my initial thoughts about AI and art 

I invite your feedback below

With AI-generated images there are a few issues that are intertwined and make the discussion sometimes difficult. I think the word "art" / "artist" should be discussed separately. Things that I describe don't have to fit necessarily to ALL forms of art. But it's about those I care about. I think it is important to state that A picture made by an AI isn't "art" automatically. I have issue with the term "AI-Art" specifically. An artist is someone who puts a unique - personally and most importantly: human - perspective in their work. Often it is a meditative process that takes a lot of work and takes time. Yes: there is definitely art that was made in seconds and is regarded as such. But I am sure in a lot of these instances the artists have condensed a lot of their personal experiences into the process that creates these works. A quote that I read somewhere (but have unfortunately forgotten where and cannot find it again) is this: "Art should never _just_ be pretty, or _just_ be political or _just_ be about impressive craftsmanship". I like it very much, because it works not by trying to define art, but rather by suggesting what art is _not_ (most of the time). with AI-art I often don't see much of the unique (subjective) perspective that great artists put on their work. There's no real story, no intent most of the time. 99% of it is pretty pictures that are "in the style of X and Y with a little bit of Z". I hear the argument about AI "just being a tool" a lot. And I think we should be more nuanced than that. There's a whole spectrum of "tool"-ness. A pencil is very much a tool. I didn't make the pencil, but I can still call myself the artist of whatever it is drawing - I have to move and do 100% of every stroke myself. I have almost 100% control over the whole creative process. There are more advanced tools where I have to do less and less myself, but I also give up control. The most extreme version would be to simply tell another person to draw a house or something similar. This would definitely not be art and not make me an artist (unless it's some weird arrangement of performance art) and I wouldn't really "own" it from a creative standpoint. I do essentially nothing and have another intelligent being that does the actual work. Like outsourcing. That would be the other end of the tool-ness-spectrum. AI is pretty far away from just being a pencil but not yet somewhere where I don't do it myself. I am still the artist, but I would argue I am less involved. Another factor that I want to mention: With the pencil every artistic inspiration and remixing of prior art is done by _me_. My art is 100% influenced by my experience and my life that shaped the neurons and synapses in my brain which make me _me_. Even if I cannot remember all my life, it is 100% me. With AI-Art I essentially take other art which has been condensed into an LLM by some outside company in a way I cannot really access or understand and delegate the image-creation to that LLM. I am less in the driver seat to put it nicely. Taking an off the shelf LLM from some company that scraped everything of the internet is neither a "personal journey" aka personal perspective (a subsection of all possible experiences in the world) nor a special one. A better way to create AI-Art in my opinion would be to train my own LLM or something or other more specialized neural networks. I think it's also important to not kid ourselves. Not everything we do and like and enjoy has to be "high art". A lot of what we enjoy everyday is also like nice tapestry or decorative ornaments without much artistic intent. I have the feeling that a lot of people who argue that AI-art is the same as "art" (whatever that may be) in midjourney- or stable-diffusion-subreddits haven't spend much time in art museums or galleries. I don't feel they've stood much time in front of great paintings. Because typically these paintings aren't just pretty, but have a _lot_ of important historical context around them that is enourmously important when inderstanding these works. They're not simply art, because they're pretty! I don't want to be snobbish or anything. But the fact that everyone with a computer can now create something that is indistinguishable from a painting by e.g. Manet doesn't make this person equal to Manet and the creations as much "art" as Manet's paintings. I think essentially creating art that is not using AI-tools makes it much more "pure" and undiluted. A really personal thing. All the inspirations and outer influences have been filtered through the artist before being put into the work. The more AI-tools are being used the less of an artist's personal experiences end up in the work. It's also often simple less "work" as it is much quicker. It's essentially a large suite of shortcuts. But true artistry is something that takes time and work to develop. That doesn't make AI-"art" nice to look at and completely worthless. It's just that ultimately (imho) art ist something that is made by humans for humans about being a human in this world. So in your case: I think it would be very easy to create many interesting images with AI. I just don't see how it would be different to what other people would create. I have the feeling a lot of AI-images will look very dated in a few years once the novely has worn off. The few ai-art-pieces that will stand the test of time will be where artists thought long and hard about how to involve them. They would also have made more of an effort than to simply pick one of the existing models and only use it as is.

CAD helped Draughtsman become more efficient, there may be a similar argument to make regarding A.I. and how we can use it. The revealing-forth of new technologies tends towards societal restructure. We could see changes to Copyright.

Good words, thank you. The monetary part can be of course painful for some, especially when corporations and cost minimizing thought patterns are involved. But eventually the feeling for true, meaningful art inside people will win. Sail the storm of change.

Any artist that currently or will rely more on AI image generation than their own talents, efforts, techniques etc are not good artists. People used to be a lot quieter about the shit they weren't good at, and I hate the argument that "AI image generation makes art more accessible" Cant draw? Cant paint? Cant do glitch art? put the effort in and learn. the satisfaction of learning a new skill and producing something you made is more valuable than some garbage a model spit out

Thank you for talking openly about this!

Thanks for taking the time to lay out these thoughts openly. I think the conversation regarding pure originality vs. derivation/plagiarism in art easily gets extremely muddled because many don’t know where to draw the line between Art (in the traditional sense) and REMIX culture, or particularly in this case, “Video Art” vs. “VJing”. The existence of sample-based music/DJing becoming mainstream in the world of 100% original compositions are the closest analogs we can look to here imo. Is plunderphonics actually Art? One’s position on that will color the entire discourse, and its hard to convince either side to look the other way. Personally, I think the entire ecosystem of art is kind of broken and impossible to find a middle ground on this topic until society’s overhauled its positions on copyright and IP to something more suited for these times. Sometimes I think about how, under the current way sampling in art is handled, this situation could have potentially led to arguably one of the only legitimate and practical applications of blockchain/Web3 tech lol. If every piece of media had a signifier attached to it that could trace when/where it was being used and tracked on a massive ledger, then the matter of credit and royalties could actually be tackled in a more meaningful way that could allow the Artist and Remixer to coexist in a more transparent way. But on the other hand, that could also be easily and very likely weaponized by IP holders and enable them to shake down every remixer and/or instantly block any creative appropriation at all, which would be hell, imo.

great video! it is def not going away, so might as well learn it as a tool

my friend works in finance and for him, using chat gpt at work is an open secret - he says that everyone in his office is using it to complete tasks but no one talks about it openly. he suggested we use it to reinvent my dj website, so we sat down for a few hours and the ideas that chat gpt generated (text and images) were so played out and "cheesy" for lack of a better word. this was my first real indepth use of a.i. and i was totally unimpressed. i have seen some cool stuff (including the b&w video you posted with the chains and roots) but i think it takes much more actual "work" and knowledge than most people are willing to put in. at the same time, what we are seeing is the beginning of a.i. and i definitely feel like i need to do more research to figure out how i can best use a.i. as an effective tool for what i do. important questions of our time. also, as a dj, i don't have intellectual property hangups in the same way as others might - "remix culture" has been a reality for a loooooong time and it is how i make a living.

I've had very similar thoughts as you regarding how far back does it go? "Creatives" usually have a desire to be credited, but in reality our work is built on the backs of innumerable people. As you said, all of the intangible influences that have made an impact on us throughout our lives, yes... but I've often wondered also about the people who developed the software, manufactured the hardware, get power to my building, built the building I work in, transported the materials there, sourced the raw materials, etc. The network of influence that is the foundation for our lives and work is incomprehensible, and none of those people are receiving credit for the very real role they played in our ability to create the things that we do.

I agree that AI art with music is fabulous. Its amazing to see how AI brush strokes patterns together to patch images on the fly based on the input/settings given. There's always been software to visualize music, AI is just another medium. =) and looks great as a backing layer

Its the same thing as any DJs music thats at core mostly samples. Become something resembling more of a piece of art or content based on the style that the person injects himself into the peice. Alot of factors at play. People have just had longer to adjust in some mediums than others. . You cant ever truly own anything. And if you're living right. You won't want to

If I created visual art and you were somebody who I took inspiration from, am I stealing from you? I feel like people are always taking ideas from inspiration and building on them. Whether its an LLM or a human learning from your work that is reproducing something in a similar style, is that stealing?

I agree with your points. When photography came to the forefront people who drew and painted felt that they were going to become obsolete. That never happened. AI art is just an evolution of the process. We’re moving at an ever faster pace and the things we create are going to change as a result. For me the idea that all artists are just an amalgamation of those that influenced them is the same as an AI. Just it takes artists years to do that. It’s just now you can expedite that process. Soon there will be a difference between good AI art and bad. People will be able to tell the difference. Appreciate this post. This is really just the beginning. Btw loved all the music and visuals but that last track ❤. It’s Menko but what was the name of it. Gotta head over to the bandcamp and get the EP.

I quite disagree with you there, it is easy to say "the most important thing is the visuals quality" and ditch away all the ethical concerns here. It is not about the outputs. It is not about remixing and the end usage being derivative or not, it is about the tech being build on stolen content. Training a model on copyrighted content is making an usage of the said copyrighted content, especially if the model is meant to have a commercial usage. And both Stability or MidJourney are making a commercial usage out of their tech, big time. Let's make an harsh comparison, let's say some product is made out of slavery work. It's a really good product. Can we say the slavery behind does not matter ? AI models trained from scrapped content are made out of theft. The tech is here to stay that's quite sure. And I do enjoy it, I'm learning it and it's fascinating, but I before integrating it to my workflow, even more commercially, I am waiting for models trained on content that is ever public domain and/or from people who opted-in. On the ethical questions of thefts, I'm not against theft in every case, like the world is unfair the inequalities are huge, if poor people steal from the wealthy that's make the world less unfair. But that's not the case here, on the training of SD or MJ models it is techbros in big companies stealing art from creators, and other creators getting some crumbles from this heist and saying "hey I got something so that's not bad". The problem is really not the results being derivatives, the problem is by using it we are supporting those companies when they do rogue actions. We can do better. And we can do something about this, like for everything if a company is doing shitty moves as consumers we can boycott their products and publicly shame them for their actions. Actual activism. We are artists, our actions are seen (quite literally for us visual artists). And if I may add, a larger problem about AI tools in general is it emerged in our capitalists society. Sure some individuals can empower themselves with those tools, but in the end for the largest part it is profiting to the people who own the means of production and who will mass deploy such tools. We already replaced a lot of workers by machines, now we are replacing many jobs that can be done by someone behind a computer. The less qualified ones first, the more technical ones later. If more profits can be made by ditching people and replacing them by a cheaper workforce gosh it's gonna be a good time for the bourgeoisie. If we do not step up against big companies when they are making unethical actions, and we still use their products, then we are supporting those actions and it makes it easier for them to do make this kind of moves.

I see AI in the video art production, almost the same way I see using samplers in music. These days it’s harder to develop a “unique” / “original” / “one of a kind” artwork of any kind. Or at least, anyone that views your artwork will have an other artist to compare yours too.. more often then not. I’d say that, personally I am pro AI, the way it’s opening creators art too more endless possibilities. And yes it depends on the end result of your work, and if it comes out looking too much like an artist you referenced from AI… that’s “lazy”. But AI can be used for good. And I can see the creative world evolving through AI. With each artist either for or against AI.

moral and legal stuff aside, if people can tell your stuff is AI they will likely call it shit, but only if they can tell.

I am not going to comply in making my self and other artists obsolete ....no more then I give auto tune any credit ...As an artist ( legally) I always retain the write to my images even if some one else owns one of my paintings I am the only one who decides who may reproduce it...what if you were born with an art talent and you took part in an art show and some guy with a computer makes digital pics and sells them for more then your paintings? you are supposed to be rewarded for your talent .to imply that some one with a computer program is more talented then some one who is a naturel born painter is laughable. that's why art is pricy. I have been waiting over 4 years for a PREPAID chromagnon. now its gonna be trash because of AI? at the same time i do realize that artists of to day are all using tools to aid them. even in the 1500s they used a candle light through a key hole to make pounce patterns. so you have opened some what of a pandora's box. if an artist can produce something really great with out AI that the public loves over something made with AI, that to me makes it more valuable then something made with AI. AI may force artists to work harder and do even better.

I've very extensively experimented around with locally hosted SD and custom models like you said with different weights of influence from different artists and models, various hot new fresh gAI solutions that gets funded every month or so because they want to sell seats to our studio, and actively use gAI adjacent tool in my art (but this tool is distinctly and specifically ethically made and licensed with consent) The conclusion I've very easily reached is, gAI as talked about in the video is a neat tool, but nearly useless in a meaningful capacity for production work, and ethically unusable. I play around with it, sure; but never would I attach it to my creative output or name. It's no different from going on Artstation, taking someone else's work, tracing it, and calling it mine. Doesn't matter if I cut up 100 different images from 100 different artists and amalgamate them into an image, it's still not my work and it'd be professionally, ethically, and creatively embarrassing. A lot of your points are just questioning if it's really stealing or if it's really derivative. Answer is yes, it is. The end results do not change that fact. Would you cut up an episode of Family Guy and use it in a set? If the answer is yes then you never really cared about theft or derivative work. Transform it all you want but the reality is you're still inputting someone else's work and masking it as your own. It doesn't matter if you ripped the episode yourself via DVD, doesn't matter if you got it off someone else uploading it to RU-vid, doesn't matter if you got if off a TikTok or a website for Family Guy clips, doesn't matter if it came from a machine that chops up the episode and creates interpolations of it. I work in the 3D industry. Your suggestion that complete shunning and rejection of gAI is impossible is honestly hilarious. Maybe 6 months ago, sure, now? Man. Nothing is technically usable, legally it's a ticking time bomb. Where the suits were quick to adopt it, realistically it never stays in the production pipeline because it's technically incompetent and near useless for any meaningful work. Outside of slop garbage work that comes to us, it's useless. For a whatever ad about effervescent pills? Sure. For our actual creative works we intend to deliver and put our name to? Never. gAI is relegated to strictly menial and meaningless work, and that's the only realistic use case for it. It's the budget option for when we don't want to care, and the results have always reflected that amount of care. Putting your brand or name to gAI output is slowly but surely being a sign of a lack of artistry and most jarringly, budget. On "inspiration like a person", AI does not learn like a human being. It apparently doesn't matter when a neuroscientist or AI scientist repeatedly confirms it is nothing like a human being, the notion that "oh gAI is like a human being" is ridiculous because we do not have a infallible index of input data nor do we interpolate that data the way gAI does. Again, gAI is just very extensive interpolation of the data stored. It has been shown to be able to replicate almost verbatim the training data used as output, even when not prompted to do so. You don't get to hide be "ohh but it's not me the AI did it!" when it turns out your generated image is a 100% compositional match to an existing image you don't have the creative liberties to use. There are ethical ways of generating assets for art that don't rely on stolen work, they already exist. An example, vocal synthesizers have existed since the early 2000s and have been adopted in Japan without problems or ethical issues since. Today the trend is moving towards AI assisted vocal synthesis, and yet the vocal synth industry isn't affected by the ethical issues that plagues gAI, ever wonder why? Why is it only when compared to the new generation of AI voice cloners does it ever occur, out of gross misunderstanding of the sheer differences between them? The point here is this, do you actually care about the ethics of the tools you use? Do you have any problems using some else's work wholesale without permission or acknowledgement? Would you take some else's entire set if you could? If not, where's the line you won't cross? How much do you obfuscate someone else's creative works until you deem it acceptable to call it your own? Are you arguing for ethical derivative use, as it's done in vocal synthesizers where the voice providers have always been well compensated if not outright celebrated with high profile releases and recognition for it, with explicit understanding of the tool they are helping create and the use case that surrounds the creative culture around it, as a creative tool with meaningful control over the result (for a more relevant example for you, let's say like using a LORA pack from someone who creates a bunch of textures you use, but only trained on their own data and only when you buy the license for it); or are you arguing for gAI that scrapes the voice of a musician you enjoy without consent, where someone out there can just take it and call it their own work without any credit, compensation, or acknowledgement of the source (a more relevant example, you just took the textures you liked and trained a LORA without any permission from the creators and called it your own creative work)? Even if there were an ethical checkpoint for SD, would you use it? Or is the entire draw of the tech so that you can slap someone else's creativity on it, but not too much because you'd be very rightfully called out of straight up plagiarism, so you slap 4-7 different LORAs of different artists to obfuscate it, THEN pull the gAI slot machine lever arm until you get a result you want?