I think that AI eventually will solve technical maintenance problems, but not human-related ones: limited attention, trust, motivation issues. And we are going to support mostly "old" projects everybody relies on, not some new AI-gen stuff.
LLMs represent a very important opportunity to force-multiply limited resources.
Now, setting aside ethical issues for a moment, open-sourced knowledge, writing, history, data, Q&A, and tech is essentially a prerequisite for a data-driven technology like LLMs, and if those turn out to be a net win for humanity, then we can directly trace the routes to initiatives like this one that can curate humanity's best contributions.
And our plan is to willy-nilly give money to everyone who asks for it with no oversight or attention to other factors or human involvement. Game over. You win.
Copyright is a fundamentally unethical concept and must die. Open source foundations should rightfully support the death of copyright, whether by LLMs or by other means.
I think this is hopelessly naive. The LLMs crapping out code are shamelessly ripping off open source code, sans copyright notice. It makes no sense for a foundation supporting open source to also support this massive copyright massacre.
Also, I think you're going to get flooded with requests to give money to vibe-coded crap, because if you have no skills or shame but want to make a little money off your AI-generated crap, why not try and extract money from this initiative? The curl guy showed this is very real.