, a giant of the Indian information technology ecosystem, argued in favour of investing in developing infrastructure and compute services for AI rather than developing its own large language models.
Big industry giants such as OpenAI has its LLM ChatGPT, search engine giant Google has Gemini and Microsoft as Copilot while Meta has its own Artificial Intelligence model Llama and Elon Musk-owned micro-blogging platform X (formerly Twitter) has its own Grok AI Services in addition to several other large language models.
However, Nilekani’s views were contradicted by Google’s India Research Director Manish Gupta, who while speaking at Bengaluru Tech Summit this month, contradicted Nilekani’s advice saying that Nilekani is not practicing what he preached. Gupta said that Nilekani developed India’s unique biometric-based identification system known as Aadhaar which is a foundational service as he did not build use cases for Aadhaar.
Today, Aadhaar is the most widely used identification system in India for availing a range of government and private services, subsidies and for identification purposes.
Gupta said: “He is not preaching what he practised. He revolutionised India’s technology landscape by starting with the basics. With Aadhaar, he did not start with use cases; he started with building foundations. We too must, using our constraints as ingredients for innovation.”
However, while responding to Manish Gupta’s criticism, Nilekani later explained why he believed that rather than building its own LLMs, India should invest its energy and money into building use cases and computing resources.
“Foundation models are not the best use of your money. If India has $50 billion to spend, it should use that to build compute, infrastructure, and AI cloud. These are the raw materials and engines of this game,” Nilekani told the media.
AI and Government
This debate on the country’s future AI strategy is not only limited to industry titans but the government is also working to harness the immense potential of artificial intelligence.
Earlier this month, India’s information and technology minister Ashwini Vaishnaw informed the Lok Sabha that the government wanted to harness the AI’s potential for national development, saying that the government believed in democratisation of technology. He said India’s AI Mission was based on seven pillars for national development and the government was setting up AI labs in medium and small towns such as Gorakhpur, Lucknow, Shimla, Aurangabad, Patna, Buxar, and Muzaffarpur to democratise the technology.
“AI’s most significant applications will be in sectors that directly influence the lives of people. Agriculture, education, healthcare, logistics, and the financial sector stand to benefit immensely from AI solutions tailored to the needs of the country,” said the minister.
Ashwini Vaishnaw also informed the members of Parliament that the government was ready to bring a law to regulate the AI sector but it required ‘consensus building’.
The minister said accountability has to be established in society and the legal framework has to change for which lots of consensus was required but the government was open to the idea of bringing a law to regulate the AI sector.
India needs its own LLMs for regional languages
Though the debate over the future direction of India’s AI strategy has sharply divided the experts, large language models built by tech giants such as Google, Meta, Microsoft, OpenAI and others are primarily designed for processing and drawing insights for information available in the English language.
India has 22 constitutionally recognized languages and more than 1,500 other dialects.
A country with multiple widely spoken languages such as Hindi, Marathi, Bengali, Odia, Tamil, Telugu, Malayalam and several other languages need their tailor-made large language models that can foster innovation and research for native speakers in their own language.
India’s homegrown LLMs
For example, India’s Ola Group, which started as a ride-hailing platform and then diverted into electric two-wheelers with Ola Electric and other services, launched its own AI platform known as Ola Krutrim AI Platform and Cloud Services.
Ola offers Krutrim AI Cloud Services, offering customizable pre-trained models that can also be used for seamless translation powered by AI.
AI Giant NVIDIA’s work on Indic language LLMs
More importantly, NVIDIA, the world’s dominant supplier of AI-related hardware and software, has launched a small languages model (SLM) in Hindi called NVIDIA NIM.
The NVIDIA NIM microservice model, which stands for Nemotron-4-Mini-Hindi-4B model as it uses 4 billion parameters has been derived from NVIDIA’s Nemotron-4 15B, a 15-billion parameter multilingual-language model developed by the Santa Clara, California based US tech giant.
NVIDIA is supporting local startups that are building language models in Indic languages.
For example, Mahindra Group’s IT services company Tech Mahindra, has used NVIDIA’s Nemotron Hindi NIM microservice to develop an AI model called Indus 2.0 for Hindi users and users of dozens of Hindi dialects.
Secondly, India’s Sarvam AI offers speech-to-text, text-to-speech, translation and data parsing models using Sarvam 1, India’s first multilingual LLM.
In addition to this, companies such as TCS, Wipro, Infosys, and Zoho Corporation, and e-commerce platforms such as Flipkart are working to use Indic language LLMs.
