> If you peek inside the LLM, the process seems different for things they don't know. This is really interesting, could you link some relevant resources? I won

> If you peek inside the LLM, the process seems different for things they don't know. This is really interesting, could you link some relevant resources? I wonder if a chatbot could at least signal at the UI level that the answer might be a hallucination.