Title should be more like, artists concerned about Stable Diffusion AI model that makes images look human-made.
replies(1):
I've read in advice for technical founders that programmers tend to want to jump and automate the wrong problems, and this feels a bit like one of those cases in retrospect.
However, visual art creates a cultural feedback loop that is constantly evolving, it is a crucial mirror that shows us stark naked. Just read the history of art.
People should not forget that there is no way this technology can make such a vital form of collective expression relevant without artists participating in it.
Of course it will disrupt a lot of things, and nobody enjoys that, but new ways of meaningfully involving humans in it will continue to emerge. Otherwise it would just cease to be relevant.
Even fine artists have likely had their practice influenced by the whims of what the algorithms on major social media outlets are willing to favor in order to get engagement. It kind of feels like many creatives have been incentivized into becoming slaves to these processes, which has in turn made them seem replaceable with AI.
People consume art because they enjoy admiring the human talent that creates it, celebrating that some individuals are capable of extraordinary feats the vast majority of people are incapable of. It’s the same reason people watch sports—they enjoy admiring the top echelon of human physical ability. Very few people would watch Olympic Games performed by realistic androids.
I do agree that tools like this could eliminate mediocre graphic designers, or anyone else creating visual products that are so mundane that their viewers never bother to consider the artist. Corporate Memphis [1] designers’ days are numbered.
[0] https://blog.paperspace.com/art-style-transfer-neural-networ...
I guarantee that if there is ever a self-driving cup, people will be mostly focused on the human engineering talent.
The starving artist who sells out to create ads or create content for commercial entities will find those opportunities dried up
A single agency using this tool effectively could "in theory" produce 500 times the artistic output from a single artist. Vastly shrinking the market for a decent paying career.
But the human element is the main thing. This is the crucial part. Take us out of the picture and there's nothing very meaningful left for other humans to see.
Sounds like the AI model should be paying royalties to every affected artist for the right to sample their work.
and
I imagine a primary motivation for investing in art-generating AI is cost savings.
disclaimer: By default, I tend to be less sympathetic to artists here. It's due to their communities' collective silence about decades of ever-ratcheting (and typically purchased) copyright laws.
One thing that makes me a little hopeful is that every image I've generated with DALL-E 2, even the best ones, would require non-trivial work to make them "good".
There's always something wrong, and you can't tell the model "the hat should be tilted about 5 about degrees", or "the hands should not look like ghoulish pretzels, thanks".
There's also this fundamental limitation that the model can give you a thing that fits some criteria, but it has no concept of the relationships between elements in a composition, or why things are the way they are. It's never exactly right.
It's like the model gets you the first 90%, and then you need a trained painter to get the second 90%.
But yeah, it will certainly devalue the craft, don't get me wrong. And anyone who is callously making comparisons to buggy whip manufacturers should consider how it would (excuse me, will) feel when AI code generators pivot to being more than a copilot, and suddenly the development team at your office is a lot smaller than it used to be, and maybe you aren't on it anymore.
If you spend a lifetime mastering some skill, and then it's just not valued anymore, it sucks, and you get pretty mad about it.
I can recall a little bit of the fuss at the beginning of digital art, "is that even real art?" I don't think this fuss has as much to stand on. It's a tool for making art; whether it's really art or not is a judgement only the individual viewer can make.
> At CC, we believe that, as a matter of copyright law, the use of works to train AI should be considered non-infringing by default, assuming that access to the copyright works was lawful at the point of input.
https://creativecommons.org/2021/03/04/should-cc-licensed-co...
There are already tools out there for people who want to automate the programming part of game dev. I don't see how that's any different in principle from an AI that generates art. This seems like a better solution to the problem of making a game without being any good at art than buying assets since your assets should be unique.
Rembrandt started out as an apprentice, with about four years starting around 13 before he opened his own studio. Picasso was trained by his art professor father from the age of seven in what sounds like a very traditionally hardcore fashion.
I would gladly trade less efficiency in art generation for more artists being employed. "AI stimulation" is not a 1:1 swap for being an actual artist.
https://en.wikipedia.org/wiki/Amen_break
Not saying AI/ML won't be damaging and cause negative effects, just pointing out that other uncompensated prior art has been used to create transformative works before we ever had DALL-E.
