Why so much hate toward AI art?

[DuniX]

Here's my opinion on the matter with a possible worst case scenario theory.

The problem is that (especially at the time of writing this) we are currently in the wild west of this technology. There are no laws or rules and no real regulations for this software.

That is not the worst case scenario.
The worst case scenario is the precise opposite of that.
Where is the idea that the Government is on your side and is going to protect you? Who is going to save you? The Democrats? The Republicans? The ever increasingly Communist European Union? Has Occupy Wall Street never teached you the lesson on the Divide and Conquer Strategy they are using now?

If "Artists" in their brilliant "wisdom" manage to successfully to Patent a "Artstyle", Art as we know it will be Over.

What they fail to realize is that the soulless corporations can legally do the same thing. In Law there are no Rules for Thee but not for Me.
What do you think Adobe and Microsoft and Google is going to do?
Corporations have no problem with Contracts and Acquisitions, they do that every day.
If you are an Artists that actual has a Job I don't see how they are going to escape from those Contracts.
Either Slow or Fast they will eventually acquire all the Data they need to Train the Models and they will Absolutely Have NO MERCY if it's going to cost them billions or trillions of dollars, They Will get a Return on that Investment.
If you play the game of Copyright Enforcement they will play a much better game at that.

I just don't understand on what Universe artists seems to live on.
Record Labels already control the Copyright of All Music, Disney already controls the Copyright of gigantic backlog of Movies and Shows.

If you can Copyright "Artsyles" that opens Pandora's Box.
More things will be Copyrighted and Enforced while True Artists will continue to be marginalized.
The rise of Corporate Artists that sell their souls in exchange to be "Permitted" to utilize an Artstyle and become part of that "Family".
And ALL the AI Users that you are now deriding will automatically become part of that and it will be their mission to End You as free artists just like you are trying to End Them.
Mickey Mouse's Boot will be stomping all over your face, forever. That is the future you chose.

Artists Themselves have to Adapt.
There is no other choice other than a much worse one in the long run.
You cannot be a Free Spirit, you cannot be a True Artist if you drink the Kool Aid.
AI Art is precisely the Wild West, it's precisely the Unknown that is being Explored in an environment that is so constantly being Censored and Sanitized for Corporate "Sensibilities".

 

[InfiniTales]

The driving idea behind Copyright was originally to protect artists, and therefore creativity. That was the theory.

Now look at reality. Who invokes copyright laws and benefits from them the most? The artists? That sure is always mentioned, without fail. But, surely that explains why every single YouTube content creator struggles with never-ending and constant copyright takedowns on their videos, right? Videos that often don't include a single note of music, or music that they wrote and performed themselves.

YouTube's systems, as any copyright implementation, are completely biased from the start to benefit the interests of corporations and publishers, not the artists and creators.

The way I see it, in practice, copyright laws choke creativity big time.

 

[InfiniTales]

That's a very bleak picture. As you say, a worst case scenario.

In its current state, I find A.I. image generation best suited to create background images, scenery. Still far from perfect, absolutely, it requires a lot of tweaking.

But here's the thing, while looking at the results and thinking about ways to change the prompts and improve the results, I am having a lot of fun. Some bad or unexpected results often gave me plenty of new ideas and inspiration. And that would then make me sometimes switch the prompts completely around. Which in turn would inspire me further, and before I knew it I'd have spent half a day on a single prompt.

The A.I is a tool, and it's always my own human creativity and ideas driving it.

It's ideas, imagination and vision that create art. I don't see A.I. having those anytime soon.

 

[Stonkss]

Here's my opinion on the matter with a possible worst case scenario theory. With the way I understand AI art, it is an algorithm that scans through the internet's collection of images and pulls and compiles data on all of it. Then, using software, prompts can be made to compile a new image based on tags that the algorithm determines based on the data it collected (tags like hair color, eye color, body type, etc.) I think it's a great tool, and shows the advancement of technology. I am a firm believer of working smarter, not harder, which is what this is.

