That is actually a point that rarely gets brought up -- we're so concerned about the dangers of AI in warfare, we don't necessarily stop to think of where they may be able to do a better job at avoiding lethal errors.