@kkarhan That article seems deeply flawed in its description of what “training” AI actually means.
Anyway, the point of my post is that OpenAI is preventing ByteDance from profiting from its machine’s output without compensation, even as OpenAI is profiting from random humans’ output without compensation. Whether or not it’s legal isn’t the point of the irony.
There are very finite ways to do things correctly - regardless if programming languages or images or art...
Those things are scientifically and mathematically quantifyable and at a certain level can't be avoided.
Unless we explicitly specify it differently, we expect a human hand to have 5 fingers and music to use harmonics and loops in programming languages to be how they are or sentences in a language to follow it's grammar...
Would you also be fine with paying perpetual residuals to the authors of your schoolbooks because apparently you learned from their works and most likely passed school exams if not more with the help of those and I assume make a living off those?
@kkarhan That is a really flawed argument, and I’m not going to even start to engage in it. Suffice it to say that just because something is finite doesn’t mean it’s inevitable. The number of ways pixels can be arranged is finite, but every artist’s creation is inarguably unique, and copyrightable, by any court’s definition. Your assertion that sentence and grammar patterns aren’t copyrightable is just…objectively false. There are millions of copyrighted books made out of those patterns.
If I were to copy entire works verbatim that'll obviously be a violation as someone who can actually create and thus violate copyrights by virtue of authorship.
OFC If I were to just let a machine do that this doesn't mean I'd get off the hook.
@kkarhan Please stop or I’ll mute you. Your arguments have many erroneous assertions and misunderstandings. I’m not in the mood to debate you on this basis.
Add comment