The secret is that it doesn’t ;)
I doubt our current statistics based AI methods ever will.
Programming requires precision. Code is an exact specification for how a machine should operate.
I don’t see how we can get perfectly precise responses from a nondeterministic process.
Not having to context switch as much is a real blessing, and I've only reached it in one language in specific contexts. If I can transfer this to many languages and many contexts, my productivity will rise a lot.
LLMs are great for that! Sometimes they’re good at translating from one language to another. They miss idioms from time to time, but they’re usually pretty good with that task.
Not so much for writing software themselves.
I would like to see it in action, and then I'll form an opinion.