Do you believe AI art is "stealing" artworks?

[Something Different]

I think this take is honestly a pretty flawed one but I see so many people sharing this opinion. By saying that AI is stealing art that makes it sound like the AI is just straight up copying the artwork and claiming that it made it. But this very clearly isn't the case, but it's the only case where the statement, "AI is stealing artworks," is true.

I know people are not claiming what I said before, what they are actually referring to is the AI using artworks to learn. But then where does the stealing part come into place? Stealing refers to taking a belonging away from someone without permission. When it comes to digital art though that definition doesn't quite work. So to steal digital art would be to copy it, then claim it as your own work, which we all know AI art generation doesn't do.

But that isn't my only problem with this take. My other problem is that what the AI is doing is essentially what a lot of other artists do as well. Artists will look at other artists' works, perhaps the artists like something about it, so they adopt it into their style. That is exactly what the AI does as well. The AI will take images, add it to it's knowledge database, then uses that knowledge to make better art. So to me it's kind of hypocritical of artists to say things like this because it's fine for them to do it, but not for AI to do it for some reason.

That is why I don't like this take all that much.

[Holografix]

Kiddiec͕̤̱͋̿͑͠at 🃏 の投稿を引用：

Holografix の投稿を引用：

... In photography, a human operates the camera. In AI image gen, a program generates the image. Different processes.

That is the false equivalence fallacy or fallacy of false comparison.

On the left-side of the comparison, you are talking about data-input that makes photography possible, but on the right-side of the comparison, you are talking about the image-processing aspect of creating A.I. generated images.

A more honest & not misleading nor mistaken comparison would compare the input of both methods, or the processing aspect of both methods, rather than comparing different stages of the process of two different methods.

Equal comparisons between Photography and A.I. image generation would be :

(data-input comparison)

In photography, a human operates a camera via choosing lighting modes, angles, & potentially staging the subject-matter before-hand; 
in A.I. image generation, a human operates an application via constructing & inputting a prompt & curating the automatic output.

OR

(image-processing comparison)

In photography, a camera generates the image via recording light on a slide or digitally in memory; 
in A.I. image generation, a computer program generates the image via a series of artificial neural networks that have been trained to imitate (but not copy) examples from large data-sets of images, and sometimes to mix imitation routines from multiple data-sets.

It should be noted that the human can be removed from both of these processes, as image generators can be autonomously fed prompts via scripts and cameras can be set to take pictures on a timer, or a time delay, or even just perpetually, such as surveillance cameras.
(Although, currently a human still has to curate the output in both scenarios before output can be both selected & accepted by the general public as art.)

You might say that surveillance cameras aren't art or artistic but I think that Jonathan Nolan would disagree with you about that.


:redcircle: :ycircle: :gcircle: :bluecircle: :pcircle:

Comparing photography with AI image generation is the false equivalence. They are dissimilar.

[Shiro♌]

Something Different の投稿を引用：

Shiro♌ の投稿を引用：

No, they're just projecting their dissatisfaction, but for a good and valid reason, especially if they've been making a living by selling their art.
I'd compare it to the massive reform a few hundred years ago when tons of people lost their jobs because machines were cheaper and performed the job better and without any flaws.

It is annoying, but that doesn't mean they shall end their career as artists.
There'll still be some people who'll always prefer art made by artists than some silly AI technology, and would be willing to overpay.

I'd even argue that those people would be willing to pay in gold for some great art.. So, their projections are justified and they're obviously exaggerating, as nothing's decided just yet.
It may increase the values of their artworks numerous times if things turn in their favor..

Yeah, I don't think AI is gonna completely kill the art industry because AI art always does look artificial

It's going to improve in the future.. But like I'm saying, there'll be people who'd pursue and collect art made by people and reject every single AI art - Which convinces me that they have nothing to fear, besides losing a portion of customers who don't care at all.

