Today there is an unprecedented opportunity to do good using public data. The main bottleneck is that we can't direct enough statisticians to work on humanitarian projects. 

MIT engineers have been developing BayesDB, an open-source platform that addresses some of the these problems. Novice BayesDB users can answer data analysis questions in seconds or minutes with a level of rigor that otherwise requires hours or days of work by someone with advanced training in statistics plus good statistical judgment. This talk will focus on what and why BayesDB is, not how it works. It will use examples from collaborations with the Bill & Melinda Gates Foundation and Boston Children's Hospital, showing how BayesDB can jointly analyze neuroimaging data and survey data collected from kids in slums in India. It will also discuss new initiatives aimed at using BayesDB to build empirical maps of poverty, inequality, and psychological suffering. Examples include data on PTSD vulnerability and resilience in US Army veterans, including data on adverse events caused by psychotropic medication, and on electronic health record data for members of poor, rural US populations. It will also include a brief review of other AI technology being developed by Vikash Mansinghka's lab, the MIT Probabilistic Computing Project.

In the future, we may post a transcript for this talk, but we haven't created one yet. If you'd like to create a transcript for this talk, contact Aaron Gertler — he can help you get started.