When I see discussion about "AI for the people" or "community-led AI" or "public benefit AI," the focus is almost always on tooling and infrastructure.
But I don't think you can have any of those things without seriously rethinking the purposes and design of AI.
It's quite possible that LLMs and image generators — which were designed to serve profit-led industries — are simply at odds with community and public benefit. So anyone looking to build ethical and communal AI needs to start not with questions like "How do we democratize existing services?" but rather "What sort of AI would serve the needs of actual communities and societies?"