Comment by magospietato

If you've ever attempted to extract a meaningful few-shot response to a non-trivial coding question from an LLM, this shouldn't be a surprise.

That said, I have worked with actual humans in the industry who perform this badly, and that is still a significant achievement for a software program.