> When you ask an LLM to tell a joke, the LLM is guessing what joke a majority of people would find funny. The result is almost never funny.

This is... false. The LLM is generating text that would be associated with "being a joke" in its corpus. It doesn't know what "funny" or "majority of people" means.

You're forgetting about RLHF.

Many of us forget that an LLM without RLHF would not respond how we think.