Making the Right Things Easier: How the UN Is Scaling AI for Public Good

How are we going to take care of our one planet and all its people, especially considering that our population is projected to reach 10.4 billion before the end of this century? One approach is by achieving the Sustainable Development Goals (SDGs), created by all the member states of the United Nations (UN): 193 countries worldwide. But there are obstacles on the way.

At GTC 2025, I attended a session titled Empowering Global Change: United Nations Adopts AI to Further Sustainable Development Across Nations. The conversation focused on a new AI-powered chatbot developed by the United Nations in collaboration with Accenture and running on NVIDIA’s platform. Its purpose is deceptively simple: make it easier for people to access information about the world’s SDGs. The execution, however, is anything but simple and it raises important questions about how we build AI for real-world impact.

The SDGs, adopted by all UN Member States in 2015, outline 17 goals aimed at addressing global challenges like poverty, inequality, climate change, and education. We’re a decade in, and according to the panel, we’re on track to meet only about 20% of them. One reason? The data needed to drive action is often inaccessible—trapped in long reports, dispersed across platforms, and hard to navigate for those without deep policy expertise. The chatbot addresses this by acting as a multilingual research assistant, surfacing specific, sourced information in response to user questions.

What struck me wasn’t just the technology, but the design choices behind it. The team emphasized accessibility, not just in terms of language, but in helping people connect with complex information in ways that are usable and trustworthy. Users can trace answers back to their original sources, preserving context and enabling validation. It’s a design decision that mirrors a broader principle I hold: transparency isn’t a nice-to-have in AI systems—it’s foundational.

The panel also acknowledged the deeper challenge of scaling AI responsibly, especially in public-serving contexts. It’s one thing to prototype a solution, another to sustain it across languages, regions, and evolving data. This project doesn’t claim to solve the SDGs. Instead, it recognizes that meaningful change begins with making knowledge accessible by removing friction between those seeking answers and the data that could help them act.

In an era where AI is often framed as revolutionary, this felt more grounded—a reminder that sometimes, the most impactful applications are those that quietly remove barriers. Not every AI project needs to dazzle, but it should deliver. This one does it by helping people move from information to insight, and from insight to action.