AI crap in compo entries?
category: general [glöplog]
Quote:
Honestly can't remember everything in De Profundis right now - but in general, a pastiche can be a powerful thing in a demo. Though that's not the same as pixeling over (as in doing a 1:1 copy).Magritte
@ham: Try this.
Quote:
Quote:Let's get back to the point about the technology itself vs. its current implementation.It is about a trend, if you can call it that, and it feels people see that.
The main problem is not remixing prior art in an automated process, but that the data corpus used to do that was obtained without the consent of millions of original authors, yes?
In other words, AI-enhanced remixing of your own material is okay, as is remixing of some properly-licensed set of material, yes?
Or is there some more general objection to the technology itself rather than its current use?
i personally have no beef with the tech behind it if the training data is ethically sourced. the problem for the AI generators is that the tech is currently only useful with unethically sourced training data. the same tech, trained on just the output of one person, will not produce usable images because it needs the millions and millions of images to come up with coherent pictures. people have tried that!
@rexbeng:
I will not talk about the prod anymore. Because someone will say that I “reiterate”. I (and perhaps some others too) thought the conversation went in more general direction. If this thread should be about the prod and is going to be about the prod, I urge people to chime in and tell me so. If so, then I’m off. I really don’t have anything (more) to say about the prod. I regret bitterly having seen it in the first place.
I will not talk about the prod anymore. Because someone will say that I “reiterate”. I (and perhaps some others too) thought the conversation went in more general direction. If this thread should be about the prod and is going to be about the prod, I urge people to chime in and tell me so. If so, then I’m off. I really don’t have anything (more) to say about the prod. I regret bitterly having seen it in the first place.
@wayfinder: Check out the example I posted earlier in this thread. It can be used for creative work even if it's trained on a small set of images.
But yeah, the thing with "hey, algorithm! gimme a nice picture of a cute animal that smokes pot" isn't really a possibility.
But yeah, the thing with "hey, algorithm! gimme a nice picture of a cute animal that smokes pot" isn't really a possibility.
Quote:
i don't appreciate these words being put in my mouth (as you're saying that's how we all think). you seem to have built up a kinda rickety thought construct in which everyone who does not in every even tangentially related discussion renounces any and all instances of morally dubious behavior is wholeheartedly condoning it instead. and not just they, bt anyone else arguing their side in an unrelated issue? ... which in turn somehow makes other transgressive behavior not just okay but actually gives the perpetrators the moral high ground? anyway, it's pretty shoddy logic and you should stop using it.Wait. So, pixeling over an AI generated image is a no-no, while painting over other people's work is totally fine? How 'craftmanship' and 'honesty' are served in the case of the latter but not the former? (hint, De Profundis X Die Toteninsel iii)
copying other people's images is a useful training exercise, but presenting them as one's own work is not good. i've made mistakes like that and i'm embarrassed in hindsight (for example, the soundtrack for fr-027 was an undeclared cover, which I thought was obvious but have learned in the meantime was not). i *think* i have learned from this event, which i still consider a dark spot on my resume.
the complexities arise around how much of a cover or remix or homage or citation or allusion exactly there is, and how much is merely an unchanged replication. the german word for it is Schöpfungshöhe, which literally translates to creation elevation - this is used in legal arguments to determine the creative input someone had when working from source material.
the challenge with the large AI generators being of course that they automate the collation of microscopically small infringements that each on its own appear to not be problematic. as i wrote earlier in the thread, credit laundering.
Quote:
Yes, the mentioned client/executive use-case. Good riddance. =)But yeah, the thing with "hey, algorithm! gimme a nice picture of a cute animal that smokes pot" isn't really a possibility.
My example was more limited to replicating your own drawing techniques in an easier way, an AI-fill-tool.
This would eliminate most of the hard-core pixelling challenge (except possibly initially) that grip and The_Sarge feel so strongly about, hence my question about the tech itself was also referring to that "optimisation".
@Wayfinder, I wouldn't say it's as complicated as you describe, nor 'manipulative'. Nor I believe the discussion here (which has shifted towards the pixel branch of demoscene graphics) is about things being legal or not.
On the one hand there's a practice that has been a thing since the scene's infancy and may be served by various methods, with generative AI being the latest (under question?) trend.
On the other hand there's the ethics surrounding this practice, and the values (creativity, craftsmanship, inspiration, you-add-your-own).
I am not sure how the two hands balance in the end, or how they should balance, but I personally find it hard to separate the contents of the first hand to different categories with different priorities.
I apologize if my choice of words may sometimes make things sound unfair; there's instances that I need to translate between Greek and English in my head, and I only get to practice the second tongue mostly with my demoscene activities, sadly. :)
On the one hand there's a practice that has been a thing since the scene's infancy and may be served by various methods, with generative AI being the latest (under question?) trend.
