The simulation of judgment in LLMs