As do humans, I guess. The interesting question is whether LLMs can be more reliable than humans using knowledge to make decisions...