“No, wait, I’m lying,…”
Could be replaced with…
“No wait, that’s not right! I was confused/mistaken…”
Nobody would say (I imagine it being in a robotic, monotone voice) the following.
“I regret to inform you that my previous statement was a complete fabrication.”
Because that’s just as valid of a statement if one is to take the phrase…
“No, wait, I’m lying.”
Literally.
I’m pretty sure you know this, but… why fret over it?
Well, I don’t see a reason to. You do you.
Like I said originally, all they have to do is say they’re wrong, or that’s wrong. Simple, fewer words, and truthful.
It annoys me more when native english speakers have complete disregard for the use of then and than. It’s as if they use them incorrectly on purpose. Same thing, albeit not to the same extent, with there/their/they’re.
And when they write of instead of have. “Would of/should of/could of”.
I probably make mistakes here and there too, but at least I try to make a semblance of effort not to.
Don’t know if that’s an unpopular opinion though.
Regarding AI Art. Yes. While it could be argued that in its early days it borrowed or “stole” images from other sources that were produced by real people, at what point did AI develop a foundation or knowledge base from which it could work off of? Doesn’t AI have the ability to learn? It has, without a doubt, improved since it began.
It does seem ironic that a great deal of human artists stole… er, were inspired by the work of others so they could (hopefully) develop their own style. (There’s potentially a larger discussion here, but I’ll stay away from that for now.)
Anyway, my point is that, given the AI image generator, can we still say that AI steals others’ works? I’d think that AI has learned enough that it’d be nearly impossible to take an AI image and say that it came directly from X or Y image produced by a real person.
While it can be said that it steals from artists in that their talent wasn’t utilized to generate whatever image, by which they would usually (or hopefully) be paid for said work, that I can understand.
Pros of using AI: Super easy, almost instantaneous, free, and will usually be of high quality.
Cons of using AI: Might have small details that aren’t right. It is still learning after all. It might take multiple attempts to get it just right if not “close enough”.
Pros of hiring an Artist: Easy in that details can be discussed with an actual, understanding human, will more likely be of a high quality, have way fewer chances for mistakes like you’d get with AI.
Cons of hiring an Artist: It could take a long time, but hopefully, they will have an idea of when to expect the work to be completed. It is, in most instances, not free depending on the size or complexity of the work requested and the level of talent the artist has.
So I guess that would mean you get what you pay for.
I believe AI image generation has surpassed the point of mere “stealing” and has learned enough to generate original images. Is it lazy? Yeah. Is it the devil? I personally don’t think so. But I’m not going to point and laugh at anyone rebuking it or thinking it’s 
the instant they find out it was produced by something lacking a soul. 
If an image objectively looks great/perfect and you immediately fault it just because it’s AI generated… Sorry, that’s not a rational argument.
Not to be all Oral Roberts or whatever but the bible literally says that sloth is a mortal sin.
The problem isn’t that AI art “looks bad” or “steals art to replicate styles”. It’s the worry that corporations will start replacing humans with AI. It may look high quality, but it can never be. It lacks what makes art art, a human touch.
AI has no life experience, AI doesn’t know what love is, AI has never known hardship. Would you seriously sit down and read a comic generated by AI? Would you read stories generated by AI? Would you play videogames where a large quantity of the creative work was produced by a cold lifeless computer? I know I wouldn’t. Not only do I not see the point of spending my time on it, I also think it’s immoral to the countless artists out there who will be/has been replaced by this dumb thing.
If you can generate an image/song/text in a second it has no value.
Fuck AI “art”.
Learning is basically what makes an AI to fulfill it’s purpose, at least it is with LLMs that we talk about here. Feeding it data is the process of learning, that did not change. It also did not change that it needs data to grow, creativity does not emerge from that.
