glad I switched to linux!
article under that one, has mentioned that Dropbox compromised content to AI.
I had used it previously for a project to send to a client, Thats dissapointing.
First thing to do would be⦠How to turn off that copilot ai thingy. From that article, M$ is treating their customers like a cash cow. Iāll pass and keep what I have till it dies.
As someone with a cartoony style that has been baked into several base AI models I can assure you AI knows how to do cartoony stuff.
Itās weird seeing your name up on a big wall of artist names with generation examples beside it so people can pick and choose what style they want.
Iād honestly call it flattering but I completely understand if others donāt see it the same.
Ugh, another Windows version. Itās annoying, especially if you have to deal with anything IT-related
Hm, but Iām wondering how pervasive this AI integration will be? I canāt imagine not being able to opt-out of it if you are based in the EU. Iām not the most paranoid user, but having all of my files automatically shared with a third party sounds like a no-go.
Hereās 2 more articles:
https://www.behance.net/gallery/159290265/No-To-AI
https://medium.com/graphic-language/my-case-against-ai-ad6489e124f2
I understand there are a lot of problems with regards AI but I have a question. One of the issues seems to be that AI examines the work of real artists and then uses their styles in new images that people want created. Isnāt that only the same as an artist being influenced by another artist and copying his/her style? I know it is an issue that the person using the AI doesnāt have to have any skills or talent and there is a lot involved in that but with regards using the work of existing artists, donāt we all do that anyway to some extent?
The problems Iāve seen around the web is that the original artistās name was put to such work and the original artist is understandably upset because the promptor sells it as work by that original artist. The AI promptors are profiting from the original artistās name/reputation.
Owch! That is bad.
Additional to @CrazyCatBirdās point, which is a very severe one:
If I ācopyā a work I have seen once, then I must have the skill to copy it from my imagination to be able to deliver a good copy. If I rely on reference-pictures, the task gets easier.
But usually we give any picture our own touch, our signature if you want, and Iām not speaking from the letters you write as signature.
Many artists may be able to paint a picture that is very near to the original, but only very few are able to paint using the same signature as the original painter did.
But we live in the world of digital art now, our pictures are not on canvas in the beginning, where you can tell by the stroke, the used paint, the canvas and a few more things, like underpaint, which painter made it. This is the reason why most copies of Rubens, Vermeer, van Gogh, Picasso, ā¦, can be revealed as a copy.
AI is able to do this with every artist it is trained on, and it is able to copy them perfectly, because the material layer, the matter, is missing, this is the difference. And for digital artists, this is the horror in person, especially in combination with @CrazyCatBirdās point from the post above mine and yours.
Michelist
When a human makes a copy they do not say that they created it. Humans are too called out for plagiarism. Nowhere there is a free reign for copying artwork in human world. So when AI proponents tell that AI is same as human then human law should affect AI as well. So if AI makes a copy then it should be called out too.
But then some people say it is not a perfect copy yes and when a human makes a copy it is also not perfect copy. Still humans are called out for plagiarism. Disney and some other companies go after artists for making fan arts.
But the point isnāt to copy. The point is to imitate the style while creating something entirely new. That happens all the time with humans in all artistic fields. How would it be different if you imitated Daliās style but created your own image? The only issues I see are when it comes to putting the original artistās name to it, when claiming it is your own work rather than AI or if AI takes work away from professionals. The use of styles doesnāt bother me at all.
The issue is not about styles. Styles canāt be copyrighted. The issue is impersonation, using copyrighted images without consent to train a machine to make profit by selling license to use the machineās output or to gain stock valuation from it. We canāt compare machines and a human amateur trying to copy dali.
We simply canāt compare how a human brain works to a machine doing mathematical computations. It is a false equivalence spread by AI proponents. Human being not just copies artwork they also draw based on many other factors such as their emotions /mood like fear, happiness, or life experiences. AI just finds/predicts the position of a pixel in the image based on probability.
The issue is targetting an artist by taking their work and specifically training Ai to damage their market value (read about samdoesart)
If the Ai companies took effort in getting the data ethically then there wonāt even be a debate.
I have said that I agree with you about impersonation but that is extremely rare. The main reason people use this technology is to create their own images. Artists use copyrighted material when they are influenced by another artist. I have looked at many copyrighted paintings and am often influenced by them when I paint. AI isnāt doing anything more than that in most cases. If you are just going to copy a piece, why bother with AI? Why not just download a copy of the original? What do you think of all the artists who have done their own versions of the Mona Lisa?
There are several things we have to consider here.
For starters, as someone else already pointed out, a human making a painting thatās too similar to another artistās work, especially without giving credit, is generally frowned upon in the art community and prosecutable by law.
The particular case of artists making their own versions of the Mona Lisa is not really a good example of what usually is happening with AI generated images. The Mona Lisa was painted way before any IP protection laws were written and, in any case, any copyright would have expired a long time ago. Also, everybody recognizes the Mona Lisa and anyone would recognise an artistās own version of the Mona Lisa as that and not an original concept.
Letās think of a different case. An artist finds a relatively unknown painter contemporary to Da Vinci, imitates his paintings and sells them as his own original work, hoping that nobody will catch on. What is your moral evaluation of that action?
