#entering my yorha era

[netverk]

AI Art is Not Collage or Sampling And Here's Why

I rarely feel motivated to make text posts but over the past few days I've seen an uptick in people deciding to defend AI Art, despite them maybe once being against it.

I'm sorry also if this post seems patronising or goes over things which you may already know, I'm writing this for theoretically all potential audiences so I feel obligated to go into detail on things. Again, sorry for the length.

I have been tracking the development AI image generation for many years. When I first saw the news about DALL-E 2 I was shocked that technology had seemingly advanced to the point that any image could be conjured from the aether, basically. However, in my own experimentation with it, I found it to be consistently disappointing in its results. As time went on, we all learnt of the staggering amount of Art Theft this, and all other Generative AIs, engaged in to power themselves. To see then, posts circulating equating AI Art with Collage/Remix/Sampling is very strange, it as if everyone forgot about the whole consensus agreement that these things stole art in a massive and blatant way. On some level I hypothesise this might be because of the reporting on the fact that AI Art may be used by The Powers That Be (Adobe etc.) to tighten copyright law to an incredibly draconian degree, making it so that individual art styles may be copyrightable, but really, I don't know! Regardless, AI Art is not engaging in sampling, remixing, or collage. It is, to cut a long post short, literally a machine to generate forgeries. Again, I thought this was widely agreed upon, but apparently not!?

In college/university I studied Photography. The debate around AI Art now has shades of the debate around the start of Photography, but not really.

------------

The instigating post here is by @whompthatsucker1981, who boldly stated:

"i think that gay sex cats is the new duchamp’s fountain"

Gay Sex Cats is an AI Generated Image of the words 'GAY SEX' resolving in a group of kittens walking towards the "camera". From a distance it resembles the words 'GAY SEX', when viewed closer it resembles those words and also a group of kittens walking towards the camera. When viewed even closer, or for basically any amount of time, the AI generated nature of the image betrays itself, with the kittens being typically malformed and strange looking.

Duchamp's Fountain is an iconic work of Dada art. It is a men's urinal, turned onto its side and signed 'R. Mutt, 1917'. In invoking Fountain, 'whompthatsucker1981' is on some level politicising dislike towards AI Art. In the tags on the post, he states

Oh, so I guess he really is outright politicising the dislike of AI Art now.

Bluntly, he is completely and utterly wrong about Fountain, and it is an insane false-equivalence argument.

Fountain is hated by Chuds largely because of what it represents. The intent can sort of be gleamed through just considering why it exists and is called 'Fountain'. It's a urinal, but its been de-familiarised and turned literally on its side and dubbed a fountain. It is suggesting that theoretically, anything can be art. If you interpreted it that way, that's basically the 'intended' read! This was the core concept, in his words: "everyday objects raised to the dignity of a work of art by the artist's act of choice".

You can read other things into Fountain. All Art Is Subjective.

Chuds hate Fountain for it doing that. For it to dare suggest that art can have meanings and readings and be, visually, just a piece of ephemera or junk the artist decided to elevate. The Trad-Chud Statue PFP crowd profoundly do not understand art, and believe that art should be purely judged on its visual qualities and in their eyes depiction of traditional and historic scenes. They cannot stand that art is not exclusively Jesus and Imperial History Fanart.

AI Art is ironically really good for people who wish that art was all just meaningless, beautiful images! It does this by literally stealing the works of countless artists present and past.

To loop this around to the first work of ""art"" mentioned, what exactly is Gay Sex Cats trying to say exactly as a work of art?

It... isn't. Of course, that is not to say people cannot read into it like whompthatsucker, as All Art is Subjective. But it is an image not made with any outright statement on AI Art, Non-AI Art, or anything really, in mind. It is a shitpost, it is two words the 'Internet' as a collective semi-conscious has agreed (for whatever reason) constitute a Low Effort Funny, there exist other images of the words 'GAY SEX' rendered in various ways. This image thus simply slots in as the latest in a series.

