petersuber on Nostr: Question for the hive mind: When #publishers negotiate to let #AI tools train on ...
Question for the hive mind:
When #publishers negotiate to let #AI tools train on their copyrighted publications, do they ever demand that the tools provide accurate citations and links to their sources?
AI companies are willing to pay to train on copyrighted content and publishers want the money. But publishers also want the traffic and credit that comes from accurate citations and links.
I'm guessing that either that publishers don't make this demand or that AI companies refuse and publishers buckle and just go for the money.
But my only evidence is that some of the most prominent AI tools will not give citations and links, even on request, when we know they've trained on copyrighted content. It looks like they're resisting demands from publishers as well as competition from tools that routinely provide citations and links.
But I'd love to learn from those with first-hand knowledge.
When #publishers negotiate to let #AI tools train on their copyrighted publications, do they ever demand that the tools provide accurate citations and links to their sources?
AI companies are willing to pay to train on copyrighted content and publishers want the money. But publishers also want the traffic and credit that comes from accurate citations and links.
I'm guessing that either that publishers don't make this demand or that AI companies refuse and publishers buckle and just go for the money.
But my only evidence is that some of the most prominent AI tools will not give citations and links, even on request, when we know they've trained on copyrighted content. It looks like they're resisting demands from publishers as well as competition from tools that routinely provide citations and links.
But I'd love to learn from those with first-hand knowledge.