Calculators wont give you completely wrong results, not even once, where "AI" does that way too often. If calculators did too, mathemeticians simply would not use them.
This is why using LLMs to generate deterministic code (with human-verified tests) is a much better idea than including LLMs directly in runtime systems.
