The problem is that AI code is painfully verbose and annoying to read. It's like having a coworker who's kinda shit at his job and who's PRs you have to review and try to get into reasonable shape by requesting changes again and again.
I don't know about you, but I'd rather just write the damn thing properly myself than have to deal with that.
The current AI generation has an additional pain point compared to the bad coworker, it's unable to learn, you can give it a rule file but it doesn't always respect it properly and it doesn't update it itself.
TLDR: the bad colleague will stop making the same mistake in the PRs the AI no
Hopefully this will change in future generations of AI
The problem is that AI code is painfully verbose and annoying to read. It's like having a coworker who's kinda shit at his job and who's PRs you have to review and try to get into reasonable shape by requesting changes again and again.
I don't know about you, but I'd rather just write the damn thing properly myself than have to deal with that.
The current AI generation has an additional pain point compared to the bad coworker, it's unable to learn, you can give it a rule file but it doesn't always respect it properly and it doesn't update it itself. TLDR: the bad colleague will stop making the same mistake in the PRs the AI no
Hopefully this will change in future generations of AI
Right, maintaining bad AI code is the new maintaining bad offshore code. I didn't want to spend my career doing the latter, either.
Maintaining bad AI code would be committing or shipping bad AI code in the first place