The library is still there, the admission is still free. The card catalogue isn’t as useful for finding the contents anymore. But, as long as you can find the tutorials, answers, etc. are still available, still free to access.
The AIs trained on the data, but they didn’t delete it when they were done training.
To be clear, those are rare books bought by the kilogram, not rare books people would want to buy.
If your aunt’s boyfriend’s sister’s dog’s mum writes a sci-fi book that’s a blatantly obvious cross between star wars and lord of the rings and gets 100 copies printed and friends and family buy 20 to support her, that’s now an ultra rare book with only 80 copies on the market and a prime target for AI companies to buy because it’s cheap and it’s new and unique training data.
There’s a library, been there for years. Across the street someone opened a book rental service, and people are flocking to the paid service for access to what they were just getting for free. Not only that but the paid service is less accurate and you still have to go to the library from time to time to check what the service is selling.
Somehow we need to stretch this metaphor to account for the fact that the book rental service can only work by sending millions of robots into all the free libraries every day, taxing the resources of the free library with all the traffic.
And the resources in the paid book store are allowed to lie and manipulate with barely any warning (maybe 0.0001pt font underneath the store sign), while the store tells you everything is trustworthy and absolutely reliable.
You forgot the part where the real answers are obfuscated by all the poor quality tutorials being generated every day, repeating wrong answers and proliferating the problem.
This keeps getting posted, but it’s still wrong.
The library is still there, the admission is still free. The card catalogue isn’t as useful for finding the contents anymore. But, as long as you can find the tutorials, answers, etc. are still available, still free to access.
The AIs trained on the data, but they didn’t delete it when they were done training.
Except that rare book thingy, where they’re debinding, scanning and trashing them.
Or the fact that a ton of indie forums had to close or go private because of the excessive scraping.
To be clear, those are rare books bought by the kilogram, not rare books people would want to buy.
If your aunt’s boyfriend’s sister’s dog’s mum writes a sci-fi book that’s a blatantly obvious cross between star wars and lord of the rings and gets 100 copies printed and friends and family buy 20 to support her, that’s now an ultra rare book with only 80 copies on the market and a prime target for AI companies to buy because it’s cheap and it’s new and unique training data.
That’s awful, but it isn’t what this post is about.
It’s actually worse.
There’s a library, been there for years. Across the street someone opened a book rental service, and people are flocking to the paid service for access to what they were just getting for free. Not only that but the paid service is less accurate and you still have to go to the library from time to time to check what the service is selling.
Somehow we need to stretch this metaphor to account for the fact that the book rental service can only work by sending millions of robots into all the free libraries every day, taxing the resources of the free library with all the traffic.
And the resources in the paid book store are allowed to lie and manipulate with barely any warning (maybe 0.0001pt font underneath the store sign), while the store tells you everything is trustworthy and absolutely reliable.
You forgot the part where the real answers are obfuscated by all the poor quality tutorials being generated every day, repeating wrong answers and proliferating the problem.
But bro I’m so sick of pretending…