People still buy hand crafted items of course, but they tend to be relegated to high-end or niche markets. The same would apply here. The larger number of people employed churning out relatively uncreative material would be displaced just like weavers and seamstresses were by textile mills.
I’m not sure where the line is for machines. Copying the signature is funny, but also fixable. Is copying the style fundamentally wrong? If yes, is it wrong when humans do as well?
Feels like an impossible line to define.
Can DALL-E make set defining breathtaking high res promo art, with fantastic composition, style and proportion? Maybe, but probably not as well as ex. Raymond Swanland right now.
Can DALL-E crank out the 50x lands and commons needed for the set? Almost certainly.
Superstar artists will be fine but I fear the bottom will fall out of the craft.
Call me a doomer, but I think this makes the possible consequences even worse.
Remember the 80/20 rule.
A lot of modern product innovation is not really about improving quality - rather, its about introducing lower-quality versions of existing products which are significantly cheaper than the original but still "good enough".
Dalle2 and friends could fall into the same bucket. If they produce artwork that is objectively worse than a human-painted version would be, but still "good enough" for many mundane usecases - stock photos, concept art, etc - we might still see a wide adoption and displacement of human artists from those usecases - along with an overall drop in quality of artworks.
The problem with being shocked by Dall-E, in my view, is that it shows an ignorance about the historical development of art and its incredible diversity of practice + the final productions and forms of art. OpenAI have sort of Warholised digital art in a way and that's just very standard in art history. People went crazy when Warhol productised art but in reality this was an overreaction and plenty more stuff came after that which completely different in its orientation towards art (e.g. something like Hans Haacke). Dalle-E is a system for producing digital art in the way that Warhol's practice was a system for producing visual art as a commercial product.
Do you think AlphaGo was motivated by a burning desire to discover better board game strategies?
A main hallmark of breakthrough artists is inventing new styles and making them their own, a feat that AI has yet to perform very well (unless you include the aforementioned blurry telltales of generation and scrambled letters). Otherwise artists are just borrowing ideas from each other anyway so AI removes the need to learn someone else's style to replicate it.
In my opinion it will hurt the bottom tier of artists, but the ones who create new styles along with their own works will continue to thrive.
I'd actually love to see DALL-E take on the greeting card industry. Now that would be fun.
I see comparatively few ethical concerns about the economic fallout professional artists and illustrators will experience as a result.
I fully grasp the concern here. It's not unlike what happened during industrialization, but on an intellectual level. I don't know what the solution is. Basic income has been floated, but I'm having a hard time evaluating all the arguments for and against.
Not to mention the 3D graphics of the past 2 decades, oh ma gawd! It's artist paradise compared to what people had 50 years ago.
THE FUTURE IS HERE AND ITS AWESOME
You will know it when you see it.
And 3D printers for paintings exist, that can replicate brush strokes and other techniques, and they will only get better over time.
So there's very little left in the human arsenal as the AI generation and AI painting techniques both improve.
Or hire the computer and you can have all three.
You're telling me I don't appreciate the genuine artifact? Well, tell me how realist painters survived film photography, and how film photographers survived digital photography. I'd ask but I don't know any anymore.
Subjectivity defines what is good. This thing makes art as reliably as 9 bucks gets you fries, a coke, and a burger. Talk about disruptive technology.
Starting e.g. an art education right now seems likely to be extremely nerve-wracking as your talents may very well be woefully obsolete by the time you graduate; the exception perhaps being those top-0.1% talents that will feed the models of the future with new material.
It's not backwards. It's the same as if a human artist studied it.
Backwards would be thinking that any creative work you make is a derivative work of many creative works that you've seen in the past which you aren't copying from.
“An obese politician smoking a cigar being held up by a crowd of starving people wearing American flag pants in the style of Norman Rockwell”
“A web page sign up form in the style of Uber”
Content violation or ??. Heck, any cartoon in the New Yorker would probably be a content violation.
The hard part, I think, is art making an “on the edge” statement. If they allowed this, it could easily - and would - be taken too far (depending on what your version of too far is).
I just don’t think human art and all the subtleties behind it can be algorithmically replaced. They weren’t algorithmically created.
Corporate soda ads… that’ll probably get replaced (but will still need layout and mixing)
UI design and video game art (not concept art in game art) seems too precise. And meaningful art seems to have too much in baked human judgments.
