The deciding factor for being able to effectively utilize LLMs and dodge hallucinations is ability to read code and intuition for how a solution should look. I think juniors are especially hesitant to just dig into understanding some source code unless they have no other choice, e.g. preferring to wait on email response from the author of the code over piecing things together.

This makes LLM tools so tempting, you don't even have to wait on the email response anymore! But of course, this is basically going in blind, and it's no wonder that they end up in hallucination mazes.