Artificial Intelligence & Data Science

Innovations in artificial intelligence, remote sensing, machine learning, and big data analytics allow for more accurate, frequent, and granular insights related to global poverty and inequality.

We are in the midst of what some have called “the fourth industrial revolution,” borne by a sea change in technologies to measure, store, and analyze human information. Satellites ply the sky, taking billions of images of Earth and its inhabitants; mobile phones, crowdsourcing tools, and social media apps generate trillions of bytes of relational data every day; machine learning algorithms and artificial intelligence (AI) predict everything from our writing styles to which households are most at risk for flooding; wireless sensors measure and report real-time traffic, air pollution, and the sound of trees being felled in protected rainforests.

At scale, these novel data sources can offer a significant potential to benefit humankind. But while AI and other data science tools are rapidly changing the world, their impact on those in low- and middle-income countries (LMICs) remains unclear—and we’re not gathering the data we need to reliably measure these their effects. AI may improve the lives of the most vulnerable through faster, more innovative, and often cost-effective ways, but will likely come with real threats to individual privacy and livelihoods.

Across its portfolios, CEGA is invested in studying how the latest innovations in AI-powered research designs and deployments, along with other frontier technologies and methodological approaches, can improve the lives of those living in poverty. We are leveraging AI and data science to build new infrastructure to make research more open, transparent, and reproducible; funding rigorous evaluations of deployments of AI-supported interventions across agriculture, health, governance, humanitarian response; and, investing in capacity building, technical resources, and workshops for actors who shape institutions and decisions. Throughout all this work, we’re thinking carefully about the consequences of getting it wrong. CEGA prioritizes the safeguarding of individual, organizational, and sovereign data privacy and advises global decision-makers about the trade-offs present in this work.

AI-Enabled Research Infrastructure

CEGA is leveraging AI to build tools that make research insights more accessible, open, and actionable for decision-makers. The results of most development interventions remain buried in PDFs and gatekept behind publisher paywalls that often have inconsistent or incomplete reporting of results, cost data, methods, and more. As a result, decision-makers have incomplete information and struggle to generalize results across contexts, limiting the true impact and potential of research.

CEGA is working to address that gap. Our evidence aggregation work joint with the Global Poverty Research Lab at Northwestern University, coined DEVIDENCE, is leveraging LLMs to automate the extraction of key information from academic papers to enable faster and more timely meta-analyses. The resulting product, a publicly available dataset of treatment effects from global development papers, will enable decision-makers to access evidence across studies with lower cost and higher accuracy.

Decision-makers also face challenges in considering the tradeoffs between various development interventions when they lack accurate estimates on the cost-effectiveness of programs. Most published research lacks information on the cost of interventions and their return on investment for beneficiaries. CEGA is using AI to create new cost modeling and automated prediction tools to fill the gap of costing data, allowing donors, practitioners, and researchers to route limited funding for global development more strategically.

AI-Enabled Deployments

Across our portfolios, we deploy, measure and improve the impact of AI and data science-backed applications in LMICs that can make it easier and faster to improve lives. We generate rigorous and timely evidence on who is benefiting most (and least) from the AI transition, and how to improve diffusion and deployments in low-resource settings. This is particularly important in LMICs, where public administrators and service providers often lack access to high quality socio-economic information on communities and where language, connectivity, institutions, and culture differ substantially from the high-income settings where most frontier models are tested at scale, and where the vast majority of development resources are concentrated.

Our network of researchers are leveraging unstructured and structured data in new applications, including advising ministries of agriculture on how to boost farming yields, strengthening state capacity to tackle fraud, estimating the cost of ending poverty globally, creating new remote and highly accurate ways of identifying where the most vulnerable individuals are located, effective delivery methods, and so much more.

AI Capacity and Field Building

Ensuring sustainable ownership and effective, responsible use of AI and data by LMIC governments, industry, and organizations is a fundamental priority. CEGA establishes direct channels between AI researchers and experts, LMIC governments, and front-line organizations to deploy, iteratively improve and govern frontier AI responsibly at meaningful scale. We deeply invest in skills development, technical resources, and curricula for social scientists, engineers, data scientists, regulators, and public administrators who shape institutions and decisions. Our data and AI labs, in collaboration with LMIC governments, help build sustainable AI and data science functions within ministries, empowering actors to integrate data-driven insights into policy-making and service delivery. Beyond these activities, CEGA also hosts AI4GD workshops and trainings, bringing together academia, industry, nonprofits and government to discuss the latest innovations and best practices, novel research and methods, and how to leverage AI, data, and tools for impact at scale.

AI Safety

The utility of AI is dependent on large datasets and, in global development contexts, that data is sensitive information about the world’s most vulnerable populations. What happens if that data isn’t safeguarded? At CEGA, we think deeply about the data privacy of individuals and we’re building and applying new privacy enhancing technologies (PETs) to allow governments, NGOs, researchers, and more to serve development beneficiaries, without compromising their safety. We’ve also spent time investigating how individuals in low- and middle-income countries (LMICs) think about data privacy and how their considerations differ from those in high-income countries. Drawing on durable partnerships with institutions and experts in LMICs, CEGA is uniquely positioned to offer expert guidance to development actors who need to leverage sensitive data.