Honestly the last 2 years make me not want to pick a brush ever again. Complete garbage and antagonistic world we live in if you are a creative or artist. We get paid peanuts and they still feel like replacing us with the same work they stole from us.
I’m a computer science student but also an artist before anything. I feel for you guys and was thinking of what we can do to protect your guy’s work. A way to encrypt / copyright your artworks from being used in a data set in the AI. This needs to be a legal issue in a court for this idea to be implemented. AI is still a new technology, and because it’s new, we haven’t yet set any limits to it. Keep speaking out on this so we can
@@oredaze "It's too late for that, the murderer is already murdered so many people. We can't punish and put laws for the murderer anymore to prevent more case in the future. Its too late" - basically, you.
thanks for the video and the guidance ! I personally am more affected mentally by the ai-art stuff than i would like to be. Its frightening because i feel like i nearly got to make this step of making a portion of my living with art and now these companies shut the door right in front of me ^^(because i think its affecting mostly the lower skillbased jobs like book-covers,cd-covers...).
I'm in the exact same position. We just became expendables and most of job positions that were there for us to learn better skills while having any mony from it are disappearing right in front of our eyes.
I really feel this too. As someone just trying to break into the field, it feels like an already difficult path just got way harder. The 'lower' jobs are how some got their start.
it's on purpose because the corporations want us all to be stuck at their dead-end jobs. That's why college is impossibly expensive, and now the few things you could train for without college, they will do anything to keep us from being able to be independent from the corporations that want to work us to death.
I am on the same boat. 8 years of effort teras and hard work only to be soul-raped like this. The moment artists entered the digital area, this is the moment when we lost the game, because it was thanks to technology, even thanks to Photoshop that we could create everything that we have created so far...
Great video. But i think you gave up on legal action too quickly. There's also music AI and developers only using copyright-free samples because they fear legal issues. Visual artists just dont have organizations to back up claims. I think court should be our main avenue of resistance, and if it fails we will lose nothing and do as you suggesting.
The technology will keep evolving, in bright daylight or underground. You can't stop it now. Keep throwing lawyers, politicians and cops at the problem, though.
No, you're misunderstanding why Music is different, and trying to fight AI from a intellectual property perspective is the LAST thing artists should be doing, and may actually end up opening up actual human artists to legal liability. I'll start with the latter before looping back to music: An AI which looks at tens or hundreds of thousands, if not millions of images, and uses them as a reference to generate a new image (excluding situations where you're asking it to specifically generate an existing copyrighted character, or just to spruce up a specific existing copyrighted image vs generating a brand new one) is almost certainly fair use, even under how strict and draconian copyright law is: The impact and influence any 1 input image has is minimal. If anything, it's more obviously fair use then a human artist looking at references, which will just be 1-2 images and would be more likely to show the influences of the original work being referenced, which is important legally to the standard of if a work is "transformative", a ley part of the fair use determination. And regarding the issue of AI replicating specific styles, like of Junji Ito or Sugimori (of Pokemon fame) that is even MORE clearly fair use, because "style" is not something protected by US copyright law, at all. So, if there if there's a ruling or law that establishes these AI's aren't fair use or transformative, or that "style" now CAN be copyrighted... well, that would have implidcations for a human artist, too: If a AI piece which isn't obviously similar to one of the original pieces it was sampled from isn't fair use, then a piece of art a human artist made where the reference used IS obvious won't be fair use either. Suddenly, you could be sued because your piece has a similar pose or lighting or composition to another. If Style is copyrighted, it's the same risk: Now you can ge sued because your piece happens to be in a similar style. You know whose gonna benefit most from that? Large corporations, the same ones everybody is worried will use AI's. People seem to have this idea that by fighting AI art, they're fighting against corporate exploitation of artists, but we have consistently seen that ANY time there's been a restriction of fair use or an expansion of copyright protections, the big corporations are the ones who benefit, and they use those expanded laws to go after more people, while they're too big to sue themselves so they can continue to break and bend those rules without consequences. As far as music, the reason why Music AI's are being handled differently is because copyright law for music itself is different: There are only a limited amount of notes that exist, vs the infinite combination of shapes and colors that exist in visual art. It is COMMON for musicians to come up with similar or identical beats or sections of music tracks as a result of that. There have been many high profile cases where songs that aren't actually likely stealing from another were ruled to be infringing as a result of this. So it is almost CERTAIN that any music generator (not even one that even HAS inputs, but just one thatr's totally random) will spit out music identical or nearly identical to some of the music it used as an input, and there'd be real risk of a infringing legal case.
@@MajoraZ so very open question here, I'd say that there should be new laws about companies defult opt in of using user generated content for such things like AI ? I don't have issues the the tool exist, but do have concerns on how it came about? I don't know anything about music either, but from what i hear, you do pay for sampling at least? what are your thoughts about moving foward on this with art? should I keep posting online? should i switch to traditional? do I just keep things local? How does this actually effect ai ? will it stagnate as its not givinn as much new stuff?
Honestly I just see this as big companies trying to monopolize visual art on a whole, When you think about it visual art is an entire industry that for a long time was very difficult for big corporations to fully exploit and there's no denying that there's money in this field despite public perception. Honestly this might sound conspiratorial but I think they just created these apps so that they could just have huge monopoly on the industry and tried to disguise it as a means for everyone to be creative. The big corporations can save money by laying off teams of artists and not hiring freelancers and at the same time make even more money by distributing these software to naive people fooling them into thinking they can make money by selling prompt art, but in reality something that takes this little effort to make will have little value as the market oversaturates and even the biased people who were all for this AI art progression will soon realize that the prompt art is worthless due to the fact that anyone can make it by typing a prompt, I mean why buy AI art anyway if you can just type in a prompt and make it yourself and it's going to be particularly difficult for your products to stand out while using AI art for example, if you have a shirt business that uses art, what's going to stop people from ripping off your designs or how are your designs going to stand out in a sea of easily generated content in an already oversaturated market like that, doesn't seem like a particularly beneficial tool for the public in the long run if you ask me, a good way for small business owners to fail actually. Meanwhile big corporations just take in all the money they can get with the exploitation of art while simultaneously killing the medium, at least as way of making a living from it. Don't know if you agree or not but honestly this really seems like the big plan they had...it may offer benefits for people who cant afford freelancers for small projects like book and album cover illustrations but with advent of AI music and literature those facets of art could ultimately meet a similar fate MAYBE with the exception of music since the copyright laws for that are incredibly strict. I just don't get how people don't realize the easier something is to make for the average person the less value it has in the long run, that's why low skilled jobs pay so much less most of the time. To really some it up in my opinion this was really sinister in it's inception or apathetic at best, lets create a database with peoples art without their permission for an "AI" that spits out a derivative while we're simultaneously profiting off all the time they spent perfecting their skills and disguise it as a useful tool for everyone. They're also well aware of how slow copyright laws are to catch up so they know by the time laws are put in place to protect artists the damage has already been done. I don't blame the casual people using it but the huge Corporations who funded it as well as the biased tech bros who keep spreading the lie by how useful it is and saying that their aiming for some non existent utopia where AI does everything for us and we won't ever have to work anymore to get by which obviously won't work economically with how our society functions and it's like putting salt in the wound for them to scoff at artists Work while using these AI when it's literally trained off artists work. I guess there's some solace in knowing this "art" is largely worthless as it becomes more mainstream and becomes oversaturated. The real question is though, why do we as humans strive to try and make ourselves obsolete, I can understand the automation of jobs that most people don't really want, but it baffles me why we strive to get rid of the careers we actually strive for. Anyone who does art as a career obviously did that because they love it (I mean we literally have the whole right hemisphere of our brains associated with creativity), and I know society is about progress and efficiency and whatever but I honestly don't see the automation of art as a means of driving society forward at all, all I see is our brains stagnating and becoming creatively bankrupt as big companies turn out content on a conveyor belt and even if you decide to make the difficult choice of somehow transitioning to a new career field, which would obviously require more training how long before an AI takes that away too? Rant aside I wont ever stop making art regardless and any of you reading this shouldn't either if you're as passionate as I am. Art made by your own human expression and life experiences will always be superior to an AI derivative so please don't give up, these corporate capitalists are going after as many fields of work as they can to increase their profits.
your worded this perfectly, as an artist taking my first ever commission from an overseas client,it made my day a little better knowing that others too think the same way about the dark inception of this whole thing thanks
@@pumpkincat7994 I'm glad this helped and I'm slightly more optimistic now that more people will see these AI for what they truely are. Soon the general public will realize in using these AI models none of the art they use for their products will stand out in this endless sea of easily generated content and it will ultimately be more detrimental to society in general if people keep using them.
we didn't strive for this, its just the rich, the people with power and influence, the big companies who benefit the most from AI art, they also have the most financial power to influence where technological development is funded, most people arent in favour of AI art and its not really a field worth research and dev into in terms of actual benefit to most humans, but with the system we live in thats just how it is, profit over human happiness.
I highly highly recommend rethinking the legality issue of the training data, dear Nathan, as many others have already pointed out. Connect with Steven Zapata, Greg Rutkowski, Karla Ortiz or AI researchers like Abhishek Gupta (Montreal AI Ethics Institute) to maybe gain a different perspective on it. We absolutely need people like you to be not just simply dismissing this.
In general I agree (like always) although the legality point seems to be wrong - in EU (and I think UK but I don't know their current law) it's illegal to take your work and reuse it without your permission plus it's illegal to take anything and tie together with your name if you didn't give them written permission. And they will start getting hit with lawsuits any time now, all is in preparation, I saw some law firms already officially confirmed that they are preparing such lawsuits. What personally bothers me the most, aside from just stealing artists' hard work to fill AI databases with them and then sell it, is people like me - beginners preparing to get their first job in any studio remotely are now expendable and won't have internships or job positions where they would be able to learn proper work (not to mention while having any salary to live off, which is especially hard now during inflation). Well-paid jobs for great artists will stay - sure! But all of us below, us who already were seriously struggling for survival, might be just gone. And I know that Schoolism or other courses would help with building skills and all that but many of us already right now can't afford it, many of us struggle really hopelessly to learn after their day job (if they're lucky enough to have one). And then I open Instagram, see some guy who says he's disabled but thanks to AI he earns over $10k a month with his AI "art" and I see interview with him where he calls all of us, artists, even including those whose art was stolen to create his "art" he sells, losers that soon will extinct thanks to amazing AI tools and that people like him "professional prompters" will replace them, selling prints and soon also commissions because he spends 2-3 hours a day constantly typing prompts and gets many abominations but at least several good pieces a day that immediately go to his print shop. At this point if more lawsuits and regulations won't hit them and if the AI protector (that I already saw is supposed to be slowly created to delete all copyrighted art and pieces created with copyrighted art from AI databases) won't work we're simply doomed and we either starve to death or give up the only job that gives us job, which is being artists. (and lets be real, doing a different day job and then struggling with mental exhaustion doesn't cut it)
I read the same interview! It got me so mad! Not only he uses others' work (and types artists' surnames) but also then he mocks them - people who worked hard for years, some of them for decades, to learn all that.
And now I saw some jerks using Sam Yang's art to make AI that replicates pictures in his style. I also saw that Luisa Preissler's "fan" wrote to her that they fed AI with all her artworks they could find and now they're recreating new pieces "in Luisa's style" and they asked her, only "out of respect" if they can start posting it on their art account. That idiot even added that just in case she says no, it's all legal.
About copying art styles, is that even copyrightable as of yet? I think I read that it isn't because otherwise we would have monopolies of styles or something like this, but I'm not really versed on copyright laws
@@Sweaper At this point it's not about copyrighting styles though but reusing one particular artist (feeding AI with all their hard work and then typing in prompts to create something in their particular style) - it's highly disrespectful and a real D-move
It's a good question. To my understanding, copyright laws allow repurposing of existing artwork. As soon as we try and define what art is and what art is not, we get into deep trouble. For instance, Andy Warhol famously repurposed copyrighted images to great acclaim and became notable in modern art history for it.
The last I've heard it it legal issue and it breaks several copyright and database laws in EU and there are already law firms preparing with artists to sue AI owners
Saying AI art is "using someone else's work" is like saying you are using someone else's work when you study the great masters. The AI is trained on them, it does not have the image inside some sort of database.
Thank you so much for making this video. I've gone through many of your courses on schoolism multiple times! Your advice is always a guiding light to me and this bit of advice is no exception.
I'm joining this conversation late. I am a career artist, formerly Disney, marvel and stuff like that. I've been watching algorithmically generated tools grow and expand since the early 2000s and knew this was coming. Natural evolution of the technology. Not one bit happy about it, but, also powerless to change it. Here's the ong and short of it as I see it. I don't see any way that this does not replace the majority of artists inside of 5 years. There are specializations within the professional art community that will hold out longer and maybe always retain a small contingent of working professionals, supported by a holdout anti AI community if nothing else. We are early on in what we are calling AI art creation, and it's exponential rate of growth is going to blow everybody's mind and very short order. No area of the broader art profession is safe. I've worked in animation modeling illustration marketing, you name it. All of this boils down to data sets. Any existing style that isn't a one-off random piece boils down to a repeatable definable data set of shape language, color palette, etc. I've spoken to a Kotaku writer about this and they have covered it somewhat, but here's the part I don't think anyone is grasping. There are probably tens of thousands of students in art schools and universities right now incurring tuition debt for a profession that will no longer exist by the time they graduate or will exist in a greatly diminished way. Regarding ai's need for artists that will always exist in order to guide and direct it. I hate to say it but you're wrong about that. The same way in art director or instructor guides and teaches a student or Junior artist on composition, use of light, negative space, etc. Those things can and eventually will be part of the AI process. I'm not a negative person and I don't want to be a doomsayer but I don't see how this doesn't transpire.
Well, i myself was working in those mid tier jobs that are basically vanishing as we speak, in 5 years span i see almost all of the art jobs going to AI, its going to be a hobby, maybe traditional art gain value that way i dont know. But since creative process is just input vs output then yeah art is dead in a way that working as an artist is dead.
