The Open(Closed)AIs of the world have millions of dollars to spend on IP datasets.
Arguing that copyright forbids training AI models without paying authors is the moat that would prevent any hope that small labs, individuals, and open source communities can ever compete with these huge corpos.
The books and other artworks they are arguing over didn't come generated spontaneously from nowhere, and it's disingenuous to refuse sharing them to inform what is basically the worlds next currency : Intelligence. Doing so is just saying that knowledge and intelligence must belong to rich corpos only and never be democratized.
This is unexpected but Meta is basically being the good guy by giving away their research papers and models weight resulting from millions of $ of training.
The alternative to open source AI is everyone's subjugation to the oligarchists in charge of Intelligence. Copyrights holders who argue against free training of open source models from their work are morally and ethically wrong here.
Arguing that copyright forbids training AI models without paying authors is the moat that would prevent any hope that small labs, individuals, and open source communities can ever compete with these huge corpos.
The books and other artworks they are arguing over didn't come generated spontaneously from nowhere, and it's disingenuous to refuse sharing them to inform what is basically the worlds next currency : Intelligence. Doing so is just saying that knowledge and intelligence must belong to rich corpos only and never be democratized.
This is unexpected but Meta is basically being the good guy by giving away their research papers and models weight resulting from millions of $ of training.
The alternative to open source AI is everyone's subjugation to the oligarchists in charge of Intelligence. Copyrights holders who argue against free training of open source models from their work are morally and ethically wrong here.