3 principles for creating safer AI | Stuart Russell — Steemit