I can see why after looking at your portfolio, nothing about your work stands out as anything inspiring or interesting. It's just employable gruntwork that shares no real human emotion. Of course you will be replaced as a cog in the machine, its time to shack up and start expressing your true emotions in art. What is your opinion? Why are you different?
Pretty much my sentiment, there's nothing an artist can do that AI can't with brute force. This is an existential issue that needs to be addressed and regulated by law, otherwise we give it a free pass to obliterate art as a job.
@@ranzu3138 Well it wont be uninvented, and our jobs wont be saved. Thing is that they just went it trained their AI's on all the work we did to just throw us out of the window. Ai's training models need to be regulated. Because they do not respect any copyright.
I just now watched it, very thoughtful and well reasoned. I have no claim on seeing the future but I suspect that how Steven and we all believe things should be, and how things actually will be, they'll probably be very different from each other. Doesn't mean we shouldn't fight for what we believe is right, but as I said at the top, I see tremendous inevitability in all of this.
A few points. The AI doesn't stitch together the source images. In the training process, the source image is completely scrambled and covered in noise. After that, the software removes the noise bit by bit, checks if it resembles the initial image and is assigned a score for it. It keeps doing that over and over again until the "unscrambled noise" vaguely resembles the original input (it never goes back to the original, because both the addition and the subtraction of noise is random). For the generation process there's no source image to compare to anymore, because it starts directly from random noise, and it's the human who fine tunes how to remove the noise and when to stop (via use of prompts and steps settings). The examples you see on display on those websites are curated. It is true that a single image takes second, but in order to get an acceptable result you need to understand what is completely wrong and what looks "off", be it shapes, proportions, framing, pose, lighting etc and fine tune the prompt and the setting in order to get closer to what you want, and it can easily take hours and hundreds of generated images to get there. In this case, artistic knowledge and expertise is essential to know how and what to fix. When people say it's a tool, they mean it can (and will) be used to give artists a good headstart and customized reference. Creative and technically expert artists are still going to come up on top, because they can direct the ai and can use its results to the fullest.
Steven Zapata did a video essay about these programs. They are not designed as tools but as a replacement for artists. And it shows, because as tools they are really lacking about user inputs and controls. Ultimately the result is not "your" art, it's a rehashing of all the works included in the datasets and it doesn't reflect your personality, you didn't earn it. Artists should take action and demand removal not from the prompts (it's next to useless) but most importantly from the datasets. These programs would not exist without the works of artists poached and scavenged in the datasets. The companies designing these programs profit from our work for free, it's unacceptable. I agree that young artists will be tempted to take the short way for how easy it is. Then how are we going to experience new styles, everything will be some rehash of already existing works. A personal style can manifest in very minute things like just the way you draw a line or a circle. There is no way someone can really develop something truly personal with these the programs alone because how loose and scattershot the control is.
As somebody who doesn't particularly care about AI's, but DOES care a lot about Fair use and copyright law reform, I had a fair amount of issues with Steven Zapata's video. I'm gonna copy the comment I left there here: "Sorry for the 8 paragraph comment: I'm not an artist nor AI art advocate, but I am somebody deeply concerned with IP laws, copyright, and fair use (IANAL, just a IP nerd), so my stake in this whole issue and debate is from that angle: I think that the fact that the datasets various AI uses is sourced from groups that have research status and then the resulting AI algorithms can be used commercially is pretty interesting, and I agree is sketchy, but i'm not sure that it's a smoking gun, either: If the datasets are LINKS to images, rather then downloaded images themselves, then it is arguable that the dataset isn't actually a "use" of any of the works itself. I say "arguable", because the act of opening a link DOES create a local copy of a file, so the training that produces an AI algorithm still would be reproducing the images. I, personally, don't think that the local temporary creation of a image or audio file etc should constitute an act of use or infringement, because practically speaking, somebody isn't actually saving an image every time they look at an image online even if technically that's how computers work: Opening any page on the internet ever would constitute copyright infringement. It is my understanding that LEGALLY speaking, this is sort of a grey area and different rulings have handled it differently depending on other specifics. I'd probably say that training an AI on images via links should constitute a "use", but that's just my opinion, not me saying that's what I think it'd be legally, I'm not sure. I do, however, think that the resulting AI would be considered fair use, even if the use of the images is considered a legal use. The legal standard for what's considered fair use is complicated and varies by country, but in the US the main question is mainly if a work is "Transformative" (though there are other factors too beyond that). What counts as transformative is ITSELF a massive, complex, abstract question, but explained as simply as possible would be if the new work significantly restructures or recontextualizes the original into a new thing to where the original's influence is nonobvious at a glance. This is, again, a VERY, very simplified explanation, to the point of arguably being misleading, but it's the best one I can give. I would personally argue that fair use determination should actually be much, much broader, but even in the current IP law status quo, I think an AI image generation algorithm would almost certainly qualify: It's not even an image itself. It's essentially an entirely brand new "thing". My best analogy would be saying that an AI isn't fair use would be like saying me typing up a text description of the mona lisa is copyright infringement of the original painting, which would be absurd, but there's even MORE separation here then that because the AI isn't a description of any 1 image, but of trends between THOUSANDS of images, where the influence of any one inputr is relatively minimal, at least as I understand how AI works. It's actually more like saying a book that describes lightning techniques is infringing on the copyrights of all the images the author looked at to see how other artists do lighting. That's obviously absurd. This is also why I think the actual output images AI make would ALSO likely be fair use in most cases: There were thousands, if not millions of images used in the datasets the AI is trained on, and the impact of any one of them on the final work would be relatively minimal. You;re correct that the AI doesn't just use references like human artists do... it does it in a way that is MORE obviously transformative then a human artist using references, where there is a lesser number of them and in many cases the influences used will be more obvious. Frankly, I found most of your arguements in this section (the one at 31:32) sort of irrelevant: that the AI is an AI and doesn't have to be skilled in the way a human is doesn't matter in a determination of if something is fair use, and it shouldn't be (Should an artist that draws a rendition of a copyrighted thing poorly be fair use while one doing it well be infringing?). "Style" is also not something that can be copyrighted (Nor should it be: Should artists be liable if they happen to draw something stylistically to another person's style?): Infringement is done on if a SPECIFIC work or character is being copied, and AI art, by it's nature, is generally spitting out a new output... ...unless you ask it do otherwise. Asking somebody to make a piece of art in the style of another artist would probably be fair use, but asking it to make an image of a copyrighted character doing something would probably not be, because CHARACTERS can be copyrighted. I, personally, think that something like noncommercial fanart SHOULD be fair use, but it isn't. Likewise, i've seen cases where an AI isn't generating a brand new image, but is filling in details on an existing image, or "uncropping" it, to fill in space outside the images current boundaries. In those cases, it's pretty obvious what the work i'ts building on is, and I suspect these would be ruled as infringing. As far as why Dance diffusion is different, I don't think it's because AI companies/authors think lesser of visual artists (tho they may, IDK), but rather because the nature of music means that there is only so many ways songs can be set up, vs the infinite variation of visual art: Two musicians will not just sometimes, but OFTEN come up withy similar or identical beats or sections of tracks in their music because there's only so many notes. There have been many legal cases where songs that probably weren't actually stealing from each other were ruled to be infringing. So ANY piece of generated music that's relatively simple in composition is almost invariably going to be nearly identical to at least one other piece of music. This is true of human made music, too. The entire way music coporyright works is sort of different as a result of this, hell there's even mandatory licensing for music, where record labels HAVE to allow people to use their tracks (Personally, I wish mandatory liscensing was a thing for visual art, games, movies, etc too; and I think that music that's only incidentally similar and not proven to be copied SHOULD be fair use, but that's just me; but even i'd admit I'm not sure how AI generated music would fit inot this) I want to loop back to something I brought up at the end of the paragraph above the last one, where I point out how trying to say skill or difficulty should determine fair use status, or that "style" should be able top be copyrighted, would create massive issues with liability for human artists. That, I think, is the bottom line here: If there WERE to be a legal ruling that the act of training an AI on other works is infringing, or that the new output work is, there would be absolutely massive unintended implications for human artists too: It would severely limit less contentious, more obviously benign instances of researchers training AI, and of human artists using references, and open both up to massive liability: Imagine being sued because somebody else thinks you used their work as a reference for a piece of yours, since the two are as similar as an AI's input image and the output. Or because your style is similar to theirs. You know who would benefit most from that? The same large corporations you're worried about here. The ones that already send out mass DMCA claims on content online. They would absolutely go after smaller artists, who can't afford to defend themselves in court; and those large corporations will also feel free to do the same things they accuse others of doing, because they're too big TO be sued by smaller artists. We see this all the time now: Fox stole some guy's youtube video, included it in family guy, then hit the original upload with a takedown. The "you wouldn't steal a car" antipiracy ad used stolen music, etc. As I said at the start of this comment, I am not an AI art advocate. But I do think trying to attack AI generated art from a intellectual property/copyright angle is a bad idea: Both because mechanically speaking I think it is often fair use and should be, but also because I think any laws or rulings would have grave unintended consequences for human artists. I'm sympathetic to the labor concerns artists have, of AI art replacing human artists and putting them out of a job, but It needs to be addressed in other ways. EDIT: Also, something I forgot to say I don't really have a good way of segueing into: Even if using copyrighted art as inputs is probably fair use, since the law doesn't really make a distinction between how much skill is involved or if there's a human creative vision in fair use dermination, to my knowledge; there IS a determination of if there was human creative input if a NEW work gets copyright protections: There's been court cases establishing a Monkey picking up a camera and taking a photo a photographer has set up doesn't give the photographer the copyright, and the photo is instead public domain. If there is a legal avenue here that is likely to succeed and not be as destructive to fair use here, it would be argueing that the output of AI is public domain"
Steven made some good but points for sure, but sure it's not made to be a tool, no one said it was, but it CAN be used as a tool. It's an artist's auto tune, and the Pandora's Box has already been opened. No going back now.
People should work together and stand up! We can battle this. The music industry had this before and they won't touch any of it because of legal issues. There is no difference in art. These companies are trying to monopoly visual art
There are already primitive generative 3D AIs out there, and given the rate of improvement to 2D generators, I don’t see it taking very much time for them to be just as effective, other than the current implosion of seemingly all the big tech companies.
I'd rather say that there should be a restriction that every single AI piece has to have visible watermarks all over - if it's supposed to be a tool artists will deal with it and ordinary thieves won't be able to use it as their own
@@briannaalejo9226 Yes, and to do so they would need a lot of time and a bit of skill. It would stop them from flooding internet with hundreds of thousands of pictures that AI makes within seconds and it would discourage most of them because they are lazy thieves
@@annabellepointer7494 It doesn't take much time to remove watermarks, nor does it take much skill. But I do agree that there needs some form of encryption/copyright for artists work that prevents their work to being used in AI. Anyone can take your artwork from instagram, twitter, art station with a simple screenshot, and make a category of your style in a data set of a collection of your work. Either artists need to copyright all of their work, or move to a platform that easily implements a form of encryption to an artist's work once posted.
Hi! professional artist here. I believe you have quite a few points Nathan. I would like to add my two cents, trying to be rational and neutral. The end game of AI is not 3D generation, but Completed animated clips in every style (3D, live action, and 2D) with AI generated voices and performances. Right now AI is getting over technical details (Hands, warped facial features..) and the next step is gonna be solving the problem of consistency (cannot generate proper comics or boards) THe last feature to conquer is the Illussion of life, and storytelling. It's extraordinarily bad at doing it, and since that it highly specialized knowledge, it may take years till they get it right. That leaves the most basic foundation of drawing and painting as the golden skill to have "Gesture, Acting, and storytelling in the form of emotional direction" Even if AI could do that, an Amateur wouldn't have a clue of how to command those features, the formation is still necessary. A Jedi can do wonders with a Laser Saber, a kid would only do havoc and predictable shenanigans, that soon will become so common, they will be perceived as "Gaudi" like the Microsoft Word illustrations back in the 90's.
AI revoluton made me stick to my comic novel even more - I need to push it out and see if people care about human created Indie comic. With all it's human imperfections.
I'd hate to be an AI user (non-artist) who submitted a piece to a publisher and was asked to make a few minor alterations. What would I do? Or a book illustrator who had to compose 32 illustrations, with the same character. AI seems good for one-off paintings but, from my dabblings with Midjourney, you have almost no control over what the software will spit out. Like Marco Bucci says, we should use our own AI - "actual intelligence".
It's interesting to hear your thoughts on the topic, thank you for sharing them! I do wonder how AI could be used as a "tool" for artists, though. I have a hard time picturing it as such, as opposed to actual tools that allow(ed) artists to express themselves in new ways and inspire new ways of creating art (such as digital art tools, 3D, or perhaps even camera obscura back in the day). And with all the shady things related to those machines collecting and processing data, I wonder if I would even consider it ethical, even if I was able to find a way to use it as a tool. Artists don't want to save their work from being devoured by those without a reason after all. That being said, if somebody would like to share views that differ from mine (respectfully) with me, I'd love to learn and see if there is something to change my mind about. Not having to worry about AI being an enemy would be lovely, but frankly, the main thing triggered by it that I see is a lack of respect for artists, which is rather ironic and doesn't make me feel very optimistic about the entire thing.
Hi! there are ways to ethically use it. I am still testing it (not claiming or posting anything) but so far, I figured a very rudimentary, ethical workflow, with a few rules. 1. NEVER use a stable diffusion trained with an artists specific work without their consent. That is crossing a line. 2. You may produce an image, or several, preferably on a secured way (downloaded SD or Midjourney private mode) and use it as an initial brainstorm like you would in any google search. These cannot be presented on professional work, only used as an idea generation for color schemes, general shapes or composition. You will have to paint the entire thing, do a line pass that is your own, and recolor from scratch. So far, I don't see any ethical difference between AI and using Pinterest or google as reference gathering. It has the plus of probably blending several styles together, which will help you further your own in a way that you are less likely to infringe someone else's copyright. If you find an ethical problem with this, please let me know. I am forming my position I can really use rational arguments that may oppose this.