On the other hand there's the ethics surrounding this practice, and the values (creativity, craftsmanship, inspiration, you-add-your-own).
I am not sure how the two hands balance in the end, or how they should balance, but I personally find it hard to separate the contents of the first hand to different categories with different priorities.
I apologize if my choice of words may sometimes make things sound unfair; there's instances that I need to translate between Greek and English in my head, and I only get to practice the second tongue mostly with my demoscene activities, sadly. :)
This is getting boooring.
Rexbeng is for almost a year trying to say that AI LLMs are only the next tool in a long succession back from using the stone to grind the seashells to produce paint for painting. I think this stance was merely naive a year ago. Nowadays, after all the conversation, all the writings (both on the wall and on paper), after watching all the development, after all the grift causing all the grief, this stance comes off as unbelievably shortsighted, uninformed, un-nuanced, un-empathic, stubborn, misaligned with reality, a petrified pose. Just a constant stream of whataboutisms and unyielding tech-bro apologism. Man it doesn’t even matter what new “thought experiments” you come up with, everyone knows what you’re gonna say : you’re gonna repeat: “Either embrace the AI or burn everything else down with it you apes. All of your toys are equally petty and evil.” You’re trying to squeeze people into thinking that ethically it’s either unhindered AI or back to the stone age. That’s so false. Perhaps you’re even not fully aware of what you’re doing. Please don’t get me wrong, I have nothing against you, I usually like talking to you, and I apologize if I’m being out of line. I just need to speak my mind openly.
I know I know, someone can go and say smth similar about me as I’ve been against prompt-to-image machines nearly from the start. But I really do my best to listen. Honestly.
I urge all who did not do it in due time to read this: https://www.pouet.net/topic.php?post=580256
I cannot find a single point in this post that doesn’t resonate both with my own thoughts and the unfolding reality.
Rexbeng is for almost a year trying to say that AI LLMs are only the next tool in a long succession back from using the stone to grind the seashells to produce paint for painting. I think this stance was merely naive a year ago. Nowadays, after all the conversation, all the writings (both on the wall and on paper), after watching all the development, after all the grift causing all the grief, this stance comes off as unbelievably shortsighted, uninformed, un-nuanced, un-empathic, stubborn, misaligned with reality, a petrified pose. Just a constant stream of whataboutisms and unyielding tech-bro apologism. Man it doesn’t even matter what new “thought experiments” you come up with, everyone knows what you’re gonna say : you’re gonna repeat: “Either embrace the AI or burn everything else down with it you apes. All of your toys are equally petty and evil.” You’re trying to squeeze people into thinking that ethically it’s either unhindered AI or back to the stone age. That’s so false. Perhaps you’re even not fully aware of what you’re doing. Please don’t get me wrong, I have nothing against you, I usually like talking to you, and I apologize if I’m being out of line. I just need to speak my mind openly.
I know I know, someone can go and say smth similar about me as I’ve been against prompt-to-image machines nearly from the start. But I really do my best to listen. Honestly.
I urge all who did not do it in due time to read this: https://www.pouet.net/topic.php?post=580256
I cannot find a single point in this post that doesn’t resonate both with my own thoughts and the unfolding reality.
Quote:
Deep Meet is super obvious with the AI images. The fish, the squids, maybe the eye. Also the starting image with the android is probably AI but pixeled over. The other stuff is just converted from AI images.
So probably close to zero hand made graphics.
But as Frost says, it's bigger than just these demos. It's a very concerning trend and the audience need to start seeing the difference between hand made stuff and AI prompted shit.
Until a few days ago, the thought never crossed my mind that AI was already being used for (Amiga) demos. I didn't recognise these graphics but maybe that's because i was naive enough to think it wasn't an issue and i wasn't looking for it.
What's incredibly disappointing in those cases (The Deep Meet, Ordo and whatever other demo i don't know about) is that it is never mentioned in the credits or the readme. At least be honest about it and inform the viewer that not all of the graphics were actually done by the graphicians. Stop taking credit for stuff you didn't do.
This Revision has been a real eye-opener for me and not in a pleasant way.
Hey, happy you usually like talking to someone petty and evil. I, much like you, can only speak my mind openly though, and I do know that language can create gaps sometimes because it just is what it is. But choices of words don't change the meaning of the things I say, I believe. If they did, you wouldn't stick all those 'un-' tags to me after all. :)
So, I stand by what I have said from the very beginning. That, yeah, I generally believe that AI may be a great tool for artists. Somebody even posted an example of an artist that trained a model with her style a few pages back I think.
Can we stop the talk about general AI now, and stick to the scene topic?
So, I stand by what I have said from the very beginning. That, yeah, I generally believe that AI may be a great tool for artists. Somebody even posted an example of an artist that trained a model with her style a few pages back I think.
Can we stop the talk about general AI now, and stick to the scene topic?
@rexbeng
Hey man, I neither think you are petty and evil nor have I ever said that.
