[luis8]

I wonder how feasible would it be to index everything locally? I mean how far are we from having the cpu and storage space to allow a search within our computers?

Do we really need to store everything ? Maybe a local "ai magic" could select the specific information we regularly search for and sync from an open public index.

In the case we need results it could sync and/or request those bits we are not indexing locally.

Just thinking out loud

[pmayrgundter]

Good q. Totally feasible imho, and I used to help build the goog search engine! :)

https://en.wikipedia.org/wiki/Wikipedia:Size_of_Wikipedia

20GB, which is tiny these days. Fits on a $10 USB key.

I think with some of that AI magic we'll soon be in a situation where most of the traditional (non-proprietary) human knowledge you ever need (text, images, 3d models of machines, anatomy, etc.) will be immediately accessible for free via an open-source knowledge engine.

With a commodity network connection, that knowledge engine can keep itself apprised of the post-traditional/proprietary knowledge by doing all the drudgerous work of reading through the world of Ads to get to today's news, etc..

Mate that with more AI magic to make it easy to conversationally interact with the engine and you get an agent who can represent you online and provide a distilled interface of online for you. Your digital twin or avatar.

[luis8]

Sounds really good.

I hope the open source community can come up with a solution for this. Maybe there are already projects with this goal, if not someone should start it.

It will improve our collective knowledge and at times like this were disinformation is really high we need technology/tools to reduce it

[istjohn]

What problem would that solve?

[luis8]

On top my head i can think of:

    Privacy no need to explain why this is important

    Resilience if google/any other search engine goes down it usually mean chaos

    You have total control of what is shown to you so no one can manipulate 
    what is shown to you(at least not that easily) assuming it is open sourced
    of course

I guess there are cons to this approach too but it looks/sounds really good :)