@@jordanagonzalezdopeso8047 Thank you for your reply! So far I think using AI this way in the brainstorming phase of the process seems ethical indeed, since you're merely using it as an inspiration, possibly also making it only a part of the brainstorming process to further personalize the potential idea. At least that's what my opinion on it is right now, will further consider it in case I'm missing an important point. I'm not sure whether the potential dangers/negative consequences of AI images are worth it, especially with how easy it seems to be to misuse it with malicious intent and get away with it, although at this point I don't think there's much that can be undone, sadly (I'd be very happy to be wrong here, though). Curious to see if somebody has a different opinion on this, too!
@@Nediame I think the most important things to remember are these quotes "Everyone has a camera but not everyone is a photographer" "Your talent is in your brain, not your fingers." Embrace the personality element of art, dont just hide behind the craft. Be someone people want to remember, it's the only thing we can do to survive as creatives as these jobs get more and more automated in music, film, and art.
@@Nediame The AI can't write jokes or tell stories very well, and its shiny quality just gives it the illusion of good art. Inject life experiences and story in your work and you'll be fine I believe. Don't rest your laurels on employable grunt work or by the numbers commissions. Be someone people get excited about, you don't need AI for any of it. If you aren't and cannot take that on as a task, you will not last as a creative. Everyone fantasizes about this career, be it acting, film, music, etc. but you have to have the stones for it (and even more so now.) What is your opinion? Why are you different? Look deep within yourself for that, it will take some time, but it is worth it. Some of the most successful art are poorly drawn comics like Cyanide and Happiness, and that was all sold on ideas over shiny craft that machines can replicate.
While it makes really intimidatingly flashy and professional work, real game/tv/film projects have too many specific parameters. I haven't found a good consistent use for it yet with my clients. There's always a level paintover, a set to hit the measurements of, camera parameters. I think it will be awhile.
I agree that any well trained artists can surpass any amateur, but the amateurs and money exploiters are becoming a majority, and that's something to worry about... In the beginning, i as an artist was hyped when this technology started appearing, I thought people would use it to generate reference material and train themselves with more specific ideas or designs Then, I started to get disappointed when I realized a lot of users weren't using it to learn anything, but instead they started exploiting it and just claiming the generated images as their own art As the time passes by, I see more and more of these people taking other artists images and using them to "replicate their style" some even start to get arrogant and make fun of regular artists for protesting Currently, i see more professional artists making videos about this topic, some say we don't have another choice but to start using AI too, others still have hopes that the pages that provide AI services will implement some way of regulation, shit, there even are some talking about wich carreers will become useless thanks to AI replacement. Yesterday, I just saw some guy on IG who clearly was using AI (not sure if it was NAI Diffusion or other model) he even got a Patreon and was begging people to commission him, the thing is that he claimed that all of his posts were drawn by himself, even when I called him out for not beign honest, he started throwing arrogant insults and then ended up blocking me, clearly he just cares about getting money with the less effort as posible And that just had me thinking, how long it will take until hundreds of these type of people start appearing, we already saw what happened with NFTs when unscrupulous people discovered that they could make money by just copy-pasting images, now that generating "art" is possible, they will be there to exploit it, 100% guaranteed And that's just depressing, because such a promising technology fell into the wrong hands, and there is no way to stop then from appearing. In the end, it's up to the consumers if they decide to support this kind of behavior
An eloquent and collected video as usual, Fowkes. You are an incredibly skilled and wise artist who I've always looked up to, and this video mirrors my thoughts exactly. A file extension to remove an art-piece from AI view is interesting, and as a computer scientist and artist both, I would certainly be on board with its development. Art is the marriage of passion, love, vision, bravery, determination, every human emotion. AI has no heart, no emotion, and will never replace us. The skill cap on AI art is moderate, and sure it will grow, but the skill cap for real artists is boundless. That fact alone is something that should drive every artist to grow, because your uniqueness and passion is what makes art, not a machine.
Haven't watched the video yet but my first thoughts of AI being able to make art was "oh great, don't tell me I've spent all that time learning to draw for nothing". Then looking more into it, there's stuff it can do well and stuff not so well. I don't think this technology is going to vanish, and it wouldn't be able to create art if man made art didn't exist in the first place so what's the best thing I can do moving forward? Make use of it I guess, maybe as a tool to help generate ideas and such. I don't think I'd ever sell what an AI straight up created though. Alright! watched the video, I very much agree with most of what you said. Going to check out the second one :)
Besides DRM technology to protect from copy to some extend there is sadly nothing that can stop people trying to use your work to just feed it in, since unless its a very obscure format people will just create converters. One thing you can try, though it may be too late, is to just use obscure file formats, however the work the ai people have to do to get around it frankly trivial. The problem is that even vector based art (SVGs) are suseptable to just reading the pixels since the way a lot of the ai works is the input can be anything. The input for things like the game of go I believe was just pixels on the screen.
This is sad A.I. takes the collective skills, abilities, and wisdom humans have churn out through the ages and can package together in a instance like some cheap Mickies burger. Through out history humans have wished to have supernatural abilities but as great thinkers realized that it's the challenge of developing those skills that can give meaning to your life. Now with A.i and transhumanism in the future humans will possess such abilities will it become a world where everyone has god like abilities competing with each other seeing who can push those boundaries? Or a world where some will possess such technology fused (transhumanism) with them while keeping a class of fully humans that will be part of a dystopian slave class? All these concept are a possibility for our species quite frightening it will impact all of the human experience and not to spread doom and gloom but looking back historically and presently we haven't always been so nice to each other.
Great Video Nathan, I completely agree. Now more than ever its important to have a strong grasp of the fundamentals. I have a few friends that are programmers I'm going to see if I we can get this going we definitely need the protection.
AI cannot be compared to the whole camera situation. A camera makes images of the real world, paintings on the other hand can do anything. So it is easy to understand how both can coexist - camera for the real world, paintings for fantasy and everything else. AI on the other hand not only replaces paintings, but 3D, animation, comics and pretty much everything else. Nothing can coexist. Not to mention that camera can only make pictures in hyper realistic "style", where as art can go anywhere on the spectrum. Now, it wouldn't be such a bad thing to make art easier and faster IF it was YOU who were making it. And this brings me to the "tool" part. AI is not a tool. Tools are things people use that help people create. When you use AI you are NOT the one creating. Imagine an artist being a chef and a person that uses AI as someone who orders food. Or more precisely a person that uses AI is like someone who commissions artists, but you "commission" the AI. Instead of an email, you write in the prompt your desired subject matter. So AI REPLACES artists. Doesn't help them create. Also it's too late to regulate it. It is already reaching pro level. Unless they delete the code, which won't happen. I don't know what can be done about it. Maybe only a few artists can stay alive through patreon or other community driven funding scenarios.
9:15 I think this would have been the perfect solution if only the technology allowed it. I would have like to see the the sources of each photo created by AI. By the way, it's still isn't clear who has the rights for an AI generated image. An idea is not copyrightable, only the expression of the idea (according to the US copyrights office). One could argue that AI prompts are just idea and that more human involvement is needed to actually register that artwork as your own.
Real art of any value, for me, can only come from the human mind and honed talent. Tools are useful but a tool creates nothing meaningful without human intervention.
it's not just artists who will be replaced by AI the sad truth is people in charge couldn't care less, in fact, that's the whole point those who tried to warn everyone were called lunatics, I don't think there's anything left to do by now, unfortunately 🤷♂🤷♂
@@kamiladamczyk2469 That isn't what they did. I believe they are using stable diffusion, but it seems to be modified and not as good as the standard release - Stable diffusion was no doubt trained on many images on DA though, to be sure. They are however at least appearing to make some efforts to spearhead ways that artists might exclude themselves from future data scraping. Once again, companies respecting tags or labelling showing artists don't want their way used is another story altogther. AI is not going to be uninvented and is not going away, so I suppose any efforts to regulate how models are trained are better than none, we have to start somewhere.
@@taskmasterblaster i agree, but they implemented a way to opt out after they trained their model so tbh this is still sketchy AF. So yeah all your future works might not be used to train next iterations of SD that they use but still they used it already. And i agree with you, that any effort to make training models more "moral" i guess is a good effort.
It certainly does feel a little bit like the suits wanting to capitalise on recent interest in AI but also come off like the good guys. I can just imagine some wanker in the board room saying " wait! What if we could make a killing off this AND look like the good guys!" The only right way to do it was to train their own model from scratch on consenting and / or non living artists hey.
Programmers may be able to make a place that keeps art from being automatically scraped for AI, but if humans can see it, humans can manually train an AI with it. Just takes more work.
It is kinda funny to hear how unfamiliar with technology some artists are. If a human can perceive it on a screen, a script can be written to collect capture it at scale.
@@zenko4187 Yeah exactly. I imagine there are ways to make it harder for people to 'harvest' your art, but unless you had your art in some gallery with security preventing all photography- there's no way to make it impossible. If someone is determined enough they could just screenshot the images and train a model with them.
the data these AI companies are scraping and using in their datasets is disturbing. you can conjure up some really disgusting images with it. We have to hold these companies responsible to standards and ethics. The only way to do this efficiently is through lawmakers.
This is currently a big topic in the art community, but I don’t think art is the only industry in danger here. I think most people are really underestimating the potential of AI. People have been talking about self driving cars for years, but we haven’t seen those much at all yet because there’s a very large safety concern. AI art is much less life threatening so releasing test models for free online is virtually harmless and allows them to basically teach themselves to refine their algorithms. Soon enough self driving cars will be released and cabs, uber and lyft, any kind of driving service will be made obsolete. As scientists work towards developing AGI; basically AI with human level intelligence, I think it is reasonable to say that AI will eventually dominate all industries. AI will make your food, clean your house, teach in school, raise and slaughter cattle and design buildings and then the machines that construct them. I don’t think there is a measurable limit to what AI can do.
How you gonna become great if you don't have job to get experienced? How you gonna pay the bills while learning? Those starter "dead" jobs that your talking about was starting point to be able to pay some bills while learning and getting experienced. Now it will be just hobby for those who have some other job and free time.
AI improve in speed of light . Several weeks ago, we criticized AI can't do perspective or coherent human proportion. But now they do. Every criticize against AI become outdated daily. It is scary and bleak time for an artist .
yeah as a concept artist with a year before graduating this is definitely really disruptive in so many ways. it feels like i have to drastically reconsider my entire education to either adapt to a workflow or system i don't even know ill want to be in or completely pivot. On a personal level it does seem very disheartening, in a very dystopian way, something thats so tied to human creativity can now be easily manufactured by a machine and seeing Ian Mcque having to defend his art from accusations that its Ai generated doesn't help. I think its really easy for our crowd to have our identities tied to our work being that it's so personal. For our hopes and careers it might be a huge blow, however, our value as people isn't dependent on whether we can produce art or not. You still matter
If I were you I wouldn't bother trying to change your career much, since AI is going to start appearing in everything and causing the same disruptions if it goes unchecked (and it probably will at least until enough fields are affected). It might be the best for us artists to market our humanity somehow, building an audience that wants to see us and our work because of who we are and how we do our artistic process, instead of strictly hiding behind our jpegs. There will always be human appreciation for human skill, and I imagine if AI takes over everything commercially people are going to be more likely to watch an artist paint or a musician perform just for the passion of it, than watch a clerical worker fill out spreadsheets or a "prompter" type out prompts that anyone can type. And who knows. Maybe we can actually advocate for some fair laws that will protect artists from this obvious corporate predation. I'm sure most "professional prompters" will change their tune quick once the AI has enough user data to automate their brain-dead easy typing tasks.
whats so creative about ripping off other artists styles and not sharing your own personal struggles and experiences through art? Whats so creative about drawing another sci fi soldier or smoky mountains? Who cares, plenty of people can do that. Can you tell stories like George Carlin or offer something unique to the world? No? Why should you be paid to express "Creativity" then, what new are you bringing to society? Your unique life and experiences are what to make and craft your art around, stop looking at the front page of instagram/artstation for inspiration... Look around you. Use this suffering and fear you feel to make something actually cool; something people can relate to, AI can't do that.
AI struggles with deeper meaning and intentional emphasis. I think it will be better to assist artists with rendering and populating detail in scenes in new ways, at the discretion of the artist.
There are three sides to the AI equation: One, which is fine is technological. It's a great tool and won't stop anytime soon and will just grow. That's a given! The two others aren't as easy. Legal and Commercial. Let's start with legal: For legal I feel that any platform that uses content or any input without the permission of the person who owns or develops that "content" is ofc misusing that information, therefor, either the AI art apps start paying for that content to be available, or they're going to start to have legal issues in the near future. In the present days we are just talking about some "unknown" to the major audiences artist but can you imagine if Midjourney starts using art from Marvel or DC?! Or any other major studio without their consent?! Or even worse if a studio publishes something from an AI artist they do not have permission or contract with from another studio?! Concepts/styles from games copied to another from AI?! Or even using movie actors or personalities?! No my friend, these people pay huge amounts to the people who make them and they're not gonna let their property be (mis)used without any penalty. The other, I've already touched, is commercial use: Just like in any military AI, the decision cannot sit on the machine alone and there has to be a human decision behind it. AI art should not and cannot be used in any commercial use by the same legal reasons I've stated before. If you start making money out of AI you are either "stealing" content or copying it from others and people will not pay for something that's not original of their own. If AI becomes as easy as going online and do a search, then why should anyone pay for ir right?! AI should only be available for educational or personal use. The moment it starts involving money (personal art for sale) or huge money (movies or games) then it will start to be a problem. The way I see it, AI has two options: Either AI art (midjourney or whatever) function the same way as any stock photage site (photography has been around for ages), paying the people who provide the content and their users (with all licenses in check) or AI will just become as pointless as someone who goes online, does a search and publishes something as their own, with the same value and recognition ...trivial, pointless and worthless.
