• The VOIDRUNNER'S CODEX is LIVE! Explore new worlds, fight oppressive empires, fend off fearsome aliens, and wield deadly psionics with this comprehensive boxed set expansion for 5E and A5E!

WotC So it seems D&D has picked a side on the AI art debate.

Status
Not open for further replies.

tomBitonti

Adventurer
They take the products of artists’ labor to use without their knowledge or permission. I don’t care if a court decides they’re allowed to do it, it’s still theft by any meaningful definition of the word. It is not the same process as human artists taking inspiration from other art because these algorithms are not thinking beings. They aren’t capable of contributing anything original to a creation, they can only recombine elements directly copied from elsewhere.

Bold added by me. This, I think, is highly debatable. Perhaps a discussion for another thread.

TomB
 

log in or register to remove this ad

tomBitonti

Adventurer
What makes algorithmically generated images not art is that they are created by totally a different process than art is.
Additional text omitted. A change of process does not conclusively change the result. I think a much deeper analysis and discussion are necessary.

Certainly, computers process information differently than biological creatures. I rather don’t think that is a sufficient reason to categorically deny them certain capabilities.

TomB
 

tomBitonti

Adventurer
You cannot have AI-generated content without a large database of human-generated content to “train” it on.
Additional text omitted. This is very probably false, while it may be a current limitation. Indeed, the example of AlphaGo, which did initially train on the historical go record, but then was retrained entirely by playing itself, provides a counter factual.
I suspect hindrances to AI art are more that evaluating good art from bad art will be difficult without training data, with the problems that art is strongly coupled to being human. Senses, emotional response, cognitive response, and other processes are not immediately accessible to an AI.
TomB
 

gamerprinter

Mapper/Publisher
I've already posted a response, but as a followup. There is no debate to me. Rather, no, I won't ever use it, don't need it, don't want it. What you do with it, I don't care at all - it doesn't apply to me. No Debate. What WotC does, I don't really care...
 

FrogReaver

As long as i get to be the frog
Sure, but we're not really talking about the practical are we? The question of whether or not an AI is actually producing genuine art, in the same way that a Human does, is more a question of philosophy than of realism.

Theres also an important nuance in that even within truly human-made art, many pieces that are produced wouldn't necessarily "count" just on the merit that a human made it and called it art.
I think trying to place such limitations on what gets defined as art shows a rather fundamental weakness in your position instead of strengthening it. Trying to separate 'art' from 'art-like images' doesn't make for a compelling argument. IMO.

Once upon a time when forums like this were more commonly populated on the internet, I was actually a part of a digital art scene that was based around making artistic images to insert into one's signature. This mostly revolved around using stock imagery as a basis and incorporating it as part of a composed piece.

And commonly, most artists (myself included) had zero issue picking up whatever bits of photography we wanted, even if we were going to unethical means to do so. We (mostly) weren't selling these images for profit, and as it was it was early 2000s internet; nobody cared about piracy and ripped versions of CS4/5 Photoshop were plentiful.

But the thing was, you couldn't just take a stock image, crop it to size, and then call it art. Nobody tolerated that sort of nonsense, and doing minimal edits weren't gonna fly either.
For me, that's not evidence that such wasn't art, it's evidence of that art not meeting the expectations of the person who hired you.

And I imagine once the dust settles as far as image AIs go, their products will be seen in the same way. If all you do is generate an image and present it with little to no actual work, you're not making art. If you utilize that image as part of a composed piece, then you are.
IMO. Being 'art' is not something dependent on the amount of effort put into the 'art'. Nor of the quality in the end.

Whether you intended to say it or not, you are implying that only this one person had the opportunity for original thought and all that followed are merely derivative.
The point is that no 'art' today is produced in a vaccum. We all have been inundated with it since we were born - which has nothing to do with original thought. Then apart from human created art, there's the world itself from which we draw inspiration from - and most often we depict objects from our world within the art we do.

ART AI's don't have sensors to sense the world with. And even if they did and produced 'something' it would likely not be something comparable to anything we produce, because the machine, even with the best sensors isn't going to experience the world in the same way as us. Thus, the way to have a machine produce something similar in form to human art is to provide it with examples of human art, not equip it with sensors.

The real question is whether after such initial seeding, can the program go beyond that. And the answer is, Yes! If allowed. The most simple case is for the program to have another algorithm that updates the image database and or forms new connections between the images. Current technology is already capable of this, although possibly not yet implemented.

But that has nothing to do with the philosophical question of where art originates from and if humans are capable of it even in a vacuum, thus distinguishing them from an AI who we all recognize fundamentally cannot function without the data they were trained on, and in fact can't even function with only its own inputs, which is another distinction humans have.
Ever seen a person that's been blind for his whole life paint? That's basically the equivalent of an Art AI without any training data.

Edit: And more to the point, if you recognize that at least one human had the capacity to make art when no art yet existed, then you are acknowledging that this is possible for humans to do.
If no human today does art that way, then why does it matter whether they theoretically could?
 

tetrasodium

Legend
Supporter
Epic
Can you, and will you, please directly agree with the statement that art should only be utilized in AIs with the express and conscious consent by their artists?

Everything you're saying at this point is just demonstrating an extreme bias against the concerns of artists, so please, demonstrate that you actually acknowledge these concerns as legitimate.

