Stuart_Armstrong comments on On motivations for MIRI’s highly reliable agent design research