Millions of physical books are being scanned to train AI models. They’re then dumped in the trash.
180 reposts251 likes
Millions of physical books are being scanned to train AI models. They’re then dumped in the trash.
180 reposts251 likes
They have an idolatrous worship of a scaling "law." Every time they make the training corpus ten times bigger, they get a unit increment in intelligence. But I don't think they're that far from having literally scanned every book. How will they find new books to get to the next factor of ten?