A year in the making…

David Lobell
Atlas AI
Published in
4 min readFeb 6, 2019


In the world of sustainable development, decisions are often made without data. As the great Kofi Annan wrote last year, “data gaps undermine our ability to target resources, develop policies and track accountability. Without good data, we’re flying blind. If you can’t see it, you can’t solve it.”

The reality is that it’s hard to understand the daily lives of people living in the poorer parts of the world. Field surveys are logistically complex and expensive — particularly in fragile and conflict-affected countries — and so they are rarely done. For example, fewer than half of all African countries completed more than a single nationally survey on incomes and wealth between 2000 and 2010.

Credit: Jean N et al, Science, 2016.

In the face of this data scarcity, a natural response is to lean on experience and intuition and hope things go well. As Mo Ibrahim recently summed up well: “Our friends in the development sector and our African leaders would not dream of driving their cars or flying without instruments. But somehow they pretend they can manage and develop countries without reliable data.”

But imagine if organizations knew exactly how to find the people in greatest need. Imagine if governments and citizens could see exactly where social programs were performing, and where they needed improvement. How many more lives could be transformed?

What if we could track which agricultural and other business investments actually bear fruit — how much more effective would investments be, and how many more people would be excited to invest?

A year ago today, with these ideas in mind, we decided to take a leap and launch a company building world-class AI solutions for the global development sector.

We are three professors at Stanford University, working at the intersection of development economics, crop science, remote sensing, and artificial intelligence (AI). We’ve shown that satellite imagery can be used to map poverty and crop yields in Africa, using a combination of economic data, space technology, and machine learning algorithms.

Credit: Sustain Lab. Data from Jean et al, Science, 2016.

We have always hoped that our discoveries would eventually change the world for the better. But before last year, we hadn’t spent much time actually trying to change the world.

Then we began talking with The Rockefeller Foundation. They challenged us to think big, about how to scale our work and serve organizations working on poverty and inequality. They connected us to people who helped us translate our ideas into a business plan, build and lead our Board, and manage initial start-up operations.

A few months later, with seed capital from The Rockefeller Foundation, we launched Atlas AI, a public benefit corporation with a mission to accelerate progress towards the Sustainable Development Goals (SDGs). The Foundation continues to be an active partner, bringing the unique perspectives and network that come from being a long-term steward of results-oriented global human development.

Sustainable Development Goals. Credit: UN

Over the last six months, we’ve begun working with nine partner organizations in Africa to test-drive our initial products, which include high resolution datasets on wealth, consumption, and farm yields across Sub-Saharan Africa. Across the region, we’re working directly with governments, NGOs, development enterprises, and researchers to make sure our technology meets their needs.

Soon, Atlas AI will launch a public platform providing open access to data and publications that we’ve produced. And we will continue to work with partners to figure out where we can provide the most value to the global community.

Credit: Atlas AI

Two principles have guided our work over the past year. First was to assemble the best team we possibly could, and let them get to work. We were able to attract many of the best scientists and engineers we’ve ever worked with, who have the technical talent, domain knowledge, creativity, and passion to make a difference. Soon, we will officially welcome our new CEO, Victoria Coleman, who brings a wealth of experience developing technologies for public good, most recently as CTO at Wikimedia Foundation.

Second was to approach the problem with humility. We are not trying to solve problems on our own, and we know from experience that people working on the ground can see the specific challenges (and opportunities) far better than we can from Silicon Valley. But listening to our partners and understanding their issues, it’s clear that better data and technology can help them do their work better, cheaper, and faster.

If 2018 was the year to take a leap, we’re hoping that 2019 will launch our sprint towards a better future. We now have the data and tools needed to be more deliberate and data-driven in the pursuit of sustainable development. The time is ripe for a revolution in how decisions get made in the global development sector. Please follow us on this exciting journey.

This post was authored by David Lobell, Stefano Ermon, and Marshall Burke



David Lobell
Atlas AI

Professor, Earth System Science, and Director of Center on Food Security and the Environment, Stanford University. Co-Founder, Atlas AI