Kaj_Sotala comments on [link] New essay summarizing some of my latest thoughts on AI safety