[Holografix]

Kiddiec͕̤̱͋̿͑͠at 🃏 の投稿を引用：

Holografix の投稿を引用：

Why concern yourself with the inner workings of the camera, when the discussion is about art? ...

That's quite the deflection.
Go back and re-read the conversation if you don't understand, but to summarize :
:red_blob: - You stated that A.I. image generators duplicate images,
:yellow_blob: - I then pointed out that cameras duplicate images,
:green_blob: - you then stated that a camera does not duplicate an image because it creates a photo,
:blue_blob: - I then pointed out that a photo IS indeed a duplication of an image, so much so that it is often treated as evidence; a search of the definitions even demonstrates such.

The point is, that the argument that A.I. generators duplicate images is flawed on two-fronts, but the semantics about definition and how the creation-process works, is addressing & refuting the specific argument that states that "A.I. generated art is not art because it is a duplication" by pointing out that ...so is photography, so much so that cameras & photocopy machines are RELIED ON every day for making copies of evidence that is observed in a court of law.

Further evidence that photography is considered duplication of images are the bans places on cameras in theaters, and places where in-development company work-product is developed, and most notably of all : in most areas of secure government facilities and military bases.

Photography is clearly & legally recognized as duplication of seen images, all over the world.
In order to make a consistent argument that what A.I. image generation is duplicating images and use that as the basis for claiming that it's not art, one must also do so with photography, otherwise the argument is inconsistent and hypocritical for demonstrating double-standards.[www.google.com]

...and for the record, A.I. image generators don't even duplicate images, they imitate, which is a significantly different process that is akin to what professional artists do all the time when emulating styles of others.


:orbtoshoot: :white_pearl: :stonesball: :steel_ball: :MindballBall:

To go on like you do, asserting that photography is mere duplication, is reductive.

A key point that you miss is that: Art is not about duplication or copying. Art is sui generis.

[Kiddiec͕̤̱͋̿͑͠at 🃏]

Something Different の投稿を引用：

Holografix の投稿を引用：

The programmer is feeding the AI imagery, or writing code that programs the AI to scrub for images online. Either way, the programmer is feeding (directly or indirectly) the images to the AI. And the AI is creating a composite image from the scrubbed/fed imagery. It's a composite copy, or in other words, derivative.

It's not art.

Then why is it called AI art then? At is literally in the name lol

Oh boy, this is a fun one.

Holografix will claim that it's not true art,
meanwhile,
I will claim that it's not true A.I. :P

Something Different の投稿を引用：

Kiddiec͕̤̱͋̿͑͠at 🃏 の投稿を引用：

...

I can definitely see that happening, AI causes changes in art laws and it ends up hurting artists more than the AI in the end.

Except this isn't even a real A.I. in the sense that it isn't actually an intelligent entity.
It's only "intelligent" in the same sense that a "smart phone" is "smart".
If changes in law occur, it's certainly not the A.I.s fault ...not today anyways.

Something Different の投稿を引用：

agu の投稿を引用：

...

I'd say there is that artificial look, AI art has these weird imperfections always, like something always looks off. Like sometimes people's limbs will look like melted plastic, there are just some things the AI can't do correctly/accurately yet.

Not always, and those will happen less over time, however, that's also one of the charms that might get lost to time if people don't preserve a copy of the current routines, before the technology gets too good at its intended goals to continue to exhibit and demonstrate these artifacts.

KyokoKirigiri の投稿を引用：

yep just like the corp that bought github and is using ai to patent computer code that ai scrapes is theft

and just like goggle and co using ai to create involuntary profiles and selling the data it scrapes and not paying the person that made the data any money for that data

Trained neural networks don't use scraped data in their generation routines,
at most, they use it only during training.

In many such systems the training data is even discarded afterwards, leaving no remaining data from anyone else in the generation process, similar to how most people learn how to draw.
The actual generation process is a bit different than how a human draws but the learned concepts contained within the neural network are very similar to how a human learns artistic concepts, and do not generally entail storing copies of the training data, and they certainly don't require storing copies of others' works after training.

