Kaj_Sotala comments on AI risk-related improvements to the LW wiki