524
‘Impossible’ to create AI tools like ChatGPT without copyrighted material, OpenAI says
(www.theguardian.com)
This is a most excellent place for technology news and articles.
Then LLMs should be FOSS
All AI should be FOSS and public domain, owned by the people, and all gains from its use taxed at 100%. It’s only because of the public that AI exists, through the schools, universities, NSF, grants, etc and all the other places that taxes have been poured into that created the advances upon which AI stands, and the AI critical research as well.
That does nothing to solve the problem of data being used without consent to train the models. It doesn’t matter if the model is FOSS if it stole all the data it trained on.
Copying is not theft or stealing.
Copying copyright protected data is theft AND stealing
Edit: this also applies to my stance on piracy, which I don’t engage in for the same reason. It’s theft
By definition you're wrong
It’s theft.
You can steal all you want, but it’s still theft. Piracy is theft, stealing data to be used as training data is theft.
Not everyone wants their creations to be infinitely shared beyond their control. If someone creates something, they’re entitled to absolute control over it.
You are only hurting yourself by adopting a rule like that.
The only way I can steal data from you is if I break into your office and walk off with your hard drive. Do you have access to something? It hasn't been stolen.