General idea, it comes up often in LLM discussion from engineers lamenting that LLMs can never solve any novel problems. This might as well be true but as said, almost any problem can be decomposed into bite sized, non-novel parts.