Anti-AI artists have yet to convince me why AI is so bad a problem
The main problem is the AI is "trained" by harvesting copyrighted content without the creator's permission, often from websites that explicitly forbid automated content harvesting
, and there's really no fundamental difference between "training an AI with my data" and "saving my data in an ultra-compressed fun-house database that can extrapolate and interpolate."
Computers don't "learn styles." They "store data." Neural networks give computers a remarkably weird way to store large amounts of data with a degree of compression that seems almost magic. But at the end of the day, you give the computer a bunch of data, it saves a large file, you give it the command, "give me the first chapter of A Song of Ice and Fire," and it reads that file to produce the first chapter of a copyrighted work about as accurately a highly compressed JPEG.
Oh, and then the AI company monetizes my work. And that's really what comes down to. I produced some content, and now a tech billionaire wants to take my content and use it to make money without paying me for it. The main arguments for OpenAI being allowed to do this boil down to:
- Unlike JPEG, MPEG, and other forms of lossy compression, neural networks aren't technology. They're magic. Since they're magic, that means when OpenAI takes uses them to take somebody's content and sell it to other people, the copyright disappears. But OpenAI still has copyright over its data files, because copyright is really something only autistic tech billionaires should benefit from.
- OpenAI's business model requires being able to take your stuff and not pay more for it. They're super innovative and awesome, and you don't want them to go bankrupt, do you? Think of the shareholders!
- China will beat us! We can't let something like the creator controlling his own work get in the way, or China will beat us. Chinaaaaaa.
Not that "Sophie" understands this, because he's a retarded pedophile. I'm sure his arguments are fucking stupid and should be used in court to sentence him to death.
Oh, and one other thing, another party that's been injured by OpenAI's practice is any company that was working on AI legally, meaning they made sure they had legal rights to any content they were using to train their models. Sam Altman didn't invent the DNN or even the DNN-trained chatbot. His main "innovation" was to declare that copyright law doesn't apply to what he's doing and do an end-run around getting permission to use others' content, while a whole bunch of suckers were out there using data with permission like total chumps, which naturally is slower and more expensive. Wish the rest of us were smart enough to realize we could just steal everything as long as we get the press to run cover for us.
