I guess most people can’t afford (both time and money) to run an LLM just to have a saner browsing experience. So i suspect the key to adoption would be for the community to band together and spawn some LLM servers for the community backed by donations or other sustainable practices.
The future is probably once again crowd computing. We could probably use the federated network to distribute LLM requests from the community and load balance among those that are running LLM endpoints.