Comment on Congress Wants Tech Companies to Pay Up for AI Training Data
General_Effort@lemmy.world 6 months agoThese open datasets are used to fine-tune LLMs for specific tasks. But first, LLMS have to learn the basics by being trained on vast amounts of text. At present, there is no chance to do that with open source.
If fair use is cut down, you can forget about it. It would arguably be unconstitutional, though.
That’s not even considering the dystopian wishes to expand copyright even further. Some people demand that the model owner should also own the output. Well, some of these open datasets are made with LLMs like ChatGPT.
wewbull@iusearchlinux.fyi 6 months ago
It’s not a case of cutting down fair use. It’s a case 9f enforcing current fair use limits.
General_Effort@lemmy.world 6 months ago
Can you give an example of something that is outside fair use?
Just in case, there is confusion here: Obviously there is no past precedent on exactly the new circumstances, but that does not put new technologies outside the law. EG the freedom of speech and the press apply to the internet, even though there is no printing press involved.