The problem is that (especially at the time of writing this) we are currently in the wild west of this technology. There are no laws or rules and no real regulations for this software. Art that people actually put time and effort into making gets pulled into the database that is then used by someone else, who very likely, put zero time and effort into creating art. Right now, the software isn't perfect, and work DOES need to be put in to fix the results, but that something time and more research can fix. One day, touch ups won't be needed in AI art.

I have used a few different services to create art (all non-published) and I can say, there is little to no effort (the only effort being using a few thoughts to come up with a prompt). To add to that, there are many people who use this service and paywall image sets that they created. This of course is a bit predatory to someone who doesn't know any better, and paints an unrealistic image of the type of artist that person is.

Additionally, if this type of software gets into a situation where it is determined legal for commercial use, it could lead to a situation where artists can't keep up, as the demand for actual artists begins to drop off. Of course to counter that, artists would have to step up their game and become better than the AI. Which sounds reasonable, but then remember, AI pulls data from art, so if the artists get better, so does the AI.

Now imagine if AI art services become official use and legal. Companies wouldn't have to hire artists anymore, when all they need to do is gain access to one of these services and just create the art themselves. Of course, artists need to earn money as well, so if an artist can't get a job by creating art, they're forced to move onto other endeavors. At some point the database that the AI pulls from becomes stagnant, and then everything starts looking the same.

This of course is a very crack dystopian theory.

This has nothing to do with this website, the people who use this website, or the views this website takes on unlawfully acquiring and redistributing software. We all use this website. We are all pirates. We are all in the wrong.

Can AI art be a good thing. Absolutely. Can AI art be a bad thing. Absolutely. AI art should be used like any other image software: a tool that enhances your own art. It should not be the end means for your art.

I agree with you about generative AI being primarily a tool for artists but I want to make some corrections. I'll be talking about Stable Diffusion but others work in a similar manner.
First of all what people call generative AI is a software that take a trained neural network file and uses process called diffusion to produce image guilded by prompt. The trained neural network (often called 'model') is just a 2-5 Gb file. Once trained on a dataset of images it stays the same. You can update it by feeding new images but the output will be the other version of the model so you won't lose the first one. There are hundres if not thousands of these models publicly available on civitai and similar sites. Almost forget to mention: you can produce new models by merging existing ones together in all kinds of way.

With just one of these models you can produce inifinite multiplied by infinite amount of images of different sizes. And that's just for a single prompt. The whole thing works offline because all the information about styles, objects, poses, colors, image compositions, all this information is packed (baked) into this single file. Even more. You can generate images over existing ones (just like I did with the image bellow) and that's called 'inpainting'.

I haven't even scratched the surface here. Possibilities are mind blowing and new tools are being invented every day.

Now about dark future. Right now there are more artists alive than during all the centuries of human history combined. And there will be even more in the future. Yes, creative industries (and not only them) will undergo dramatic changes because of this AI-revolution. But things have to be this way.

 

[A B C]

How would you feel if you knew that countless people can now do their job faster and with great results? This has nothing to do with garbage saturation or "sweat" and hard "work", but rather with money and the fact that now countless people have the possibility to create something and earn from it. Why would anyone want that? Obviously they don't want competitors or people making money from it.

This happens every time a new technology emerges. I am just surfing and enjoying this new technology and will do the same when the new one is released. Those who are crying are the ones who have something to lose.

 

[d7anime]

How would you feel if you knew that countless people can now do their job faster and with great results? This has nothing to do with garbage saturation or "sweat" and hard "work", but rather with money and the fact that now countless people have the possibility to create something and earn from it. Why would anyone want that? Obviously they don't want competitors or people making money from it.

This happens every time a new technology emerges. I am just surfing and enjoying this new technology and will do the same when the new one is released. Those who are crying are the ones who have something to lose.

