AI for the Last Billion: Building Useful, Local, and Inclusive Systems

​Event Description

Much of the current conversation around AI assumes always-on connectivity, high-quality data, and sophisticated digital infrastructure. But for billions of people, especially across parts of Africa and Asia, these assumptions do not hold. The risk is not just exclusion, but the creation of AI systems that are fundamentally irrelevant to how most people live and work.

This session focuses on what it takes to build AI that actually works in low-resource, last-mile contexts. From voice-based systems that operate through basic phones, to offline and small language models that function without reliable internet or electricity, the conversation will highlight practical approaches to making AI accessible, usable, and trusted.

At the same time, we will confront a critical gap in the sector: the lack of investment in the foundational data systems and infrastructure that make meaningful AI deployment possible. Without this “plumbing,” even the most promising tools fail to deliver real impact.

Drawing on examples from education, agriculture, health, and financial inclusion, the session will bring together funders, technologists, and practitioners to explore how to move beyond hype toward locally grounded, sector-specific AI solutions that deliver real-world value.

Speakers

• ​Mareike Schomerus is VP at Busara, working at the intersection of behavioral science, policy, and social research across the Global Majority.

• ​Sid Ravinutala is Chief Data Scientist at IDinsight, building AI and data-driven solutions to address global development challenges.

• ​Sulakshana Gupta is Vice President of Philanthropy Partnerships at Viamo, scaling voice-based AI solutions for underserved communities.

​Discussion Questions
What does it take to design AI systems that work in low-resource and offline environments?
How can voice, local languages, and simple interfaces expand access to AI for last-mile communities?
Why are data systems and infrastructure still underfunded, and how do we change that?
How can funders and implementers collaborate to build AI solutions that are both responsible and practical?

​Location & access
Accessibility matters deeply to us, and we do our best to choose spaces that reflect that. That said, some of our Oxford venues are in historic buildings without lifts. This room is unfortunately not accessible for wheelchair users, those with mobility challenges, or anyone needing step-free access.