> LLMs tend to stay very generic, to have a lower chance of being "wrong".
It depends entirely on how you prompt them.