The latest in preference elicitation and preference learning in AI [links] — LessWrong