Something Different の投稿を引用：

... photography, as the other guy is saying cause that only takes one press of a button to take a photo.

As someone who has done both, I can appreciate photography & the work that goes into it too.
Good thing I already debunked the issue with this oversimplification. :P

https://steamcommunity.com/discussions/forum/12/3729575042628880226/?ctp=2#c3729575042629532217

...but yes, they're both somewhat involved processes where a machine does most of the "actual creation work" and the human primarily does staging & directing.

Kiddiec͕̤̱͋̿͑͠at 🃏 の投稿を引用：

...
Now, so as to not misrepresent or mislead... you do point a camera lens at something before pressing that button and people often don't give enough credit to the effort entailed in angling, positioning, finding coordinates, and dealing with weather and traffic conditions in real life.

...but this is similar to how people don't give credit to the fact that A.I. generation requires a prompt be written, that is often carefully considered and run through multiple iterations that are curated, and whatever the A.I. generates is not identical to anything else.
...

[Kiddiec͕̤̱͋̿͑͠at 🃏]

Holografix の投稿を引用：

Kiddiec͕̤̱͋̿͑͠at 🃏 の投稿を引用：

...

Comparing photography with AI image generation is the false equivalence. They are dissimilar.

They both generate images.

You only claim that they are dissimilar on the basis that you assert that one is art and the other is not, however, this claim fails to address the reality that they both generate images and therefore there IS something quantifiably the same between them to compare.

Holografix の投稿を引用：

Kiddiec͕̤̱͋̿͑͠at 🃏 の投稿を引用：

...

To go on like you do, asserting that photography is mere duplication, is reductive. ...

Except that I have never actually stated that it is "mere" duplication.

The art of photography usually entails staging a scene that is then recorded, and that recording entails duplicating the image of the scene from the physical world onto a slide or digital memory.
However, photography need not actually entail staging at all, as one can merely snap a photo of something that they found interesting or beautiful in the natural world and still have that widely accepted as art. If art is supposed to be a uniquely human-expression, then on that basis of reasoning, a photograph of a sunset is not art - but good luck getting many other people to agree to that.

On the whole, photography is more than mere duplication, as many things which are photographed may be carefully planned & staged, however, in some applied usage, it attempts (successfully enough to be recognized as an intellectual property violation & even national security threat in a court of law) to duplicate documents and imagery EXACTLY as they were seen before being copied.

To claim that A.I. image generation is mere duplication, however, is much more reductive, as the A.I. image generation systems imitate images, concepts, and styles, without actually duplicating them. They MIGHT be capable of duplication in some situations but this has never been challenged in court yet and I believe that even when someone pushes it so far that it eventually does get challenged in courts, lawsuits & charges over photographic copies will always remain more frequent than lawsuits over A.I. image generation for the very simple fact that claims & charges about theft via photographic duplication consistently have merit and do not need to attempt to extend the definition of copyright to have a broader scope than it currently does. We will see this continue to be evidenced over time in the future.

[Holografix]

Kiddiec͕̤̱͋̿͑͠at 🃏 の投稿を引用：

Holografix の投稿を引用：

Comparing photography with AI image generation is the false equivalence. They are dissimilar.

They both generate images.

You only claim that they are dissimilar on the basis that you assert that one is art and the other is not, however, this claim fails to address the reality that they both generate images and therefore there IS something quantifiably the same between them to compare. :seewhatyoudid:

Holografix の投稿を引用：

To go on like you do, asserting that photography is mere duplication, is reductive. ...

Except that I have never actually stated that it is "mere" duplication.

The art of photography usually entails staging a scene that is then recorded, and that recording entails duplicating the image of the scene from the physical world onto a slide or digital memory.
However, photography need not actually entail staging at all, as one can merely snap a photo of something that they found interesting or beautiful in the natural world and still have that widely accepted as art. If art is supposed to be a uniquely human-expression, then on that basis of reasoning, a photograph of a sunset is not art - but good luck getting many other people to agree to that.

