I’m not using text-to-speech engines, I am bad at writing all by myself
The book needs to be not copyrightable at the moment, then you need to have a copy, then you need to put work into scanning and OCRing it, then to place it in somewhere accessible. It’s a pretty narrow window. Some books that get published now get no digital release or it’s put behind a heavy price tag.
The point of bookburning is that it would only be accessible to AI, bringing forth it’s monopoly on said first edition. No one would upload it to IA in the process. It is deleting one source of knowledge out of existence.
Fair use loophole, they claim they convert one physical book into one digital with no illicit copies, while also supporting the delusion of AI being an equivalent to a human reader. It is weird on so many levels but no one of noticeable weight asked them wtf.
I’m not sure that I follow the story right.
They specifically looped the LLM to infere with itself and use terminal WHILE some target metric is false? And it bruteforced it’s way out of containment?
At the time they were pretty useless though. Moreso than Meta’s ones. Some hipsters from Silicon Valley bought it following the fad but then found they can’t do anything with it.