Building an Inclusive AI Future – 81st Session of the UN General Assembly

Hello, everyone. Welcome back to the SDG Media Zone. We are live from UN Headquarters during the 81st session of the General Assembly. Our next session is about to start. I'm going to hand it over to the moderator, Teresa Mondria Terol from NPR. Thank you so much.

Thank you very much. And thank you very much, everyone, for attending this session on building an inclusive AI future. I am joined here today by Alexander de Kruif, the administrator of the UNDP, Kate Calo, founder and CEO of Amini, an AI startup building data infrastructure in Africa and the broader global South, and Big Ocean, the world's first hard of hearing K-pop band. Today, as AI reshapes economies, a real test is whether its benefits reach the communities most at risk of being left behind. The session brings together three perspectives on building inclusive AI. The UNDP's role as an ecosystem builder, expanding compute access, and enabling policies where AI is least available. An African entrepreneur whose data infrastructure is opening AI in the global south. and the world's first hard of hearing K-pop group, whose use of AI in performances shows inclusion in practice. Administrator de Cueva, I'm going to start with you. As a development organization that works across 170 countries and territories, how does the UNDP see the impact of AI on international development? And what is the UNDP doing to support countries to address these opportunities and challenges alike?

Well, first of all, for us, AI, artificial intelligence is an incredible accelerator of development and the capacity to upscale. to bring solutions at places where we couldn't bring it. It just helps us to jump some of the traditional obstacles that you have in development. And I mean, quite a substantive share of the projects that we do today, there are elements of AI or on our operational side or in bringing the solutions that we need to bring. The challenge that we see is that the speed of adoption of AI is much higher than the speed of countries building governance to do so. That gap is widening. Now, there is a discussion that has been fueled over the last weeks, or should we slow down AI or not? We actually think that we should speed up the speed of governance. I think it's going to be hard to slow down the speed of technology playing its role. That doesn't mean that we should just let everything happen the way it is happening now. But the big gap that we see is that countries don't have the right capabilities to govern AI. Speeding that up is one of the other domains on which we work.

Thank you. Kate, as an AI entrepreneur, what do you believe needs to be in place for local innovators to build AI that serves their own communities?

Thank you for having me, and I think it's a great question. You know, I spent the last 15 years building AI for big tech here in the US, lastly at NVIDIA, and then realizing that there was a fracture that was being created. On one side, the technology was making so much advancements, but on the other side, a lot of the global South was still being left behind. Very recently, a lot of companies have been pitching AI diffusion, increasing access to the models, But for me, the real work that needs to happen is we need to start treating infrastructure and capability as a responsibility. If we're only focusing on the diffusion and access of models, but not focusing on building the infrastructure that will allow developers to build solutions for their own communities and their own countries, we won't make any progress. Because what will happen is we're running through a world that will look very similar to one part of the world, but adapt solutions that are actually not localized to the local context. And I think we need to reverse this curve, and that only happens if we build the infrastructure and we train people locally to be able to build for themselves.

Thank you for such an interesting approach. Big Ocean, in your music, you use AI as an assistive tool to perform and connect with your audiences. What does AI inclusion look like from your side? And what can technologists and policymakers learn from that about designing AI for people in an inclusive way?

OK. Firstly, thank you for having us. And we are the world's first K-pop AI group. And we sing in three different styles-- Korean style language, American style language, international style. Our slogan is "Feel music, feel beauty." Thank you so much.

So to answer your question, thank you. We use AI in three main ways. All three came from things we really needed. First is vocal training. So AI shows us the pitch we cannot hear. So we can correct it ourselves. And second, it's rhythm. We use a metronome. We use a metronome. So that place, so we can't catch the rhythm during this system. And so, yes, third. When we make a song, the voice, we use AI voice technology. So the system learned our voices and the guide vocal, and then it changes the guide vocal into our own voice. So the final voice you hear is based on our voices.

And not only that, at the technology project, we found the tools or make them our salvage because we needed them without the tools coming signal would not have been possible for us. And with them, it is possible. And for us, this is for inclusion through AI. Looks like.

And something surprised us. A lot of people want to use this tool too. We made one metronome system for deaf people. Oh, sorry, I'm so nervous. OK. Yeah, we made one metronome system for deaf people. But we learned that it can also help lot of people who are learning music for the first time. So. So seeing and feeling the beat have them too. So our company is looking at looking. Looking at ways. To. To develop it to create. Yes. This is why accessibility shouldn't, should not, oh, should not, okay. This is why should not, this is why accessibility should not, should not, Sorry. So the night. Yeah. So the night. So that. That's important. That is. Yes. This is. This is very important. Important point. Part of the design from the start. And when you start. And when you start with people, when you start with people who are often left out, left out, you can often, you can often something, something, um, something create, uh, better for everyone.

Thank you.

These are very illustrative examples of building an inclusive AI future. As a closing question, I would love for each of you to give me one thing that gives you hope about this potential to create a more equitable future.

First of all, what we see is in any of the countries where we work is an incredible appetite with a young population to contribute to the development of their own country through entrepreneurial ventures. I mean, yes, you can do that in civil society, you can do that as a politician, but especially in finding solutions for the day-to-day challenges that people live with, technology is an incredible enabler, and AI will be a very, very important lever to do so. So that is to me a big reason of optimism. We are not blind for the problems of access. You're not blind for the geopolitical tension that is around AI. But for us, on the day-to-day application, it's an incredible potential that we see there. So let's tap into that potential and make sure that it can be done in a safe way.

What gives me hope, I would say, is the millions of developers that are right now in the Global South who already have been using AI for a very long time, who are quite savvy in data science and machine learning, and have actually been driving AI innovation in their own countries. So really, for me, it's about enabling them, meeting them where they are, and continue supporting them for them to be able to build the future of their economies.

Yeah. Wow. Okay. To answer your question, yeah, I believe many more things became, you know, become possible. Let's share example from this week. At the store, at the store at UNGA event, yeah, all our cash was installed in the room. So it's a technology we introduced. For people who use hearing aid, the hardest part is not often the volume. So it's telling different sound apart. So when music and crowd noise come in at the same time, it can be very hard to hear the music clearly. So, but the all cast send the sounds from stage straight to the hearing aid. This makes a new kind of listening possible.

And that is a very small example. Without AI, our training would have been much harder. Yeah. And without AI voice technology, we could not even release our beauty girl. Yeah. But I think, I believe AI can solve most of them. Thank you.

Thank you everyone for joining and let's hope for the future that you all painted here today. Thank you.

Thank you.