I am as afraid of this as I am of Github copilot. Meaning, not much.
DALL-E is super cool, but like AI becoming conscious, I am not holding my breath.
All my friends and family will get AI cards now!
* feels like there will be a subtle shift from (corporate) designers as creators to designers as technical sheepherders and retouchers. senior designers who have a real nose for art direction and understanding the fundementals will be fine. interns and juniors may get hit with a widening canyon of blindly using these tools without understanding why the "base" styles/logos look good. the act of creation is a learning experience that this removes.
* I feel like phi-y in a different comment has the most accurate read on the situation, given how this will not replace the superstar artists who've already gone through the years of training and sweat-shop work to refine their craft, but the beginner artists who "need" the sweatshop years to get a real-world sense of their craft and the money from the more easily reproducible works.
* there is a difference from using digital art as a tool (say, an apple pencil) to digital art as a resource-crutch (photobashing) when crunched for time or resources. I have mixed feelings about photo bashing as an art but those who rely on the resource-crutch sign may get hit hardest versus the ones that already have the raw illustration skills and can adapt/art-direct on the spot.
* copyright laws are gonna be interesting when people start trying to "base" results on logos like disney and coca-cola and get lawyers sent after them if it's too obvious. My fear is this creates yet another power imbalance between those who have the money to sic lawyers on others versus not.
* I'm mixed on DALL-E because on one hand and to be perfectly blunt, I feel like the most easily reproduced work is ugly/same-y as sin, and lacks all originality compared with the more creative/taboo-art circles who are the first ones to push stylistic and thematic boundaries. I'm not worried about DALL-E replacing styles that are inherently harder to reproduce.
On the other hand ... a sense of grace should be extended to artists who are affected. Being slowly replaced/treated as obsolete is a hard thing to experience, much like how natural age is rough on people's emotional states to the point where it's a type of trauma. It's messy, and it's hard, and I acknowledge there's going to be casualties.
That is absolutely not what the OP is complaining about. They're not saying that because AI is good, they won't find work. They are complaining that in training AI for art generation, builders took works from living artists, without consent from them, and that in so doing allowed generators to make new art in the style of said artists.
The example given is that Stable Diffusion even tries to reproduce logos/signatures of living artists.
If I produced a rubbish search engine that bore a malformed "gigggle" logo using Google colors, how long do you thing I would survive before being sued out of existence by an army of Google lawyers?
But that's exactly what many AI generators are doing here.
Edit: the first version of this comment confused Stable Diffusion with OpenAI, and stated that OpenAI was owned by Google. OpenAI has a strong partnership with Microsoft. Stable Diffusion is not OpenAI. Sorry for the errors.
Does the line between novel and graphic novel blur, as more authors "direct" AIs to depict what's in their brains?
Is this how we get to infinite generated worlds in the metaverse?
you're not afraid of DALL-E, you're afraid of an army of fiverr workers stealing your job. Stock photos and low quality art have already been commodified. Very few people go and commission bespoke stock art from the individual working artist, they get a subscription from one of the gazillion content stock photo factories for a few cents.
I can bring up a form of art that is not captured by this conceptual scheme. Carsten Holler's SOMA:
The exhibition Soma was installed at the Hamburger Bahnhof in Berlin in 2010. Its main element were 12 reindeer in two pens running the length of the former railway station. Half of the reindeer were fed the fly agaric mushrooms in their food, which are part of their customary diet in the wild, and turn their urine into a hallucinogen. The reindeer urine was collected by handlers and then stored in on-site refrigerators for use. The experiment was extended to canaries, which were housed in two hanging cage pieces, to mice, and to flies. A mushroom-shaped Elevator Bed was installed in the middle of the space, and visitors could spend the night on the premise for a fee.
Do you see what I mean by this digital art vis a vis the collapse of all art being an overreaction? Carsten Holler is a real artist, the art is good, and its not reproducible in the way digi-physical stuff is. It's experiential art and presented in a gallery. Conceptually, its about as far away from DALLE-E as possible but its still art and not "captured". I think art is not over and DALL-E is not poison for artists.
But framing it as a matter of art does sound like gatekeeping. It'd be "that cannot be art, because x" when all they mean to say is "please don't stop paying me for this".
