Thanks Emar and it's really great to be here at this Indiaspora event in Bangalore and to be speaking to all of you. It's really a wonderful two or three days that you're having here. So I'll talk briefly about the work we have done in the past, which is what we call digital public infrastructure, and then of course how it leads to AI deployment. The idea of DPI was that if we can create open, interoperable public rails at population scale, then we can create a much more economically progressive and inclusive society. We can enable people to formalize and so on. So those are the kind of ideas.
It began with Aadhaar, which I actually led the project, joining the government and working there for five years. Today, Aadhaar has 1.4 billion people who have an Aadhaar, and it's used about 80 million times a day. Somewhere, somebody is doing an online authentication with Aadhaar. So there was a fundamental digital public infrastructure and a foundational infrastructure which was required because at that time, many people in India did not have any ID at all. Because when they were born, they did not have birth certificates, and not having an ID became an impediment for their life. They couldn't go anywhere, board a train, answer the cops, or get a job. So Aadhaar was essentially that foundational infrastructure which is used everywhere.
Now, on top of that, we built the ability to link Aadhaar to bank accounts, and we enabled instant bank account opening with something called Aadhaar eKYC, or electronic know your customer. That led to the world's largest financial inclusion program, the Pradhan Mantri Jan Dhan Yojana, which essentially got more people into the banking system in nine years what would normally have taken 45 years. So we were able to accelerate financial inclusion, and a few hundred million people entered the banking system. That laid the foundation for the world's largest direct benefit transfer, where money was directly transferred into people's accounts. That changed the whole game of how money reaches people in a clean way, without any fraud or corruption, directly entering the bank account.
The same channel is used for remittances when people come to the city and want to send money back to their family. And then around that, we built a thing called the Aadhaar Enabled Payment System, which allowed people to withdraw money with a small business correspondent in a village. So the whole thing transformed the way money was put into bank accounts and how money was taken out of bank accounts. This whole other system was done by an organization in the government called UIDAI, of which I was the chairman.
Now, around the same time that UIDAI started in 2009, another organization was set up in India called NPCI, the National Payment Corporation of India, which is set up by the Reserve Bank to look at payments. We worked together on doing all this Aadhaar, bank accounts, and KYC. So I was also an adviser to NPCI, and we said if everybody has a bank account and everybody is able to withdraw money from the bank account, why should they even withdraw the money at all? Why can't they just pay from the mobile phone to somebody else? That led to the birth of UPI. UPI was an idea conceived in 2013, implemented in 2016, and launched in April of 2016, a banner year for India. By October of 2016, it had reached about 100,000 transactions a month.
But then two things happened. One was the demonetization, which led to a huge push on digital payments. And then of course there was the whole COVID situation, where people wanted to give money to others without touching them or the money, and therefore digital payments became very popular. So today, UPI, which is run by this amazing company called NPCI, is the world's largest payment system. It's bigger than all of the others put together. It does about 21 billion transactions a month and has about 500 million users, and has transformed the way that we do things in this country.
So these were two big things that enabled this digital transformation. And that was then further—we said how do we—and a lot of other things were built. I mean, we're talking about DigiLocker. Many of you are using DigiLocker; millions of people are using DigiLocker. DigiLocker is again built on the same foundations of digital infrastructure, Aadhaar, and so on. We have built India's single access toll system called FASTag, which does again billions of transactions and has dramatically changed the efficiency for roads, a single system across the country.
And then, leveraging the data which is there in people's records of their expenditure and so on, they can now use that data to get access to credit. So we have built a whole system called the Account Aggregator ecosystem, which allows people to use the data to get access to credit. So what that means is somebody who has nothing can get an Aadhaar. With the Aadhaar, they can get a bank account, they can get a mobile connection, they can start a small shop, they can start selling, they can take payments in UPI, and then they can use the history of UPI payments to get access to credit. So this is really a whole sequence of things, all at population scale. The idea is that if a billion people have access to all these things, that's how the country is going to be inclusive.
So this has been a very fundamental change that has happened over the last several years. And this idea of DPI has gone global. We have a number of institutions today which are taking the DPI idea global, and our goal is to get to 50 countries in five years. We are already at about 39 countries which are using some kind of the DPI that we have built here, and we think that this will be a big thing. Just a few months back, we had a global DPI summit in Cape Town where we had 1,200 delegates from 109 countries. So it's a global movement of implementing DPI.
Now, what we learned from DPI is that it's all about diffusion. It's not about technology. Technology is there. How do you make it work? How do you change a society? How do you get people to adopt it? How do you make it frugal? How do you make it high volume, low cost? How do you get a billion people to use it? So those are the lessons that we learned in our DPI journey. And when the whole AI thing came along—and I know a lot of people have spoken about AI—but it's all again about diffusion. How to make it work for people, otherwise it doesn't mean anything. How does it actually make a difference to people's lives?
