The field of AI Safety has provided technology for explaining AI predictions, as well as quantifying how much any given prediction should be trusted. In this talk, Ilya will introduce some of that technology, and show how it can be used to improve the way AI predictions are consumed to enhance their impact.