A big part of my appreciation towards the arts is the creativity behind it. Obviously I am not impressed with an internet-searching AI when compared, unless I am specifically interested in the code of the AI
How much do you know about the code of AIs? There seems to be an idea that they just photoshop together stuff they find off the internet, which is nothing like the actual situation.
As a CompSci major I at least have a great idea of what’s under the hood, so I can appreciate the creator of the AI and those that helped push the field to its current limits.
Well, you should know in that case that AI art generators don’t “search the internet” when they generate images. I have a local installation of Stable Diffusion and it works perfectly fine with the internet disconnected.
To be fair: to create said local model on your pc it had to get trained on thousands if not millions of accurately described images. You only install what the model has learned on your computer- not the training data.
Sure, but the person I’m responding to literally described AI art generators as “internet-searching AI” and that’s a common misconception of how they work. Even among people who confidently claim to know how these things work.
Yeah - didn’t read his first comment… CompSci Major…
Sorry…
It’s okay, there are a lot of misconceptions out there and what you said was also informative and true. So hopefully it all helps.
It appears I did a lot of damage by over-generalizing in my comment. My bad.
Either way, that Stable Diffusion was trained on shared data, not that it literally requires Googling and scanning pictures.
I wasn’t trying to make it a lesson on AI so my bad! Thanks for the info!
Comp sci majors are perfect examples of the dunning Kruger effect when it comes to anything tech related.
I judge art more by how it affects me. But then again i do separate beauty / something that i can find meaning in from art by an intention. And AI has no intention, or an intention to create at least. A spider web covered in morning dew is beautiful and can contain meaning but its not art.
Art is incredibly subjective and comes in many forms, I consider AI art to be just another form of art. And as a person who can barely draw a stick figure, a form of art that I can actually use to get ideas out of my head for once lmao
How do you judge how much “creativity” is behind any given image? You say you’re “obviously” not impressed with AI-generated art, but how is it so obvious to you which images are AI generated and which are not? There are plenty of quizzes online that challenge people to identify AI art, this one for example, and it certainly doesn’t seem obvious to me.
Based on my use of AI chatbots for story writing (chatgpt not included for different reasons), I personally have a somewhat negative view in general of the fact that, to me at least, the stories are average quality. Another downside is they pretty much all end with a happy little moral like message like “Benny found from that day forward he would work extra hard to make sure to do the right thing after he realized his mistakes.” Almost always, except with my early tests with chatgpt, ended with anywhere between a sentence or a short paragraph with a happy little message.
Also, I am also one of those people who is less likely to like your art if it looks similar to an AI generated image/style because I absolutely do not like how soulless it feels compared to art an actual artist has made.
It’s because you are using a chat/instruct fine tuned version of the LLM.
Don’t think that represents what is actually currently possible for creative writing with a SotA LLM.
Part of their fine tuning step was literally to make it come across as more soulless than it was in the pre-release version, which was still a chat/instruct model (just an earlier version) and yet coming across as too human-like to users.
deleted by creator
Thing is, most art by people calling themselves artists is still soulless, pretentious shit. With AI at least it’s aesthetically pleasing.
Also, your experience with AI chatbots might be a problem with prompt engineering. Thing is, these generarive AI’s results are directly correlated to the effort you put in precisely explaining what you want in a way that it understands.
6 fingers ._.
I’ve definitely been trying with my prompt engineering, but I guess I’m just not good enough.
AI created stuff is sometimes quite good and tops low level writing in pretty much every case … I personally think that there are many ways we can profit from this culturally.
We have to fight all those greedy assholes though, that think they can now put wages under even more pressure because of AI.
It’s always a gamble reading comments on posts like this.
You’re almost guaranteed to find comments stating/implying that ALL AI art generators have the exact same coding, for better or for worse.
These comments tend to either be along the lines of “AI never copies other images” or “AI only copies other images”.
Just like any other type of program, YMMV. One program might have had a lot of time put into coding and creating it, but the next program on the list might have been rushed. One program might have copyright protections in mind, but another program might come from somewhere with little to no copyright laws.
Never change, internet.
It’s a tool to automate low level stuff. I think it’s good to be honest since any newcomer would now automate these things and focus more on the creative decisions.