Yep, I literally have formal education in art and seeing artists seethe like this comes across as pretentious. They are the ones most poised to take advantage of this too. All their ‘theft’ arguments boil down to loss of capital (understandable but blame the economic system). If anythin, we should be rallying around discussions on how we can maintain our current lifestyles after most things become truly automated.

My prediction is that all the fears will pass and new artists will emerge that create bigger masterpieces for others to consume.

also, as a fun reference, when I first got to college, most of the older teachers HATED digital anything, especially anime shit because it didn’t adhere to their old world views. By the time I graduated, most people, including the dean had a tablet in class and used it for figure drawing and painting kek

 

[d7anime]

I agree with you about generative AI being primarily a tool for artists but I want to make some corrections. I'll be talking about Stable Diffusion but others work in a similar manner.
First of all what people call generative AI is a software that take a trained neural network file and uses process called diffusion to produce image guilded by prompt. The trained neural network (often called 'model') is just a 2-5 Gb file. Once trained on a dataset of images it stays the same. You can update it by feeding new images but the output will be the other version of the model so you won't lose the first one. There are hundres if not thousands of these models publicly available on civitai and similar sites. Almost forget to mention: you can produce new models by merging existing ones together in all kinds of way.

With just one of these models you can produce inifinite multiplied by infinite amount of images of different sizes. And that's just for a single prompt. The whole thing works offline because all the information about styles, objects, poses, colors, image compositions, all this information is packed (baked) into this single file. Even more. You can generate images over existing ones (just like I did with the image bellow) and that's called 'inpainting'.

I haven't even scratched the surface here. Possibilities are mind blowing and new tools are being invented every day.

Now about dark future. Right now there are more artists alive than during all the centuries of human history combined. And there will be even more in the future. Yes, creative industries (and not only them) will undergo dramatic changes because of this AI-revolution. But things have to be this way.

View attachment 2775810

100%. I think we’re in an upheaval moment where established big screen creators who push nothing but their own agendas will see their old audience take media creation into their own hands.

 

[cohenber101]

Yep, I literally have formal education in art and seeing artists seethe like this comes across as pretentious. They are the ones most poised to take advantage of this too. All their ‘theft’ arguments boil down to loss of capital (understandable but blame the economic system). If anythin, we should be rallying around discussions on how we can maintain our current lifestyles after most things become truly automated.

My prediction is that all the fears will pass and new artists will emerge that create bigger masterpieces for others to consume.

also, as a fun reference, when I first got to college, most of the older teachers HATED digital anything, especially anime shit because it didn’t adhere to their old world views. By the time I graduated, most people, including the dean had a tablet in class and used it for figure drawing and painting kek

They didn't hate anime. Instead, they want you to gather details and maximize references to the fullest first. They understand that you can draw anime, but they also want you to explore other techniques. Do you think they told you to sketch cylinders and spheres in under an hour for fun? They taught you from the ground up, and all you get from them is that they dislike your anime drawing?

Is this just another "formal artist" who will become another "figurehead" in a political party?(His work also fucked, all his landscape building has perspective line fucked up all over the place)

Just kidding, but I strongly disagree with people who said this.

 

[Doublefrost]

My game started out with a lot of AI weirdness that you're talking about, but I'm slowly trying to replace the worst of it and patch things up. Technology is advancing rapidly. Here's a few examples of newer images I generated/seen others generate and included/will include in the game:
View attachment 2351272 View attachment 2351273 View attachment 2351277

I'll be honest with you, man. AI art makes cum look like it's frosting on a donut. Which is both hilarious and making me hungry. Especially that third one. Now I want another apple fritter. Thanks!

 

[d7anime]

They didn't hate anime. Instead, they want you to gather details and maximize references to the fullest first. They understand that you can draw anime, but they also want you to explore other techniques. Do you think they told you to sketch cylinders and spheres in under an hour for fun? They taught you from the ground up, and all you get from them is that they dislike your anime drawing?

