LLMs have great applications in domains where they augment the weaknesses of algorithmic techniques and where their outputs can be automatically checked, e.g.