@jbqueru@fosstodon.org I’ve had this experience virtually every time I try to use one, even for C#, Swift, and especially AppleScript. It boggles the mind how anybody is supposedly getting near-perfect code.

The only time I've had semi-useful results is when I ask for something that matches one of the existing questions on SO, and then the result is usually a mostly-identical version of the top answer.

In effect, LLMs are almost like better search engines, if they weren't also prone to being wrong so often.