Improvements come from increasing the amount of data or improving the internal networks. There are theories that these types of models will hit a ceiling soon because they simply already pumped pretty much all data there is through them. Instead with AI stuff flooding the Internet, you get some backward loop where AI gets trained by AI stuff which is generally undesired as this can lower the quality.
Yes we do, I don’t think AI ever made it obvious from which artists the image was influenced from the most unless the user told it to. Nothing changed here. Just the amount of fingers are more often correct. 
Also, just because more sophisticated AIs exist do not make the less ones disappear. I still see more of the latter simply because they are free more often.
If we allowed AI art but only if you prompt it to mix it many styles and not an iconic one, that’d be impossible to enforce. And if the forum does end up being sued by an artists or their right societies then philosophical questions do not save us. I read here and there that the legislation catches up on the matter where at least AI art is not considered art in the legal sense.
In Germany we even have laws that assume you must have used copyrighted music and you have to actively proof that you used free-to-use audio at events or else you have to pay general fees anyway. (did not find an english source for that, it is called “GEMA-Vermutung”) With such an approach it does not even matter which artist the AI makers stole from when you are sued by a legal body representing enough artists.
Hell, even a single artist could sue us given how likely it is their art is in the training data of all big AIs and your work surely was influences in some tiny tiny percentage.
The whole matter of using AI art or hiring an artist for professional work is another topic, though personally I find it unethical to not only use AI art, but to make money with it. Saying “but I will be less likely to be caught by which artist in specific I used ‘stolen goods’ from” is, well, quite an insight if that is what you mean. 
I find it interesting that the people who defend “AI Art” as “making art more accessible”, seem to forget that to generate those images, you have to have either and expensive phone or expensive computer. You can get a pencil and paper from the dollar store. Or even take a pencil from Ikea and doodle on a napkin from a fast food joint. Or find some mud and make shapes with your hands.
Art has always been accessible. It might not be great, but it has always been accessible to everyone. No matter the skill level.
Even poor people are artists!
See the joke is that many artists are “struggling” artists, see
AI will never replace humans with anything.
It’s not a living being, it’s basically not real.
The word “use” could have a much broader meaning (or be more nuanced), but I do agree that making money from anything AI produces is scummy.
I’m certain that’s not what I mean in regards to AI-produced images. Again, in that AI image generators have LEARNED (in as much as any machine can “learn” anything) to make an original image that is not solely nor explicitly derived from any number of other images, regardless of their origin (be it other AI programs or humans). I realize that might sound contradictory, but in regards to anything that’s “similar” that’s getting into styles.
For an example, Norman Rockwell never painted a picture of Dinosaurs with cybernetic augmentations firing lasers on a glowing grid pattern battlefield with galaxies in the background (this is what I mean by being “derived from”) but an image with those items could be made to appear in a Norman Rockwell style.
If we’re going to say this and that looks similar to X, therefore “stolen” - I can’t agree with that reasoning. Take human artists who make photorealistic images. Would they have issues with each other claiming everyone else stole their art style? I would hope not.
Now… I did like your post up until this point because I felt it was misrepresenting what I was saying.
If tomatoes grow from a vine, and I have a tomato plant at home that bears fruit, that’s almost like someone saying I stole a tomato from the store simply because anyone can get 
from the store too. false equivalency? Anyway…
Similar to if someone were to use an AI image generator to make an image then someone else says the image was somehow stolen… from what? I don’t know.
Or are we suggesting that if/when someone uses an AI image generator that they’re doing it with malicious intent to pass off as their own, thinking they’ll get away with it? That’s why when you used the word “insight” I wasn’t sure if you were thinking I was justifying that as a reason to use AI.
Well, it wasn’t. I’m doing away with the notion of “stolen” and saying that images are created. Yes. They have a database of styles, but what else would there be to build off of or use as a reference when a human using it asks for (for example) “Something in the style of Picasso”.
If that’s enough to reason that the resulting image is then “stealing” from Picasso… Then I don’t know what to say. I refer to my Futuristic Robo-Dinos a’la Rockwell example.
