It's good at code generation, feature wise, it can scaffold and cobble together shit, but when it comes down to code structure, architecture, you essentially have to argue with it over what is better, and it just doesn't seem to get it, at which point it's better to just take the reins and do it yourself. If there's any code smells in your code already, it will just repeat them. Sometimes it will just output shit that's overtly confusing for no reason.

There's always going to be some overlap, wanting to use a new skill/library in a production system, but maybe in general it's best to think of learning and writing/generating production code as two separate things. AI is great for learning and exploration, but you don't want to be submitting your experiments as PRs!

A good rule of thumb might be can you explain any AI-generated design and code as well as if you had written it yourself? If you don't fully understand it, then you are not in a good position to own it and take responsibility for it (bugs, performance, edge case behavior, ease of debugging, flexibility for future enhancement, etc).