Cormac McCarthy said that books are made out of books. Interesting to think of a book as a sort of palimpsest. And maybe that’s what these LLMs are too? Strange, layered things that are “scraped again.”
It's a fascinating area but I think there are serious issues regarding how LLMs use both copyrighted and personal data to train their AI without acknowledgement or payment. I wonder what Wordsworth would think?
It will be interesting to see the outcome of other lawsuits - the Anthropic agreement to pay authors $1.5 billion was only because of the use of *pirated* texts in their training data. You make an excellent point about *acknowledgment* - if the defence is "fair dealing" / "fair use", one has to show that the copyrighted material used is duly acknowledged and it's hard to see that thre AI companies can show this.
Cormac McCarthy said that books are made out of books. Interesting to think of a book as a sort of palimpsest. And maybe that’s what these LLMs are too? Strange, layered things that are “scraped again.”
This should be interesting
It's a fascinating area but I think there are serious issues regarding how LLMs use both copyrighted and personal data to train their AI without acknowledgement or payment. I wonder what Wordsworth would think?
It will be interesting to see the outcome of other lawsuits - the Anthropic agreement to pay authors $1.5 billion was only because of the use of *pirated* texts in their training data. You make an excellent point about *acknowledgment* - if the defence is "fair dealing" / "fair use", one has to show that the copyrighted material used is duly acknowledged and it's hard to see that thre AI companies can show this.
A host of ‘stolen’ daffodils ??
Or "An AI did my spirit steal / I had no human fears / It seemed a thing that could not feel / The touch of human years" ?!
"The Grok is too much with us; late and soon,
Swiping and spending, it lays waste our powers." ? :)