AI is forcing the hand of the Demoscene.
category: general [glöplog]
4gentE: just ignore them all, not worth it.
Oh sorry, I thought that "let's suppose" part of
was clear enough. As in "let's pretend". As a mental exercise. Obviously it was not.
It was not meant as a "manipulative strategy" to fish out a deeper understanding behind your words, I assure you. I only did it to set up a thought experiment in which I could pose a question that followed.
Quote:
So let’s suppose I am a crazy old luddite fart
was clear enough. As in "let's pretend". As a mental exercise. Obviously it was not.
It was not meant as a "manipulative strategy" to fish out a deeper understanding behind your words, I assure you. I only did it to set up a thought experiment in which I could pose a question that followed.
Quote:
It was not meant as a "manipulative strategy" to fish out a deeper understanding behind your words, I assure you.
...I mean it wasn't meant originally, but it kinda turned out that way tho... So, sorry about that.
@pengan
You seem not to have any understanding of GRASSROOTS vs ASTROTURF. Whatsoever.
Therefore, let me try to explain: Demoscene, as you are well aware of, is a GRASSROOTS subculture. As in grown from below, against any authority. LLM AI fad is being pushed FROM ABOVE, so it's a clear example of ASTROTURF. There. That's the difference.
The LLM owners/pushers are the exact equivalent of the"copyright giants" you mention.
Quote:
I think demosceners are a group of highly digitally literate individuals who originated from the warez community. They were once fearless in challenging copyright giants and breaking the restrictions imposed by hardware and software vendors. But why are these rebels so anxious and lacking confidence when it comes to the topic of AI?
You seem not to have any understanding of GRASSROOTS vs ASTROTURF. Whatsoever.
Therefore, let me try to explain: Demoscene, as you are well aware of, is a GRASSROOTS subculture. As in grown from below, against any authority. LLM AI fad is being pushed FROM ABOVE, so it's a clear example of ASTROTURF. There. That's the difference.
The LLM owners/pushers are the exact equivalent of the"copyright giants" you mention.
I'm not sure what do you mean by LLM owners. It's based on a framework (e.g. transformer architecture) that nobody claims copyright on. The framework was created by affiliated researchers, sure, but AFAIK it was not immediately recognized as a game changer.
The "bosses" who owns/push stuff from above are usually playing "wait and catch pray" strategy. They always on a lookout what's new and promising in labs, open source communities, grassroot movements etc...and when something cool starts emerging they try to put their hand on it as fast as possible to simply take over.
Then, of course, if you want to secure your "investment" (aka your "catch") you fuel marketing machine and build narrative around it. But your thesis that LLMs or AI is just pushed from above is a little bit incomplete.
The "bosses" who owns/push stuff from above are usually playing "wait and catch pray" strategy. They always on a lookout what's new and promising in labs, open source communities, grassroot movements etc...and when something cool starts emerging they try to put their hand on it as fast as possible to simply take over.
Then, of course, if you want to secure your "investment" (aka your "catch") you fuel marketing machine and build narrative around it. But your thesis that LLMs or AI is just pushed from above is a little bit incomplete.
@tomkh
Sure.
I was trying to compare Demoscene and AI LLM fad in an attempt to guess the answer to the posed question that went something like "why are demosceners suspicious of embracing LLM AI". So, in this scenario, it's pretty obvious that the "big players" back in the day did not want the scene to crack and distribute the warez, as it is obvious that today the "big players" actually want you, encourage you to use LLM AIs. Again, that was my guess as to what causes the amount of pushback from the scene. As you know, I personally don't have to go that deep, I have my own environmental/societal reasons, which perhaps not that many people in the scene share.
Sure.
I was trying to compare Demoscene and AI LLM fad in an attempt to guess the answer to the posed question that went something like "why are demosceners suspicious of embracing LLM AI". So, in this scenario, it's pretty obvious that the "big players" back in the day did not want the scene to crack and distribute the warez, as it is obvious that today the "big players" actually want you, encourage you to use LLM AIs. Again, that was my guess as to what causes the amount of pushback from the scene. As you know, I personally don't have to go that deep, I have my own environmental/societal reasons, which perhaps not that many people in the scene share.
Oh, I agree with you that it doesn't make sense to embrace any of the concrete AI product, like Midjourney or ChatGPT, because you will be supporting big players this way and moreover, doing it for free.
This, however, doesn't mean you should block yourself from toying with LLMs or stable diffusion on your own.
You could probably still do it in a demoscene spirit "against authorities".
Whether or not demoscene is really sending a big effs to authorities I actually doubt. It feels like at least part of the demoscene is just game engine players bitch.
But going back the topic...