Therefore the message of Gay Sex Cats is simply to exist, to an elicit a reaction from the viewer, presumably the creator hoping for a little laugh and a like-and-reblog. It is not making any concrete statement about what art means, or really doing anything more than just being a shitpost.

------------

The thing is, though, whompthatsucker1981 didn't just make a really terrible comparison between two entirely different things made for two entirely different reasons, he also painted Gay Sex Cats. Like, he physically painted the AI image.

This act, interestingly, is quite clearly designed to make a statement 'on' AI Art. Implausibly, the arguments presented in his post responding to various rebloggers commenting in the tags. Let's take a careful look through his statements!

He starts by saying that many of the reblogs are 'missing the point', if you look through the screenshots of tags he included in his post, you'll find that the takeaway of a lot of people is 'hooray, I can reblog that shitpost now and not feel bad about it'. He does note that yes, he was taking a silly meme seriously and was 'trying to make people think about it'.

He goes on to ask the rebloggers:

The thing that 'whompthatsucker1981' fails to get is that it inherently is less of an act of Art Theft by virtue of being hand painted by him. Even should the base image he is practically tracing be made from the mulched together essence of thousands of other images, to make your own work with paint basically means that you... aren't stealing art.

This is the common defense of AI Art Defenders, that it is completely harmless. No one is losing money, no one is actually getting hurt, so why should we dislike a funny image, a fun tool, by virtue of AI.

In literally the same day I encountered this post, I encountered another post pushing the exact same argument, basically. Generative AI is fine, for memes, as memes are free and just posted around! This post was by a bunch of different people all replying to one-another but the main 'top two' levels of that are the ones that matter.

https://machine-saint.tumblr.com/post/729656361855713280/remember-when-the-cool-nostalgic-music-trend-was

The majority of Nightcore uploaded to YouTube quite specifically names the track that has been Nightcored in the title. With remixing, the base track is still often credited somewhere.

Most Nightcore doesn't really alter the base track that much, but one could say that in the choice of song 'Nightcored' there is still some echo of the uploader's spirit. Even so, this argument isn't really worrying about Nightcore/memes making a statement or being something personal to the creator. This argument is simply saying: "Oh, but you steal images and sounds and media constantly to generate internet culture, so how is AI Art any different?" This is why I interjected this post in here, because it puts forward the same argument as the other.

It has been said that 'Everything is a Remix', I make a lot of works sampling other things. One of my favourite emotes in Fortnite uses a song which samples a house track from the 90s. Sampling is very much a normal and accepted part of cultural creation. Sampling and Collage are very much the same things in different mediums. Collage is when a still work collages things from disparate sources, these elements are conspicuously from different places. Much like Duchamp's Fountain and Readymades, the collage artist is taking typically ephemeral imagery and elevating it to art. Fundamentally, in their process they are creating, they are imbuing whatever it is they are doing with a part of their soul, their mind.

An AI is not doing anything like that. At the start of the post I mentioned how I studied Photography. In the early days of photography, there was a similar debate over how it can capture images perfectly, how it is mechanical, how it may not be 'art' by such a virtue. In time, it became accepted that photography was art because in the act of using the machine, a photographer is still exerting some degree of creativity. Choosing the precise composition, waiting for 'the decisive moment', or time of day, or staging and lighting a studio a specific way. Photography is a litany of choices, and although due to capturing reality they don't quite have the hyper-distinct mark of an author quite like illustrative arts (unless you put in the work to cultivate a specific aesthetic that becomes linked to you, as with Wes Anderson), they are still works guided and driven by the human soul.

One could argue that AI Art has a human element. Someone typed words into a prompt box, someone used img2img and attempted to steer the generation a certain way. Just like with photography, the artist's primary tool is a machine, just this machine captures that which isn't seen rather than that which is. The thing is, this is not how Generative AIs really work.