The fact that AI can steal the style of other artists is a problem. If Ai art looked as something new and different, I guess it would be more okayish. But some of the arts in the midjouney stuff I can look and feel it's a copycat of a human artist. And that's a huge ethical problem.
AI is nothing more than a machine that does mass copyright theft and obfuscates it. There is little to no way that this will be considered fair use in court. The only reason people are overwhelmingly for AI is because this provides them with a massive gain. For the first time in history, they can produce high quality artworks with no talent or experience, and an infinite amount no less. Any negative sentiment against such a massive gain such as that will be met with an extreme amount of backlash and defensiveness, as I'm sure you have already experienced. But the methods that they are using to achieve that are clearly in the wrong. We just have to be patient and in time, I'm sure this will become illegal. The only problem is that the legal system is so abysmally slow that the damage would have already been done by the time that this happens. But I am just trying to be optimistic here, and I do not think that this will be the end of human created art.
Glad to hear someone who's being positive! As an artist with a bit of experience (over 10 years), I'm hype to change up my workflow and collaborate with AI to create much more than I ever thought was possible. I love hearing you praise the quality of the AI images, your honesty is a breath of fresh air and I feel the same way, AI blows my mind! I will look forward to see what you can create using it and adding your painting skills in the mix.
This video is problematic in that it fails to address the fact that AI is not in fact a tool, but a replacement. You think "high level art" will safeguard your job? That's just arrogance. Ai went from an undecipherable mess to beautiful images in 4 years. FOUR YEARS! It is far more likely that the only art that will stand out from the AI mess is amateur work, as it's not as desirable. Also, so long as your art can be viewed, it cannot be posted online without it being accessible to AI. no matter the file format, no matter the image. if it can be displayed, it will not be safe. This even holds true of physical traditional art! All it takes is a photo!
"beautiful" "images" bruh they're just training the ai on a specific image or artist. It's really just laundered tracing and I guarantee you that it ALWAYS looks worse than the original works.
As a non-artist, the scariest thing to me about AI art is that it would discourage people from developing their skills and making their own art. The idea of the world being filled with soulless, made for profit works makes me deeply sad. Love your work by the way.
@@DragoonBoom Technically AI just mix style and color of the artist. But you can get unlimited variation from it if you think about it. Not too mention you can train AI with another AI work and create even crazier things. The only human factor is choosing and put a limit.
Artist rely on tools to be able to create their art. But it takes skills and personal abilities to do that. A.I. isn't a tool just a mere tool it's an extension of humanity that will become it's own entity and some will try to fuse with it (transhumanism). As for me I will stay human and rely on my own abilities no matter what direction technology takes us. In a way even though having always seem impossible to create such technology it has always been pointing in this direction where humans through their designs and technologies will be able to solve all the issues that arise in the human experience. But I also see it as great dangers laying ahead in our existence it would be naive not to think that's a possibility.
Despite what should or shouldn't be done about AI, I see it as likely unstoppable, short of images that are made invisible to AI as I mentioned in the talk... Somewhere in the world it will be made available despite top down enforcement against it, and in that case many will use it thru a VPN. Then more will jump on AI in an effort to not be left out, to be on a level playing field. I'm a professional artist but not a tech or legal pro. For me it always comes down to finding ways to be relevant and of value to our audience.
I don't think it's possible to make a image file format that "is invisible". It's possible to make proprietary file/image formats but than who can see or not see the file will be decided by the owner. Truth to be told it's already possible to publish material in proprietary file formats... But I think the more you go down this path the more you get lost. Point is: all data can be processed. The only way to make it impossible to use something is to not digitalize it or not put it on the net. Therefore the only way to make a picture not usable from AI, in theory, is to publish it under a licence that exclude from the use of such picture in AI databases. But I think there's a ay better way to handle the problem, though I'm not completely sure what's your point when you say AI should not be trained on your pictures or other artist's pictures. I mean: would an AI generated image be good is it wasn't generated by using in the prompt "in the style of X artist"?
I don't care about AI's: But I am somebody who cares a lot about fair use and copyright law reform: Fighting AI's from a copyright angle could easily backfire and open up human artists to legal liability: An AI which looks at tens or hundreds of thousands, if not millions of images, and uses them as a reference to generate a new image (excluding situations where you're asking it to specifically generate an existing copyrighted character, or just to spruce up a specific existing copyrighted image vs generating a brand new one) is almost certainly fair use, even under how strict and draconian copyright law is: The impact and influence any 1 input image has is minimal. If anything, it's more obviously fair use then a human artist looking at references, which will just be 1-2 images and would be more likely to show the influences of the original work being referenced, which is important legally to the standard of if a work is "transformative", a ley part of the fair use determination. And regarding the issue of AI replicating specific styles, l ) that is even MORE clearly fair use, because "style" is not something protected by US copyright law, at all. So, if there if there's a ruling or law that establishes these AI's aren't fair use or transformative, or that "style" now CAN be copyrighted... well, that would have implidcations for a human artist, too: If a AI piece which isn't obviously similar to one of the original pieces it was sampled from isn't fair use, then a piece of art a human artist made where the reference used IS obvious won't be fair use either. Suddenly, you could be sued because your piece has a similar pose or lighting or composition to another. If Style is copyrighted, it's the same risk: Now you can ge sued because your piece happens to be in a similar style. You know whose gonna benefit most from that? Large corporations, the same ones everybody is worried will use AI's. People seem to have this idea that by fighting AI art, they're fighting against corporate exploitation of artists, but we have consistently seen that ANY time there's been a restriction of fair use or an expansion of copyright protections, the big corporations are the ones who benefit, and they use those expanded laws to go after more people, while they're too big to sue themselves so they can continue to break and bend those rules without consequences. As far as music, which I see people point to as proof that the people making AI's hate artists and are skirting having to use public domain input images: the reason why Music AI's are being handled differently is because copyright law for music itself is different: There are only a limited amount of notes that exist, vs the infinite combination of shapes and colors that exist in visual art. It is COMMON for musicians to come up with similar or identical beats or sections of music tracks as a result of that. There have been many high profile cases where songs that aren't actually likely stealing from another were ruled to be infringing as a result of this. So it is almost CERTAIN that any music generator (not even one that even HAS inputs, but just one thatr's totally random) will spit out music identical or nearly identical to some of the music it used as an input, and there'd be real risk of a infringing legal case.
Hmm, may I ask, let's say a person with little to no art experience generates AI images which copy a certain artist's style on purpose, mentioning them in their prompts and whatnot, and makes profit out of them, with no consent from the original artist - would that artist be protected by law in any way? And if not, do you consider it fair? AND if not, do you perhaps have any ideas how to prevent people doing such things from getting away with it because of technicalities? Asking genuinely.
ai actors are already replacing actors, news anchors, social media influencers, etc. ai is unstoppable. I am not defending ai, im just stating the obvious. People can't tell the difference.@@user-ch9yb2ox4v
the problem is that humans can no longer tell the difference between ai art and art made by humans. I know this for a fact because I am selling ai art. I am able to generate ai images that traditional artists and digital artists can't tell what's what. Of course, most ai art is obvious because it has a digital look but ai can be used to make art that looks traditional. Now in 2024 ai has come so far in two years that I am positive in two more years nobody will be able to tell the difference. @@TheShinorochi
Nathan, the internet at large seems to be thinking of this all in an adversarial way. I've been following this stuff since AI images were barely intelligible - a couple of years - i've used all the platforms and have stable diffusion running on my PC, and I am certain that the future is going to be collaboration with AI, especially in the entertainment and concept art space. I am already seeing concept artists jumping in and experimenting - just as they were the first to find a balance with photobashing, and kitbashing 3d to create artwork ( which was for a good while called cheating but now an industry standard). You should look into img2img prompting. Starting with an image and running it through AI with a prompt to modify it. I am seeing artists now creating artwork through a process of iteration by painting, image 2 image AI, and repeating this sometimes dozens of times before arriving at a final result. I've played around with this myself, it's pretty fun. It's is akin to art directing the AI. It takes cues from what you have painted and what you describe and can offer some neat ideas and nuances which you can then develop through painting and then hand back to it again.
I hate to be pessimistic but most artist are not and won't be a match for AI: First, the "advantages" we still have are things that AI will eventually learn with ease. Composition, color, iteration, anything that may be "unique" to artists is not. AI has come this far and will go farther. Even if we still had "advantages", those advantages will only help high level artists. Entry to medium level jobs are an easy win for AI and most artists won't be able to keep up, and they won't be able to climb to the top of there's nothing to climb in the first place. AI can improve by brute force, it is designed to do so and with the amount of resources it has no amateur artist will ever catch up. All this will make it so new artists are scarce, and only people who can afford it would eventually develop the necessary skills to match AI. There's tons of things to automate, and the whole point of automation was to let people concentrate on art. But it seems art has already been targeted, so either AI is regulated and legal action is taken, or art will be made by robots while people become slaves to work, money, and mindless consumption.
The funny thing is AI has been around for more than 50 years yet now it has come to a point now artist care about it. The funny thing is now it matters when no one had anything to say when back in the 50's and 60's this tech was created. Most artists are creating art that was not created directly from their own mental memory that is comprised of no artists style they learned from. We did what AI is doing but now it can be done in minutes. Additionally artist tend to claim ownership over things that they don't legally copyright or have or could trademark. Because you can't trademark a style because no style is original. Also people have been stealing from artists for centuries and continue to do so via crowdfunding. AI Art does not bother me because I am not doing art in fear or AI or tech. Because just like digital, 3D or any CG art seasoned eyes can tell art not created from an artist perspective. The impurities and the failures in art produced by humans it can't recreate. But people will complain about anything because it is easier to look at an out instead of getting better as an artist.
Reductionism at work. Either you know very little about art or you are arguing in bad faith against artists. (Edit: for clarity this comment is about your first paragraph )
I would also like to add a few reasons for which artists of every field should DEFINITELY learn how to use AI. When the pipeline will be secured (right now is extremely vulnerable) and the legislation will protect us, having your body of work trained into an AI will mean INSURANCE. Something happens to you as a physical being ( vision loss, central nervous system degradation, parkinson..etc) and you will be able to have a backup of your skill level at your peak. A good that can be potentially inherited in case you pass away in an untimely manner, and a contract must be competed. It's SERIOUS black mirror stuff, and we must think about it now, before companies do and present us with abusive contractual clauses that do the same.
Nathan Fowkes is a good artist but a bad teacher. If you look at the majority of his students don’t know how to paint. Look at schoolism students its the worst. A good teacher must go down to the level of a beginner students to mmet their level. You cant just show off your bad ass skill and say unity with variety and keep repeating unity and variety. Are you kidding me? That is a work of a scammer. Preying newbie students to enroll color and light because the name sound so catchy. I will give you a tip. If you want to earn big money just put color or light in your title and boom you will earn money. If you want to be a starving artist go ahead take schoolism.
I was already getting my mind blown one year ago when AI art looked mostly like barely recognizable messes... right now, I am 100% sure that AI art will change everything forever.
What change are you talking about? The best this can achieve is more volume of images being made but for what? We already are saturated enough by social media feeds and this just opens the floodgates to a tsunami of thoughtless "content".
@@BombaJead Art will become a lot more accessible, anyone will be able to bring their vision to life just by using their computers. Being a professional artist will become even more difficult than it already is.
@@boitahaki yeah what will differ people using AI and those who actually can produce images is intent, platform and story telling abilities. Because you can bring a vision to life, but why does your vision matter? Or who cares? I think people misunderstand the art creation process and how it evolves on canvas when you build it. You dont build anything with AI you just chose what you like best out of few iterations and then you go to the next set, nothing else.
@@boitahaki art has never been inaccessible, people have been making art since we were cavemen so I don't get your logic there. And as for being professional artist ultimately it will depend on what people value but yeah if people value more ai art, earning money out of art would be nearly impossible (and this ironically includes ai art).
@@BombaJead So if art has always been accessible to everyone, then why are you so upset over AI art and wanting to ban it? Did you know scarcity is what make things being worth money?
For more arguments against A.i, you should come and search Steven Zapata Art and his vid: “The end of Art: An argument Against A.i Art”, might help you. Goodluck Artist!
Honestly the last 2 years make me not want to pick a brush ever again. Complete garbage and antagonistic world we live in if you are a creative or artist. We get paid peanuts and they still feel like replacing us with the same work they stole from us.
Have a nice life, you weren't cut out for this field in the first place. Coward.
I’m a computer science student but also an artist before anything. I feel for you guys and was thinking of what we can do to protect your guy’s work. A way to encrypt / copyright your artworks from being used in a data set in the AI. This needs to be a legal issue in a court for this idea to be implemented. AI is still a new technology, and because it’s new, we haven’t yet set any limits to it. Keep speaking out on this so we can
It has been taken to court, But the community needs to stand up for itself. The music Industry wouldn't let this slide. They had this battle before
It is too late for that, unfortunately. The AI is already trained quite well enough. Now we are about to see the fallout.
@@oredaze Nothing is too late lmao. If this is illegal those mfs gotta delete all data sets they stole and start anew.
@@theoutlinist1085 Somebody probably has to sue several companies (and win) for that to happen.
@@oredaze "It's too late for that, the murderer is already murdered so many people. We can't punish and put laws for the murderer anymore to prevent more case in the future. Its too late"
- basically, you.
the midjourney app couldn't have done it in 10 seconds without stealing from your decades of experience.
thanks for the video and the guidance !
I personally am more affected mentally by the ai-art stuff than i would like to be.
Its frightening because i feel like i nearly got to make this step of making a portion of my living with art and now these companies shut the door right in front of me ^^(because i think its affecting mostly the lower skillbased jobs like book-covers,cd-covers...).
I'm in the exact same position. We just became expendables and most of job positions that were there for us to learn better skills while having any mony from it are disappearing right in front of our eyes.
I really feel this too. As someone just trying to break into the field, it feels like an already difficult path just got way harder. The 'lower' jobs are how some got their start.
its true we can deny it as much as we like but the industry is crewed ..who wants to be in on this art now?
it's on purpose because the corporations want us all to be stuck at their dead-end jobs. That's why college is impossibly expensive, and now the few things you could train for without college, they will do anything to keep us from being able to be independent from the corporations that want to work us to death.