Hey man, I neither think you are petty and evil nor have I ever said that.
Quote:
My example was more limited to replicating your own drawing techniques in an easier way, an AI-fill-tool.
This would eliminate most of the hard-core pixelling challenge (except possibly initially) that grip and The_Sarge feel so strongly about, hence my question about the tech itself was also referring to that "optimisation".
I've drawn with my hands basically for as long as I've lived, and pixelled for almost as long, I can say that I wouldn't do away with any part of my creative processes. If i didn't like the hard-core pixelling I would be a total idiot if I did it anyway. It's part of the process and it's very important, at least for me.
Actually, I kind of hate pixelling... All those countless hours spent manually dithering crap to Ginnunga and back is no joke, but It is something that gives me reason to do it. I challenge myself and my sanity (lol) to overcome myself, in a sense, and I wouldn't trade that away. Without it, it would be meaningless. The same goes for all tedious and boring parts of my creative processes. It's a part of the whole and my art, and without it it wouldn't be the same. It shapes me and the things I do.
I second Frost here. If one is not personally driven to get on fights with the pixels, why would one want to go ahead and do it in the first place?
Quote:
I generally believe that AI may be a great tool for artists
So, the tool that’s been dreamed up for clients/execs so that they can circumvent the creative can be a great tool for artists? Perhaps. Probably.
So can an AK-47. And a piece of bone of a neighbours dead daughter. Or, wait, someone’s suffering can also make for a great tool/vessel for someone else’s art.
Fine. Let me put it this way. I strongly believe that AI will become a great tool for artists once it becomes open and democratic (I mean really democratic, not the marketing BS we get) and free from corpos. I wish it will.
With regards to 'so can an AK-47'. Yes, it can. See https://www.boredpanda.com/gun-country-installation-usa-map-michael-murphy/
Quote:
I strongly believe that AI will become a great tool for artists once it becomes open and democratic (I mean really democratic, not the marketing BS we get) and free from corpos. I wish it will
Cool. In the meantime, let’s watch it move fast and break stuff, let’s watch it run amok, destroy lives, squash creatives, all while hoping for a better future, for a better, more democratic toy for “true artists”.
Let me put it this way. I strongly believe the hungry will cease to be hungry once the wealthy buy food and give it to them. I wish they would.
That argument "democratic" is pure bullshit. Everyone can draw (OK, almost everyone, but hey, there's people out there creating awesome art with their feet, mouths and whatnot!) Paper and pens costs next to nothing. You only have to do it, and if you don't then it's probably not your thing anyway.
Quote:
That argument "democratic" is pure bullshit. Everyone can draw (OK, almost everyone, but hey, there's people out there creating awesome art with their feet, mouths and whatnot!) Paper and pens costs next to nothing. You only have to do it, and if you don't then it's probably not your thing anyway.
Yeah, that “democratizing” is the lamest bullshit about marketing of these machines. So somehow, some tech that only the wealthiest companies with access to sh*tloads of unsolicited data and processing power plus top talent, heavily centralized is more democratic than a pencil and a paper. The argument that this “democratizes art” is so dumb that it’s not even funny. This is the “marketing BS” rexbeng mentioned.
Those who seem to be in favour of AI in art seems to be those who don't do visual art. I get it, it opens a new field for you with infinite art just by typing a short prompt and you are done and have some amazing burning skull with a fish in the background for your demo.
But we who do art are mostly against AI making art.
You can ask yourself if you want yourself to be redundant. Freed from human contact and no-one wanting your creative input or expertise and playing around with ideas, all built up learning the why's and what ifs. This is for whatever field you are working in. This applies to code, music and art. The holy magical threesome that makes a good production. The ones that favours AI in whatever field are against human contact in a sense and miss all the communication, friendship, laughs and mistakes within the process that making art together means.
And also to those who do AI prompt, convert it and put your name in demos, fuck you.
But we who do art are mostly against AI making art.
You can ask yourself if you want yourself to be redundant. Freed from human contact and no-one wanting your creative input or expertise and playing around with ideas, all built up learning the why's and what ifs. This is for whatever field you are working in. This applies to code, music and art. The holy magical threesome that makes a good production. The ones that favours AI in whatever field are against human contact in a sense and miss all the communication, friendship, laughs and mistakes within the process that making art together means.
And also to those who do AI prompt, convert it and put your name in demos, fuck you.
The thing with techbros and their insane need to invent solutions to problems that doesn't exists, so they also have to invent the problem to market their solution.
Pfffffttttttttt... Speak for yourself please, in case you haven't noticed, it's only a handful of folks who are arguing in this thread from the arts angle, and the rest isn't quite as worried as (mainly) the two of you apparently... Not to the point that they'd bother to give their opinions in this endless shitstream of irrelevant opinions anyway.
Fuck AI, and fuck the tech-bros pushing it.