As was already described during the last time people discussed AI Art, Generative AIs take a huge corpus of imagery and artwork. Rather than bringing these elements together in collage, the AI kinda dreams up a new image that resembles all the things it is trained on, so you can describe a room and it will generate an image roughly like that. This result is not akin to any sort of prior form of art. There is no more creative input than pulling for 10 in a gacha game and hoping you get what you want. Even with things that can 'steer' it, such as img2img and whatever the heck the thing that made Gay Sex Cats is called, you are still functionally at the mercy of RNG. The images that train the AI are not acquired with any degree of consent, they are scraped en-masse. If you have any delusions that an AI isn't an art forgery machine, simply toggle one of the 'filters' on Playground.ai and then look at the generated image in the history. In an experiment I did, I applied the 'Filter Style' of 'Ominous Escape'. Only when looking in my history did I see that basically meant that it appended the following to the prompt:

elegant, highly detailed digital painting, concept art, smooth, sharp focus, illustration, in the style of simon stalenhag, wayne barlowe, and igor kieryluk.

Again, why is this suddenly now considered OK, when just a few months earlier it wasn't?

Generative AIs have been fed the sum-total of imagery online almost, these images have passed through their insanely convoluted multi-stage digestive systems. You ask it to create something, and it shits out a thing that vaguely looks like what is in your mind, if you're lucky.

Yet at no point in this process is there any true transfer of you into it, your mind, your soul. The prompt is an instruction, but you cannot steer how the machine interprets it too finely. There is no creation, no artistry, no effort. Just the digested mulch of all that came before, shat out, polluting the internet and the world.

This all conveniently answers the last question of whompthatsucker1981's original post.

Yes, quite literally it is worth more because he spent time on a piece of paper making it. It is that simple. AI Art has no value because it is shat out by the digestive system of a stack of algorithms, it is mindless, artless junk. Even for making 'shit'posts AIs aren't OK. In using them you are functionally endorsing how the AIs were created: Through mass art theft on a gigantic scale. It doesn't matter if what you're making is 'harmless and not for sale'. It is not what is being made that is the problem here, it is the literal nature of the technology itself. To reiterate:

Generative AI is powered through non-consenting Art Theft, devoid of any soul or human input beyond a prompt or an image to draw from. The most it should be used for is to quickly brainstorm and iterate on ideas, for inspiration on a personal and individual level, by feeding it your old works. Even then, it runs into the inherent ethical issue of 'Using It At All'. To support AI Art is to support the eventual destruction of the field of 'Arts', from Visual Development and Concept Art down to even actual things on screen. The people developing this technology do not have your best interests in mind, and only wish to create a world where small artists in the production chain of media do not need to exist anymore. The writers may have won a victory against Generative AI, but the battle is not over yet. As I said above, to even use them for silly images is on some level an endorsement towards the vision that this technology is inevitable and 'the future'. Immediately before this, NFTs were touted as inevitable and the future, yet then everyone listened to artists concerns and they never happened.

This technology does not have to be a part of our future.

#ai art #ai #artificial intelligence #entering my yorha era

[landoftheway]

FFXIV Nier Raid Connections

Since it seems most folks have had time to play the raid by now, I’m gonna speculate on the connections between the raid series and the Drakengard/Nier franchise. Under a cut for both length and spoilers:

Before I get into the details, I wanna preface this with two things:

The themes of Yoko Taro’s works have always been more important than the plot details, and I’m confident saying that holds true here. There’s a bunch of speculation I could get into about what this raid series is thematically about and how that relates to Taro’s other works, but that’s a separate matter I may or may not talk about. All I’ll say for now is that the central theming of this raid series and its associated story has been “belonging” from what I can tell.

As of the time of writing this, the story of the raid is technically not finished as we still have an unknown number of weekly lore updates to go through. I’m writing this now mostly just because I have thoughts I wanna express based on what’s in the raids themselves, but the new info we get in the coming weeks may change some of the conclusions I’ve reached. In that event I’ll either update this post or just delete it and make a new one.

Alright, with that out of the way let’s get into the connections themselves. If you’re not already familiar with the raid series, please either play it for yourself if possible or go look up footage of their contents in the order of their release, as I’ll be jumping all over the place with my thoughts.

The Breach Coin we get for clearing the third raid has the following description: “This round bit of metal, discovered within the machines' mind-bending tower, is stamped with the words "The Arc."” This obviously has a connection to the Arc that was shot from the original tower in Automata in Ending D, but I don’t think it’s proof that the forces we encounter in the raid came from the Arc itself. Given that the tower is clearly a recreation of the one from Automata as well as the entire point of the Arc in Automata being non-hostile, I don’t think it would make a lot of sense for there to be a causal link between the Arc going off into space and the forces that arrive in FFXIV. My best guess here is that the coin has this message simply because that was the original tower’s purpose and thus the recreated tower has surface-level similarities.

2B’s final message to the WoL states “We discovered that our enemies- the white-clothed androids, the tower, that colossal girl- came from that white orb. As did 9S and I.” Given what we know about the orb producing duplicates, I think it’s safe to assume that the 2B and 9S we encounter in FFXIV are also duplicates of the originals; this would also track with how much of their behavior and attitude seems more in line with their personalities near the start of Automata as opposed to the very end. This begs the question of how and why they and the duplicate Anogg would oppose the other forces produced from the orb, particularly considering that it very definitely appears to be a Seed of Destruction with all that that entails. It also begs of the question of how 2B will “save the memory of [her and 9S’s] reunion- and of the time we spent in this world, with you.”, considering that if they are indeed duplicates then they should presumably cease existing once the Seed is gone. My best guess is that 2B, 9S, and Anogg will continue to exist in some form on the “other side” of the Seed, and that perhaps the upcoming weekly lore will give us some more info on that. Though I do think all of this makes perfect sense thematically, so even if we never get a conclusive answer I wouldn’t consider this a big deal.

With the additional info we now have on Seeds of Destruction, I think we can pretty definitively work out both what they are and how this raid got kickstarted story-wise. The Seeds are “planted” in different worlds by the God/Gods (depending on translation) of the Drakengard universe and are the means by which their forces, the Watchers, can invade and wipe out humanity/mortal life. In the case of FFXIV, a Seed got planted on the First presumably by pure chance and began manifesting duplicates of the machine lifeforms from Automata as an initial invasion force meant to fully activate the Seed. Duplicates of 2B and 9S likely were not intended and may simply have manifested as a result of their deep connection to the fate of the machines, or they may have been produced by the Seed in order to create the false YorHa as a more effective invasion force than the basic machine lifeforms. Either way, the Seed is fully activated when the Red Girl duplicate enters it at the end of the third raid, becoming first a semi-formed and then fully-formed Mother Watcher (AKA a Grotesquerie Queen or Mother Angel, again depending on translation). Why exactly it took this long and this much effort to “activate” the Seed is unknown; we just don’t know enough about their inner mechanics to say for sure, though given how it took the magical seals in Drakengard breaking to unleash them it may be that they need a sufficient infusion of magic in order to fully activate. It’s also possible that Konogg and Anogg being the first to make contact with the Seed influenced what it drew on as inspiration for its invasion force; them being technology-obsessed could have influenced the Seed mimicking technological forces it already had something of a connection to (given the connection between the formation of Automata-era tech and the magic that was introduced to the world of Nier via the ending of Drakengard). But this is entirely speculative on my part, and I’m hopeful the upcoming lore will at least give us a little more info on this front.

Given how the Red Girl duplicate appears to “become” the Mother Watcher rather than simply summon her, I think it’s relatively safe to assume at this point that Mother Watchers are themselves born from Seeds, and given the ending of Drakengard 3 and how exposure to the Flower also leads to a similar transformation, I think we can also speculate that the Seeds are either born from or at least directly connected to the Black Flower. Given the symbolism of the terminology involved, the Seeds may literally be the seeds of the Black Flower (as many have speculated for a while), the flower itself being the God/Gods’ primary tool for destroying humanity or even a direct manifestation of them.

#final fantasy xiv #ffxiv #nier