What if we don't need to solve AI alignment? What if AI systems will just naturally learn human values as they get more capable? John Wentworth explo…| www.lesswrong.com
