AI Generated Art is a hot button issue in the cyber spaces I frequent. The argument goes AI is trained on the artistic styles of real people which then can generate art like that person but isn’t copyrighted. A human artist inherently gives consent for humans to be influenced by their work, but did not give consent for their work to train an AI to recreate more of their work.
Replace ‘AI’ with ‘photographic film’ and I see no substantial difference. Sure, museums don’t want you walking around with flash photography because they’d rather you to hit the gift shop on the way out but artists still paint and now we have art shows with photography as well. I would not say photography has damaged the artistic world in anyway and I doubt AI Art will as well for similar reasons.
I just can’t tell my other group this, but thought I would try playing with this idea here.
Do you want to scope this conversation to just one narrow aspect of AI generated art (which looks to be the matter of authorised/unauthorised reproduction of existing works)? Because that’s just one tiny facet of a big collection of issues, and there are lots of ways in which it’s nothing like photography, but I am not sure if you want to talk about those.
Meanwhile, here are a few ways that AI generated art is different from photography just within the scope that you mention
With conventional photography, you can reproduce something that resembles the original to some degree limited by the conditions and equipment; the best possible outcome is something that looks like an exact duplicate.
With AI generated art, you can create someting that resembles the style of a specific artist, but is not an exact copy and thus (in a hypothetical best case scenario), be able to pass that output off as the genuine work of that artist.
You can also, in creating and marketing a huge volume and variety of those somethings, compete with the original artist, despite possessing none of the talent that artist possesses; Or in other words, you can use someone’s hard work to put them at a commercial disadvantage.
You can also, as an end user of AI art generation, inadvertently commit what is essentially plagiarism of an original work, without ever knowing that the image you now clutch in your hands closely resembles that thing you have never seen firsthand.
Or in other words, AI generated art is in some, possibly many ways different from photography. If it was the same in every way, it would be photography.
Yeah, that’s not really an answer, is it? Is your mind made up beforehand, or can you answer @Mangetout’s specific points?
There are a lot of artists who deal in media other than physical paint, including digital art. There are (as I understand it) ways to treat digital art so it is possible to tell a copy from an original.
However, there is no way to prevent an artist with a recognizable style (say, Lichtenstein, to pick someone who is fairly well known) from having his style copied in new works which are then marketed as his work. A human being might be able to do that, if they have some measure of talent. But if AI can be trained to do it, it no longer takes talent to produce a “Lichtenstein” piece of art, and you can make as many new and different ones as you want. That degrades the value of the actual art, like counterfeit money degrades the value of real paper money. I can certainly see that as a problem that has nothing to do with photography.
Actually, the main reason is that flash photography contains UV light, which damages pigments over time. One or two photos is no big deal, but if you have people taking flash pictures all day long, eventually the damage becomes noticeable.
Ever notice that if you leave something out in the sun, it often fades or gets bleached? That’s mostly UV light doing the damage. That is what museums are trying to avoid.
The fact that they can then use this to ream you at the gift shop is just an added bonus. It’s not the primary reason for it.
I do.
If you want to make an AI image similar to something that one of the classic artists created, that’s probably no biggie. You’re not going to deprive Leonardo da Vinci or Salvadore Dali of any income. They aren’t exactly in any kind of condition to cash a check these days. But what if you are a modern artist with a distinct style like Boris Vallejo or H.R. Geiger? Should you be forced to give up your income because someone trained an AI on your style and now it can spit out new works faster than you can?
That doesn’t seem at all similar to photography to me.
This is especially true when you consider that AI doesn’t make copies. It makes new works based on the same style.
Cribbing off of your own art is called being a hack. Daring to strike out from anything you have done before, particularly if any result could be identified as your previous work, is the very duty of artists.
What you are talking about is Illustrator Work and how AI affects that. That is a different angle, and one I will have to ponder.
I don’t understand what you mean. Why, in half or more of the photographs I take of people they have six or more fingers per hand regardless of the number they may actually possess.
This is an entire additional debate. I’ll start with you don’t get to decide what artists ought to do; or, put another way, even “hacks” have a right to profit from their own work/distinctive style. I also think the statement is just plain untrue about pretty much every well-known artist out there.
The artists who get studied often have a distinctive enough style that an unsigned or previously unknown work can be identified as theirs.