Lastly, and this has already been pointed out, likening the AI image generation to an artistās creative process is a false equivalence. Letās not kid ourselves here, we donāt really know much about how the brainās creative process works. Even if you paint your own version of the Mona Lisa, there are many non-copyrightable factors that influence you, your own emotional response to the original art will change how you interpret it, and you have a plethora of life experiences and real-world observations, to name a couple of things not accessible to an AI, to inform your creative process. On the other hand, the only thing an AI has to inform its ācreativeā process is billions of images stolen from the internet.
Your arguments are flawed. Monalisa is not under copyright. Moreover I already mentioned an artist copying work is still called out. Nobody is giving free pass to humans doing this. And you getting influenced by painting to create your own unique painting is not the same as AI. AI is not in any way similar to your brain function.
You also conveniently left out the point about corporation making profit by selling the service which is built upon the work of artists without consent, it rules out fair use.
Yes, unfortunately the AI discourse is full of flawed arguments making parallels to how a humans learn from other artists. Itās all nonsense.
The main motivation for the generative AI is profit. If someone canāt make the art themselves, they would have to seek out someone who can. Now they donāt have to, they can create a generated image, or ask someone who knows how to do it at a fraction of the price and effort. There are so many slimeballs out there trying to make a quick buck in whatever way possible.
Another issue is that of scale and efficiency. The tech is disruptive because of how easy to use and quick it is. If someone can generate dozens of decent quality images that could be passed off as the real thing, then itās no longer a harmless gimmick.
If the AI tool will get as good as a professional illustrator, and it will be possible to achieve the desired design and pose with it, make no mistake, no artist out there will be safe. I understand that some jobs become obsolete and new jobs replace them, but I think it would really be a crying shame in this case.
One thing here is that these computational monsters are only able to imitate artist X, Y or Z because they have been trained on āillegally acquiredā image material, i.e. stolen from artist X, Y or Z.
If the pixel conglomerate generators had only been trained on legally acquired material, then the manufacturers of these pixel conglomerate generators would probably still be looking for a sufficient amount of training material. But above all, they would not be able to imitate modern styles and art movements, because the rights to current image material, in contrast to da Vinci, Rubens and co. have not yet expired. And if the heirs of Picasso and other modern artists were to have the rights to the pictures painted by their ancestors extended, as many do, then the manufacturers would still need ages to teach their pixel conglomerate generators their styles.
Here comes the point where the snake chases its own tail to feed on itself, if it did so, it would be suicide.
Because if the manufacturers of the pixel conglomerate generators had wanted to hire enough artists to try to imitate these works in sufficient quantity to create training material of a usable quality in the quantity they needed initially, they would not have had the money to pay the armies of artists needed for this, because they would still have to sell their pixel conglomerate generators in order to earn this money to pay the artists. It probably wouldnāt even be enough for a sufficient number of third-rate artists, who in turn could hardly deliver the necessary quality, at least not en masse.
Who would work on something of which success is not certain, in the hope of being paid on success, if success serves to make you redundant and therefore unemployed from that day on? Morons?
Because any artist who can paint pictures in the quality and quantity required for training would be foolish not to sell them to well-paying customers instead of producing them in large quantities for dumping wages in order to make themselves superfluous in the long term. The manufacturers of the pixel conglomerate generators would hardly find enough artists who are good enough and yet completely stupid enough to willingly and with all their might participate in their own abolition.
So how could they spend something that they still have to earn?
At a time when it was not clear whether it would be a success, vast amounts of money would have had to be invested? For pictures painted for the dump? As we all know, they did not pay a penny.
To avoid committing suicide, it went on:
But the manufacturers of the pixel conglomerate generators took the easy way out and, in order to be able to earn money at some point, opted for theft, then committed the biggest art theft in history.
Since then, with the support of those who profit from it, they have been trying to make this theft-based product appear good, useful and worthy of legalization.
And as you can see, they always find people who donāt attach any value to this crime, because maybe they could profit from the result.
Itās all about acquiring something in the cheapest possible way, without personal commitment/effort. It doesnāt matter how it came about, only the creators stole it, Iām just using it. There are thousands of attempts to trivialize it, why do people defend it? Better I excuse it now, maybe it could be useful one day, right? Itās not a bad thing, is it?
An āargumentā like: āNow anyone can create artā is the joke par excellence, everyone could do it before, only many were dissatisfied with their results and preferred to do something else because they were too lazy to practice, practice, practice. Iāve been practicing since I was a child, and Iām still below average, so what?
Those who prefer such generators do so either
to make money commercially by replacing flesh and blood artists with PCs,
or to create something that they themselves are incapable of or too lazy to do,
and the ātoughestā do it to create something they themselves are incapable of and claim āI created thisā,
and the brazen criminals among the ātoughestā do it in order to then sell the āoutputā at a high price as an unknown work by well-known artists.
Only very rarely do artists who have trained the generators on their own visual material use it. But here, too, there is still a very big BUT hanging over it, because this is only possible because todayās generators have been optimized long enough on the basis of stolen material, only possible because of that, so that today smaller amounts of data are sufficient for training, but an artist must still have painted vast quantities to achieve something appealing. Thatās why images based on AI trained with its own material have a salty aftertaste.
The creators of these generators have made sure from the beginning that EVERYTHING produced with the generators has a very bad taste, probably for a very long time. Even things that may be okay.
So now we have millions and millions of wannabe artists flooding the art platforms with their computing results, pictures of real artists get drowned in these floods, making the artists unemployed and their pics worthless, we have agencies that have replaced parts of their workforce with computers ā¦
ā¦
Michelist
This sounds interesting: