I think of AI as just another abstraction layer, somewhat similar to what high-level programming languages provide compared to writing machine code. Deciding how deep to understand the abstraction layers is a choice the user has to make, which could be optional if they don't really need to.

Nevertheless, the responsibility of whatever a human produces with AI is still on the human.

With that said, knowing how to use AI the way it's right for you can give you a huge advantage. You don't have to though. And there is not a standard way of doing it.

What I recommend to everyone is give it a try and see if and how it could help you. At the end, you have to make the decision based on your constraints and what you're aiming to and can sacrifice, including but not limited to speed, accuracy, learning, etc.