I fear this becoming a reality too. Ultimately, that is up to corporations if they want to do good or evil.
It’s not so much that computers can work fast as it is efficiently. Yes. It seems “fast” to us humans, but I’d presume there are a LOT of calculations involved.
If superheroes… let’s say a ‘Speedster’ was real, and he’s also an artist. If they can work so fast and make whatever in what seems mere seconds, does it still lack value?
So the idea that if something is made (relatively) fast it automatically lacks value is, imo, not an argument.
I mean, what if a human artists gets the full idea of a work of art in their mind first? It might be a flash of inspiration or worked out over a period time, but there comes a point when the work is essentially finished. The only thing left to do is apply it to a canvass, or chisel it out of a giant slab of marble, or physically express it in the medium of their choosing. To take it from concept to reality.
You can be assured if said artists could create it in an instant - they most likely would.
I agree.
Rocks are not living things, but they’re real. And sorry to break it to you. Somewhere on the hard-drive of a server there are a series of 0s and 1s that together function as they were coded. Whether it’s a calculator or an AI Chatbot or Image Generator. Or maybe I’m misrepresenting what you believe the word “real” to mean. Is computer code real?
(Don’t answer if you don’t want to. It’s just something to ponder)
AI is nothing more than a tool. Photoshop is a tool, MSPaint is a tool.
We can worry about machines all we want, but it will always be the human factor that matters in the end. It can be used as a toy, but I don’t think it should be used to replace anything a human can or would be able to do that would benefit society.
Or would that be wrong? Like if an AI program discovers the cure for cancer? 
Sure, if superheroes were real and could draw insanely fast, that would have value. That isn’t the case with AI though. It really isn’t about the speed of generation, it’s just that AI can never deliver quality and by extension it’s worthless. When you can generate a billion images in no time, how could they ever have value?
Then it wouldn’t be art. I’m fine with using computers to screen for medical issues and develop cures if that’s a possibility.
Interesting aspects, though I feel I need to get it down to a more technical perspective to point out the issues with AI art:
Human artists get influenced by many things in nature, life, media and so on. If they look at a Picasso, they get in touch with his work on ways where Picasso (or any other, to make it less abstract, living artist) does often not get by-passed in their rights.
For example a museum likely has the permission of artists (or their descendants) to show these pictures. Or the website you saw it on often also is allowed to share the image in that resolution under that specific license. Important here is that it is not a sure fact that your source is providing said images without consent of the right holder.
While we don’t know what a soul or consciousness is, we assume that the human artist brings in a personal perspective into it which makes the art emerge. In contrast to a person who plagiarizes specific paintings that do not look different to the source.
AI models may work different, depending on your stance of if there really is a soul. But what surely is different is the role of the AI creators.
AI these days are working by artificial neural networks. You probably saw images like these before:
The left circles represent the user’s input in some encoded form. The blue cycles take the input, do mathematical calculations based on them and the weight value of themselves to pass the data on. Usually there are many more columns and rows of blue nodes where one’s output is another’s input. Green is the produced image.
(if you cannot see it well, the connections are arrows pointing to the right)
The training process of an AI is to give it images and textual descriptions of said images, usually crawled from the same source and, due the amount of the data, without much manual verification. That way the AI learns which text results which image.
The result of said training are the weight values of the node.
So when you say AI got so good that you cannot always tell which artists it was influenced by, or that it even is AI art, you basically describe the benefit of having many blue nodes and very fine tuned weight values.
That’s it. That is the main difference to cheap AI models. Not more, not less.
Now, that might be not that much different to the human brain in principle. (Though not in fact, our brain has magnitudes of more neurons - in number, complexity, and connections)
What does make the difference here is that the AI was trained by people with purely commercially interest and without personal creativity on what the AI is used on to create.
