As a young software engineer with a lot to learn, I would have been better off with ChatGPT or Claude than I was with experts exchange, reading manuals and banging my head on the wall until something worked. Often the SDKs I had to work with were inconsistent, buggy or required unsafe/undocumented features to accomplish basic things. I would not categorize the time I spent struggling with those arcane tools as productive learning. It was just "shit we had to deal with" to do the job.
So today, if you are a young engineer feeling like you are way behind, feeling like an imposter, feeling like you can't catch up to the industry: welcome to the club. I've been doing this for 20 years and those feelings are never far away. Instead of trying to lean on LLMs as a crutch, though, use your imagination! The tools we have now are what make us so much more productive. Use them, but don't let them use you. If you are learning especially, write the code, and let the LLM critique your work. Otherwise, give the LLM problems and ask it how it would solve it, and learn about the concepts that come out. Treat it like a Google search that just works way better and (for now) has no ads.
It's literally the same argument as how to use IDEs. The more you understand what it's doing, the better you will be at your job.