People overestimate the ability of LLMs and programming. It’s very useful at times. I use it sometimes, but I have to be very careful because your can very easily go down a rabbit hole of doing things that are VERY wrong and not industry standard with lots of issues. And if you spot them and try to get the LLM to sort it you often go round in circles in solutions that do not work.

I now mostly only use it to point me in the right direction of something I currently don’t know about, but I make a conscious effort to look at it’s sources rather than taking it’s word for it.

And the other really good use of it is for debugging errors. That I have no qualms with. Errors are usually well documented online with solutions on how to fix them so LLMs mostly know what they’re talking about with them and can point you in the right direction of sorting it out way quicker than trying to find that info yourself due to how shit search engines are now.