There are multiple ways of doing it, like developing your own AI tech, not releasing sources to it and showing what it can do "in their face".
This, however, doesn't mean you should block yourself from toying with LLMs or stable diffusion on your own.
You could probably still do it in a demoscene spirit "against authorities".
Whether or not demoscene is really sending a big effs to authorities I actually doubt. It feels like at least part of the demoscene is just game engine players bitch.
But going back the topic...
There are multiple ways of doing it, like developing your own AI tech, not releasing sources to it and showing what it can do "in their face".
I don't know. As I said, I know why I'm personally boycotting LLMs for reasons repeatedly stated, reasons which don't need to go into this big players vs independents territory.
Perhaps the promised 'user friendliness' of MidJourney or ChatGPT just rubs the scene the wrong way. And justifiably so IMHO.
The 4k, 64k or oldschool categories for example were self-imposed restrictions. They were not meant to enhance "user friendliness", exactly the opposite.
Because people, in the scene especially, by now have come to understand that marketed "user friendliness" has for a long time in fact been exactly the opposite : concealed "user hostility" with the intention of control and deskilling. A poisoned gift.
The ultimate demoscene way (at least the way I see it) would be : Wanna write? Then learn how to write, don't use Chat GPT. Wanna code? Then learn how to code, don't use commercial render engine. Wanna paint/pixel? Then learn how to paint/pixel, don't use Midjourney. Not as a rule of course, more like a loose shared ethos with plenty of room to do exactly the opposite and be appreciated nevertheless on your own merit.
And of course, there's the "cultural image" issue. I think of the scene more as a deliberately obscure black metal band, while LLMs these days are more like Britney Spears.
Perhaps the promised 'user friendliness' of MidJourney or ChatGPT just rubs the scene the wrong way. And justifiably so IMHO.
The 4k, 64k or oldschool categories for example were self-imposed restrictions. They were not meant to enhance "user friendliness", exactly the opposite.
Because people, in the scene especially, by now have come to understand that marketed "user friendliness" has for a long time in fact been exactly the opposite : concealed "user hostility" with the intention of control and deskilling. A poisoned gift.
The ultimate demoscene way (at least the way I see it) would be : Wanna write? Then learn how to write, don't use Chat GPT. Wanna code? Then learn how to code, don't use commercial render engine. Wanna paint/pixel? Then learn how to paint/pixel, don't use Midjourney. Not as a rule of course, more like a loose shared ethos with plenty of room to do exactly the opposite and be appreciated nevertheless on your own merit.
And of course, there's the "cultural image" issue. I think of the scene more as a deliberately obscure black metal band, while LLMs these days are more like Britney Spears.
For me as a coder, AI art might be an extension of procedural art, if done right. It might give new possibilities. As a childhood dream, I was thinking to create infinite planets just from formulas...maybe AI can help here. Just of course without relying on commercial tools or someone else's artwork - that's lame. But I wouldn't entirely dismiss some of the ideas boiling "in the field". Maybe we just have different perspective.
Different perspective? Exactly. I feel the LLM AI as a creative "art" tool is imagined totally upside down. When you say "AI art might be an extension of procedural art", I immidiately think "Nooooooo. The LLM AI art is the exact opposite of procedural/generative". It's trained on real, scanned in art, not on pure mathematical formulas. Formulas are employed in "transfiguration" (for lack of a better word) of already "rendered" stuff into other already "rendered" stuff. As opposed to procedural art, nothing is actually being generated from pure formulas here.
Like, there's this AI based thing called "Wonder Dynamics". It's a subscription based service that lets you upload a camera footage. This camera footage gets analysed. Then you select a 3D model, a robot or something. That model gets tracked, lighted and keyed-in into your footage, mimicking exactly what the original actor in original video was doing, all with camera movement and all. Now, if you asked me ten or so years ago, I would say that pretty soon we will be able to use 3D rigged models, whole 3D scenes with lighting and all rendered so realistically as to be able to completely replace video. This would have been parametric (almost procedural) you see. But this - what Wonder Dynamics is doing is completely the other way around, it's completely upside down. To a classicaly digitally educated mind it sounds near impossible. That's why it needs such a complex technology, so much data & grunt. Because it's been imagined the wrong way around. Why was it imagined the wrong way around? Because it was imagined to be used in the end by a trained monkey, a bored TikTok kid. And because the "AI" underneath was programmed to be universal, it was programmed by people who knew next to nothing or very little about possible future end use scenario. Now, who could applaud that? Who with any connection to the scene could possibly applaud that?
