If you trained a model on a single copyrighted work, then that would be a copyright violation because it would inevitably produce output similar to that single work.
But if you train it on hundreds of thousands of copyrighted works, that’s no longer a copyright violation, because output won’t closely match any single work.
How is something a crime if you do it once, but not if you do it a million times?
One of my first tasks in my game development career was to change the data type used for the main currency in [Famously Addictive Farm Simulator Game], because a user had exceeded the maximum value.
I eventually found out approximately how much IRL money this person had spent on this game…
6 figures. And not barely 6 figures.
People don’t spend that much because they’re just having fun.
There is absolutely something different about these kinds of games. It’s abusive and dangerous, and we should consider it a health hazard.
Yep. Same problem we have with AI use of free-to-view literature and art. The author’s intent is often to invite others to participate in a collective effort, and start an ongoing conversation where works can be shared back and forth and everyone improves as a result.
Corpo use of FOSS — and especially ML training on free-to-view works — often takes the fruits of the collective effort and then sprints directly away from the community, refusing to participate and sometimes even wrapping a thin for-profit layer around the free underlying tools.
In the case of AI, this for-profit wrapper is so comprehensive and so thoroughly obscures any reference to the source material that not only can it replace the original communities very effectively, but it denies any ability to navigate through to the original communities even if you wanted to.
Dirt, dust, dead skin? Oils? Gut bacteria? Dental fillings? Food you just ate? Oxygen in your lungs? Oxygen in your blood? Implants for sure, right? What about hair, or nails?
I can imagine a scenario where someone tries this ability for the first time only to wind up naked, perfectly clean-shaven, bleeding profusely from every orifice and extremity, breathless and doubled over in pain, convulsing on a pile of shit, hair and other gross, getting their back sliced open by disembodied toenails.
I love seeing state-level law about consumer protections.
As the article notes, it tends to force sellers to treat it like a national law or deal with the headache of splintered markets… but it’s way easier to get state-level proposals passed!
How does your ISP have anything to do with port forwarding, or wired vs. wifi?