Is this just another "formal artist" who will become another "figurehead" in a political party?(His work also fucked, all his landscape building has perspective line fucked up all over the place)

Just kidding, but I strongly disagree with people who said this.

oh I 100% agree. Understanding the fundamentals is key when it comes to illustrating and design. I think freshmen should still have to take traditional figure drawing and perspective classes, but there shouldn’t be a reactionary hate to the incorporation of new tech.

For example, since master copies were a big focus in our early years, I can see a student interested in AI being forced to generate an image and copy it traditionally to ensure they understand what’s happening. It’ll allow them to see where AI still lacks( composition, shadows, anatomy), while picking up on what it usually does right (color, values, etc)

Also it would be hilarious to see AI art extremists born from a lack of acceptance from traditional artists

 

[Meaning Less]

there shouldn’t be a reactionary hate to the incorporation of new tech.

Your mistake is to assume those hating on AI are artists themselves.

If anything this hate comes mainly from users that don't like the final result and don't even want to deal with the wave of inconsistent low effort cgs that are bound to flood the market.

Fortunately they already added a tag for it so people can filter it out/in as they wish.

 

[MoonlitMatcha]

Because anyone can basically make "good looking" art now. And if everyone can make good art, then can it still be called good?

 

[InfiniTales]

Because anyone can basically make "good looking" art now. And if everyone can make good art, then can it still be called good?

For me, that'd be a full, "Yes. Absolutely. Why wouldn't it?!".

If you start associating/confusing exclusivity with quality, then that 2008 mobile app "I am rich" that cost US$999.99 and did absolutely nothing would be the greatest piece of software to ever grace the App Store.

 

[smitty1989]

You must be registered to see the links

 

[Ximpactor]

You guys gotta realise that "AI art" doesn't mean that there's an actual sci-fi sentient robot painting pictures for whoever asks for it, nor is is a "collage" of pre-existing pictures. It's a VERY light piece of software scripts that learns how to depict something by looking at images to guide it's digital paintbrush, just like any real person wanting to create art would. Claiming that Stable Diffusion is "collaging" or using existing images is preposterous, original dataset of 5 BILLION images is 240 TERAbytes heavy! And Stable Diffusion models sizes? In 4-8 GIGAbytes ballpark. And no, it's not "connecting" anywhere to get info, or "search internet real-time", local UI's capable of working completely offline.

Stable Diffusion is but a TOOL, a neural network that you use for synthography. And already, at this very moment, you can use it to produce original art indistinguishable from artworks of highest quality with consistent style and characters - cause it can be TRAINED to create what you need it to create, from perfect hands and specific characters faces to sexual poses and dynamic action. And as you might have guessed - it requires time, skill and technical know-how. All that said - good luck trying to ban or track it, it's gonna be simply unenforceable at this point.

I'm gonna attach a simple explanation image as to what "diffusion" means in regard to creating art, but you more then welcome to doing bare minimum and googling "how stable diffusion works" before you start spouting nonsense when discussing "AI art".

This person is extremely dumb, his own attachment explaining Stable diffusion disproves this. Please don't listen to this. The basic idea for stable diffusion is, you start of with an image that is just random noise, the AI has previously been trained to remove something known as gaussian noise in successive steps, it will then attempt to remove the 'noise' from the image, resulting in a final image. To see if the image is correct, a GAN trained on other artists work is used to direct AI in the right direction. This is an extreme over simplification. He knows this since its covered in the shitty attachment he posted.

When this person says:

It's a VERY light piece of software scripts

He means automatic111's webui stable diffusion is a light piece of software. Actually training the GANs and stable diffusion models takes serius GPU time, everyone I know has used something like google collab which lets you buy virtual GPUs. automatic111's webui is a clever bit of software that essentially uses a trained model (hence the reason, you NEED to provide a training model) to generate images. Nothing here is similar to how a human would generate art. Real artists don't start from a picture of random noise and work backwords to generate images.

