Comment by ChaoStatiX

I still use LLMs, but strictly for sanity checks or quick summaries. Trusting them blindly to debug complex systems often leads to a rabbit hole.