Agreed that something must and will change, and that things will be litigated. Also agreed that the authors and artists have a valid complaint that’s worth a lot more consideration.
Personally I’m not sure I agree with your assessment on fairness. An AI is not (for the foreseeable future) a person, and can read and imitate much faster and much more effectively than a human can, so yes, it’s a new thing.
But up until now there has never been copyright protection on style. There has never been a prohibition on humans reading someone’s work and trying to write something in a similar style. There has never been a (need for a) clear definition of style.
If we want to change that, (and maybe we should!) copyright doesn’t really fit what we’re after. Extending copyright enough to prohibit AI style imitation would also seriously inhibit human art production, because you could always (if you have deep pockets for lawyers and for hiring whoever counts as an expert on style) say that someone’s new work is in some (combination of) pre-existing style(s), and therefore is either infringing (if it’s in a protected style) or not eligible for protection (if it’s in a public domain style). The same restrictions that are why you can’t copy more than a snippet of a copyrighted work now, would most likely start to work against you if anything more than a snippet is in a style you didn’t originate.
As an example, if Tolkien had been able to copyright his style of high fantasy, it would still be protected today. SO MUCH other fiction could never have been published if that were the case. And if he doesn’t count as deserving style protection, then who does? And then the better AI gets, and the more human authors use it in the course of their own actually-creative work, the worse the problem becomes and the blurrier the lines get.
AI-generated content reminds me of the evolutionary process that produces “mimicry” in insects and other species. On the surface a bug may outwardly resemble a leaf or a caterpillar may resemble a venomous snake, but no matter how good they get at creating that first impression a deeper look will reveal the fundamental differences between the mimic and the genuine article.
Just as a leaf-camouflaged insect can’t photosynthesize nutrients for the plant it sits on, AI-generated literature can’t make a reader think and feel the same way human-created literature can because you need a human mind to do that.
i agree with all that. the difference of course between george martin and chatgpt however is that one of those two is a human; that one of those two is engaged in an act of creativity and the other is not.
we don’t know that these algorithms will get better. see also full self driving. so while it’d be good to look somewhat ahead - we don’t necessarily have to. we can decide that the current use is infringement and go from there.
it’s very easy to distinguish - not by the output ( ex. style ) - but by the inputs. and that’s the part here that matters imo re: fairness.
they could have only used public domain works. they didn’t because they knew it would be less useful
I think this just proves to me that AI is just another scheme to devalue labor’s significance in production. Also, most AI is just smoke and mirrors where technicians still have to constantly manage/tweak/override unexpected outputs to ensure something akin to a regularly functioning system. It’s really frustrating seeing how much effort is employed to avoid paying creators their due and not being piss babies about it. Seriously, just pay creators to create things and stop making up stuff that isn’t really good in the first place.
I’m currently traveling, so please excuse the delay (and the quality of my unfinished thoughts).
I agree that commodification - turning something into a tradable entity - was and is not balanced (neither financially nor otherwise). I could also argue it is alienating artists from their artwork.
What prompted my knee-jerk reaction is that I do not see this:
I didn’t read his piece as productive, and perceive his tone as negative as well as somewhat pompous. Hence the quote upthread, including the tongue-in-cheek allusion to Camus’ Sisyphe. As I stated, finding meaning in striving reminds me very much of what the existentialists have to say about life (including art). But the text seems to ignore the inherent meaninglessness of said striving. And hearing it in the voice of someone who is, among other things, a rather successful (financially and artistically) old white man makes me very suspicious and triggers my bullshit detectors, and thus my sarcasm glands excreting some bad text.
I’m still not sure about my own argument, though.
I think that everyone could use LLMs in a creative way. And I think the models themselves are a result of creativity. What they are not is a replacement for human creativity. I agree with many criticisms in that direction
They are a tool, and what can be done creatively with them remains to be explored and reflected in ways which I cannot forsee. What I can see, and what I came to doubt that Cave and other critics can see: these tools are based on statistics, i.e. mathematically described rules of probabilities. Artificial intelligence is a misnomer, and I am under the impression Cave falls for this. AI is not intelligent, and not creative in itself. It is a thing, not a being.
To conclude, I certainly do not think it is an existential evil which must be defended against at all costs. And it’s not the end of everything, may not even the end of anything. It’s just another technology. It is not inherently Bad, nor is it inherently Good. It is. What we DO with it matters.
Oh, f*CK me: I’m triggering my own bullshit detectors now. Did I just write something which can be summed up by a comic character?
Bingo, but the capitalist class doesn’t want to hear it in that they want to keep fantasizing that these LLMs will be easily left unattended making them billions in new IP rather than admitting they’ll most likely be used in work that needs fluff and basic instructions like customer support. It’s not as good as copying the utopian socialist “chicken roasts flying into mouths effortlessly” kind of vibe. I’ve had this discussion so many times with folks that it just seems some really do believe that labor is disposable or unnecessary to production of wealth. And it’s quite an unusual opinion to be held in my experience since so many things are dependent on the intent of the labor, even simple minor changes of repetitive actions, to make anything (a service or product or both).
very true. ( although i think women who are engineers, and women of color who are engineers, are likely well aware of what century old literature is like )
That’s why they swear upon realizing that’s all they were using. They didn’t actually consider the data set until just then, since after all they are still engineers.
This cartoon gets to the point of my query, which is how the AI companies were allowed to put vast amounts of copyright material into their machines without any let, hindrance or compensation to the original creators. Of course they could actually have contacted people and agreed to pay them for their work, but it’s a lot harder to make a profit when you have to fairly compensate the workers.
Keep in mind that there have been plenty of cases of large corporations have been trampling all over the copyrights of small creators for years now and have rarely been held accountable for it. Copyright is there to the protect the rights of corporations, not of creators…
It’s also possible the company purchased a copy of the relevant texts from Amazon and then used them as they saw fit. They certainly hijacked stuff that had been posted on public or semi-public sites wholesale. Presumably a LLM “trained” with nothing but texts approved by Moms for Liberty would give drastically different results.
that’s not how laws like copyright work. if i buy a book on amazon, i don’t then have the right to duplicate it and sell the copies as my own. it’s the copying that’s the problem. ( if i stole it to copy it, then that would be an additional but separate problem )
yup. please see the entire conversation above with posts by multiple people explaining why that’s a problem.
you introduced a new argument saying it was okay because “purchase” and i was pointing out how that doesn’t change anything. purchase or not world be a separate issue
I know the situation is a shitshow for authors and other creators. However, having a creator determine every final use of their works is also a stranglehold on creative activity. The use of samples in music mixes, just about any transformative or parody work, even work deemed too directly derivative would all be out of bounds.
Once I purchase an item, I can use it pretty much as I want. I know that Atwood would prefer that her works not be used in this way, I heard her say so in person. I don’t see how it can be prevented under current law if the work in question wasn’t acquired illegally.
AI created books being published under author names for sale on various websites without the knowledge, much less consent, of those authors should be illegal, but there seem to be problems with getting the titles removed.
