Reply to thread

Message

<blockquote data-quote="Umbran" data-source="post: 9089303" data-attributes="member: 177">The basic issue with generative AI is not simply about loss of jobs.It is about how generative AI depends upon prior art.&nbsp; Generative AI does not function if not trained on real data, which to date has generally been acquired by breaking copyright.&nbsp; They are, in effect, stealing from artists to make themselves independent from artists.That's where comparison to most other automation fails.&nbsp; Automation that replaces, say, physical labor, does not actually need prior physical labor to function.If those who trained these AIs had asked for, and paid for, proper license to use prior work, this would be a different conversation.&nbsp; If Disney pointed a generative AI at its own library of content, that it owns, there'd be little basis for complaint.And it is kind of telling that apparently much of the typical training data as been scraped off the internet, from smaller artists who have little effective way to protect their rights, rather than taken from, say, Disney.&nbsp; If this was entirely unquestionable, ethical, and legal, there'd be no need to avoid content from companies with significant legal departments.</blockquote>

[QUOTE="Umbran, post: 9089303, member: 177"] The basic issue with generative AI is not simply about loss of jobs. It is about how generative AI [I]depends upon[/I] prior art. Generative AI does not function if not trained on real data, which to date has generally been acquired by breaking copyright. They are, in effect, stealing from artists to make themselves independent from artists. That's where comparison to most other automation fails. Automation that replaces, say, physical labor, does not actually need prior physical labor to function. If those who trained these AIs had asked for, and paid for, proper license to use prior work, this would be a different conversation. If Disney pointed a generative AI at its own library of content, that it owns, there'd be little basis for complaint. And it is kind of telling that apparently much of the typical training data as been scraped off the internet, from smaller artists who have little effective way to protect their rights, rather than taken from, say, Disney. If this was entirely unquestionable, ethical, and legal, there'd be no need to [I]avoid[/I] content from companies with significant legal departments. [/QUOTE]

Verification