ChatGPT provides incorrect solutions to most iOS prompts relating to UI stuff that I gave it.
They would look correct to someone who knows how to program, but does not know iOS. Most of them even compile. But they're very wrong, even when I tried dozens of different inputs in an attempt to get the correct solution from it.
But are we confident GPT4 will fail in the same ways? What about 5? I've been coercing it to write code all day and it's honestly pretty good. It can get stuck, sure, but when it does and I provide it the compiler error it can fix in 50% of the time...
They would look correct to someone who knows how to program, but does not know iOS. Most of them even compile. But they're very wrong, even when I tried dozens of different inputs in an attempt to get the correct solution from it.