How India is rewriting the AI playbook

What I find it really fascinating is a. completely different model, which is. what India is doing. And in fact we had a piece about this in. the world ahead, the annual that I edit. And it was written by Nandan Nandan. Nilekani. Here he is. And. he is the co-founder of Infosys and now. the chairman of Infosys. But. I think he's also known as the sort of. architect of the Indian tech stack. And. the Indian model is to see AI as part of. the government tech stack, which is. quite a I can't think of anyone else.

who's doing. doing it that way, can you? No, so this. is taking the AI models that are built. by your Indian companies and embedding. them inside things like I mean I guess. the big ID system is one, right? Well, exactly. I mean basically that they. started off so he he's best known I. think outside Infosys for building. Aadhaar, which is the the ID system. So. it's a biometric ID system and the. government then uses it to you know pay. benefits and all this kind of stuff. And then they built a payment system on. top of that. And the model in both cases.

has been that both the government and. private companies can then use those. rails to build other services. And. that's that's been very successful. And. it means that if you're you know an. Indian. e-commerce company, you can use the the. payment system that's already there. You. don't have to you know reinvent it and. reinvent the wheel. And essentially. the piece that that Mr. Nilekani wrote. for us, he explains that that's what. they're doing with AI. They are they are. building and this is something that. is a I think a collaboration between. industry and academia. So IIT is.

involved in this. But things like I mean. there are hundreds of languages in. India, but there are 22 official. languages. And so you really want the. all the government services to be able. to deal with all of those languages. And so they have built a model that. they've gone out and gathered loads and. loads of data. and then they've built a model that can. translate really well between those. those different languages. And so you. can then use all these IT services and. whatever a government services in. whatever language you choose. Another. thing they've done is providing.

agricultural advice to farmers. And. they're doing that with. you know farmers in in some parts of. India are not going to have smartphones. They're not going to have internet. access, but they are going to have. feature phones and they can make phone. calls. And so you can give them. essentially agricultural advice via. those phones using a text-to-speech. system. And so the government can issue. central agricultural advice or it can. have policies. But then you can have a. chatbot that essentially. gives tells people that information in. different languages. And again, the idea. is that this sort of these AI rails that.

help with things like translation can. then be used by both the public and the. private sector. So they really are. seeing this as a stack. And one of the. ambitions that India has with this. government tech stack is to export that. you know to build essentially to you. know send Infosys and other companies in. to build. versions of this in other countries that. are building their own digital. government infrastructure. I don't know. how far they've got with that, but. that's certainly an aspiration. But you. know on that model you would then have. well let's we'll make you a as part of. that stack a a model that can handle you.

know whatever languages you need. to be handled by the by the stack in. that country. Which does mean that the. the real interesting thing for the. Indian model is less that their system. speaks 22 languages and more whether or. not they've got a repeatable process. that can do that for an arbitrary other. country. >> I suppose so. I think that's probably the easier bit. I. The other point he makes in this in this. piece is. you know it's it's about what AI means. for India, but also what India means for. AI. Because he says you know if you can.

if you can show that AI has value to. farmers in in you know remote rural. areas, then you know then I mean it's. much easier to show that it has value to. white collar workers. But he's saying if. we're really trying to suggest that AI. has these sort of broad societal. transformational powers, then the. ultimate test of that is is in India. And he's saying to you know he issued he. sent an invitation to companies do come. to India and try try running your your. stuff here. And they're also doing stuff. with with AI in the education system as. well. So sort of personalized tutoring. And and again, they they consider that.

to be part of this sort of broader These. are government services and this is. government infrastructure just like you. know health care or. you know other other bits of. you know government infrastructure. So. they do see AI as sort of potentially. playing a transformative role in that. And I don't I don't think I see any. other countries doing something on this. scale. No, well presumably a big part of. it is this idea of low resource. languages, right? And so India having. somewhat uniquely like 22 languages each. of which have millions and millions of.

speakers, but most of which have very. few speakers outside of India, it would. be very easy for the Indian government. to just turn around and go well look, English is one of the languages that we. have in the national language. There's. loads of really good English language AI. models. We'll get them in very cheaply. And so I guess there must be an element. of social responsibility of you've got. to serve all of these languages equally. and so that. >> Also I think it's I think it's it fits. in with the kind of you know the way. India brands itself on the world stage.