Good morning, everyone. This is Chaitanya.
As Ranjit mentioned, the idea here is to explain what we are trying to do and to inspire you. That's the task that's been given to me, so let me see what I can do.
First, how many of you know what we're trying to do here? Has somebody explained anything to you so far? Or are you just here for an internship? What is Vishwam? Does anybody know?
(Audience member: Trying to develop a Telugu LLM)
Okay, so I understand you're here to do an internship. I'll just explain the concept of Vishwam a little, then I'll explain why we are doing it and what we expect from you. I'll just take maybe 15-20 minutes.
Why Vishwam?
The idea of Vishwam started last year when we realized that AI is going to change the whole world. And we realized that we—meaning India in particular—were getting left behind. We know that the best AIs out there, like ChatGPT and Claude, are from the US. We also have models like DeepSeek from China. Unfortunately, we don't have anything from India.
So the premise was: what do we need to build our own LLM, our own AI? But before that, the first question is, should we even build our own? Even prominent people like Nandan Nilekani have said that we don't need to build our own LLMs; we should just build applications on top of existing ones.
At Vishwam, we don't agree. We believe that we have to build our own AI, our own LLMs.
A Lesson from History
The example I always give is the Industrial Revolution. We all learned about it in social studies, right? What was the Industrial Revolution? What was the one-line answer? It was the steam engine(audience member). The fundamental premise was that anything a human muscle could do, we could replace with a machine. People found that you can harness power through steam. Whether it was breaking a rock, weaving clothes, or pushing something, we used human muscle. They realized you could replace every human physical action with a machine.
And who found it first? The British. Because they were the first to find it and we didn't adapt at that time, they were able to colonize us. They had the machines, and we had to either buy the machines from them or give them our raw materials, which they would use their machines on and send back to us as finished products. We all know this history.
We are at a similar stage now. We've realized that AI is at a level where it can replace the human brain. Right now, what is the muscle that we, as a country, use? It’s the brain. That's the single most important muscle we use for IT outsourcing. We learn computer science—C, Java, or whatever programming language—and we are ready to take on work from outside. Our brain is the muscle, and we deliver the finished product. That's how the whole of India has grown in the last few decades.
That is at risk now because of AI. Just like the steam engine replaced human muscle, if AI can replace the brain, what do we do? Especially if we don't have our own AI? We will have to depend on others. Even now, if you want an airplane, you have to get an Airbus or a Boeing. We don't have our own. Same with many other forms of manufacturing. The same thing will happen if we don't have our own AI. That is the first reason we need to build it. There is no other option.
Why It Matters More to You
Unfortunately for you, you are the generation on which this responsibility falls. We have been lucky. For the last 25-30 years, from 1995 to 2025, my generation could learn a simple skill like building a web page with HTML and JavaScript and get paid 20-30 lakhs a year. We were a lucky generation; we made a lot of money. You don't have that option. Now, you have to stand up, and we are here to walk you through it. We have to be ready for a world that will not pay you 30 lakhs to build a simple web page. So, we need to build our own AIs.
Why Telugu?
The second most important reason is that AI today works mainly for English. In this world, a person who knows English is already five steps ahead of a person who doesn't. We all know this. Some of us are really good programmers, but we are not fluent in English, and that's a problem. We see people who speak good English get placed, even if they don’t know programming as well. Communication is critical.
Now, imagine that same gap, but amplified. A person who knows English also has access to AI. How do you communicate with AI? You have to ask it in English. You can try it in Telugu or Hindi, and it works at a certain level, but there is no way it works at the level of English. You've all experienced it. And if you can't talk to the AI, it means the person who knows English is now not five steps ahead, but fifteen.
What else will happen? People will stop learning our own languages, like Telugu. Everyone says AI is already working superbly, that we are at AGI (Artificial General Intelligence). But if AI really worked, I shouldn't need to learn English. I should be able to talk in Telugu. If you have to write an email to your supervisor, why should you struggle with English? You should be able to say it in Telugu, and the AI should draft it perfectly. It doesn't work that way yet. We are not there.