On the whole, photography is more than mere duplication, as many things which are photographed may be carefully planned & staged, however, in some applied usage, it attempts (successfully enough to be recognized as an intellectual property violation & even national security threat in a court of law) to duplicate documents and imagery EXACTLY as they were seen before being copied.

To claim that A.I. image generation is mere duplication, however, is much more reductive, as the A.I. image generation systems imitate images, concepts, and styles, without actually duplicating them. They MIGHT be capable of duplication in some situations but this has never been challenged in court yet and I believe that even when someone pushes it so far that it eventually does get challenged in courts, lawsuits & charges over photographic copies will always remain more frequent than lawsuits over A.I. image generation for the very simple fact that claims & charges about theft via photographic duplication consistently have merit and do not need to attempt to extend the definition of copyright to have a broader scope than it currently does. We will see this continue to be evidenced over time in the future.


:redcircle: :ycircle: :gcircle: :bluecircle: :pcircle:

Are you blurring the distinction between photography and AI image generation so no distinction can be made between them?

And, for what?

Seems as tho you're just grinding my comments, but not making any assertions of your own. Parasitic tactic.

[Kiddiec͕̤̱͋̿͑͠at 🃏]

Holografix の投稿を引用：

Kiddiec͕̤̱͋̿͑͠at 🃏 の投稿を引用：

...

... (argument from consequence is a logical fallacy) ...

Well, you call it that but there are at least 2 (major) flaws with that claim.

First, let's look at the definition of the fallacy that you're talking about & alleging, so that it can be properly addressed :
https://www.google.com/search?q=argument+from+consequence
"Concluding that an idea or proposition is true or false because the consequences of it being true or false are desirable or undesirable."

(I could elaborate further (and have already started a draft of further elaboration), but I'll keep this short for now.)

First, the "argument to consequence" fallacy requires that a value statement of something being good or bad actually be made, and I did neither, I merely stated observable events and how they relate to definitions and practical application.

Secondly, the "argument to consequence" fallacy follows this format :

    If P, then Q will occur.
    Q is undesirable.
    Therefore, P is false.

However, what I pointed out does not even require the outcome to either be good or bad in order to become an observable truth, in addition to the fact that I did not assert whether it was a good or bad thing.

Edit: 
It took me until now to realize this but I just want to note this down somewhere 
(for future reference in matters of reasoning) : 

The "argument from consequence" fallacy states that if an outcome is bad then it must be false, however, my position is that regardless of whether a consequence is good or bad (especially when its bad) you can't argue with the reality of consequence, even if you don't like it because to do so is to be in denial about the reality around you.

Holografix の投稿を引用：

Kiddiec͕̤̱͋̿͑͠at 🃏 の投稿を引用：

...

Are you blurring the distinction between photography and AI image generation so no distinction can be made between them?
...

Clearly not.

Holografix の投稿を引用：

Kiddiec͕̤̱͋̿͑͠at 🃏 の投稿を引用：

...

...
Seems as tho you're just grinding my comments, but not making any assertions of your own. Parasitic tactic.

Do you have an actual counter-argument or just this empty dismissal?

I don't even mind if your counter-arguments are a little rude as long as they provide something that can be debated / debunked or considered & agreed upon, primarily because it does bring up issues, even if flawed claims, that other people are also going to be thinking, and thus gives one the opportunity to consider the counter-position and then either debunk it or acknowledge where it provides helpful new ideas worth conceding to.

Anyways, we may not agree on this subject matter ...like... at all (lawl *chuckle*) and I certainly think that you're wrong,
however,
I respect your position because it challenges me to make better arguments and really think about the subject matter, even though I think that you actually believe these arguments and aren't just playing devil's advocate.

---

You also wanted someone to describe the relevance of some contemporary artists & 20th century artists and while that's difficult to do while remaining on-topic, I've already started a draft addressing that request that you made early in the topic as well, which extends my views on some of the artists that I've name-dropped already because of how they relate to the subject matter.

