I believe most people use AI to help them quickly figure out how to use a library or an API without having to read all their (often out dated) documentation instead of helping them solve some mathematical challenge
If the documentation is out of date, such that it doesn't help, this doesn't bode well for the training data of the AI helping it get it right, either?
Unfortunately, it often hallucinates wrong parameters (or gets their order wrong) if there are multiple different APIs for similar packages. For example, there are plenty ML model inference packages, and the code suggestions for NVIDIA Triton Inference Server Python code are pretty much always wrong, as it generates code that’s probably correct for other Python ML inference packages with slightly different API.
I often find the opposite. Documentation can be up to date, but AI suggests deprecated or removed functions because there’s more old code than new code. Pgx v5 is a particularly consistent example.