I am on the same boat. 8 years of effort teras and hard work only to be soul-raped like this. The moment artists entered the digital area, this is the moment when we lost the game, because it was thanks to technology, even thanks to Photoshop that we could create everything that we have created so far...
Great video. But i think you gave up on legal action too quickly. There's also music AI and developers only using copyright-free samples because they fear legal issues. Visual artists just dont have organizations to back up claims. I think court should be our main avenue of resistance, and if it fails we will lose nothing and do as you suggesting.
I'm so glad we have Djamila Knopf and some other artists being very vocal about our rights and fighting for those who can't
The technology will keep evolving, in bright daylight or underground. You can't stop it now. Keep throwing lawyers, politicians and cops at the problem, though.
No, you're misunderstanding why Music is different, and trying to fight AI from a intellectual property perspective is the LAST thing artists should be doing, and may actually end up opening up actual human artists to legal liability.
I'll start with the latter before looping back to music: An AI which looks at tens or hundreds of thousands, if not millions of images, and uses them as a reference to generate a new image (excluding situations where you're asking it to specifically generate an existing copyrighted character, or just to spruce up a specific existing copyrighted image vs generating a brand new one) is almost certainly fair use, even under how strict and draconian copyright law is: The impact and influence any 1 input image has is minimal. If anything, it's more obviously fair use then a human artist looking at references, which will just be 1-2 images and would be more likely to show the influences of the original work being referenced, which is important legally to the standard of if a work is "transformative", a ley part of the fair use determination.
And regarding the issue of AI replicating specific styles, like of Junji Ito or Sugimori (of Pokemon fame) that is even MORE clearly fair use, because "style" is not something protected by US copyright law, at all.
So, if there if there's a ruling or law that establishes these AI's aren't fair use or transformative, or that "style" now CAN be copyrighted... well, that would have implidcations for a human artist, too: If a AI piece which isn't obviously similar to one of the original pieces it was sampled from isn't fair use, then a piece of art a human artist made where the reference used IS obvious won't be fair use either. Suddenly, you could be sued because your piece has a similar pose or lighting or composition to another. If Style is copyrighted, it's the same risk: Now you can ge sued because your piece happens to be in a similar style.
You know whose gonna benefit most from that? Large corporations, the same ones everybody is worried will use AI's. People seem to have this idea that by fighting AI art, they're fighting against corporate exploitation of artists, but we have consistently seen that ANY time there's been a restriction of fair use or an expansion of copyright protections, the big corporations are the ones who benefit, and they use those expanded laws to go after more people, while they're too big to sue themselves so they can continue to break and bend those rules without consequences.
As far as music, the reason why Music AI's are being handled differently is because copyright law for music itself is different: There are only a limited amount of notes that exist, vs the infinite combination of shapes and colors that exist in visual art. It is COMMON for musicians to come up with similar or identical beats or sections of music tracks as a result of that. There have been many high profile cases where songs that aren't actually likely stealing from another were ruled to be infringing as a result of this. So it is almost CERTAIN that any music generator (not even one that even HAS inputs, but just one thatr's totally random) will spit out music identical or nearly identical to some of the music it used as an input, and there'd be real risk of a infringing legal case.
@@MajoraZ Really high quality comment. Breath of fresh air after reading so many ignorant comments trying to use the cops against the AI companies.
@@MajoraZ so very open question here, I'd say that there should be new laws about companies defult opt in of using user generated content for such things like AI ? I don't have issues the the tool exist, but do have concerns on how it came about? I don't know anything about music either, but from what i hear, you do pay for sampling at least? what are your thoughts about moving foward on this with art? should I keep posting online? should i switch to traditional? do I just keep things local? How does this actually effect ai ? will it stagnate as its not givinn as much new stuff?
Honestly I just see this as big companies trying to monopolize visual art on a whole, When you think about it visual art is an entire industry that for a long time was very difficult for big corporations to fully exploit and there's no denying that there's money in this field despite public perception. Honestly this might sound conspiratorial but I think they just created these apps so that they could just have huge monopoly on the industry and tried to disguise it as a means for everyone to be creative. The big corporations can save money by laying off teams of artists and not hiring freelancers and at the same time make even more money by distributing these software to naive people fooling them into thinking they can make money by selling prompt art, but in reality something that takes this little effort to make will have little value as the market oversaturates and even the biased people who were all for this AI art progression will soon realize that the prompt art is worthless due to the fact that anyone can make it by typing a prompt, I mean why buy AI art anyway if you can just type in a prompt and make it yourself and it's going to be particularly difficult for your products to stand out while using AI art for example, if you have a shirt business that uses art, what's going to stop people from ripping off your designs or how are your designs going to stand out in a sea of easily generated content in an already oversaturated market like that, doesn't seem like a particularly beneficial tool for the public in the long run if you ask me, a good way for small business owners to fail actually. Meanwhile big corporations just take in all the money they can get with the exploitation of art while simultaneously killing the medium, at least as way of making a living from it. Don't know if you agree or not but honestly this really seems like the big plan they had...it may offer benefits for people who cant afford freelancers for small projects like book and album cover illustrations but with advent of AI music and literature those facets of art could ultimately meet a similar fate MAYBE with the exception of music since the copyright laws for that are incredibly strict. I just don't get how people don't realize the easier something is to make for the average person the less value it has in the long run, that's why low skilled jobs pay so much less most of the time. To really some it up in my opinion this was really sinister in it's inception or apathetic at best, lets create a database with peoples art without their permission for an "AI" that spits out a derivative while we're simultaneously profiting off all the time they spent perfecting their skills and disguise it as a useful tool for everyone. They're also well aware of how slow copyright laws are to catch up so they know by the time laws are put in place to protect artists the damage has already been done. I don't blame the casual people using it but the huge Corporations who funded it as well as the biased tech bros who keep spreading the lie by how useful it is and saying that their aiming for some non existent utopia where AI does everything for us and we won't ever have to work anymore to get by which obviously won't work economically with how our society functions and it's like putting salt in the wound for them to scoff at artists Work while using these AI when it's literally trained off artists work. I guess there's some solace in knowing this "art" is largely worthless as it becomes more mainstream and becomes oversaturated. The real question is though, why do we as humans strive to try and make ourselves obsolete, I can understand the automation of jobs that most people don't really want, but it baffles me why we strive to get rid of the careers we actually strive for. Anyone who does art as a career obviously did that because they love it (I mean we literally have the whole right hemisphere of our brains associated with creativity), and I know society is about progress and efficiency and whatever but I honestly don't see the automation of art as a means of driving society forward at all, all I see is our brains stagnating and becoming creatively bankrupt as big companies turn out content on a conveyor belt and even if you decide to make the difficult choice of somehow transitioning to a new career field, which would obviously require more training how long before an AI takes that away too? Rant aside I wont ever stop making art regardless and any of you reading this shouldn't either if you're as passionate as I am. Art made by your own human expression and life experiences will always be superior to an AI derivative so please don't give up, these corporate capitalists are going after as many fields of work as they can to increase their profits.
your worded this perfectly, as an artist taking my first ever commission from an overseas client,it made my day a little better knowing that others too think the same way about the dark inception of this whole thing
thanks
Wow, I literally couldn't have said this better. I really hope more people see this comment and change their minds about AI (from good to bad)
@@pumpkincat7994 I'm glad this helped and I'm slightly more optimistic now that more people will see these AI for what they truely are. Soon the general public will realize in using these AI models none of the art they use for their products will stand out in this endless sea of easily generated content and it will ultimately be more detrimental to society in general if people keep using them.
@@GalaxColor honestly I've been posting this comment on a lot of videos hoping it will shift people's perspective
we didn't strive for this, its just the rich, the people with power and influence, the big companies who benefit the most from AI art, they also have the most financial power to influence where technological development is funded, most people arent in favour of AI art and its not really a field worth research and dev into in terms of actual benefit to most humans, but with the system we live in thats just how it is, profit over human happiness.
I highly highly recommend rethinking the legality issue of the training data, dear Nathan, as many others have already pointed out. Connect with Steven Zapata, Greg Rutkowski, Karla Ortiz or AI researchers like Abhishek Gupta (Montreal AI Ethics Institute) to maybe gain a different perspective on it. We absolutely need people like you to be not just simply dismissing this.
In general I agree (like always) although the legality point seems to be wrong - in EU (and I think UK but I don't know their current law) it's illegal to take your work and reuse it without your permission plus it's illegal to take anything and tie together with your name if you didn't give them written permission.
And they will start getting hit with lawsuits any time now, all is in preparation, I saw some law firms already officially confirmed that they are preparing such lawsuits.
What personally bothers me the most, aside from just stealing artists' hard work to fill AI databases with them and then sell it, is people like me - beginners preparing to get their first job in any studio remotely are now expendable and won't have internships or job positions where they would be able to learn proper work (not to mention while having any salary to live off, which is especially hard now during inflation).
Well-paid jobs for great artists will stay - sure! But all of us below, us who already were seriously struggling for survival, might be just gone.
And I know that Schoolism or other courses would help with building skills and all that but many of us already right now can't afford it, many of us struggle really hopelessly to learn after their day job (if they're lucky enough to have one).
And then I open Instagram, see some guy who says he's disabled but thanks to AI he earns over $10k a month with his AI "art" and I see interview with him where he calls all of us, artists, even including those whose art was stolen to create his "art" he sells, losers that soon will extinct thanks to amazing AI tools and that people like him "professional prompters" will replace them, selling prints and soon also commissions because he spends 2-3 hours a day constantly typing prompts and gets many abominations but at least several good pieces a day that immediately go to his print shop.
At this point if more lawsuits and regulations won't hit them and if the AI protector (that I already saw is supposed to be slowly created to delete all copyrighted art and pieces created with copyrighted art from AI databases) won't work we're simply doomed and we either starve to death or give up the only job that gives us job, which is being artists.
(and lets be real, doing a different day job and then struggling with mental exhaustion doesn't cut it)
I read the same interview! It got me so mad! Not only he uses others' work (and types artists' surnames) but also then he mocks them - people who worked hard for years, some of them for decades, to learn all that.
And now I saw some jerks using Sam Yang's art to make AI that replicates pictures in his style. I also saw that Luisa Preissler's "fan" wrote to her that they fed AI with all her artworks they could find and now they're recreating new pieces "in Luisa's style" and they asked her, only "out of respect" if they can start posting it on their art account. That idiot even added that just in case she says no, it's all legal.
@@annabellepointer7494 "Out of respect" 🤣🤣🤣
About copying art styles, is that even copyrightable as of yet? I think I read that it isn't because otherwise we would have monopolies of styles or something like this, but I'm not really versed on copyright laws
@@Sweaper At this point it's not about copyrighting styles though but reusing one particular artist (feeding AI with all their hard work and then typing in prompts to create something in their particular style) - it's highly disrespectful and a real D-move
9:10 Why shouldn't it be a legal issue? Profits are made by using someone else's work.
It's a good question. To my understanding, copyright laws allow repurposing of existing artwork. As soon as we try and define what art is and what art is not, we get into deep trouble. For instance, Andy Warhol famously repurposed copyrighted images to great acclaim and became notable in modern art history for it.
Yeah why shouldn`t it be i want to know to?
The last I've heard it it legal issue and it breaks several copyright and database laws in EU and there are already law firms preparing with artists to sue AI owners
Saying AI art is "using someone else's work" is like saying you are using someone else's work when you study the great masters. The AI is trained on them, it does not have the image inside some sort of database.
@@sisi3565 It does not break any laws, politicians have no idea how it works.
Thank you so much for making this video. I've gone through many of your courses on schoolism multiple times! Your advice is always a guiding light to me and this bit of advice is no exception.
Thank you Nathan for entering into and engaging this tough subject for artists. A lot to consider here.
I'm joining this conversation late. I am a career artist, formerly Disney, marvel and stuff like that.
I've been watching algorithmically generated tools grow and expand since the early 2000s and knew this was coming. Natural evolution of the technology. Not one bit happy about it, but, also powerless to change it.
Here's the ong and short of it as I see it. I don't see any way that this does not replace the majority of artists inside of 5 years.
There are specializations within the professional art community that will hold out longer and maybe always retain a small contingent of working professionals, supported by a holdout anti AI community if nothing else.
We are early on in what we are calling AI art creation, and it's exponential rate of growth is going to blow everybody's mind and very short order.
No area of the broader art profession is safe. I've worked in animation modeling illustration marketing, you name it. All of this boils down to data sets. Any existing style that isn't a one-off random piece boils down to a repeatable definable data set of shape language, color palette, etc.
I've spoken to a Kotaku writer about this and they have covered it somewhat, but here's the part I don't think anyone is grasping.
There are probably tens of thousands of students in art schools and universities right now incurring tuition debt for a profession that will no longer exist by the time they graduate or will exist in a greatly diminished way.
Regarding ai's need for artists that will always exist in order to guide and direct it. I hate to say it but you're wrong about that.
The same way in art director or instructor guides and teaches a student or Junior artist on composition, use of light, negative space, etc. Those things can and eventually will be part of the AI process.
I'm not a negative person and I don't want to be a doomsayer but I don't see how this doesn't transpire.
Well, i myself was working in those mid tier jobs that are basically vanishing as we speak, in 5 years span i see almost all of the art jobs going to AI, its going to be a hobby, maybe traditional art gain value that way i dont know. But since creative process is just input vs output then yeah art is dead in a way that working as an artist is dead.
I can see why after looking at your portfolio, nothing about your work stands out as anything inspiring or interesting. It's just employable gruntwork that shares no real human emotion. Of course you will be replaced as a cog in the machine, its time to shack up and start expressing your true emotions in art. What is your opinion? Why are you different?
Pretty much my sentiment, there's nothing an artist can do that AI can't with brute force. This is an existential issue that needs to be addressed and regulated by law, otherwise we give it a free pass to obliterate art as a job.
