Benjamin Recht's research group studies how to make machine learning systems more robust to interactions with a dynamic and uncertain world. They are particularly interested in making machine learning more scientific and safe by recognizing where conventional wisdom is incorrect and by establishing reliable benchmarks and baselines to measure performance. For example, they have published papers showing that conventional machine learning theory mischaracterizes how deep networks work and how many results in personalized medicine are overstating their utility. Their work is enriched by collaborations with researchers from applied fields including computational imaging and robotics.