Many users in the SunVox community and Mastodon have already known I am studying Artificial Intelligence because I have been being interested such topic for quite awhile, while I have found some problems to the current AI models, having some idea to solve; however, there is a topic I don’t really like, and in fact, I am quite against it:
The Infamous AI Arts
Some Wojaks: “NOOOO, YOU CAN’T DISLIKE AI ART IF YOU ARE AN AI MAJOR MASTER STUDENT!”
Yeah… nah. I have two distinct reasons why I don’t like them:
The first problem of AI art is how they obtain training materials. For each artwork done by a human artist, they already have the ownership because ownership exists once the artworks is created, and you can’t assume that are non-existent; however, AI models treats copyright like nothing, ripping millions of copyrighted works for training data, which is a shitty move because unlike jobs require little or no skills, artists need years and years of training to master the skill to draw something impressive, while they take many hours of brainstorming to come up with their cool ideas and stories. AI arts not only nullify all the hard-work from artists, but also give them a big middle fingers by not crediting them. (while sadly, it is impossible to locate what exactly art were taken due to the quantity.) The most important quality of Engineering are not only how precise the calculation, but also ethics. The most important rule about being ethical is NOT to do something to hurt yourself, your team, your customers and public. Clearly, there is only one stakeholder have benefits, the company itself which they have received profit; meanwhile, government need to face AI generated fake news because those company don’t care about how the model are used, and artists being threaten for their job security and copyright because how those companies steals artworks, publics need to face the enormous amount of low quality generative contents. I don’t really seems to see any good about AI content for now.
Let’s assume if they are a complete saint, and they have credited all artists and being the most ethical company in the universe, AI art still don’t click me because… they feel… empty. For some reasons, I can sometimes spot which one is created by AI, and oftentimes, they are a meh for me. What makes an art “art” is about the ideas, stories behind the artist. From traditional art like Mona Lisa, to some simple modern fanart, they always have a story, from as big as some historical events, to as small as a fan having the passion to their franchise; nevertheless, AI audio and images lack all of these. For example:
This is a black dragon standing on a plain at the full moon night, and…
.
.
.
.
That’s it, other than that, I can’t think of any story for this Drageneric Art * beyond what it has shown. I can’t feel how the artist were thinking, how they brought the idea into life, how and what they have learnt during drawing their artwork, or what kind of challenge or hardship they were facing. What remains, is a lump of impressive visuals putting together like a Frankenstein, with no personality and character because everything is interpreted from the same neural networks or diffusion models.
Some people have taken the meh-ness into the next level, by batch generating the same input repetitively. As you can see, someone created a large quantity of “Dark Knight with pointy helmet holding a gold growing sword, with greenish, fragmented background”. Everything looks like dark souls, but there is no soul within because they all lack a story unlike how people used to design their OCs with background, stats, and abilities; it doesn’t matter if you generate 1 image, or 1 Gogool images, the lack of story to the AI contents are always the problem.
You may ask: “How about giving some stories to the AI art, isn’t that will give meaning to it?” I don’t know, this is the question I can’t really answer because so far, most of the AI images and audios I have seen are either batch generated, low effort spams, or they have significant flaws and style that are enough for me to conclude that is an AI work, ruining the whole experience. Even if they are hided carefully, there must be some ways to tell if they are created by an actual human, turning from seemingly awesomeness into a generic meh-ness.
Anyway, these reasons above are why I won’t use AI for my art because Midjourney, DallE, Suno or Udio are not me, nor any artists which they never know whoy ** we think and our experience. They can become so proficient on every topics, but they can’t experience our experience since everything behind is just some complex calculations and selections over an enormous data set, with no sign of creativity thinking and story telling. If they can make a music that is close to my style, they can only interpreting my story, but never really have the full representation.
I used to hear people saying “I don’t like <genre> because it has no soul.”, and I used to quite against the quote because everything had to be made by someone, which the music can reflect their thinking at the time, even pop music (Their desire of becoming successful, dance for celebration, the frustration of breakup, or etc, which I can actually name quite a few even I am not a fan of them), but now, I can’t believe the quote is valid due to the existence of AI Audio and Images.
Hopefully, I wish others to know that even I am an AI student, I stand for the artist side because the so call AI “arts” not only lose the real meaning of arts, which there is no sense of story telling and connection to artists, but also they are spamming everywhere without creating any values but distractions.
* Drageneric – Draconic Meh-ness and disappointment
** Whoy – What, How, Why in superposition
amen! what ai art and music will always lack is the human emotions and experiences that led to it. it’s less like art and more like data analysis. just soulless
Bingo! These are also the reasons why I don’t really afraid of those generative AI because I see making music like writing my own diary. Can ChatGPT replaces your diary? I don’t think so. btw: I think I have finally fixed the comment section; hopefully, texting experience will be smoother.