@@ranzu3138 Well it wont be uninvented, and our jobs wont be saved. Thing is that they just went it trained their AI's on all the work we did to just throw us out of the window.
Ai's training models need to be regulated. Because they do not respect any copyright.
@@bwzarchive708 Well thats a good point, i think it illustrates perfectly where we as human artists should stand.
Just want to redirect people to a Steven Zapata video ''An Argument Against Image AIs''.
Completely obliterated by Royal Skies.
I just now watched it, very thoughtful and well reasoned. I have no claim on seeing the future but I suspect that how Steven and we all believe things should be, and how things actually will be, they'll probably be very different from each other.
Doesn't mean we shouldn't fight for what we believe is right, but as I said at the top, I see tremendous inevitability in all of this.
A few points.
The AI doesn't stitch together the source images. In the training process, the source image is completely scrambled and covered in noise. After that, the software removes the noise bit by bit, checks if it resembles the initial image and is assigned a score for it. It keeps doing that over and over again until the "unscrambled noise" vaguely resembles the original input (it never goes back to the original, because both the addition and the subtraction of noise is random). For the generation process there's no source image to compare to anymore, because it starts directly from random noise, and it's the human who fine tunes how to remove the noise and when to stop (via use of prompts and steps settings).
The examples you see on display on those websites are curated. It is true that a single image takes second, but in order to get an acceptable result you need to understand what is completely wrong and what looks "off", be it shapes, proportions, framing, pose, lighting etc and fine tune the prompt and the setting in order to get closer to what you want, and it can easily take hours and hundreds of generated images to get there. In this case, artistic knowledge and expertise is essential to know how and what to fix.
When people say it's a tool, they mean it can (and will) be used to give artists a good headstart and customized reference. Creative and technically expert artists are still going to come up on top, because they can direct the ai and can use its results to the fullest.
Steven Zapata did a video essay about these programs. They are not designed as tools but as a replacement for artists. And it shows, because as tools they are really lacking about user inputs and controls. Ultimately the result is not "your" art, it's a rehashing of all the works included in the datasets and it doesn't reflect your personality, you didn't earn it. Artists should take action and demand removal not from the prompts (it's next to useless) but most importantly from the datasets. These programs would not exist without the works of artists poached and scavenged in the datasets. The companies designing these programs profit from our work for free, it's unacceptable. I agree that young artists will be tempted to take the short way for how easy it is. Then how are we going to experience new styles, everything will be some rehash of already existing works. A personal style can manifest in very minute things like just the way you draw a line or a circle. There is no way someone can really develop something truly personal with these the programs alone because how loose and scattershot the control is.
Totally agree
As somebody who doesn't particularly care about AI's, but DOES care a lot about Fair use and copyright law reform, I had a fair amount of issues with Steven Zapata's video. I'm gonna copy the comment I left there here:
"Sorry for the 8 paragraph comment: I'm not an artist nor AI art advocate, but I am somebody deeply concerned with IP laws, copyright, and fair use (IANAL, just a IP nerd), so my stake in this whole issue and debate is from that angle: I think that the fact that the datasets various AI uses is sourced from groups that have research status and then the resulting AI algorithms can be used commercially is pretty interesting, and I agree is sketchy, but i'm not sure that it's a smoking gun, either: If the datasets are LINKS to images, rather then downloaded images themselves, then it is arguable that the dataset isn't actually a "use" of any of the works itself. I say "arguable", because the act of opening a link DOES create a local copy of a file, so the training that produces an AI algorithm still would be reproducing the images.
I, personally, don't think that the local temporary creation of a image or audio file etc should constitute an act of use or infringement, because practically speaking, somebody isn't actually saving an image every time they look at an image online even if technically that's how computers work: Opening any page on the internet ever would constitute copyright infringement. It is my understanding that LEGALLY speaking, this is sort of a grey area and different rulings have handled it differently depending on other specifics. I'd probably say that training an AI on images via links should constitute a "use", but that's just my opinion, not me saying that's what I think it'd be legally, I'm not sure.
I do, however, think that the resulting AI would be considered fair use, even if the use of the images is considered a legal use. The legal standard for what's considered fair use is complicated and varies by country, but in the US the main question is mainly if a work is "Transformative" (though there are other factors too beyond that). What counts as transformative is ITSELF a massive, complex, abstract question, but explained as simply as possible would be if the new work significantly restructures or recontextualizes the original into a new thing to where the original's influence is nonobvious at a glance. This is, again, a VERY, very simplified explanation, to the point of arguably being misleading, but it's the best one I can give. I would personally argue that fair use determination should actually be much, much broader, but even in the current IP law status quo, I think an AI image generation algorithm would almost certainly qualify: It's not even an image itself. It's essentially an entirely brand new "thing". My best analogy would be saying that an AI isn't fair use would be like saying me typing up a text description of the mona lisa is copyright infringement of the original painting, which would be absurd, but there's even MORE separation here then that because the AI isn't a description of any 1 image, but of trends between THOUSANDS of images, where the influence of any one inputr is relatively minimal, at least as I understand how AI works. It's actually more like saying a book that describes lightning techniques is infringing on the copyrights of all the images the author looked at to see how other artists do lighting. That's obviously absurd.
This is also why I think the actual output images AI make would ALSO likely be fair use in most cases: There were thousands, if not millions of images used in the datasets the AI is trained on, and the impact of any one of them on the final work would be relatively minimal. You;re correct that the AI doesn't just use references like human artists do... it does it in a way that is MORE obviously transformative then a human artist using references, where there is a lesser number of them and in many cases the influences used will be more obvious. Frankly, I found most of your arguements in this section (the one at 31:32) sort of irrelevant: that the AI is an AI and doesn't have to be skilled in the way a human is doesn't matter in a determination of if something is fair use, and it shouldn't be (Should an artist that draws a rendition of a copyrighted thing poorly be fair use while one doing it well be infringing?). "Style" is also not something that can be copyrighted (Nor should it be: Should artists be liable if they happen to draw something stylistically to another person's style?): Infringement is done on if a SPECIFIC work or character is being copied, and AI art, by it's nature, is generally spitting out a new output...
...unless you ask it do otherwise. Asking somebody to make a piece of art in the style of another artist would probably be fair use, but asking it to make an image of a copyrighted character doing something would probably not be, because CHARACTERS can be copyrighted. I, personally, think that something like noncommercial fanart SHOULD be fair use, but it isn't. Likewise, i've seen cases where an AI isn't generating a brand new image, but is filling in details on an existing image, or "uncropping" it, to fill in space outside the images current boundaries. In those cases, it's pretty obvious what the work i'ts building on is, and I suspect these would be ruled as infringing.
As far as why Dance diffusion is different, I don't think it's because AI companies/authors think lesser of visual artists (tho they may, IDK), but rather because the nature of music means that there is only so many ways songs can be set up, vs the infinite variation of visual art: Two musicians will not just sometimes, but OFTEN come up withy similar or identical beats or sections of tracks in their music because there's only so many notes. There have been many legal cases where songs that probably weren't actually stealing from each other were ruled to be infringing. So ANY piece of generated music that's relatively simple in composition is almost invariably going to be nearly identical to at least one other piece of music. This is true of human made music, too. The entire way music coporyright works is sort of different as a result of this, hell there's even mandatory licensing for music, where record labels HAVE to allow people to use their tracks (Personally, I wish mandatory liscensing was a thing for visual art, games, movies, etc too; and I think that music that's only incidentally similar and not proven to be copied SHOULD be fair use, but that's just me; but even i'd admit I'm not sure how AI generated music would fit inot this)
I want to loop back to something I brought up at the end of the paragraph above the last one, where I point out how trying to say skill or difficulty should determine fair use status, or that "style" should be able top be copyrighted, would create massive issues with liability for human artists. That, I think, is the bottom line here: If there WERE to be a legal ruling that the act of training an AI on other works is infringing, or that the new output work is, there would be absolutely massive unintended implications for human artists too: It would severely limit less contentious, more obviously benign instances of researchers training AI, and of human artists using references, and open both up to massive liability: Imagine being sued because somebody else thinks you used their work as a reference for a piece of yours, since the two are as similar as an AI's input image and the output. Or because your style is similar to theirs.
You know who would benefit most from that? The same large corporations you're worried about here. The ones that already send out mass DMCA claims on content online. They would absolutely go after smaller artists, who can't afford to defend themselves in court; and those large corporations will also feel free to do the same things they accuse others of doing, because they're too big TO be sued by smaller artists. We see this all the time now: Fox stole some guy's youtube video, included it in family guy, then hit the original upload with a takedown. The "you wouldn't steal a car" antipiracy ad used stolen music, etc.
As I said at the start of this comment, I am not an AI art advocate. But I do think trying to attack AI generated art from a intellectual property/copyright angle is a bad idea: Both because mechanically speaking I think it is often fair use and should be, but also because I think any laws or rulings would have grave unintended consequences for human artists. I'm sympathetic to the labor concerns artists have, of AI art replacing human artists and putting them out of a job, but It needs to be addressed in other ways.
EDIT: Also, something I forgot to say I don't really have a good way of segueing into: Even if using copyrighted art as inputs is probably fair use, since the law doesn't really make a distinction between how much skill is involved or if there's a human creative vision in fair use dermination, to my knowledge; there IS a determination of if there was human creative input if a NEW work gets copyright protections: There's been court cases establishing a Monkey picking up a camera and taking a photo a photographer has set up doesn't give the photographer the copyright, and the photo is instead public domain. If there is a legal avenue here that is likely to succeed and not be as destructive to fair use here, it would be argueing that the output of AI is public domain"
Steven made some good but points for sure, but sure it's not made to be a tool, no one said it was, but it CAN be used as a tool. It's an artist's auto tune, and the Pandora's Box has already been opened. No going back now.
Great to hear your thoughts on this, Nathan! Thanks!
Excellent analysis and projection for the future.
People should work together and stand up! We can battle this. The music industry had this before and they won't touch any of it because of legal issues. There is no difference in art. These companies are trying to monopoly visual art
There are already primitive generative 3D AIs out there, and given the rate of improvement to 2D generators, I don’t see it taking very much time for them to be just as effective, other than the current implosion of seemingly all the big tech companies.
A watermark that marks an image as 'invisible' for AI seems like an interesting idea
I'd rather say that there should be a restriction that every single AI piece has to have visible watermarks all over - if it's supposed to be a tool artists will deal with it and ordinary thieves won't be able to use it as their own
@@annabellepointer7494 They can just photoshop the water marks out
@@briannaalejo9226 Yes, and to do so they would need a lot of time and a bit of skill. It would stop them from flooding internet with hundreds of thousands of pictures that AI makes within seconds and it would discourage most of them because they are lazy thieves
bro they already trained it on 5,8 billions of images.
@@annabellepointer7494 It doesn't take much time to remove watermarks, nor does it take much skill. But I do agree that there needs some form of encryption/copyright for artists work that prevents their work to being used in AI. Anyone can take your artwork from instagram, twitter, art station with a simple screenshot, and make a category of your style in a data set of a collection of your work. Either artists need to copyright all of their work, or move to a platform that easily implements a form of encryption to an artist's work once posted.
Hi! professional artist here. I believe you have quite a few points Nathan. I would like to add my two cents, trying to be rational and neutral. The end game of AI is not 3D generation, but Completed animated clips in every style (3D, live action, and 2D) with AI generated voices and performances. Right now AI is getting over technical details (Hands, warped facial features..) and the next step is gonna be solving the problem of consistency (cannot generate proper comics or boards) THe last feature to conquer is the Illussion of life, and storytelling. It's extraordinarily bad at doing it, and since that it highly specialized knowledge, it may take years till they get it right. That leaves the most basic foundation of drawing and painting as the golden skill to have "Gesture, Acting, and storytelling in the form of emotional direction" Even if AI could do that, an Amateur wouldn't have a clue of how to command those features, the formation is still necessary. A Jedi can do wonders with a Laser Saber, a kid would only do havoc and predictable shenanigans, that soon will become so common, they will be perceived as "Gaudi" like the Microsoft Word illustrations back in the 90's.
"It's a tool in the way you would compare a Screwdriver to a Lightsaber" - brilliant point at 3:07
AI revoluton made me stick to my comic novel even more - I need to push it out and see if people care about human created Indie comic. With all it's human imperfections.
I'd hate to be an AI user (non-artist) who submitted a piece to a publisher and was asked to make a few minor alterations. What would I do? Or a book illustrator who had to compose 32 illustrations, with the same character.
AI seems good for one-off paintings but, from my dabblings with Midjourney, you have almost no control over what the software will spit out. Like Marco Bucci says, we should use our own AI - "actual intelligence".
thank you for this Nathan, Honestly all this AI crap really depressed me, i stopped painting.... but this makes me re think that desicion..
It's interesting to hear your thoughts on the topic, thank you for sharing them! I do wonder how AI could be used as a "tool" for artists, though. I have a hard time picturing it as such, as opposed to actual tools that allow(ed) artists to express themselves in new ways and inspire new ways of creating art (such as digital art tools, 3D, or perhaps even camera obscura back in the day). And with all the shady things related to those machines collecting and processing data, I wonder if I would even consider it ethical, even if I was able to find a way to use it as a tool. Artists don't want to save their work from being devoured by those without a reason after all.
That being said, if somebody would like to share views that differ from mine (respectfully) with me, I'd love to learn and see if there is something to change my mind about. Not having to worry about AI being an enemy would be lovely, but frankly, the main thing triggered by it that I see is a lack of respect for artists, which is rather ironic and doesn't make me feel very optimistic about the entire thing.
Hi! there are ways to ethically use it. I am still testing it (not claiming or posting anything) but so far, I figured a very rudimentary, ethical workflow, with a few rules. 1. NEVER use a stable diffusion trained with an artists specific work without their consent. That is crossing a line. 2. You may produce an image, or several, preferably on a secured way (downloaded SD or Midjourney private mode) and use it as an initial brainstorm like you would in any google search. These cannot be presented on professional work, only used as an idea generation for color schemes, general shapes or composition. You will have to paint the entire thing, do a line pass that is your own, and recolor from scratch. So far, I don't see any ethical difference between AI and using Pinterest or google as reference gathering. It has the plus of probably blending several styles together, which will help you further your own in a way that you are less likely to infringe someone else's copyright. If you find an ethical problem with this, please let me know. I am forming my position I can really use rational arguments that may oppose this.
@@jordanagonzalezdopeso8047 Thank you for your reply! So far I think using AI this way in the brainstorming phase of the process seems ethical indeed, since you're merely using it as an inspiration, possibly also making it only a part of the brainstorming process to further personalize the potential idea. At least that's what my opinion on it is right now, will further consider it in case I'm missing an important point. I'm not sure whether the potential dangers/negative consequences of AI images are worth it, especially with how easy it seems to be to misuse it with malicious intent and get away with it, although at this point I don't think there's much that can be undone, sadly (I'd be very happy to be wrong here, though).
Curious to see if somebody has a different opinion on this, too!
@@Nediame I think the most important things to remember are these quotes
"Everyone has a camera but not everyone is a photographer"
"Your talent is in your brain, not your fingers."
Embrace the personality element of art, dont just hide behind the craft. Be someone people want to remember, it's the only thing we can do to survive as creatives as these jobs get more and more automated in music, film, and art.
@@Nediame The AI can't write jokes or tell stories very well, and its shiny quality just gives it the illusion of good art. Inject life experiences and story in your work and you'll be fine I believe. Don't rest your laurels on employable grunt work or by the numbers commissions. Be someone people get excited about, you don't need AI for any of it. If you aren't and cannot take that on as a task, you will not last as a creative.
Everyone fantasizes about this career, be it acting, film, music, etc. but you have to have the stones for it (and even more so now.) What is your opinion? Why are you different? Look deep within yourself for that, it will take some time, but it is worth it.
Some of the most successful art are poorly drawn comics like Cyanide and Happiness, and that was all sold on ideas over shiny craft that machines can replicate.
While it makes really intimidatingly flashy and professional work, real game/tv/film projects have too many specific parameters. I haven't found a good consistent use for it yet with my clients. There's always a level paintover, a set to hit the measurements of, camera parameters. I think it will be awhile.
I agree that any well trained artists can surpass any amateur, but the amateurs and money exploiters are becoming a majority, and that's something to worry about...
In the beginning, i as an artist was hyped when this technology started appearing, I thought people would use it to generate reference material and train themselves with more specific ideas or designs
Then, I started to get disappointed when I realized a lot of users weren't using it to learn anything, but instead they started exploiting it and just claiming the generated images as their own art
As the time passes by, I see more and more of these people taking other artists images and using them to "replicate their style" some even start to get arrogant and make fun of regular artists for protesting
Currently, i see more professional artists making videos about this topic, some say we don't have another choice but to start using AI too, others still have hopes that the pages that provide AI services will implement some way of regulation, shit, there even are some talking about wich carreers will become useless thanks to AI replacement.
Yesterday, I just saw some guy on IG who clearly was using AI (not sure if it was NAI Diffusion or other model) he even got a Patreon and was begging people to commission him, the thing is that he claimed that all of his posts were drawn by himself, even when I called him out for not beign honest, he started throwing arrogant insults and then ended up blocking me, clearly he just cares about getting money with the less effort as posible
And that just had me thinking, how long it will take until hundreds of these type of people start appearing, we already saw what happened with NFTs when unscrupulous people discovered that they could make money by just copy-pasting images, now that generating "art" is possible, they will be there to exploit it, 100% guaranteed
And that's just depressing, because such a promising technology fell into the wrong hands, and there is no way to stop then from appearing.
In the end, it's up to the consumers if they decide to support this kind of behavior
humanity is getting the worst ending for sure...
An eloquent and collected video as usual, Fowkes. You are an incredibly skilled and wise artist who I've always looked up to, and this video mirrors my thoughts exactly. A file extension to remove an art-piece from AI view is interesting, and as a computer scientist and artist both, I would certainly be on board with its development. Art is the marriage of passion, love, vision, bravery, determination, every human emotion. AI has no heart, no emotion, and will never replace us. The skill cap on AI art is moderate, and sure it will grow, but the skill cap for real artists is boundless. That fact alone is something that should drive every artist to grow, because your uniqueness and passion is what makes art, not a machine.
Haven't watched the video yet but my first thoughts of AI being able to make art was "oh great, don't tell me I've spent all that time learning to draw for nothing". Then looking more into it, there's stuff it can do well and stuff not so well. I don't think this technology is going to vanish, and it wouldn't be able to create art if man made art didn't exist in the first place so what's the best thing I can do moving forward? Make use of it I guess, maybe as a tool to help generate ideas and such. I don't think I'd ever sell what an AI straight up created though.
Alright! watched the video, I very much agree with most of what you said. Going to check out the second one :)
Besides DRM technology to protect from copy to some extend there is sadly nothing that can stop people trying to use your work to just feed it in, since unless its a very obscure format people will just create converters. One thing you can try, though it may be too late, is to just use obscure file formats, however the work the ai people have to do to get around it frankly trivial. The problem is that even vector based art (SVGs) are suseptable to just reading the pixels since the way a lot of the ai works is the input can be anything. The input for things like the game of go I believe was just pixels on the screen.
so basically you want to pay them to not steal your art???
This part also bothered me
This is sad A.I. takes the collective skills, abilities, and wisdom humans have churn out through the ages and can package together in a instance like some cheap Mickies burger. Through out history humans have wished to have supernatural abilities but as great thinkers realized that it's the challenge of developing those skills that can give meaning to your life. Now with A.i and transhumanism in the future humans will possess such abilities will it become a world where everyone has god like abilities competing with each other seeing who can push those boundaries? Or a world where some will possess such technology fused (transhumanism) with them while keeping a class of fully humans that will be part of a dystopian slave class? All these concept are a possibility for our species quite frightening it will impact all of the human experience and not to spread doom and gloom but looking back historically and presently we haven't always been so nice to each other.
Great Video Nathan, I completely agree. Now more than ever its important to have a strong grasp of the fundamentals. I have a few friends that are programmers I'm going to see if I we can get this going we definitely need the protection.
AI cannot be compared to the whole camera situation. A camera makes images of the real world, paintings on the other hand can do anything. So it is easy to understand how both can coexist - camera for the real world, paintings for fantasy and everything else. AI on the other hand not only replaces paintings, but 3D, animation, comics and pretty much everything else. Nothing can coexist. Not to mention that camera can only make pictures in hyper realistic "style", where as art can go anywhere on the spectrum.
Now, it wouldn't be such a bad thing to make art easier and faster IF it was YOU who were making it. And this brings me to the "tool" part. AI is not a tool. Tools are things people use that help people create. When you use AI you are NOT the one creating. Imagine an artist being a chef and a person that uses AI as someone who orders food. Or more precisely a person that uses AI is like someone who commissions artists, but you "commission" the AI. Instead of an email, you write in the prompt your desired subject matter. So AI REPLACES artists. Doesn't help them create.
Also it's too late to regulate it. It is already reaching pro level. Unless they delete the code, which won't happen.
I don't know what can be done about it. Maybe only a few artists can stay alive through patreon or other community driven funding scenarios.
what was the name of the artist at 5:35 ?
Giovanni Battista Tiepolo
Tiepolo
9:15 I think this would have been the perfect solution if only the technology allowed it. I would have like to see the the sources of each photo created by AI. By the way, it's still isn't clear who has the rights for an AI generated image. An idea is not copyrightable, only the expression of the idea (according to the US copyrights office). One could argue that AI prompts are just idea and that more human involvement is needed to actually register that artwork as your own.
Real art of any value, for me, can only come from the human mind and honed talent. Tools are useful but a tool creates nothing meaningful without human intervention.
it's not just artists who will be replaced by AI
the sad truth is people in charge couldn't care less, in fact, that's the whole point
those who tried to warn everyone were called lunatics, I don't think there's anything left to do by now, unfortunately 🤷♂🤷♂
Deviant art is already sort of kind of but not really thinking about the invisibility shield.
you mean when they implemented their own AI and literally trained it on everyones art on their platform?
@@kamiladamczyk2469 That isn't what they did. I believe they are using stable diffusion, but it seems to be modified and not as good as the standard release - Stable diffusion was no doubt trained on many images on DA though, to be sure. They are however at least appearing to make some efforts to spearhead ways that artists might exclude themselves from future data scraping. Once again, companies respecting tags or labelling showing artists don't want their way used is another story altogther. AI is not going to be uninvented and is not going away, so I suppose any efforts to regulate how models are trained are better than none, we have to start somewhere.
@@taskmasterblaster i agree, but they implemented a way to opt out after they trained their model so tbh this is still sketchy AF. So yeah all your future works might not be used to train next iterations of SD that they use but still they used it already.
And i agree with you, that any effort to make training models more "moral" i guess is a good effort.
It certainly does feel a little bit like the suits wanting to capitalise on recent interest in AI but also come off like the good guys. I can just imagine some wanker in the board room saying " wait! What if we could make a killing off this AND look like the good guys!" The only right way to do it was to train their own model from scratch on consenting and / or non living artists hey.
Looks like the only consolation prize for artist is to become a key promt word.
Programmers may be able to make a place that keeps art from being automatically scraped for AI, but if humans can see it, humans can manually train an AI with it. Just takes more work.
It is kinda funny to hear how unfamiliar with technology some artists are. If a human can perceive it on a screen, a script can be written to collect capture it at scale.
@@zenko4187 Yeah exactly. I imagine there are ways to make it harder for people to 'harvest' your art, but unless you had your art in some gallery with security preventing all photography- there's no way to make it impossible. If someone is determined enough they could just screenshot the images and train a model with them.
@@Trishazim people are doing literally that right now to All the popular artists out there.
@@kamiladamczyk2469 Yes, exactly :(
the data these AI companies are scraping and using in their datasets is disturbing. you can conjure up some really disgusting images with it. We have to hold these companies responsible to standards and ethics. The only way to do this efficiently is through lawmakers.
This is currently a big topic in the art community, but I don’t think art is the only industry in danger here. I think most people are really underestimating the potential of AI. People have been talking about self driving cars for years, but we haven’t seen those much at all yet because there’s a very large safety concern. AI art is much less life threatening so releasing test models for free online is virtually harmless and allows them to basically teach themselves to refine their algorithms. Soon enough self driving cars will be released and cabs, uber and lyft, any kind of driving service will be made obsolete. As scientists work towards developing AGI; basically AI with human level intelligence, I think it is reasonable to say that AI will eventually dominate all industries. AI will make your food, clean your house, teach in school, raise and slaughter cattle and design buildings and then the machines that construct them. I don’t think there is a measurable limit to what AI can do.
How you gonna become great if you don't have job to get experienced? How you gonna pay the bills while learning? Those starter "dead" jobs that your talking about was starting point to be able to pay some bills while learning and getting experienced. Now it will be just hobby for those who have some other job and free time.
AI improve in speed of light . Several weeks ago, we criticized AI can't do perspective or coherent human proportion. But now they do. Every criticize against AI become outdated daily. It is scary and bleak time for an artist .
yeah as a concept artist with a year before graduating this is definitely really disruptive in so many ways. it feels like i have to drastically reconsider my entire education to either adapt to a workflow or system i don't even know ill want to be in or completely pivot. On a personal level it does seem very disheartening, in a very dystopian way, something thats so tied to human creativity can now be easily manufactured by a machine and seeing Ian Mcque having to defend his art from accusations that its Ai generated doesn't help.
I think its really easy for our crowd to have our identities tied to our work being that it's so personal. For our hopes and careers it might be a huge blow, however, our value as people isn't dependent on whether we can produce art or not. You still matter
If I were you I wouldn't bother trying to change your career much, since AI is going to start appearing in everything and causing the same disruptions if it goes unchecked (and it probably will at least until enough fields are affected). It might be the best for us artists to market our humanity somehow, building an audience that wants to see us and our work because of who we are and how we do our artistic process, instead of strictly hiding behind our jpegs. There will always be human appreciation for human skill, and I imagine if AI takes over everything commercially people are going to be more likely to watch an artist paint or a musician perform just for the passion of it, than watch a clerical worker fill out spreadsheets or a "prompter" type out prompts that anyone can type.
And who knows. Maybe we can actually advocate for some fair laws that will protect artists from this obvious corporate predation. I'm sure most "professional prompters" will change their tune quick once the AI has enough user data to automate their brain-dead easy typing tasks.
whats so creative about ripping off other artists styles and not sharing your own personal struggles and experiences through art?
Whats so creative about drawing another sci fi soldier or smoky mountains?
Who cares, plenty of people can do that. Can you tell stories like George Carlin or offer something unique to the world? No? Why should you be paid to express "Creativity" then, what new are you bringing to society?
Your unique life and experiences are what to make and craft your art around, stop looking at the front page of instagram/artstation for inspiration... Look around you. Use this suffering and fear you feel to make something actually cool; something people can relate to, AI can't do that.
Sadly, people need money, and if art is no longer a viable source of income for new artists, then you either starve or abandon art as a job.
it's over...AI will leave millions without a job, too late already
people like Nathan, will have a place, but next generations don't have a chance
AI struggles with deeper meaning and intentional emphasis. I think it will be better to assist artists with rendering and populating detail in scenes in new ways, at the discretion of the artist.
There are three sides to the AI equation:
One, which is fine is technological. It's a great tool and won't stop anytime soon and will just grow. That's a given!
The two others aren't as easy. Legal and Commercial.
Let's start with legal: For legal I feel that any platform that uses content or any input without the permission of the person who owns or develops that "content" is ofc misusing that information, therefor, either the AI art apps start paying for that content to be available, or they're going to start to have legal issues in the near future. In the present days we are just talking about some "unknown" to the major audiences artist but can you imagine if Midjourney starts using art from Marvel or DC?! Or any other major studio without their consent?! Or even worse if a studio publishes something from an AI artist they do not have permission or contract with from another studio?! Concepts/styles from games copied to another from AI?! Or even using movie actors or personalities?! No my friend, these people pay huge amounts to the people who make them and they're not gonna let their property be (mis)used without any penalty.