That said, I'm also not too impressed with many contemporary "artists", especially the type this person is afraid of being replaced with AI.
What I see here is that artists that have turned themselves into art making robots are vulnerable to being replaced by robots that are making art.
As a counter example I present the work of Ben Kovach [0], a generative artist, who makes art with robots essentially. All of his work is generated from computer programs but there are many (and certainly not all) pieces that capture a genuine sense of art that I have not seen in either the slew of human artists churning out robotic work to satisfy customer demand or current generation AI generated art.
Kovach's work reminds me of the work of Factory Record's designer Peter Saville, probably most famous for the cover of Joy Division's Unknown Pleasures [1]. In a sense Saville is pushing form the opposite direction of Kovach, trying to use human skills to create art that feels somehow robotic and sterile (whereas Kovach does the opposite, create something that feels more human than your typical machine generated art).
I don't see current generation a threat to art, but rather the trend towards the reduction of art towards being a reproducible, automatic process. That is the art that can be replaced with AI was already art created by an ad hoc "artifical" artist.
1. https://m.media-amazon.com/images/I/812FS2R2v6L._SL1500_.jpg
If people are happy with "good enough" they generally don't hire a digital artist in the first place (the whole reason DALL-E can exist is because there's a lot of digital imagery on relevant subjects/objects available to it to train, and there's even more an internet search away) or if they do, they get one off Fiverr.
For mocking up quick concepts, that might be different, but that's a workflow improvement.
Excerpt follows:
"We are suffering from the ruinous competition of a rival who apparently works under conditions so far superior to our own for the production of light that he is flooding the domestic market with it at an incredibly low price; for the moment he appears, our sales cease, all the consumers turn to him, and a branch of French industry whose ramifications are innumerable is all at once reduced to complete stagnation. This rival, which is none other than the sun, is waging war on us so mercilessly we suspect he is being stirred up against us by perfidious Albion (excellent diplomacy nowadays!), particularly because he has for that haughty island a respect that he does not show for us
.
We ask you to be so good as to pass a law requiring the closing of all windows, dormers, skylights, inside and outside shutters, curtains, casements, bull's-eyes, deadlights, and blinds — in short, all openings, holes, chinks, and fissures through which the light of the sun is wont to enter houses, to the detriment of the fair industries with which, we are proud to say, we have endowed the country, a country that cannot, without betraying ingratitude, abandon us today to so unequal a combat."
If it's a 1:1 copy, I agree. If it's a "that looks vaguely like the style that xyz likes to use", I disagree.
And I assume you'd run into plenty of situations where multiple people would discover that it's their unique style that is being imitated. Kind of like that story about a hipster threatening to sue a magazine for using his image in an article only to find out that he's a hipster and dresses and styles himself like countless other people, so much so that he himself wasn't able to tell himself apart from another guy.
Ultimately you are talking about art as a product and that means you think all art is captured by Warhol's system of art. Which is what all digital artists overreacting to this also think. It's not the totality of what art is. I like DALL-E. Good work, but now what else is there beyond digital art? Or by pushing digital art to some more extreme outcome?
I can see this eating into the commercial artwork market, which is bad news for the artists that rely on that for income, but it's not an existential threat to the meaning of art.
I'm not sure what you mean by this. Reproducibility has virtually nothing to do with interest in my opinion. I was an admirer of plenty of artists before I'd ever seen their work in person. Perhaps I am missing the point.
This is not something new. https://www.realistartresource.com/the-tradition-of-copying
Copying of existing works is part of how an art student learns. That this one happens to be a math model at its core is an interesting philosophical problem. The other people's work to serve as your training set is exactly what art students do - and the works they copied and the works that they have yet to produce are not royalty encumbered.
Nor should a mathematical model. It happens that the developers working on this problem have gotten it so that it can do its learning and creation many times faster than an art student in a gallery... but it still can't get hands and faces right.
All of us artists woke up in the world where the thing we were learning to produce all of our lives became 100x cheaper. I've been kinda waiting for this to happen for truck drivers or translators and was honestly surprised that it got to the artists first. But, there's nothing to do now.
