LLMs don't and can't do math. They don't calculate anything, that's just not how they work. Instead, they do this:
2 + 2 = ? What comes after that? Oh, I remember! It's '4'!
It could be right, it could be wrong. If there's enough pattern in the training data, it could remember the correct answer. Otherwise it'll just place a plausible looking value there (behavior known as AI hallucination). So, you can not "trust" it.
Yes? Dismissing criticism by playing victim and attacking the critic's motive and painting them a Very Bad Person™ is one of the lowest effort cards out there. It's constantly employed everywhere, regardless of what group the user might represent.
You guys are seeing big numbers in your notification box?