India might not have yet created any indigenous LLMs of note, but its largest company is now on the case.
Reliance Jio is working with IIT Bombay to launch BharatGPT, a large language model which will be tailored to Indian needs. The announcement was made by Jio Chairman Akash Ambani while attending IIT Bombay’s TechFest.
“We have been working on a project with IIT Bombay to launch a Bharat GPT programme,” Ambani said. Ambani further outlined Jio’s ambitious AI plans, aiming to integrate it “not only as a vertical inside our organization, but also horizontally across all our sectors.” This includes launching AI-powered products and services in media, commerce, communication, and even devices, he said.
“At present, we have just scratched the surface with large language models and generative AI,” Ambani said. “The next decade will be defined by these applications,” he added. AI is going to “consume everything”, Ambani said, describing how his brother-in-law showed him an AI app to control the temperature of his mattress. “To me, yes AI stands for Artificial Intelligence, but it also stands for All Included,” he added.
BharatGPT won’t be Reliance’s first foray into the burgeoning AI space. In September, Reliance had partnered with NVIDIA, whose chips are powering much of today’s AI revolution, to make AI supercomputers in India. The two companies had said they would collaborate to develop India’s own foundation large language model (LLM) along the lines of ChatGPT, which would be trained on India’s diverse languages and tailored for Indian use-cases. The companies had said they would work to build an AI infrastructure that is an order of magnitude more powerful than the fastest supercomputer available in India today.
It’s not clear if BharatGPT and the NVIDIA project will lead to the same LLM, but it seems apparent that Reliance is seeing the potential in the AI space. And for good reason — ChatGPT maker OpenAI is now worth nearly $100 billion within 7 years of being founded, nearly half of the $200 billion market cap that Reliance has managed to achieve over the last six decades. Also, the world might be only scratching the surface of the potential of AI models right now, and there’s no telling how sophisticated — and ubiquitous — these models might become in the coming years.
And Reliance is perhaps well placed to create a high-quality LLM of its own. Training LLMs can take large sums of money, and no company in India has pockets as deep as Reliance’s. Reliance also seems to have the right connections — it’s partnered with the top institute in India in IIT Bombay, and the top chipmaker globally in NVIDIA to power its AI ambitions. And if BharatGPT can end up seeing the light of day, it’ll not only reduce India’s reliance on foreign LLMs for tasks that AI might take up in the future, but also send a signal about India’s tech prowess to the world at large.