I think a big problem I have with AI write now is that the context window can get messed up and it has a hard time remember what we talked about. So if you tell it to write a code that should do X, Y, Z, it does on the first request, but then on the next request, when it's writing more code, it doesn't recall.
Second, it doesn't do well at all if you give it negative instructions, for example if you tell it to: "Don't use let! in Rspec" , it will create a test with "let!" all over the place.
Second, it doesn't do well at all if you give it negative instructions, for example if you tell it to: "Don't use let! in Rspec" , it will create a test with "let!" all over the place.