The book 'Superintelligence' proposed 'philosophy with a deadline,' prioritizing questions needed to safely navigate the AI transition over eternal questions