[LINK] AI risk summary published in "The Conversation" — LessWrong