That is the world we are looking at—a world where a person doesn't need to translate their ideas from their mother tongue to a new language just to solve problems. Right now, to solve any problem in your career, you have to perform an extra step: you convert the problem statement from English into your mother tongue to understand it, formulate the answer in your mother tongue, and then translate it back to English. We want to get rid of this, and to do that, we need Telugu AI systems.
What is missing?
So, how do we build it? Luckily for us, we know the systems work. The algorithms, the hardware, the GPUs—it's all there. The only thing we don't have is data. We don't have data in Telugu like the English language has, not even close. The internet ensured that English language data is available everywhere. AI will only work for things that are on the internet. That data is not available in Telugu.
This has been the problem we—Praveen, Ramesh, the HODs at IIIT, all of us—have been discussing for the last year. How do we solve this data problem? We can’t just create it overnight. So we started thinking, what does India have? India has one thing in abundance: youth. We are the country with the highest number of young people. How do we harness that power—your power—to build this data?
That discussion led to this program. We know you need something in return. So we said, "Okay, we will train you on AI. We will get the best content, the best teachers, and we will give you this training for free. In return, you help us collect the data—your own data—which will be used to build the system."
Nowhere else in the world is this being done. Only here in Telangana are we asking people to crowdsource data to build our technology and save our culture. What are we asking you to do? Take a microphone—or better yet, just use your phone. Every phone has a recording app. You guys talk all the time! During those conversations, just collect some data. Go talk to your grandparents. Go talk to your parents—in Telugu, not English. Collect as much data as you can.
I can speak much more fluently in English than in Telugu. See? That's why I'm communicating in English with you. We have to stop this, otherwise our language will fade away. We need you guys to collect this data.
The most important component-You.
Now you understand what we're trying to do here, the impact of it, and how important you are. You are everything. All I can do is come and give this "gyan"; I can't do anything else. Every single one of you here is important. Every 10 minutes of data, 5 minutes, even 1 minute of data is important. Everything is a contribution. Just make sure you collect as much as you can and have fun while doing it. Go back to your college, have some jam sessions, talk to your friends, talk to your parents, and get that data.
That’s the data collection piece. The second piece is the internship, where we want to train you. And here again, you guys are the most important part. Content is not a problem. MIT and Stanford have their courses available online for free. In this day and age, content is not the problem; it's you who have to decide to learn.
Let me tell you one small thing that helped me. I was just like anybody else. I followed the typical Telugu kid's process: you either become an engineer or a doctor. We choose engineering. Then GRE, then to the US on an F1 visa, then H1B, then a green card, and finally, citizenship. This is the process, we all know it.
But the difference was, when I went there to do my masters, I worked my ass off. I was studying so hard to get a 4.0 GPA because if I didn't, my scholarship would be gone. I was the same student who, back here, when my teacher gave me an assignment, would copy it from my friend. We all know it, that's okay. But there, I was so afraid of copying because I knew they would kick me out and I'd lose my scholarship. That single fear made me do my assignments myself. And that single act of doing my own assignments made me learn so much. They don't teach you anything in the US; they tell you what you have to learn and give you an assignment. You have to figure it out.
The point I'm trying to make is that I was the same guy. You are the same person who, if you go to the US, will study for 8 hours a day while working a job at a gas station. It is in you. You just have to decide to do it.
I'm telling you this because, in this internship, we have given you the course and the content. We will give you the assignments. I don't want only two people in the class to do them. Everybody, please spend time on it. It’s one month. For one month, don't do anything else. Concentrate. It's for your career.
We are spending our time here. Every one of us is a volunteer. This is coming to you for free. Nowhere else in the world is this happening. Give respect to our time, and give respect to your time. I just hope there are no big movies coming out for one month! Just make sure you concentrate on this internship. Do the exercises. Get stuck. Talk to your team leads. Don't hesitate. If a team lead doesn't answer, go higher up. We will provide all the help you need because we are also committed for this one month.
So, two things: data collection, and making sure you learn. I want Hyderabad and Telangana to have the best AI engineers going forward. We can't avoid this wave. If we lose this, we will always be labor.
Let's be leaders, not laborers.
Thank you, everyone.