LLMs are only capable of performing a finite amount of computation within a single forward pass. We know that much. They are also known to operate on high leve

LLMs are only capable of performing a finite amount of computation within a single forward pass. We know that much. They are also known to operate on high level abstracts and concepts - unlike systems operating strictly on formal logic, and very much like humans.