Like, there's this AI based thing called "Wonder Dynamics". It's a subscription based service that lets you upload a camera footage. This camera footage gets analysed. Then you select a 3D model, a robot or something. That model gets tracked, lighted and keyed-in into your footage, mimicking exactly what the original actor in original video was doing, all with camera movement and all. Now, if you asked me ten or so years ago, I would say that pretty soon we will be able to use 3D rigged models, whole 3D scenes with lighting and all rendered so realistically as to be able to completely replace video. This would have been parametric (almost procedural) you see. But this - what Wonder Dynamics is doing is completely the other way around, it's completely upside down. To a classicaly digitally educated mind it sounds near impossible. That's why it needs such a complex technology, so much data & grunt. Because it's been imagined the wrong way around. Why was it imagined the wrong way around? Because it was imagined to be used in the end by a trained monkey, a bored TikTok kid. And because the "AI" underneath was programmed to be universal, it was programmed by people who knew next to nothing or very little about possible future end use scenario. Now, who could applaud that? Who with any connection to the scene could possibly applaud that?
Sorry man, I don't understand what you are trying to say.
Classical procedural art was to me about capturing the "real signal" say 3d terrain, clouds or maybe rocks using formulas of similar characteristics, say Perlin or Worley noise, and then tuning thosw formulas by hand until the output look real. Using ML/AI you could start with photogrammetry, which is *not* violating anyone's artistic style - it's just capturing reality as is and automatically find the right parameters, formulas, grammars etc...
Do you find it morally reprehensible?
Classical procedural art was to me about capturing the "real signal" say 3d terrain, clouds or maybe rocks using formulas of similar characteristics, say Perlin or Worley noise, and then tuning thosw formulas by hand until the output look real. Using ML/AI you could start with photogrammetry, which is *not* violating anyone's artistic style - it's just capturing reality as is and automatically find the right parameters, formulas, grammars etc...
Do you find it morally reprehensible?
Quote:
Maybe we just have different perspective.
You mentioned “AI art”, so I was thinking of existant “AI art machines” of course.
When I said ‘parametric’ or ‘procedural’ art I meant there is no ‘input’. No ‘captured real signal’, no ‘photogrammetry’. It’s very possible that I don’t know how to properly express that.
Like a picture drawn in a vector program (like Illustrator) which is not a picture at all, it’s a bunch of formulas that only get rendered into picture at the end, and at any resolution because it’s parametric. As opposed to a scanned-in bitmap, manipulated in Photoshop or similar software, then spit back out.
Like a 3D model that is modelled from scratch in a 3D modelling software, which is not a model at all, but a bunch of formulas that get rendered into a model or a picture at the end. As opposed to data scanned in with a 3D scanner.
Vector images and 3D models are FORMULAS. Bitmap images, motion capture and 3D scans, on the other hand are DATA, not FORMULAS. That’s what I meant. Sorry, I do’t know if I’m communicating this right.
When I said ‘parametric’ or ‘procedural’ art I meant there is no ‘input’. No ‘captured real signal’, no ‘photogrammetry’. It’s very possible that I don’t know how to properly express that.
Like a picture drawn in a vector program (like Illustrator) which is not a picture at all, it’s a bunch of formulas that only get rendered into picture at the end, and at any resolution because it’s parametric. As opposed to a scanned-in bitmap, manipulated in Photoshop or similar software, then spit back out.
Like a 3D model that is modelled from scratch in a 3D modelling software, which is not a model at all, but a bunch of formulas that get rendered into a model or a picture at the end. As opposed to data scanned in with a 3D scanner.
Vector images and 3D models are FORMULAS. Bitmap images, motion capture and 3D scans, on the other hand are DATA, not FORMULAS. That’s what I meant. Sorry, I do’t know if I’m communicating this right.
Hmmm...I don't speak your language.
I'm talking about code-generated assets - when you use random number generators and bunch of formulas/operators to transform them into "consumable" 3d assets, e.g. as game scenery that your render or interact with in a computer game.
The only moral or even legal question is what data do you use to derive your code-generated or AI-generated assets. If you use photogrammetry you captured yourself or maybe just legally bought, I see no problem.
I'm talking about code-generated assets - when you use random number generators and bunch of formulas/operators to transform them into "consumable" 3d assets, e.g. as game scenery that your render or interact with in a computer game.
The only moral or even legal question is what data do you use to derive your code-generated or AI-generated assets. If you use photogrammetry you captured yourself or maybe just legally bought, I see no problem.
Yes, I think I’m not expressing myself accurately or adequately, sorry for that.
Forget about moral, legality and such, we’re not talking about that now. We are talking about what is the meaning of ‘procedural’.
I’m trying to say that if something is derived by apllying processes to some kind of externally gathered data, then this is not truly procedural. Procedural to me means created parametrically out of pure formulas, pure code. You know, like procedural gfx. They don’t have input DATA.