There will be a lot of takes trying different ways to deny this reality. Like if we agree that this thing is immoral or anti-cultural it will go away and our potential customers will forget about it. I think that it's a waste of time The industry is not completely gone, but it will go though a massive transformation. The basic hierarchy of artists will probably stay the same - people who were good in the pre-ai era will still be good working with the ai. If you've learned your basic skills right - they will still work.
I've personally already changed my workflow to use ai art for the sketches, ideas and references. And then polish them to commercial level using my old experience and skills. It's funny how this thing is so new, but I've gotten so used to it that it's already hard for me to come back to my old set of tools. And I do recommend all other artists to do the same and to try to find their place in the new reality - it really seems like there's no other way.
A second or third generation model may really understand relations between objects, depth, lighting, etc.
It's both amazing and scary. A little behind the curve, but expect similar developments in music generation.
Given the speed of all of this, the socio-economic discussion on AI should really get prioritized, as very little cognitive work is going to be safe.
I think DALL-E is a win-win by pushing digital art to some further extreme and forcing people who are open to art beyond the digital to try new things. Digital art had reached a point of maximum inertia before this imo.
But! Seeing DALL-E artwork from a bird’s eye view, I am more and more convinced that it’s just another tool, not a replacement: It is amazing good at producing unexpected results, and just like GPT3, it does so by regurgitating what is already out there, remixing it in somewhat interesting ways. This is an especially useful tool to quickly produce mood boards, or communicate an artistic intent. However it is not a replacement for the kind of thinking required to produce good artwork.
This is a statement that is pretty quickly disproven if you actually pay attention to the generated art. Lately I've been seeing TikTok videos where people are using DALL-E to create "new aesthetics" - "vampwave", "neon apocalypse", etc.
>interns and juniors may get hit with a widening canyon of blindly using these tools without understanding why the "base" styles/logos look good. the act of creation is a learning experience that this removes
Surely with digital art more than most professions the understanding of why the end product works is pretty orthogonal to the actual technical implementation details anyway? You don't understand composition or style or taste or brand identity or proportions by creating textures and shapes from scratch or knowing your way around a tool. To that extent, spending their days filtering mediocre compositions not created by them and fixing their flaws might give them a much keener sense of what isn't quite working than trying to judge mediocre compositions they've put a lot of effort into themselves objectively. I can see bigger issues with the thought that might be comparatively dull work for a lot of people.
From what? Until strong AI comes about, good human artists will always be able to compete on originality and relatability, and even afterwards, human art could be valued for its "authentic flavor", or something like that. Until then, among humans, only the hacks need fear these systems.
Yeah, art is vast. It includes me saying "Kerflaffle!!" while I ride a unicycle and fart out a candle in a dimly lit room. But that doesn't usually pay bills and most artists don't do this type of art. It serves a different audience/consumer. It has a different market cap. Let's not confuse oranges and grapefruit.
These models are not learning or being inspired in the same sense humans are. The laws tgat apply to humans should not be applied to them.
Since this thread is about art, I figured people might read the name Basquiat in your comment and think that he wrote the quote you cite, which would in my mind give it greater relevance to this discussion than it has.
Those discussions aside, what I meant by social reasons was people wanting to see some tech go away because it's automating jobs.
Software development is heavily labor-constrained, if copilot can make everyone a 10x developer, we'll get slightly less than 10x the features-per-year on an industry-wide basis after contributors shuffle around.
The effect will be most pronounced in application development, where a team of 1-5 is about ideal for a coherent app made with taste, and that team could produce the output of 10-50 developers. Not such a bad thing.
Unfortunately this is unlikely to be true for visiual art, I don't predict that making artists ten times as productive will meet a latent demand for ten times as much art. Could be wrong, but my sense is that about as much art is purchased as people want to buy.
It’s not a human artist, and it can only regurgitate mash-ups of work stolen from others.
It is not.
The AI model can only regurgitate stolen mash-ups of other people’s work.
Everything it produces is trivially derivative of the work it has consumed.
Where it succeeds, it succeeds because it successfully correlated stolen human-written descriptions to stolen human-produced images.
Where it fails, it does so because it cannot understand what it’s regurgitating, and it regurgitates the wrong stolen images for the given prompt.
AI models are incapable of producing anything but purely derivative stolen works, and the (often unwillingly) contributors to their training dataset should be entitled to copyright protections that extend to the derivative works.
