Raw LLM Responses

Sorry but I think this video is very harmful. I was hoping that this one would cover the actual problems that come which AI. This would have been a great opportunity to think about the future and how we as humans will deal with machines taking over the one thing we thought of as inherently human: creating art... But no, not even 2 minutes into the video and there is already pile of straight up misinformation and the spreading of a dangerous narrative. I know this sounds like a bold claim, but it all comes down to this main issue: The idea of violating copyright because the model was trained on a dataset that includes copyright-protected art and using these images without the artist's permission is absolute bullshit. If these images were actually part of the AI and if it would take parts of these copyright-protected artworks to produce new images, I would agree with you. But that's not how these models work. Every output is a completely original piece made only by the AI itself. The model was only fed with the dataset to _analyse_ certain parameters of the images in combination with the description of the image to learn how specific things look like and how it has do draw them. So the whole process of training the model is in no way different to how we as humans learn how thing look like and how to draw. Now to the argument of breaking copyright/"stealing" artworks/treating artists unfairly because of using their data without consent: As I already explained, the AI does not directly use artworks in its output images, instead it works exactly how we humans learn. So an AI shouldn't be allowed to analyze (publicly!!) available data, then it would also be illegal for us humans to look at _any_ other people's art or search for inspiration online as these artworks would also affect our own ones. This argument is especially hypocritical as almost all artists took inspiration and learned from others and I find it incredibly funny that this argument often comes from Youtubers who love to draw famous characters (that are often also copyright-protected in some way!) or make videos in which they draw these characters in different art styles from various artists and art studios etc. This brings be to the next point: "Copyright-protected" does not mean untouchable for everyone else. It only means that you have _certain_ rights to your artwork and these can vary at different degrees of copyright protection. For example, if you took a picture privately, noone but you or someone who has your consent is allowed to make it public. But if you already uploaded a picture, this type of protection obviously doesn't apply anymore. By uploading your images to the internet, you give up some of your rights. But even this can vary from platform to platform. Platforms like Youtube, Instagram, Whatsapp etc. have terms and conditions that could allow them to take more rights from you and -for example- sell the rights to your image to third parties. But generally, if you share your art on a place where everyone can view it, everyone _is_ also legally allowed to view it, including AI. Additionally, even if there is copyright-protected art, you may use it under certain conditions, usually under "Fair Use" or other conditions of artistic freedom (which by the way actually exists and is a basic right). If there are cases that these datasets themselves violated copyright by publicating private data, this would of course be a legitimate issue, but among the billions of images in these datasets, I assume it's only a tiny, tiny amount. Now, there is still an argument that you could use against me: What about the AI almost identically replicating famous artworks? Yes, it was shown in the video that an AI is capable of giving us almost identical copies of a famous painting if asked for it. First of all, many use this as "proof" for AI directly "stealing" artworks. This, however is incorrect. As I explained, the AI cannot simply take images and spit them out as a result as the model only _learned_ from it. So it's not like the AI is pressing ctrl. c + ctrl. v, it's more like an artist recreating a painting from his memory. But absolutely agree that this is an issue and we need regulations to prevent AI from directly recreating a piece of art. We can all agree that a tool as powerful as an art AI needs regulations to prevent misuse and potential harm. That's why DALL-E for example doesn't let you generate some "inappropriate" or offensive contents. It is definitely important that we as society have discussions about how we should deal with these new technologies that have the ability to start a new new era of human progress but could also cause great harm. But the spreading of misinformation and false narratives that only serve to discredit the opposing side and retain one's own position without actual argumentative superiority does only harm this debate. I'm sorry if I sounded too impolite or insulting at the beginning, I tried to keep the rest of the comment as formal as possible, but I'm just very annoyed by how many people actually believe in this misinformation and resist any logic because they are so emotionally affected by this topic. Anyways, if you have actually read my comment this far, thank you and have a nice day!