The simulation of judgment in LLMs | Dark Hacker News