If you won't, or if you're just going to continue to deflect to hammering me over something I was never even talking about, then you're only admitting to that bias.
I don't feel this question is relevant because it tries to frame itself in a very loaded manner that excludes the weight of responsibility from one side. The effort is especially unreasonable for reasons already detailed, back in post 124 I included a link to how a nonpaying user of a free service can opt out their images to prevent those images from being shared for the training of AI as the terms of service otherwise allows the host to share them for. Consent is given in the terms (also linked & partially quoted earlier). A service is provided in return for that consent. An opt out option that goes beyond that is even offered yet you want to shift to add additional layers of consent while denying that it was already obtained & opt out options that were not required are allowed.
.

For example @Charlaquin said "“Improve Pinterest” is not synonymous with “train Dall-E” (or whatever other algorithm)." above, but if by feeding those images in as training data they are able to improve their search tools or or provide new options it absolutely does.
 

I think trying to place such limitations on what gets defined as art shows a rather fundamental weakness in your position instead of strengthening it.

I think this is rather ironic, given what you're arguing actually diludes the concept art into meaninglessness. Which, as it is, tracks.

As Ive mentioned before, I've had this conversation more than once. You don't sound any different from people who have a very cynical, contemptous attitude towards artists.

For me, that's not evidence that such wasn't art, it's evidence of that art not meeting the expectations of the person who hired you.
No, its evidence that it wasn't art, because cropping a stock image is not artistic.

Again, we tread the same waters of cynicism and contempt for what artists actually do.

Being 'art' is not something dependent on the amount of effort put into the 'art'. Nor of the quality in the end.

Then you're not aware of what actually goes into artistic works.

Another fun anecdote from my life; as a kid I had a pretty substantive magician phase, and being the the likely ADD riddled perfectionist I was, I got pretty good at it too despite being between 8-10 years old when I started.

Over time, I eventually gravitated towards card flourishing and eventually the whole "XCM" phenomenon at the time, and I again got pretty good, and to this day I still retain the muscle memory to perform more or less at the same level.

The thing about card flourishing, and really sleight of hand work in general, is that it takes a tremendous amount of practice just to be able to do certain moves for the first time, and it takes a considerable amount more time and skill to be able to do them in a way thats not only aesthetically pleasing as part of a routine, but is also consistently perfect.

Pretty much anyone can make a cruddy looking card fan with a deck. Not everyone can make one thats sufficiently large, displays all the pips, is evenly spaced, and can be thrown from one hand to another without the cards spilling everywhere.

And yet, some people, who carry that same cynical, contemptous streak, liked to say that what I could do with a deck of cards wasn't impressive; that I was merely flipping cards around.

Naturally, if I handed them the deck not only were they incapable of any of the things I could do, but they'd often end up ruining the deck to some degree just because they don't know how not to grip a deck of cards like you're suffocating it.

Thats a clear example of why artistry can't be so reduced to meaninglessness, because theres more to it than you're giving it credit for by trying to assert any random thing can just be called art.

The point is that no 'art' today is produced in a vaccum.
Which as explained is a non-sequitor to what was being talked about.

Ever seen a person that's been blind for his whole life paint? That's basically the equivalent of an Art AI without any training data.

Blind people have 4 remaining senses. Art isn't just a visual medium, and even a blind person can accomplish quite a lot without sight:

Screenshot_20230317_140531_Chrome.jpg


Esref Armagan didn't need to be born with sight to paint that, and definitely never needed to see another painting either.

If no human today does art that way, then why does it matter whether they theoretically could?

For one, some humans do. See Armagan.

And for two, as already said, that discussion is specifically about what sets apart AIs and Humans in regards to whether or not AIs are actually capable of creating art.

Examining whether or not a human can create art when no art exists is a thought exercise that demonstrates the clear dividing line between the AI and the Human.

A human does not need to see art to produce an example, and does not need to see art to improve upon their own works. AIs do, and thus, they are not producing art.
 

I don't feel this question is relevant because it tries to frame itself in a very loaded manner that excludes the weight of responsibility from one side. The effort is especially unreasonable for reasons already detailed, back in post 124 I included a link to how a nonpaying user of a free service can opt out their images to prevent those images from being shared for the training of AI as the terms of service otherwise allows the host to share them for. Consent is given in the terms (also linked & partially quoted earlier). A service is provided in return for that consent. An opt out option that goes beyond that is even offered yet you want to shift to add additional layers of consent while denying that it was already obtained & opt out options that were not required are allowed.
.

For example @Charlaquin said "“Improve Pinterest” is not synonymous with “train Dall-E” (or whatever other algorithm)." above, but if by feeding those images in as training data they are able to improve their search tools or or provide new options it absolutely does.

So you've chosen to continue hammering on something I was not talking about and have no desire to talk about.

Understood.
 

tetrasodium

Legend
Supporter
Epic
So you've chosen to continue hammering on something I was not talking about and have no desire to talk about.

Understood.
Why did you ask in 187 if I'm willing to talk about consent if you aren't willing to talk about it? Is your reluctance simply because it's difficult to credibly make some of the claims you've made without ignoring the actual state of consent?
 


Status
Not open for further replies.

Voidrunner's Codex

Remove ads

Top