I've noticed some people having a lot of trouble with AI/LLM and always getting bad results. Came across this blog post that resonates with some of my comments in this thread: https://www.hackerneue.com/item?id=43775358