I see these sorts of comments repeatedly, yet they are so far from my experience it's like they are from a different world. The code generated by AI's is so wrong for me, it's not worth my time to review it. Why review something you are 9 times out of 10 going to throw away? Since I've never personally seen AI work well, all I have is youtube videos of people using it successfully. Mostly, there people are developing web apps. Sometimes they are developing in Python. Neither are languages I use a lot of now, but nonetheless there is a recognisable pattern.
The AI's seem to memorise code snippets they've seen on the web. They combine that with a remarkable ability to recognise and process the English to adapt those snippets to the context you've supplied. It's sort of like using Stack Overflow, but you don't have spend 15 minutes googling, you don't have to copy and paste, and you don't have to adapt the code mung the code to your organisations style. It's an intelligent templating engine if you like, and watching it work (when it does work) on youtube is pretty memorising. On the downside, it's wrong far more often than the Stack Overflow answer google finds, and all the context and background that tends to accompany the Stack Overflow answers (ie, the bit you learn from) is gone.
The other downside is it only works if it has seen a lot of examples of the sort of code you are say you are after. If it hasn't it will still give you an answer, but it will be so wrong you would have saved time if you hadn't seen it. So it works well if you are doing something very similar to something that's been done 1000's of times before, and posted to the web. But that's not something I do - it would bore me silly. Talking to my peers (all senior software engineers), that's not something any of them does.
Nonetheless, you 10 person shop spends their day writing your typical web app that throws up forms, collects the data, and displays it in various ways I can well imagine it works well for you.