I just tried to use ChatGPT to generate a switch for comparing a byte[] with a known set of strings, switching on the length of the string and then doing equality checks. It produced code that looked right, except all of the sizes were totally wrong.
Lesson learned: don't use LLMs for anything numeric.