Innovation at Rockefeller Foundation: Data science for social good

SAN FRANCISCO — Since joining The Rockefeller Foundation in September, Tariq Khokhar, managing director and senior data scientist, has started with relatively easy changes. “The truth that folks working on innovation don't like to share much is [that] often the most impactful innovation work is the least glamorous,” he said. “You need to balance doing the unglamorous stuff with the shiny stuff.” Khokhar said he advises new innovation leads to start with easy wins, such as using predictive analytics to spot trends in a large batch of grant proposals, in order to demonstrate the value of data and technology and pave the way for bigger investments. “That’s the strategy: Hook people on the unglamorous stuff and free up their imagination to think about what else is possible,” he said. Khokhar was formerly senior data scientist and global data editor at the World Bank, where he established its Data Innovation Fund, investing in early-stage and scale-up projects on data for development. Since joining Rockefeller, he has worked with Zia Khan, vice president for innovation, to bring data and technology to its programs. Khokhar said they share an interest in building the field of data science for social good, so that more people with his skills see the social sector as a destination. “Everyone talks about the risk of using big data and artificial intelligence, and there are lots of scary things that can come up, but there is an equally big risk of not using big data and AI,” he said. While Khokhar understands why the private sector is far ahead of the social sector in using data and analytics to achieve its objectives — because it is motivated by profits and even modest efficiency gains can benefit the bottom line — he noted that the lag from the social sector translates into missed opportunities to save lives. “The only responsible thing to do is to make this stuff work,” he said. Khokhar outlined three ways the foundation is working to do that, inside the organization and for the sector as a whole. The first is working to ensure that some of the same tools available to the commercial sector in markets including North America and Europe are available in markets such as South Asia and Africa. The combination of satellite imagery and learning has led to rich economic models providing answers to questions like, “exactly where you should put next supermarket or place the next x, y, z,” he said. By investing in areas such as real-time maps of agricultural yields, where commercial investors are not demonstrating an interest but the development sector has a real demand, The Rockefeller Foundation hopes to prove there is a market. The second is expanding the availability of training data for AI and machine learning to parts of the world that currently lack the level of data they need to benefit from these emerging technologies. “Unless you have the data to train these models, or a data foundation to build on, none of this technology works,” Khokhar said. The Rockefeller Foundation is currently exploring ways to build up training data for crop disease and building damage assessments. Third, the foundation is working on connecting data science talent to data science problems. “These skills don’t exist in great quantities in the development sector,” he said. “There is a mismatch between the availability of talent and the availability of problems that talent can tackle.” --— Tariq Khokhar, managing director and senior data scientist, The Rockefeller Foundation Most professionals in the sector have a hard time articulating the problems they face in a way that data scientists like Khokhar could come in and help solve them. “Half the challenge is knowing that you have a data problem and being able to articulate that to someone who can help you solve it,” he said. “So there is a mismatch between the availability of talent and the availability of problems that talent can tackle.” The Rockefeller Foundation is supporting DataKind, an organization that brings top data scientists together with social sector leaders, to explore ways it can scale its work to serve the social sector. “Their model at the moment is they pair up data scientists from industry with well-developed problems from the social sector and do that matchmaking function and see amazing results,” Khokhar said. “We want to help them move from a project-by-project approach to a sector-by-sector approach so common problems get solved in a repeatable fashion.” DataKind could be a key partner in this effort to build the field of data science for social good and draw more people such as Khokhar to the sector. “Big data and technology are part of our day-to-day work whether you like it or not,” he said. “The opportunity is to take that to the next level — and incorporate advances, whether it’s predictive analytics and machine learning, or newer technologies becoming faster, easier, cheaper to deploy at scale, but thinking strategically about how they can be used.” Khokhar said there are already ways that data and technology can advance each of the foundation’s focus areas — health, food, power, jobs, and cities. One example is the work his colleagues on the Power team are doing with a group of AI and machine learning experts to access historic data from utility companies in Africa, in order to make smarter investments in minigrids. “There is a greater need just to have more skilled data scientists within the sector as a whole. The Rockefeller Foundation is a relatively small part of a much bigger system. So our goal is really to try to support the system as a whole when it comes to attracting the right talent,” he said.