The Exchanges

Q And you moved from Infosys, you retired first, and then Aadhaar came your way?

A No, what happened was in 2009, in 2008, I wrote a book called Imagining India. So sort of projecting how India could be in, in the framework of ideas. Why ideas matter and stuff like that. And among the ideas in that book was, you know, digital ID and so on. So, at the same time, the government was planning a digital ID program, and they had got cabinet approval for that project. And, uh, I was, uh, I had a chance to take up that role. So, and then my Infosys co-founders were very kind and said, yeah, you can go and do some national service. So that's how I ended up in Delhi in July of 2009 to build India's digital program, ID program.

AI assessment note: “No, what happened was in 2009, in 2008, I wrote a book”

Q So when you say beckon is a protocol, you mean it's a bunch of rules? Yeah. Of the language Beckin, in a way.

A No, it's a bunch of rules for transactional stuff, for people to buy and sell things. And it can be used in commerce. So, ONDC uses Beckin as an underlying protocol. And ONDC today does twelve billion transactions a month. It's used in mobility with Namayatri, which does a hundred and 50,000 transactions a day, where people are ordering auto rickshaws. Again, You know, it's a different model than an aggregator because you pay directly to the taxi auto guy. And it's being used in energy. We are building something called unified energy interface. People can buy and sell electricity. So any, anywhere that you have to buy and sell things between different people in a consistent way, Beacon is the way to do it. Beacon is a way to do it.

AI assessment note: “No, it's a bunch of rules for transactional stuff, for people to buy and sell”

Q Is that because now each person had a unique ID per se?

A No, it could have multiple. The UPI was actually more sophisticated in the sense that you could have a UPI ID. So, uh, it could, it could connect to a, it's called a, you know, virtual ID. It can connect to a bank account. Uh, you could use a account number. You could use a mobile number as the address. You could use a virtual name as the address. So, Nikhil at some access bank or something. So we, all these possibilities are there. So it just made it Easy to use. And real time. So, and then two big things happened, right? So demonetization gave the first Boost to digital payments because people had to make payments and, you know, cash was not there. And then the pandemic also, people didn't want to hand over money. They were rather, you know, point to a QR code and send money. So both these two tailwinds also drove the, and of course, the ease of use and the amazing benefits people got from it.

AI assessment note: “No, it could have multiple. The UPI was actually more sophisticated”

Q Post the success of Infosys. What drove you on this journey, and is that why population scale matters?

A No, because at Infosys, I had done a set of things. But when I got a chance to do Aadhaar, I genuinely believed that digital technology used properly can make a big difference to lives of people. So I had that belief. And it's there in my book, two, two and eight. So that, and then I got a chance to practice those beliefs, right? I mean, you know, it's not, it's all of, we can have an idea, but somebody said, okay, fine, great idea. Now go and do it, right? So it's a very different thing. So I got a chance to get it done. And then I realized that actually we have to do a number of things. So the last 15 years, we have built different, uh, parts of this infrastructure. And I think it's made a difference because it's made it More inclusive. Everybody is in the system. It's helped to formalize. It's driving economic growth. You know, all the, all of that is happening because we have this way of thinking about digital infrastructure.

AI assessment note: “I genuinely believed that digital technology used properly can make a big difference”

Q I've heard you speak of AI in a manner where, uh, you firmly believe that large language models and building a model is going to be commoditized and nobody should spend energy on that. But you speak a lot for the Indian language and the nuance of it and building rappers on top of these models to cater to that. Uh, can you elaborate on that? What do you think?

A Yeah. So there's a couple of two, three different points here. One is large language models in my view will become a commodity. They are already becoming a commodity. The drivers for the Western guys to spend billions of dollars is essentially because they're in a competitive land grab situation. So they all have to invest because not to invest is dangerous for them. So they collectively are investing two hundred billion dollars a year, buying all, you know, chips, putting data centers, all that. But they have to do it because if they don't do it, they'll fall behind. So that's, they have a compulsion to do it. But the good news for us is all that is going to lower the cost and access to AI over time. It's already happening. And then open source AI is coming. People like Meta have just launched Lama three. And so, and they're going to give it away free. So I think this building a LLM is a mug's game for us. It's great for somebody there, but not for us. Our job is to leverage all this technology in a very low cost way to deliver value to our people.

AI assessment note: “Our job is to leverage all this technology in a very low cost way”

Q Can't that be scammed? Like whenever I looked at the crypto world, if I were to think regulation is everybody viewing transactions verified either by proof of work or proof of stake or even proof of time. If the chain in itself is not large enough, can't it be easy, easily scammed or conned into going in a certain direction?

A I don't know enough to answer that, but They took a technology, but on top of that, they layered currency, and they layered ideology. So they said, we'll create our own currency. And they created some 15,000, everybody creates some coins, ICOs, and all that. And they said, we don't want authority, we don't trust the state, so we'll do it through software. So that got them into all their issues. So what we have done is we said, look, we should separate all this. There is some great technology here, which we can use for high volume, low cost transactions. And if we can somehow bring that high volume, low cost transaction capability to the current world, By a common way of doing it, then suddenly we can take our mainstream world and turbocharge it. That's exactly what Finternet does. So Finternet allows you to essentially airdrop like a modern engine into the existing system.

AI assessment note: “I don't know enough to answer that, but”