Seems like with all AI-enabling and just works out of the box experiences with VSCode and alike, makes GNU Emacs absolete. I'm aware of AI packages for GNU Emacs, but don't think is worth the investiement so much; I would mostly save it for org mode, TUI, and some other few packages. But for programming, it doesn't seem lile worth the investment, and use VSCode instead.
Certainly knowing things will always be valuable - but the effect of assistants and LLMs may be to change what it is valuable to know by devaluing a great heap of current generation’s programmers’s stock and trade.
As an addenda: by value in the above I mean “instrumental value” or more specifically, valuable to the rich who want to exploit the skills of others to become yet richer. There is always intrinsic value to knowing for the people who love to know.
Copilot's integration with VSCode is slick. But so were ActiveX plugins back in the day, and any number of other technologies from them. Microsoft isn't doing this out of the goodness of what passes for a heart. There will be an inevitable enshittification process, one that you can avoid by not getting into their ecosystem in the first place.
On the plus side, I just ask chatgpt to generate code and I copy/paste it and fix it up in emacs. It's not as slick but it works fine. It's honestly probably better than being able to just hit enter and accept whatever is generated. I've been on a screenshare and seen people hit enter or whatever to accept copilot's suggestion, and then spend a bunch of time debugging the subtly wrong code because making it that easy meant they didn't think about it.
Also, it's only a matter of time before it becomes practical to run your own good-enough model locally and have an Emacs package that integrates with that. The hardware and models are already ready for it, there just needs to be integration work.
Maybe they will pay a little fine---and you won't be getting the money either---while their profits skyrocket. It's always like this, so I don't even bother.