The other, I've already touched, is commercial use: Just like in any military AI, the decision cannot sit on the machine alone and there has to be a human decision behind it. AI art should not and cannot be used in any commercial use by the same legal reasons I've stated before. If you start making money out of AI you are either "stealing" content or copying it from others and people will not pay for something that's not original of their own. If AI becomes as easy as going online and do a search, then why should anyone pay for ir right?! AI should only be available for educational or personal use. The moment it starts involving money (personal art for sale) or huge money (movies or games) then it will start to be a problem.
The way I see it, AI has two options: Either AI art (midjourney or whatever) function the same way as any stock photage site (photography has been around for ages), paying the people who provide the content and their users (with all licenses in check) or AI will just become as pointless as someone who goes online, does a search and publishes something as their own, with the same value and recognition ...trivial, pointless and worthless.
But it is a matter of legality.
You cant depend on single companies.
It's a systemic issue.
It is in its infancy and AI already killed artist (by preventing most and most of making a living as time goes), the future is dark
Why do we have copyright laws in the first place? Answer the questions and immediately you know what to think of AI if you don't have an agenda.
The fact that AI can steal the style of other artists is a problem. If Ai art looked as something new and different, I guess it would be more okayish. But some of the arts in the midjouney stuff I can look and feel it's a copycat of a human artist. And that's a huge ethical problem.
AI is nothing more than a machine that does mass copyright theft and obfuscates it. There is little to no way that this will be considered fair use in court.
The only reason people are overwhelmingly for AI is because this provides them with a massive gain. For the first time in history, they can produce high quality artworks with no talent or experience, and an infinite amount no less. Any negative sentiment against such a massive gain such as that will be met with an extreme amount of backlash and defensiveness, as I'm sure you have already experienced. But the methods that they are using to achieve that are clearly in the wrong.
We just have to be patient and in time, I'm sure this will become illegal. The only problem is that the legal system is so abysmally slow that the damage would have already been done by the time that this happens. But I am just trying to be optimistic here, and I do not think that this will be the end of human created art.
Glad to hear someone who's being positive! As an artist with a bit of experience (over 10 years), I'm hype to change up my workflow and collaborate with AI to create much more than I ever thought was possible.
I love hearing you praise the quality of the AI images, your honesty is a breath of fresh air and I feel the same way, AI blows my mind! I will look forward to see what you can create using it and adding your painting skills in the mix.
This video is problematic in that it fails to address the fact that AI is not in fact a tool, but a replacement. You think "high level art" will safeguard your job? That's just arrogance. Ai went from an undecipherable mess to beautiful images in 4 years. FOUR YEARS!
It is far more likely that the only art that will stand out from the AI mess is amateur work, as it's not as desirable.
Also, so long as your art can be viewed, it cannot be posted online without it being accessible to AI. no matter the file format, no matter the image. if it can be displayed, it will not be safe. This even holds true of physical traditional art! All it takes is a photo!
"beautiful" "images" bruh they're just training the ai on a specific image or artist. It's really just laundered tracing and I guarantee you that it ALWAYS looks worse than the original works.
As a non-artist, the scariest thing to me about AI art is that it would discourage people from developing their skills and making their own art. The idea of the world being filled with soulless, made for profit works makes me deeply sad. Love your work by the way.
@@DragoonBoom Technically AI just mix style and color of the artist. But you can get unlimited variation from it if you think about it.
Not too mention you can train AI with another AI work and create even crazier things. The only human factor is choosing and put a limit.
Artist rely on tools to be able to create their art. But it takes skills and personal abilities to do that. A.I. isn't a tool just a mere tool it's an extension of humanity that will become it's own entity and some will try to fuse with it (transhumanism). As for me I will stay human and rely on my own abilities no matter what direction technology takes us. In a way even though having always seem impossible to create such technology it has always been pointing in this direction where humans through their designs and technologies will be able to solve all the issues that arise in the human experience. But I also see it as great dangers laying ahead in our existence it would be naive not to think that's a possibility.
Despite what should or shouldn't be done about AI, I see it as likely unstoppable, short of images that are made invisible to AI as I mentioned in the talk...
Somewhere in the world it will be made available despite top down enforcement against it, and in that case many will use it thru a VPN. Then more will jump on AI in an effort to not be left out, to be on a level playing field.
I'm a professional artist but not a tech or legal pro. For me it always comes down to finding ways to be relevant and of value to our audience.
I don't think it's possible to make a image file format that "is invisible". It's possible to make proprietary file/image formats but than who can see or not see the file will be decided by the owner. Truth to be told it's already possible to publish material in proprietary file formats... But I think the more you go down this path the more you get lost. Point is: all data can be processed. The only way to make it impossible to use something is to not digitalize it or not put it on the net. Therefore the only way to make a picture not usable from AI, in theory, is to publish it under a licence that exclude from the use of such picture in AI databases. But I think there's a ay better way to handle the problem, though I'm not completely sure what's your point when you say AI should not be trained on your pictures or other artist's pictures. I mean: would an AI generated image be good is it wasn't generated by using in the prompt "in the style of X artist"?
I don't care about AI's: But I am somebody who cares a lot about fair use and copyright law reform: Fighting AI's from a copyright angle could easily backfire and open up human artists to legal liability:
An AI which looks at tens or hundreds of thousands, if not millions of images, and uses them as a reference to generate a new image (excluding situations where you're asking it to specifically generate an existing copyrighted character, or just to spruce up a specific existing copyrighted image vs generating a brand new one) is almost certainly fair use, even under how strict and draconian copyright law is: The impact and influence any 1 input image has is minimal. If anything, it's more obviously fair use then a human artist looking at references, which will just be 1-2 images and would be more likely to show the influences of the original work being referenced, which is important legally to the standard of if a work is "transformative", a ley part of the fair use determination.
And regarding the issue of AI replicating specific styles, l ) that is even MORE clearly fair use, because "style" is not something protected by US copyright law, at all.
So, if there if there's a ruling or law that establishes these AI's aren't fair use or transformative, or that "style" now CAN be copyrighted... well, that would have implidcations for a human artist, too: If a AI piece which isn't obviously similar to one of the original pieces it was sampled from isn't fair use, then a piece of art a human artist made where the reference used IS obvious won't be fair use either. Suddenly, you could be sued because your piece has a similar pose or lighting or composition to another. If Style is copyrighted, it's the same risk: Now you can ge sued because your piece happens to be in a similar style.
You know whose gonna benefit most from that? Large corporations, the same ones everybody is worried will use AI's. People seem to have this idea that by fighting AI art, they're fighting against corporate exploitation of artists, but we have consistently seen that ANY time there's been a restriction of fair use or an expansion of copyright protections, the big corporations are the ones who benefit, and they use those expanded laws to go after more people, while they're too big to sue themselves so they can continue to break and bend those rules without consequences.
As far as music, which I see people point to as proof that the people making AI's hate artists and are skirting having to use public domain input images: the reason why Music AI's are being handled differently is because copyright law for music itself is different: There are only a limited amount of notes that exist, vs the infinite combination of shapes and colors that exist in visual art. It is COMMON for musicians to come up with similar or identical beats or sections of music tracks as a result of that. There have been many high profile cases where songs that aren't actually likely stealing from another were ruled to be infringing as a result of this. So it is almost CERTAIN that any music generator (not even one that even HAS inputs, but just one thatr's totally random) will spit out music identical or nearly identical to some of the music it used as an input, and there'd be real risk of a infringing legal case.
Hmm, may I ask, let's say a person with little to no art experience generates AI images which copy a certain artist's style on purpose, mentioning them in their prompts and whatnot, and makes profit out of them, with no consent from the original artist - would that artist be protected by law in any way? And if not, do you consider it fair? AND if not, do you perhaps have any ideas how to prevent people doing such things from getting away with it because of technicalities? Asking genuinely.
absolute bullshit.
AI can already create exact the same "Castle" at different angles,.. your info is already outdated
Ive been making my living selling paintings in galleries for 16 years.... this will put most galleries and gallery artists out of business.
it will not replace artist, Art is a way to communicate in visual created by human to human
ai actors are already replacing actors, news anchors, social media influencers, etc. ai is unstoppable. I am not defending ai, im just stating the obvious. People can't tell the difference.@@user-ch9yb2ox4v
the problem is that humans can no longer tell the difference between ai art and art made by humans. I know this for a fact because I am selling ai art. I am able to generate ai images that traditional artists and digital artists can't tell what's what. Of course, most ai art is obvious because it has a digital look but ai can be used to make art that looks traditional. Now in 2024 ai has come so far in two years that I am positive in two more years nobody will be able to tell the difference. @@TheShinorochi
The point is if we won't be able to get paid by doing art, we still will be doing it? Craftsmanship just for the purpose of self growth?
Nathan, the internet at large seems to be thinking of this all in an adversarial way. I've been following this stuff since AI images were barely intelligible - a couple of years - i've used all the platforms and have stable diffusion running on my PC, and I am certain that the future is going to be collaboration with AI, especially in the entertainment and concept art space. I am already seeing concept artists jumping in and experimenting - just as they were the first to find a balance with photobashing, and kitbashing 3d to create artwork ( which was for a good while called cheating but now an industry standard). You should look into img2img prompting. Starting with an image and running it through AI with a prompt to modify it. I am seeing artists now creating artwork through a process of iteration by painting, image 2 image AI, and repeating this sometimes dozens of times before arriving at a final result. I've played around with this myself, it's pretty fun. It's is akin to art directing the AI. It takes cues from what you have painted and what you describe and can offer some neat ideas and nuances which you can then develop through painting and then hand back to it again.
Agreed. Add in that SD is being integrated via plugins into Krita, photoshop, and blender and its become a fun part of the workflow.
I hate to be pessimistic but most artist are not and won't be a match for AI:
First, the "advantages" we still have are things that AI will eventually learn with ease. Composition, color, iteration, anything that may be "unique" to artists is not. AI has come this far and will go farther.
Even if we still had "advantages", those advantages will only help high level artists. Entry to medium level jobs are an easy win for AI and most artists won't be able to keep up, and they won't be able to climb to the top of there's nothing to climb in the first place. AI can improve by brute force, it is designed to do so and with the amount of resources it has no amateur artist will ever catch up.
All this will make it so new artists are scarce, and only people who can afford it would eventually develop the necessary skills to match AI.
There's tons of things to automate, and the whole point of automation was to let people concentrate on art. But it seems art has already been targeted, so either AI is regulated and legal action is taken, or art will be made by robots while people become slaves to work, money, and mindless consumption.
kind of a negative nanny aren't you?
The funny thing is AI has been around for more than 50 years yet now it has come to a point now artist care about it. The funny thing is now it matters when no one had anything to say when back in the 50's and 60's this tech was created. Most artists are creating art that was not created directly from their own mental memory that is comprised of no artists style they learned from. We did what AI is doing but now it can be done in minutes.
Additionally artist tend to claim ownership over things that they don't legally copyright or have or could trademark. Because you can't trademark a style because no style is original. Also people have been stealing from artists for centuries and continue to do so via crowdfunding.
AI Art does not bother me because I am not doing art in fear or AI or tech. Because just like digital, 3D or any CG art seasoned eyes can tell art not created from an artist perspective. The impurities and the failures in art produced by humans it can't recreate.
But people will complain about anything because it is easier to look at an out instead of getting better as an artist.
Reductionism at work. Either you know very little about art or you are arguing in bad faith against artists. (Edit: for clarity this comment is about your first paragraph )
I would also like to add a few reasons for which artists of every field should DEFINITELY learn how to use AI. When the pipeline will be secured (right now is extremely vulnerable) and the legislation will protect us, having your body of work trained into an AI will mean INSURANCE. Something happens to you as a physical being ( vision loss, central nervous system degradation, parkinson..etc) and you will be able to have a backup of your skill level at your peak. A good that can be potentially inherited in case you pass away in an untimely manner, and a contract must be competed. It's SERIOUS black mirror stuff, and we must think about it now, before companies do and present us with abusive contractual clauses that do the same.
Nathan Fowkes is a good artist but a bad teacher. If you look at the majority of his students don’t know how to paint. Look at schoolism students its the worst. A good teacher must go down to the level of a beginner students to mmet their level. You cant just show off your bad ass skill and say unity with variety and keep repeating unity and variety. Are you kidding me? That is a work of a scammer. Preying newbie students to enroll color and light because the name sound so catchy. I will give you a tip. If you want to earn big money just put color or light in your title and boom you will earn money. If you want to be a starving artist go ahead take schoolism.
I was already getting my mind blown one year ago when AI art looked mostly like barely recognizable messes... right now, I am 100% sure that AI art will change everything forever.
What change are you talking about? The best this can achieve is more volume of images being made but for what? We already are saturated enough by social media feeds and this just opens the floodgates to a tsunami of thoughtless "content".
@@BombaJead Art will become a lot more accessible, anyone will be able to bring their vision to life just by using their computers. Being a professional artist will become even more difficult than it already is.
@@boitahaki yeah what will differ people using AI and those who actually can produce images is intent, platform and story telling abilities. Because you can bring a vision to life, but why does your vision matter? Or who cares?
I think people misunderstand the art creation process and how it evolves on canvas when you build it. You dont build anything with AI you just chose what you like best out of few iterations and then you go to the next set, nothing else.
@@boitahaki art has never been inaccessible, people have been making art since we were cavemen so I don't get your logic there. And as for being professional artist ultimately it will depend on what people value but yeah if people value more ai art, earning money out of art would be nearly impossible (and this ironically includes ai art).
@@BombaJead So if art has always been accessible to everyone, then why are you so upset over AI art and wanting to ban it? Did you know scarcity is what make things being worth money?
For more arguments against A.i, you should come and search Steven Zapata Art and his vid: “The end of Art: An argument Against A.i Art”, might help you. Goodluck Artist!
Yes, burn, beaches
We're doomed :'(