Do not use ChatGPT if you are beginning coding. It does not always produce valid or good code, and if you are not experienced, you may pick up on bad habits that some people's code in ChatGPT's training set has built.
It's not great for blindly generating code to use, but it's great for asking for help -debugging, planning etc. I ask it a lot of 'dumb' questions and I feel like I've learnt a lot more than I have in the past. It's given me code that doesn't work every now and then, but I'm yet to find an instance of it explaining a concept wrong.
> but I'm yet to find an instance of it explaining a concept wrong.
How do you know for sure? LLMs output is often plausible-sounding but incorrect - usually it's fairly obvious, but it can be subtle enough that I would not suggest using it until you've learned the old fashioned way and can better judge whether the LLM is wrong.
