Straight to the point, India needs its own LLM

People are not talking about this; artificial intelligence feels like a party to which we’ve only been invited as guests. The real work, the creation of the most powerful AI models, happens behind closed doors in other countries. These models are built by giant companies with almost unlimited money and massive banks of computer chips called GPUs that most of us can only dream of accessing.

This creates a huge problem for India. Because we are not building this technology ourselves, we are forced to be consumers. We use AI tools that were designed for other places. They don’t truly understand the richness of our many languages, the nuances of our culture, or the specific challenges of our society. Think about the biggest issues we face improving education in remote villages, delivering better healthcare to everyone, and helping farmers get better yields. The AI from abroad is a generic, one-size-fits-all solution that often doesn’t fit our problems at all.

This situation has created two big gaps for us.

First, there’s the Capability Gap. The door to truly building and understanding AI is locked, and the key is expensive hardware. This means our brightest students and our most innovative small companies are locked out. They can use AI, but they can’t create it. This is why India doesn’t have its own strong, homegrown AI models. We have the talent, but our talent doesn’t have the tools.

Second, there’s a Knowledge Gap. Many people think AI is just about using a chatbot or plugging an API into an app. But that’s just the very surface. Truly building AI requires a deeper understanding. How do you prepare good data? How do you train a model efficiently? How do you know if it’s working correctly? When we only learn how to use the end product, we miss the core skills needed to build our own. We become forever dependent on others, unable to create technology that is truly made for India’s needs.

So, what’s the solution? How do we go from being users to being makers?

We need a clear and practical vision. We cannot and should not try to copy the West by trying to build a single, gigantic AI model. That’s their game, and they have a huge head start. Our path is different. Our vision is to focus on smart efficiency, local data, and open sharing.

Instead of chasing raw power, we must become masters of efficiency. We need to learn and use techniques that allow powerful AI to run on simpler, cheaper computers. Techniques with names like distillation (making a small, smart model learn from a big, clumsy one), quantisation (shrinking a model down to size), and pruning (cutting away unnecessary parts). We can build many small, specialised models that each do one job perfectly for India, instead of one giant model that does everything poorly for us.

Most importantly, our fuel must be our own data. We need to collect, clean, and share high-quality datasets in all Indian languages. This data is the foundation. Without it, any AI will be blind to India. And we must do all of this out in the open, sharing every success and every failure so that everyone can learn together.

This isn’t about waiting for a big government program or a giant company to step in. This is about what we can do right now, ourselves.

Here is how we start, with simple, clear steps:

Learning Groups: Find a few friends or colleagues online. Pick a topic to learn about each week and hold each other accountable.
Run Tiny Experiments: Your goal isn’t to change the world in a day. It’s to learn. Try to fine-tune a small AI model on a specific task. Then, share the code online for others to see.
Organize Data Sprints: Pick one Indian language. Gather a group for two weeks to find, clean, and organize data for that language. Publish it for free for anyone to use.
Share Computer Power: Ask university labs or local startups if they can donate a little of their unused computing power for community experiments. Every little bit helps.
Document Everything: Write short, simple reports on what you did. What worked? What didn’t? How much did it cost? This shared knowledge is how we all move forward together.

This is a grassroots movement. It requires patience and hard work. But by focusing on efficiency and community, we can achieve anything if we start together. By focusing on the AI industry and its core technology, we could build our own LLM (large language model). We need a clear vision. Nothing is impossible; for every problem, there is also a solution we just need to find it. We can start building AI that solves our problems and reflects our values. It’s time for India to stop just using AI. It’s time for India to start building it.