[SEQ RERUN] Artificial Mysterious Intelligence — LessWrong