To archive an cutting edge with their AI model, the creators need to feed it as much data as possible. Interests of the artist are so much in opposite to that that they are ignored. That is part of the reason not only the weights but also the training data is such a well kept secret for a company as many assume the data is basically an illegal collection of copyrighted material.
But given how important AI is to the economy (of a country) there is not much interest to protect the right-holders against that.
So tl;dr, the AI is just a machine that in practice was made from stolen goods and is commercially offered for use. There is no creativity or creation from thin air that is not based on these stolen goods.
On an abstract level, the set of weights represent the copyright infringements, like an illegal copy of a song is represented by the bits of the hardware medium it is stored on. So when you make an AI picture, these weights are inevitably producing your image! It is entirely unimportant if you make it create a Picasso or a Norman Rockwell variant of a cyberpunk dinosaur.
The counterexamples you list do not have this aspect, for example tomato seeds are distributed in the world willingly. For some goods that aspect may be questionable, like how silkworms were basically stolen from China. But just like history made that “okay”, AI generated art may be declared “okay” as well. It is just that HMF does not want to rely on that and we also decided to have a personal stance on the matter as well.
That the AI model users do not have any bad intent or did not pay for that (sometimes), that surely has an effect for them on a legal level. Ethical level aside for now. But this platform here is adding to the distribution to the images. The forum software does even mirror embedded images, so we even host AI art if we allowed that.
Regardless of how innocent you are, it can be problematic for us if stolen goods somehow, in some form, found it’s way in what we knowingly provide to the public.
That we simply don’t want that in respect of the artists comes on top of that.
That actually is an interesting argument, but not in the way you think!
I read an article that argues that AI art should be legal because it behaves like photographs of nature:
The photographer did not arrange the mountains, the trees or the rivers they take a picture from. They also did not place the sun in the sky for the lightning. They also did not ask for permission of these things to be photographed. They “just” aimed the camera and took the shot.
The timing, position and rotation of the camera is not that complex, you could store that information in a small amount of numbers. Digital cameras even do that automatically.
How can that be art?
How can AI images, that also were generated by something that was simple taken, based on it’s own set of numeric values (here the user input, not the weight values), be not art?
That was the gist of the article. However, I ended up not agreeing to it because I see a difference between photographed art and photographed nature. The latter just is, it has no artistic value before it is added by the photographer. But the first has that.
When a landscape photographer walks into a museum and does a photo of the landscape that is a well-centered Picasso, then that is not a landscape photo. And the photographer has not the right to sell that photo as their own creation.
That AI images are not a well copied original art is, I think, just a technical difference between AI and cameras. Not an aspect of art.
I think you can transfer that to photo-realistic art as well.
That is a perfect example that the technology itself is not bad or good. Already today machine learning provided indispensable results in protein-folding which is used in making medicine. I am almost certain that AI will save billions of life in that sector.
Another example that is already tested successfully is the control of plasma in fusion reactors.
The difference to AI art here however is the training data is entirely different. Protein structures or sensor/actor data are not copyrighted goods.
Chop! Chop! Dad slave!
Kinda sorry for quoting/posting this here, but it would’ve been off-topic where it was originally posted. I just have to say that my jaw dropped when I read this.
Heck! If the person who was complaining about their parent being a “hard-ass” expecting them to get a job, the least that person could do is pull their weight and take care of chores. I don’t know if they do already, but that’s not the point. It’s implying it’s the job of their parent. Like they don’t already have an occupation.
I could go on, but then I can’t help but to change the focus on the person that was complaining (and saying what they should or shouldn’t do). And I’d rather not get into that. Because that’d be like giving advice, and they apparently have no interest in getting any.
The idea here is, once someone is an adult, it’s none of their parents’ damn business what’s going on with their life. And if they’re still living with them and feel their adult child should contribute and they’re not, they can kick them out. The point being, they can worry about their own affairs, and take care of their own matters, which they would have to do anyway if they didn’t have the free labor of their children around.