[Holografix]

Kiddiec͕̤̱͋̿͑͠at 🃏 の投稿を引用：

Holografix の投稿を引用：

... (argument from consequence is a logical fallacy) ...

Well, you call it that but there are at least 2 (major) flaws with that claim.

First, let's look at the definition of the fallacy that you're talking about & alleging, so that it can be properly addressed :
https://www.google.com/search?q=argument+from+consequence
"Concluding that an idea or proposition is true or false because the consequences of it being true or false are desirable or undesirable."

(I could elaborate further (and have already started a draft of further elaboration), but I'll keep this short for now.)

First, the "argument to consequence" fallacy requires that a value statement of something being good or bad actually be made, and I did neither, I merely stated observable events and how they relate to definitions and practical application.

Secondly, the "argument to consequence" fallacy follows this format :

    If P, then Q will occur.
    Q is undesirable.
    Therefore, P is false.

However, what I pointed out does not even require the outcome to either be good or bad in order to become an observable truth, in addition to the fact that I did not assert whether it was a good or bad thing.

Holografix の投稿を引用：

Are you blurring the distinction between photography and AI image generation so no distinction can be made between them?
...

Clearly not.

Holografix の投稿を引用：

...
Seems as tho you're just grinding my comments, but not making any assertions of your own. Parasitic tactic.

Do you have an actual counter-argument or just this empty dismissal?

I don't even mind if your counter-arguments are a little rude as long as they provide something that can be debated / debunked or considered & agreed upon, primarily because it does bring up issues, even if flawed claims, that other people are also going to be thinking, and thus gives one the opportunity to consider the counter-position and then either debunk it or acknowledge where it provides helpful new ideas worth conceding to.

Anyways, we may not agree on this subject matter ...like... at all (lawl *chuckle*) and I certainly think that you're wrong,
however,
I respect your position because it challenges me to make better arguments and really think about the subject matter, even though I think that you actually believe these arguments and aren't just playing devil's advocate. :uzippie:

---

You also wanted someone to describe the relevance of some contemporary artists & 20th century artists and while that's difficult to do while remaining on-topic, I've already started a draft addressing that request that you made early in the topic as well, which extends my views on some of the artists that I've name-dropped already because of how they relate to the subject matter.


:redcircle: :ycircle: :gcircle: :bluecircle: :pcircle:

Your verbosity is a weapon. Perhaps you don't realize that you obfuscate the issues with your verbiage. I understand that it is your primary style, but you wrote all these long posts just to assert that if AI gen images end up being called 'art' for dint of being ubiquitous, it's ok for me to claim that AI gen images aren't art.

If your argument concedes a space for dissenting opinion, what is the point of your argument?

[AdahnGorion]

Half the time when I make art, the imspiration for said art comes from someone else.

[Haruspex]

Darkie の投稿を引用：

Half the time when I make art, the imspiration for said art comes from someone else.

I would argue that all of the inspiration for the art you made came from somewhere else. You pull from the totality of your experience.

AI doesn't have experience beyond what training data we feed it. What artists who rely on their art to make a living fear is that it's much more efficient at crunching that "experience" into images.

It takes a human years to learn how to art and hours or days to create one good image.

It takes AI hours to learn how to art and seconds to create one good image.

AI is the mechanized looms of the 19th century, and commercial artists are the skilled weavers who formed the "Luddites" and destroyed those looms.

In the end, history shows that the machines win.

[sotaponi]

8bitbeard の投稿を引用：

Darkie の投稿を引用：

Half the time when I make art, the imspiration for said art comes from someone else.

I would argue that all of the inspiration for the art you made came from somewhere else. You pull from the totality of your experience.

AI doesn't have experience beyond what training data we feed it. What artists who rely on their art to make a living fear is that it's much more efficient at crunching that "experience" into images.

It takes a human years to learn how to art and hours or days to create one good image.

