That doesn’t mean the answer is right.
Are you saying that there are some cases where 1 + 1 does not equal 2?
It is a tool, but it doesn’t always give you the right answer if you don’t use it right. Exactly like an LLM.
As I said, an LLM will give you different outputs for the same input. It's non-deterministic. It has nothing to do with whether or not you're "using it right". That's just how LLMs function. Whether or not it gives you the right answer is a dice roll. Sometimes the odds are very good, and sometimes they aren't, but you never have a 100% chance.