It raises the floor. You can’t do the bare minimum of doing the basics right and coast by since Ai is there to take care of the basics.
Hah! Of course we do. It’s like how people love cheap swill poured into an expensive wine bottle.
An apt comparison given that “wine tasting” is bunk.
When it comes right down to it, subjective experience for inherently subjective things like taste should be all that matters. Million-dollar price tags for art or bottles of wine are purely performative, things having high value simply because everyone has collectively agreed that it should have high value (often for purposes of tax evasion or social standing).
If there’s a bottle of wine or a piece of art that I enjoy the experience of, what does it matter how it was created?
so you’re saying human made art is cheap swill?
I mean, the study is kind of saying that. I don’t have a particular reason to doubt the result, since I’ve seen some pretty spectacular AI content.
Edit: On further thought, maybe it’s more that human content is a mediocre wine, and AI content is a similar but maybe slightly better wine, but “human” is the expensive label.
it’s definitely better than the average “artist”, even the average artist, but the top artists are still leagues above. Anyone saying otherwise is lying to themselves, as evidenced by this study.
What I hate most is that this discussion is so black and white. If any amount of AI is involved it’s allegedly garbage. You can create 99% of an artpiece on your own, add some detail with AI and it’s automatically garbage.
Makes no sense. People will just have to accept that the nature of making art has changed forever and realize that it’s not about the process, but about the end result. And if you like doing art 100% manually, no one is stopping you.
You average person won’t be able to create good art even with AI, as evidenced by the AI art communities on Lemmy. Being an artist means having an eye for beauty. You can’t teach that.
That’s one of the main reasons why people think AI art is garbage, because it’s mainly created by people with skill or zero taste. In the hands of real artist, AI art turns out amazing. I wish I had anywhere near that aesthetic gift.
Any idiot like me can tell you if art is good or bad, but creating good art is an entirely different beast. It’s extremly rare, even among professional artists.
That still requires the human touch. AI can’t come anywhere close.
How have we arrived at the situation where machines are making art while people are toiling in shit warehouse jobs? Shouldn’t the machines be working for us while we relax and make art?
I mean it makes total sense when you look how much wages are in those two areas. Creative fields tend to pay well when you’re successful at it while the warehouse jobs are low paying. Makes sense for a corporation to want to get rid of those higher paying jobs if possible.
There is some logic to that argument but it rarely applies to the top management. In my field, ten thousand engineers will make some cars but ten thousand managers will make zero cars. It’s something I regularly remind them of.
We’ve been told there is something special about our management since the dawn of time. It used to be warlords, barons, kings and queens. We were told they were backed by divine authority. Meanwhile, people like Shakespeare were creating the real stuff of value and we don’t know what he even looked like.
AI is better but it’s made by cheating. No good model was trained only on content that was public domain or properly licensed for use, so the images it produces are effectively a blend of traced artworks and never an original work.
Good thing artists never look at other peoples art.
Going to an art gallery is literally plagiarism!
Humans, unlike computers, are incapable of completely recreating entire complex images from memory. Stable diffusion can, without even needing to be explicitly prompted to do so.
completely recreating entire complex images from memory. Stable diffusion can, without even needing to be explicitly prompted to do so.
Learn how stable diffusion works then.
For starters, it can’t recreate a piece of art 1 to 1 even if you go img to img.
Yeah this doesn’t shock me. Generative AI is gonna be trained on the best art possible, so of course you’re gonna get good looking output… until you realise the thing that created it doesn’t actually understand 3D space, or find other imperfections that reveal it for the thorough cargo-copy it is.
u tol you realise the thing that created it doesn’t actually understand 3D space
You might find the following paper interesting as the reality is a fair bit more nuanced than you might think:
Will give that a read in the morning, thanks. I am only talking about the generated art I’ve seen, which often features a clear lack of understanding of 3D space. When I see generated art that shows understanding, I’ll be impressed.
Ah, you mean diffusion models (which are different from transformer models for text).
There’s recent advances in that as well - you might not have seen Stability’s preview announcement of their offering here, and there’s big players like Nvidia and dedicated startups focused on it as well. Expect that application of the tech to move quickly in the next 18 months.
Yeah, I didn’t thin LLMs did art generation.
Actually, the transformer approach was just used with some neat success for efficient 3D model generation:
That looks cool :)
