@Anilyan13

1 year ago

I have watched your previous videos on the topic, and while I agree with a lot of what you said, I think that you are not addressing two important concerns regarding AI art: one is about how AI art was fed so much copyrighted artworks from other people, and the second is more related to this video since by trying Midjourney instead of Stable Diffusion, you seem to think that your ideas are too specific. With Stable Diffusion, you can get that specificity, from determining the pose, to facial details, to the light direction and all of that stuff, and you can also make it iterate over pictures you provide it and control how much it deviates from the original, so... I think what you said in this video would probably change if you try out that one. Which doesn't disregard what you said about creativity, but still. I'm not the best artist out there (and not a professional, since I'm actually a programmer), and I like to write and tell stories too so my attention in what skills to develop is strongly divided, but I actually first learned about AI art thanks to an artist that did exactly what you suggested in a previous video: generate very specific references. It was a traditional artist, so we can say for sure that the final piece was 100% made by that artist, and because of that my first impression of AI art was really positive, and also everything I ever wanted: it was the perfect tool to make my art faster, and perhaps allow me to create things that fuse my many interests. Suddenly, the possibility of making a comic was within the free time I had, among many other things. But the more I heard the concern from artists (even though most sound like a broken record and don't really go in-depth, a few do), the more my perspective changed. I'm not going to discuss work and job opportunities because that shift will happen to probably every industry. To me, the one problem that is exclusive to artists is the fact that the AIs were trained on works from other people. For some reason, the same wasn't allowed for the music industry, and AIs that generate music were way more careful with the pool of works fed into it. Why are visual artists treated as inferior? I would say that a lot of what AIs generate still falls under transformative art, but... it's really visible that they know how to imitate the best artists out there, and things get worse when users can decide to ask the AI to imitate the specific style of an artist. Even worse is when the users intentionally steal, like what happen with this Raiden Shogun fanart (summary: an artist was streaming their progress, someone took a screenshot, finished the piece with Ai, posted it before the artist and then accused the artist of stealing from them since they posted first): https://www.youtube.com/watch?v=GgU7CipUH38&list=PLXdEvsQaYu-yoICo6n541Z3HOMbm_Y5u8&index=16. It's true that the blatant theft part is more a problem with the people than with the tool. But even so, the more I think of it, the more this situation sounds akin to appropriation. The way non-artists are calling artists of gate-keepers while taking their years of work and effort just to create derivative stuff, pretend they drew it, sell it, or post so many high-quality works so quickly on social media that there is no way human artists can compete for the attention of the algorithm are very concrete problems. If we let AI artists submit artworks in art competitions, they will likely win against human artists and that's not fair. People were already selling stolen art on etsy, but this adds another layer of complexity for the artists targeted to take that down. Some people using AI art are calling themselves artists just because they had the idea behind the prompt and imagination has its own value, and while I've been tempted to agree to an extent, now I realize that would be like saying that a client that commissions an artist to draw a certain thing is the actual artist - simply not true. And as for people who use parts of what AI generates in photobashing, I don't even know how to classify that since it still incorporates a lot of work. The name AI art itself is a misnomer, since this is actually deep learning. Again, it relies on the works it was fed and not on any intelligence of the machine. It's not thinking - it just learned that people have a preference for works similar to this and this artist, and so does its best to copy their style and apply it to the prompts. Of course some artists are tired and consider it unfair, even those that actually teach people how to imitate their style - because other artists would always bring out their own flair and every style, despite similar, would be personal. AIs don't have that limitation, so they are taking something personal and making it cheap and available to everyone. If the AIs being currently available hadn't been developed that way, it would be an entirely different story. For example, Adobe is developing their AI based on copyright free stuff and I fail to see the problem with that AI specifically, even though some of the problems above about art competitions, algorithms and such still persist, it's not a problem of identity theft or appropriation - just a problem with the priorities of society. There are many layers to this discussion and I wish more people were able to treat them separately instead of assuming a stance of "ai art good" or "ai art bad". I do think artists can make use of this. But I also think it's important to recognize that the current AIs available stand on a problematic foundation, and the more we rely on those, the more we are perpetuating the problem.

161 |