That’s true whether we’re discussing dall-e or GitHub copilot.
“Training” an art student and training an AI model are vastly different, and your equating the two is, frankly, nonsensical and absurd.
An art student isn’t a trivial weighted model capable only of mapping stolen text prompts to stolen image representations of them.
> It happens that the developers working on this problem have gotten it so that it can do its learning and creation many times faster than an art student in a gallery
It hasn’t learned anything.
It correlates stolen textual descriptions with stolen images, and then regurgitates mash-ups of the same.
This type of AI model cannot produce anything other than purely derivative work stolen from others.
Not sure the people selling paintings knocked out in an afternoon to tourists who visited their studio and chose to buy an original instead of the visually indistinguishable print for a fraction of the price have much to worry about...
Some of them really are outstandingly good. Beyond what I expected, and I have had access to Dalle2.
Also, since you like to emphasize the 'digital' part of these works, don't forget we're probably not much further than one Master-level project for some mechanical engineering students from having AI-generated physical paintings.
If that happens, I agree that the top 100 human artists in the world will likely have jobs, but they won't be successful in the sense that their work is uniquely valued by society. We pay to see the very most talented humans perform tasks that have been successfully automated, such as chess and lifting heavy objects, not because we need the service they provide but because we get an emotional kick out of seeing other humans perform way outside the normal range of human abilities.
Union boss: "I don't know Mr Ford, but how will you get those robots to buy your cars?"
Like others have said, the first ones have issues, but late stage is when things get problematic.
What you describe can only happen with general intelligence, not these fancy neural nets. If anything, they will become powerful tools to help artists augment their creativity.
If I was to dabble in sci-fi art and made something that fit in the art style of Steward Cowley ( https://archive.org/details/terrantradeauthorityhandbookstar... ) do I need to credit the art?
When it comes to playing around in blender - my designs are obviously derivative of others - do I need to credit those artists? Even the ones that I don't remember more than a "I saw this print at a comic art show once..."
How original does my own work have to be before it isn't a mashup of stolen images that I half remember?
Probably, yes.
> When it comes to playing around in blender - my designs are obviously derivative of others - do I need to credit those artists?
Again, probably, but nobody is likely to care if you’re not actually selling your work.
> Even the ones that I don't remember more than a "I saw this print at a comic art show once..."
Then that’s not the prompt you should be starting with if your goal is to produce an original work.
> How original does my own work have to be before it isn't a mashup of stolen images that I half remember?
How original does it have to be before it’s not plagiarism?
Now, remove your ability for individual creativity, such that you cannot come up with an original idea. All you can do is plagiarize.
That’s the difference, here. This isn’t an AI trained to have creative thought, a genuine understanding of what it’s making, and original ideas. It’s an AI trained to regurgitate mashups of plagiarized works based on weighted correlation between the prompt and the (also plagiarized) descriptions of the works it’s regurgitating.
Instead of hiring 100 artists, you could have this 1 model generating images, and then 10 artists touching up AI's results to take them from 90% to 99% on quality.
It facilitates more capital consolidation.
I think this will change up to the extent that the majority of people won't care.
We all stand on the shoulders of giants. If you're very dismissive, I think it's easy to say the same about most artists. They're not genre-redefining, they carve out their niche that works (read: sells) for them.
Nobody will care where their art comes from any more than you care about how your food's field was plowed; and all their lives will be better for it.
Keep your comment somewhere. Come back to it when you look at a new tool that promises to merrily trample on your entire field’s income and provide an endless source of “usable, I guess” substitutes. Let it provide solace as you stare into a future with no room for the craft you’ve spent a lifetime honing.
I claim humans can be inspired.
I don't claim to know how human inspiration happens, or if neurons have anything to do with it. (They may, but I make no claim). Not being able to describe the process by which human inspiration happens doesn't invalidate either of my claims.
If there is a satisfactory non-bit based explanation to how computer algorithms achieve inspiration, I would accept that to. We have the advantage with computers, that their activity is conveniently summarized by their programs which are represented in bits, so expecting an explanation in that form I think is reasonable.
The defense if the claim of human inspiration is (1) we have that word for the concept (2) we have thousands of years of thought, philosophy and literature giving support and definition to the concept.
