If it feels like magic, it’s probably garbage. Anything worthwhile takes work.

"LLM has produced these results, I have just picked the fitting ones, adjusted the prompts and corrected errors. Behold: intelligence!" I mean, there *is* intelligence involved, but it's not where expected. Smoke & mirrors, almost literally.

It is quite good for rubber duck / smarter than a rubber duck. There is value in it for some tasks, but over hyped and capabilities claimed aren't there. It isn't a time will fix it issue, it is the wrong approach. 2008 I started working with AI/ML & NLP that have good capabilities when used well.