In that area, now we are working in three areas. One is in language, because India has so many languages—22 official languages, a few hundred dialects—and that is actually an impediment. Because if somebody is not able to use English and they only know their native language, they are not able to interact with a computer. So if you can actually crack language, you crack inclusion of a different kind. A few years back, I funded a group at IIT Madras, AI4Bharat. Sujatha is here; her husband was part of that. And the idea was that if we can make Indian language data available widely, then many people can build on top of that. And now that has actually been a foundation which also laid the thing for Bhashini. And now Vivek and Pratyush have set up Sarvam, which has built India's first 100 billion parameter model and so on.
So a lot of stuff on language, because if you can get language to work, and if you can get language to work in voice—where you can speak to the computer in your dialect or your mixture of Hindi, Tamil, and English or whatever you do—then suddenly you've made computers accessible to a billion people. And that's the idea of using language in AI, simply to make it more accessible to everybody. Voice AI. So that's one big initiative we have.
Another initiative is in education. We have now four states trying to improve basic literacy and numeracy using AI, because it's not about for people who are already literate and numerate—they can use any smartphone app. But the big challenge is those who are not even literate or numerate, and you have to get them into society by fixing basic literacy and numeracy. And we have a program called ASER, which is running in four states and which is actually making a difference to the learning of children. So that's in education.
And the big push we have is in agriculture, because India is again a country with a few hundred million farmers. And if you can make farmers more effective—if they can earn better money, they can plant newer crops, they can learn about sustainable planting and so on—then that's a big deal. So the agriculture initiative is there. We're working—the first time we worked was with the state of Maharashtra, which launched something called MahaVistar, which is an agricultural application which is not just using an LLM but actually using data from different sources: the pricing data from the markets, the warehousing data from the warehouses, the weather data from meteorologies, and so on. And bringing all that together at the fingertip of the farmer in Marathi and so on.
So this is actually working for millions of farmers today live in Maharashtra, and that's very successful. And we're also bringing it to much lesser-known languages. For example, the Bhil tribe, which are in Nandurbar and so on in Maharashtra, speak Bhili, and we have now brought a Bhili version of this technology. So people can speak in Bhili and get agricultural knowledge. So there's an example of how we're using AI to actually make a difference to a farmer in his own mother tongue and get data and answers relevant to their business, their work.
Now, this took nine months at the EkStep Foundation. It took us nine months to do this. And then there was a request from Ethiopia to implement something similar, and that was done in three months. So again, the lesson here is that it's not about technology. Once you have the playbook for doing these things, then you can do it much more rapidly. And then, in January, I had a meeting with the Prime Minister where he said, 'Why can't we do this for cows? Because cows can't tell you when they're not feeling well.' And he put us on to Amul. And Amul launched an app called SarahBen, which was launched in three weeks. So in three weeks, they rolled it out using the know-how we had built.
And Amul has 2.6 million farmers. It has 22 million cows and cattle. It has 1,400 vets. So there's simply not enough vets to deal with the challenges of so many cows. And now with SarahBen, the farmers are able to get information about cows, the lactation, the drop in production, if they have any health issues, and so on. So this again is a direct application of AI to the use of the common man, to the farmers, and so on. And this is built in three weeks. So once again, I think learning from the DPI experience, we learned how to build AI DPI at scale.
Now we're bringing the same learning of scale AI implementation, trusted implementation, including institutions, and then taking it at scale again. And we are doing that now in education, in agriculture, and so on. And we think this is the way to do it, because AI is only as good as it makes a difference to the benefit of people. Otherwise, it's fine. It's all technology. You can have so many models and so on, but it really is about usage. And India will therefore, in our view, be the place where AI actually gets leveraged. And we will do it at scale for a billion people. We'll do it in a way that it's frugal and cheap, so that the costs are lower—cost of building models, cost of inferencing, cost of reasoning, everything will come down. It has to be affordable to everybody in India.
And we will do it with institutions, because to make AI work at scale is not just about technology. Technology is one-third of it. It's about institutions, about systems, about trusted data. It's about designing in a way that you don't get locked in, and so on. So that is the kind of thing which is happening. And so now we have launched a mission along with many global partners. We launched it at the AI summit called 100 AI Diffusion Pathways. So we're saying if we can work collectively around the world and build a hundred ways of actually making AI work in different sectors, and then sort of create a playbook of how that can be done, and then make that playbook available to others, that we think is the fastest way of getting AI adoption in the real world.
And so the AI Diffusion Pathways is something we're working with companies, with international institutions like the UNDP. Many people are all part of this coalition. It's the grand tent. We're all coming together to create these 100 diffusion pathways for AI by 2030. The idea is to do it also in a time-bound manner by 2030. And that's something we welcome all of you also to think about and to participate in, because if we can do that, then we can really make it useful. And my real worry is that if we don't do this, AI will be a race to the bottom. It'll be a race to more and more AI that people are actually not benefiting from, people are hurting from. And you can also already see in the US there's a lot of antipathy towards AI.
And we have to also at the same time build really useful cases of AI which make a difference to the common man. And that's what we are trying to do here. We are using our cumulative experience of building at scale, actually building it for a billion people, which has made a difference in people's lives, and bringing the same approach, the same diffusion techniques, the same trusted approach, the same attention to high-volume, low-cost transactions, and actually making a difference. So this is the plan we have, and I think we'd like to welcome all of you to join us in this journey of applying AI in a way that it matters for people. Thank you very much.