 

[desmosome]

This person is extremely dumb, his own attachment explaining Stable diffusion disproves this. Please don't listen to this. The basic idea for stable diffusion is, you start of with an image that is just random noise, the AI has previously been trained to remove something known as gaussian noise in successive steps, it will then attempt to remove the 'noise' from the image, resulting in a final image. To see if the image is correct, a GAN trained on other artists work is used to direct AI in the right direction. This is an extreme over simplification. He knows this since its covered in the shitty attachment he posted.

When this person says:



He means automatic111's webui stable diffusion is a light piece of software. Actually training the GANs and stable diffusion models takes serius GPU time, everyone I know has used something like google collab which lets you buy virtual GPUs. automatic111's webui is a clever bit of software that essentially uses a trained model (hence the reason, you NEED to provide a training model) to generate images. Nothing here is similar to how a human would generate art. Real artists don't start from a picture of random noise and work backwords to generate images.

Using anthropomorphic language to describe what the 'AI' is doing is silly, but the basic points of what that guy was saying stands.

The images the AI used to train are publicly available stuff online. Web crawlers like LAION created datasets that points to an image URL and some keywords associated with the image. These images are not stored anywhere in the AI's code. The image is successively broken down as you say into noise while training and in the end, what the AI retains is some incomprehensible relationship between the keywords and various stages of the image/noise. There is literally no data in the AI that can reproduce a training image.

So on the technical side, I'd say they are legally just fine. On the ethical side, I don't see why automating and digitizing an act that is fine for humans to do is suddenly amoral because a machine does it on a huge scale. Given enough time, I can do whatever stable diffusion is doing and be completely in the clear legally and morally. I can browse publicly available artwork of my favorite artists online and learn whatever I can of their styles and what not. The morality of an action doesn't suddenly change based on the scale. A bad thing can become a horrible thing when done in scale, but a neutral/good thing doesn't suddenly turn into a bad thing. So it could be argued that if you are against AI art due to ethical hangups, you'd have to be against artists learning from their predecessors and the existing artworks and techniques. That would be a preposterous stance.

 

[DuniX]

So it could be argued that if you are against AI art due to ethical hangups, you'd have to be against artists learning from their predecessors and the existing artworks and techniques. That would be a preposterous stance.

That's because it always was a double standard, rules for AIs but not for me.
There isn't one artist alive that doesn't have reference material on their drives.
If Google Images and Bing will ban you the same for crying about AIs using internet images you will see how fast the will shut up.

 

[Murrmor]

I agree with you about generative AI being primarily a tool for artists but I want to make some corrections. I'll be talking about Stable Diffusion but others work in a similar manner.
First of all what people call generative AI is a software that take a trained neural network file and uses process called diffusion to produce image guilded by prompt. The trained neural network (often called 'model') is just a 2-5 Gb file. Once trained on a dataset of images it stays the same. You can update it by feeding new images but the output will be the other version of the model so you won't lose the first one. There are hundres if not thousands of these models publicly available on civitai and similar sites. Almost forget to mention: you can produce new models by merging existing ones together in all kinds of way.

With just one of these models you can produce inifinite multiplied by infinite amount of images of different sizes. And that's just for a single prompt. The whole thing works offline because all the information about styles, objects, poses, colors, image compositions, all this information is packed (baked) into this single file. Even more. You can generate images over existing ones (just like I did with the image bellow) and that's called 'inpainting'.

I haven't even scratched the surface here. Possibilities are mind blowing and new tools are being invented every day.

Now about dark future. Right now there are more artists alive than during all the centuries of human history combined. And there will be even more in the future. Yes, creative industries (and not only them) will undergo dramatic changes because of this AI-revolution. But things have to be this way.

View attachment 2775810

this looks like every garbage ai art i have seen, overly glossy anime girl.... next

 

[129587328]

its shovelware

 

[Kevodi]

tbh $150 for 1 CG is actually a good price. I'm planning for a 2D game later, and I just hope I can find a good artist with rates that reasonable

not if you live in Brazil...