How should India’s AI model look like?

Generative Artificial Intelligence (GenAI), led by ChatGPT, is creating waves all over the world. Its astonishing capabilities to generate conversations, art and videos led ChatGPT to race to 100 million subscribers in two months. The Large Language Models (LLMs) and transformers that power GenAI are ground-breaking technologies that promises to be as disruptive as the internet or the smartphone. Google CEO Sundar Pichai likens AI to fire, Satya Nadella of Microsoft heralds it as a platform shift, and Goldman Sachs believes GenAI will add $7 trillion in extra annual global Gross Domestic Product over the next decade. However, this excitement is accompanied by some unease around the loss of jobs and human agency, and the looming threat of an AI superintelligence. There are also worries about bias, environmental damage, plagiarism, and a threat to democracy itself.

Predictably, most GenAI action is happening in two countries, the United States (US) and China, and both have vastly divergent structures and worldview. In the US, Big Tech is leading the charge, as it has through all the technology waves of the internet, search and social networking. The China way is different — with Chinese companies, working with the government, creating their own GenAI models built with informational and contextual safeguards. Other large geographies are carving out their own space, albeit in a more limited way: The European Union is focusing on regulation to make GenAI ethical and responsible, and the United Kingdom aims to lead the world in global AI governance.

In this melee, India seems left out. There has been a lot of talk, and some recent announcements, of creating its own LLMs, though OpenAI CEO Sam Altman actively discouraged it. We have absolutely no doubt that both the government and corporate India have the capability to build a local LLM – fine-tuned with Indian data for Indian languages, context, and unique use cases. We believe, however, that it is more important to consider the objective and aim for such an exercise. Should India follow the western capitalist method, or the Chinese State-controlled one?

We assert that there is a third path, one that India has shown the world in recent years. Our proposition, however audacious it may seem, is that India should consider building Generative AI as a digital public good – we call this JanAI or GenAI for the people.

The amazing success of the Indian Digital Public Infrastructure (DPI) story is well known. Riding on IndiaStack, this digitisation at population scale has led to 1.4 billion Indians with a digital biometric identity with Aadhaar, simplified payments at scale, with UPI leading to almost half the world’s digital payments, the world’s largest vaccination campaign, and many other achievements. Services built on top of this stack have considerably eased health care, logistics, e-commerce, government subsidies and have led to inclusive societal growth across the country. Innumerable startups have leveraged open APIs of the stack to create innovative and at-scale services for Indian citizens. IndiaStack is already growing global, with countries such as Singapore, France, United Arab Emirates, and others signing up. In fact, the global spread of DPI and IndiaStack is a strong idea that India is championing through its leadership of the G20, including at the leaders’ summit this week.

The true power of the stack has been realised by offering it as a digital public good (much like clean air, defence or law enforcement), so that it reaches each citizen; and making it open enables large companies and startups to build value and wealth on top of it. We believe that GenAI should be treated similarly. India could build its own BharatLLM: Trained on the rich data that IndiaStack generates, and finetuned for Indian languages and context to solve India-specific use cases and problems. JanAI could be a set of LLMs built as another layer of IndiaStack, where the key is that it will be offered as a public service. Thus, it will bridge the inevitable digital divide, and provide benefits for the entire population, much like Aadhaar and UPI have done. This will also give India the opportunity to build guardrails and safeguards around privacy, bias, and other ethical AI concerns, using Indian notions of collective and societal privacy and trust which are sometimes quite different from the western concept of individuality-oriented privacy.

We believe that this could happen through a tripartite partnership between a proactive government, our world-leading IT industry, and some leading technical institutions like the IITs. Indian companies could then spin out their own specific finetuned LLMs from BharatLLM; startups could leverage this through open APIs along with ChatGPT and others to build innovative India-specific products; and millions of individual creators could use its awesome generative powers to build content- and creativity-led businesses. This is how India could combine the two biggest movements in digital technology — generative AI and digital public goods — into a uniquely Indian third way to create JanAI out of GenAI, and, once more, serve as a GenAI for All model for the world.

Jaspreet Bindra is managing director and founder, Tech Whisperer Limited UK, and Sudhir Tiwari is Global Head of the Digital Engineering Centre at Thoughtworks India. The views expressed are personal.