I’ve mostly avoided commenting on the news that Meta got caught stealing everyone’s books—including about 30 of mine—to train its AI.
Maybe it’s the decade I spent writing a tech column, but I don’t share the fear and loathing that many in my field have toward AI. I get why they feel that way, but my take is that large language models are just probability engines. They can’t think, they can’t feel—and if you can’t think or feel, you’re never going to be a good writer. So, I’m not worried about being replaced. That’s not a commonly held opinion, so I left it to my colleagues to do the rounds of media interviews, sharing their outrage and existential dread.
What did catch my interest, though, was hearing about a couple of class action lawsuits already spinning up in the U.S.
The thing with Facebook’s IP theft is that it’s way more blatant than anyone else’s. Thanks to leaked internal comms, we know they were sitting around saying things like, “We need high-quality IP to train this model. All these crappy Facebook posts and scraped tweets from Elon’s fever swamp just aren’t cutting it.” And then someone pipes up: “Well, there’s this pirate site full of books written by actual authors, edited by professionals. We could just… y’know… steal those?”
“What steal them as in commit a crime.?”
And then—of course—they did.
So, that feels like a pretty solid case to me.
I’m reminded of an old girlfriend who joined a class action against the State Bank of New South Wales about 30 years ago. She walked away with eleven grand, which was some decent coin back then. I think she bought a car with it.
The big hurdle for me is that I’m here, and the lawyers are over there. I’m not even sure I can join a U.S. class action. But if one kicks off over here, you can bet I’ll be putting Zuck in a headlock and shaking him until the jingly stuff falls out of his pockets.