Oddbean new post about | logout
 nostr:npub1g0587hzzckcncxfm78n0996qe2s58nspy29wf02tqcj5sdzcpj4q6j40hv I appreciate your work although listening to this I felt you might lack a bit of understanding as to how AI ( basically Large Lamguage Models ) work.

There's no mystery to their workings and one can simply see them more like files that are a compressed version of a lot of information fed in that can be accessed using natural language. 

Yes a threat exists as it provides a lot of information leverage and can be used a highly effective manipulation and censorship tool but the direct threats from AI seem unfounded.

 It's a tool of information compression and generation. A highly effective tool like a gun that can be used both for good and bad.

An AI model trained on your work could be used for good in relaying your work to majority of populace in their language, audio or video formats.

I share your concerns around AI being centralized, however open source models are equally competent today day and google/open AI actually have no moat.

'They' are afraid of open source AI being a liberating tool for humanity and an end of big tech empires and are thus painting a picture of it being too powerful and thus calling for it's regulation. Regualtion is the only way they can slow down truth from coming out at even more fatster, stronger and efficient forms. 

But the genie is out of the bottle now and we will win, not without a fight though.

Yes they are going to use AI against us but the only way to fight at this game is to train our own models that channel the truth we want out at a much faster better stronger rate then they push propaganda.

There is no way we win this war without having our AI guns on our belts.

nostr:npub1z6207puz6qy98h4hemuxusl94new8407ejg45rz2deal27cypk4qgg4w7p