This video by Robert Miles makes a levelheaded argument for taking existential AI Safety seriously as rapidly as possible.