I don’t believe for a second that whoever trained the model doesn’t know what it was trained with. If they really don’t, they have no business in doing research.
Now, it’s not always possible for other people to know, but that doesn’t mean the data doesn’t exist.
I would really like a professional’s view on what counts as art piracy, really. (You could generate an artwork in someone’s style, it might even include their signature. What if you just use that without telling anyone it’s made with AI, thus making people assume it’s done by the artist it copies the style from? What if you sell it letting the buyer assume it’s done by the artist it copies the style from?)
As long as you haven’t claimed it’s someone’s work, any presumption made in that regard are squarely on the buyer. Caveat emptor and all that.
Counterfeit goods infringe on trademarks, copyrights and patents. Not style. Selling something that merely “looks similar” but doesn’t involve these elements is not illegal.
edit: also, buyers of knockoff products are generally well-aware what they’re buying are fakes and knock-offs and not the originals. It’s really different from “With those mangled feet i though i was buying original Rob Liefeld but never bothered to actually ask.”
It’s unfortunate, but it’s probably impossible to protect an artist’s “style” and that might actually be for the best. An artist trying to protect their corner of the market by essentially “patenting” their own style (because I think it would have be this granular) would probably have a negative overall effect on the creative field as a whole. How could this possibly be enforced without discouraging artists from engaging with other artists and their work?
Sometimes the market just gets flooded with a popular visual language because other artists really loved it and then incorporated it into their own work, either intentionally or completely organically over time (though some probably took a cynical approach and said “this is what I have to do to be noticed”).
What sucks about the fact that AI makes this replication of style so easy, is that it can put artists in the unfortunate position of having to compete with cheaper and cheaper competition for certain kinds of work. But trying to restrict the use of an artist’s “style” has the potential to do more harm than good, when “style” would be nearly impossible to define and enforce without freezing the natural cross-pollination of ideas and significantly altering the creative fields.
Now, if someone was straight up pretending to be the artist and putting out work under their name, that would be very different (but I doubt this would be a widespread problem).
That wasn’t what I was thinking about, really. More like, if I buy an asset pack that has works from multiple artists, or a comic book, or a set of postcards or something, and I recognize the style, my first assumption wouldn’t be “obviously this is AI” (unless I recognized it as the style of a specific AI, but that’s beside the point).
I guess then it’s just “buyer beware”? I think that’s well within the realm of the average scam and we just need to keep our wits in certain spaces of the internet.
One of my friends once thought she was buying a vintage magazine or something and when it arrived it was literally a low-res printout of the cover image.
I accidentally bought so many bootleg Naruto DVDs on ebay when I was a teen (though at least those had the actual anime on them).
Not in my wildest imagination as a kid would I have imagined that AI robots would threaten the job of artists. AI was suppose to take the jobs of blue collar workers
In all seriousness, I think the industry is still young and is trying all sorts of things. Maybe as it matures it’ll find its footing somewhere else like in medicine. But I do think big companies like Disney will have an in-house AI image tool to reduce costs.
I’m not (at all) AI-philic like Starwish, but I think they’ve got the better side of a couple points in the last few days: if you’re only hostile to robots taking art and writing jobs, your position reeks of classism; and the parallels between anti-AI-ism and veganism, as ethical movements, are pretty close.
Also, I am personally firmly against Luddism of any sort. If robots (either completely independently or in concert with human editors) can do something better (including faster and cheaper) than humans without bots, then they should do it. (This statement does not touch on issues of plagiarism or intellectual property, which are separate topics.) If people are put out of work by this, the solution is taxation and welfare, not featherbedding.
I am honestly curious: do you see this happening? AI taxation, and by “welfare” I assume you mean something like the UBI - do you think it’s possible?
My issue with statements like these is ironically not the AI itself, but the human factor. Humans, at least certain ones at the top, are quite happy to screw others over if they get to keep the status quo. Do you believe that concessions, such as AI taxes, will be given? And do you believe an UBI can exist that’s good enough for a person to live of off, and what would it be financed by?
I think it’s possible enough that it’s worth fighting for.
As for funding an UBI, the flip answer is “eat the rich,” and the less flip answer is that it’d have to be part of a sweeping social reform - the kind of thing you’d need to get Sanders and Warren on, and would include Medicare for All (because massively expanded preventative medicine means a reduction in total medical expenses - fancy that!), fixing capital gains tax loopholes to make Trump and co. pay their taxes, folding Social Security over into the UBI, and yes - a robot tax to recapture some of the profits of automation.
Which if reasonably accurate with its analysis, is a bit frightening. Does it give those people who write their own job applications an advantage or disadvantage, and how can the recruiters pick the applicants who have used ChatGPT?
Having seen recruiters arbitrarily throw out half of the applications without even looking at them, setting ridiculous standards for experience and education for positions they don’t understand, and being the ones responsible for applicants needing to use buzzwords recruiters incentivized in the first place, I would say them complaining about chatgpt making their job difficult is pretty rich.
Just one example of how they treat applications even when they know they’re being recorded, how they justify their decisions even when their actions aren’t in line with those decisions.
They also throw out the applications from anyone who admits to a disability, I’ve had to do that, plus back in the day before everyone had a mobile/cell phone, if they didn’t answer because they were at work, out went their application.
When a company I worked at was looking for an administrator, they went through the photos (and social media, if available) of the applicants and just threw out applications w/o photos lol. The amount of snide comments made about everything from clothing style to makeup was insane.
So you don’t even need a disability to be refused, just a wrong haircut!