On reflection, the arguments against AI Art are equivalent to saying, “CGI affects hurt the stuntman’s pocketbook.”
Like, yeah, but it’s all Hollywood Magic as far as I am concerned. Why should I care as a consumer?
Stuntmen are still around so many years after the death of Buster Keaton so it doesn’t seem so bad.
The stunt people are openly credited, as are the CGI people, the costumers etc. At NO time does the studio put forth that the actors did their own stunts. Do you also think that students should be able to buy or steal papers online, change things around a bit and pass it off as their own? After all, they went through the trouble of modifying it a little bit.
On the art side, you’re welcome to take non-flash photos at museums, but you couldn’t take a picture of a current (non-public-domain) artist and try to make money off of it. AI is taking in all those artists’ works and making money for the purveyors of AI tech.
If you only see art as something to be consumed then there is probably no hope of getting you to understand the problems inherent in AI art. If art isn’t something to be celebrated as a unique endeavor of humanity, if artists don’t deserve to be supported for their work making art, then AI art is the best thing to ever happen.
I agree with the OP, and it’s a point I’ve made before. The advent of photography gradually made obsolete the skills involved in producing realistic paintings and portraits, but it did not obsolesce the visual arts. It may have influenced them by perhaps accelerating the move to more abstract styles, but it didn’t kill them.
@Mangetout makes the point, if I understand it correctly, that AI can be used to essentially create art forgeries. This makes it different from photography but not substantively different from what can already be done without AI, using tools like Photoshop to manipulate images to achieve a certain look. Photoshop has a diverse set of filters and other tools for that purpose, including tools that simulate particular methods of drawing and painting. AI just kicks it all up to a new level.
To use the OP’s analogy: if I take a photograph of your artwork and sell it, you can sue me for copyright infringement. Is this a way that AI-generated art is no different from photography?
Last time this came up, I provided multiple legal citations that the point of copyright was to protect the labor of artists, in order to make that labor valuable enough that people would continue to engage in it. Our current legal structure around AI quite literally devalues the labor of human artists, to the point that the very sorts of works on which AI are trained are unlikely to continue to be made.
I maintain that as a society we have a vested interest in human artists being able to make a living making art, and that we need to restructure our legal codes to account for the new technology. This isn’t to say that AI art users are dirty rotten scoundrels, any more than Gutenberg was a scoundrel for making book copies before copyright law was formalized. Rather, it’s to say that new technologies need new legal structures.
As long as artists can sell their works to AI companies, I’m fine with it. If AI companies can use artist work without compensation for their labor, I’m not fine with it.
What!? What a strange notion - and, IMHO, a wrong one. One reason that artists produce similar pieces is an attempt to explore fully the subject. I counter that that is the very duty of artists.
So for the record, you don’t appreciate Van Gogh, Rothko or Georgia O’Keeffe?
I’m a firm believer of ars gratia artis, “art for the sake of art”. Art created by humans for humans using AI tools is still human art, at least until we have sapient AI. And that art should be considered on its own merits as a work of art.
Beyond that, is the business of art. Artists deserve to be compensated for their work. Period. Likewise, people have a right to partake of art. The tricky part is the interaction of those sometime competing interests. I have no solution, but can only pushback against one-sided “fixes”.
For example, in mine opinion, the current copyright regime is a failure. It protects neither artists nor partakers of art. We need a system that give artists fair compensation of their art while allowing others fair use of that art.
The AI technology is simply putting a spotlight on how inadequate our current system is. The blame focused on the artists who use AI is a divide-and-conquer distraction which the art industry will use to devalue all artists and further shackle them with copyrights “for their own good”.
We’ve already seen art being denied copyright because of its provenance. I won’t be surprised if the next step is an onerous system that artists will have to satisfy in order to “prove” their art is “original” enough to deserve copyright protections. All while reducing the compensation for artists who can’t meet those requirements.
Kicking it up to a new level is potentially the problem.
I mean, I don’t think it’s the same as anything - all analogies are inexact. What we’re dealing with is new, and is not exactly like anything that has come before. Similar perhaps, but not exactly the same. It might be similar enough that existing decision frameworks can be hammered to fit it, or it may not.
But the prolific capacity for this technology to do whatever it does, in massive wholesale, could be enough of a problem even if it seems like the principles resemble something else.
It’s not the primary reason for it.