Forget about moral, legality and such, we’re not talking about that now. We are talking about what is the meaning of ‘procedural’.
I’m trying to say that if something is derived by apllying processes to some kind of externally gathered data, then this is not truly procedural. Procedural to me means created parametrically out of pure formulas, pure code. You know, like procedural gfx. They don’t have input DATA.
So you talk about what has bigger value to you: a code that is written "by hand" to best approximate the coder's own "memorized" perception of the reality or code that more automatically optimizes generation parameters to explicitly given images captured by camera.
In both cases, coder has to do a lot of work to get a good result.
But, of course, you have full right to be personally impressed more by the former, but I don't see what gives you the right to judge the moral side of the latter.
In both cases, coder has to do a lot of work to get a good result.
But, of course, you have full right to be personally impressed more by the former, but I don't see what gives you the right to judge the moral side of the latter.
@tomkh
So you say that the definition of “procedural” in i.e. procedural gfx is not a real thing at all, but a product of my personal preference or even better “moral”.
I came to think there is a strong chance you may be really really crazy. Or trolling like crazy, so … yes, crazy.
Whichever it is, I have to say that I’m done talking to you.
I didn’t realize this until now, I was really engaging this in a good faith. I poured a fair amount of time into it too, even after being advised good-heartedly not to by some nice people.
Or maybe you are completely sane, and I’m crazy.
Luckily, there is this whole forum for people to see for themselves.
I’m sorry if I (please believe me, unwittingly) angered you or something, good bye, have a nice life, and please don’t blow me up with a bomb.
So you say that the definition of “procedural” in i.e. procedural gfx is not a real thing at all, but a product of my personal preference or even better “moral”.
I came to think there is a strong chance you may be really really crazy. Or trolling like crazy, so … yes, crazy.
Whichever it is, I have to say that I’m done talking to you.
I didn’t realize this until now, I was really engaging this in a good faith. I poured a fair amount of time into it too, even after being advised good-heartedly not to by some nice people.
Or maybe you are completely sane, and I’m crazy.
Luckily, there is this whole forum for people to see for themselves.
I’m sorry if I (please believe me, unwittingly) angered you or something, good bye, have a nice life, and please don’t blow me up with a bomb.
With personal attacks such as "you may be really really crazy" it's definitely hard to continue the discussion. But, no worries, you didn't angered me.
…aaaand here we go some more…
https://www.engadget.com/mason-city-iowa-school-district-ai-book-ban-censorship-202541565.html
Hooray!
https://www.engadget.com/mason-city-iowa-school-district-ai-book-ban-censorship-202541565.html
Hooray!
You are not passing the Turing test today, 4gentE. :]
To be fair, what's in the engadget article is not a bad thing. I mean, the using of AI to verify huge amounts of data (whatever this data may be; from legal papers to city planning fe). The only problems I see in that article is "US" and "Republicans".
It does indicate the way AI use is going to go - or is going - and given that apparently it took only a couple of months to realize that AI isn't as good at things as it was advertised, using it for book ban suggestions is going to make a despicable thing worse.
There was an article some time ago which I can't find anymore (links welcome) that posited that a good chunk of AI hype visible on social media wasn't excitement to create new things, but a vindication that "the elitist creatives" lose their jobs.
There was an article some time ago which I can't find anymore (links welcome) that posited that a good chunk of AI hype visible on social media wasn't excitement to create new things, but a vindication that "the elitist creatives" lose their jobs.
@Gargaj, 'the way AI use is going to go', you mean that it's going to be used by governments? Ofcourse and that's to be expected.
And, again, morons on social medias shouldn't be the benchmark here. Elitist creatives won't lose their jobs because of AI, no matter what who-knows-for-what-reason-vindicatists get excited about. (Interesting though; where there really people who expressed such thoughts? Do you remember any other details? Pure curiosity asking...)
And, again, morons on social medias shouldn't be the benchmark here. Elitist creatives won't lose their jobs because of AI, no matter what who-knows-for-what-reason-vindicatists get excited about. (Interesting though; where there really people who expressed such thoughts? Do you remember any other details? Pure curiosity asking...)
Quote:
you mean that it's going to be used by governments? Ofcourse and that's to be expected.
No, that it's going to be used to destroy, destabilize and suppress, rather than to build - governments or no governments.
Quote:
Elitist creatives won't lose their jobs because of AI
Have you noticed there's a writers / actors strike?
Quote:
(Interesting though; where there really people who expressed such thoughts? Do you remember any other details? Pure curiosity asking...)
The Stability CEO, for one.
So essentially because of some criminals, you are rejecting the whole AI field? If this is not a one giant straw man I don't know what is.