It takes AI hours to learn how to art and seconds to create one good image.

AI is the mechanized looms of the 19th century, and commercial artists are the skilled weavers who formed the "Luddites" and destroyed those looms.

In the end, history shows that the machines win.

In regards to inspiration, if you were to follow sentiments such as string theory in the context of vacuum energy (be vacuum energy and the idea of strings merely born from a measurable dimensional illusion or w/e), you could say that we are "weaving" the illusion of phenomenal reality into existence by referencing e.g. universal forms or essences such as the eye or the galaxy/universe, which "we" are trying to describe, then also coinciding quantum physics and substructures such as particles, in the context of macro qualia such as gravity/dark energy and dimensions such as time.

Likewise would art and language then be attempts at describing forms and concepts by "weaving" them into existence. Indeed, the formation of proper language seems to coincide a consciously coherent self around the ages 3-5.

Of course these (e.g. platonic) forms or ideas are then eternal and we didn't "create" them at least in an essential sense, unless we are God. However, even if we are "copying" these eternal forms, it is still our mind that allows these forms to be observed. Which seems to be highlighted by phenomena such as NDEs and DMT, completely obliterating modern computers, all the while we are "tripping" by referencing eternal forms such as the eye or the universe. (Maybe as part of the universe, "tripping" on intergalactic phenomena coinciding big bangs, such as fusion, black holes, dark energy, etc)...

If you then draw the analogy to an incoherent chatbot and the equally incoherent artbot (let's not refer to these bots as "intelligences")... simply the computational power should tell you that they aren't actually speaking a language. Because they have no qualia or understanding of anything. They have no intellect. (I guess you could argue them as stealing the computationally complex entropic labour of real painters having to navigate reality and paint on a real canvas and everything, as well.)

Etc. You could again elaborate further here. In regards to these chatbots and artbots not being coherent even to themselves. Which also matters because people think that AI can be "objective." That is physically impossible with at least classical technology and what makes conscious actors special, that they have a sense of objectivity and objective measures (cogito ergo sum) at all. And is also why all of these chatbots/artbots will ultimately appear akin to a dementia patient.

Then, finally, the post also seems to drift into a false dichotomy. As there are these "luddites" who are against looms, must certainly also believe earth under a dome, etc. Whereas the other side is "educated" and can see that trivially improved robots will reign supreme, earth is round, etc. It's kind of why today's west is so extremely obnoxious in general.

[Haruspex]

sotaponi の投稿を引用：

*Gigantic wall of over-flowery text*

I mean no offense, but you aren't writing a high school paper with a minimum word count. Your point could be made more concise.

Luddites don't have to believe the Earth is flat. They just didn't like the mechanized looms because they threatened their livelihoods, so they destroyed them. In the end though it didn't matter. The looms won. Not many people still hand-weaving textiles for a living these days.

Those in support of the looms aren't necessarily smarter. In fact, it takes a lot of skill to do hand weaving. For commercial purposes though... Would some Joe still pay an artist for Sonic the Hedgehog art when and AI can do a decent enough job for free and in seconds?

[Q-T_3.14.exe]

From what I have seen the AI converts images into anime style. That's all I know.

[lightwo]

Aside from the theoretical stuff, there's something to note; if you generate a model using images you don't have a right to use as such, copyright issues may arise.

[Haruspex]

Q-T_3.14.exe の投稿を引用：

From what I have seen the AI converts images into anime style. That's all I know.

Yes. When trained on anime.

It's trained on images "tagged" with text. So if you feed it hundreds of images containing a banana, with the tag "banana", it makes connections as to what constitutes "banana-ness". Then it can output something with the essence of "banana-ness" from pure noise though several steps of refinement.

Of course it will also make connections with several other dimensions, but the quality of output is limited by the quality and quantity of input. If you feed it nothing but crayon drawings by a 5 year old, it will only he able to generate images that look like crayon drawings by a 5 year old.