Well, LLM networks don't have a 'back track and error correct' component in the design, AFAIK.