Another interesting take is that a lot of human made art isn't art either. Art is supposed to be questioning or critiquing social convention, not be "pretty pictures" which the large majority of output have been for the last few decades.
In the end, painting is 99% a practiced skill, and 1% original human thought. And this 1% is arguably comparable in effort to what is provided as input to an AI tool.
I mean, according to a physicalist's view of the human condition, yes. You are a function of the haploid cells your parents conceived you with, and the environment those cells responded to over your lifetime.
actually, that stat is completely and totally wrong. If you take the most similar parts of human and chimp DNA- just the well-aligned proteins- then yes, we differ by 1%. But that's not a metric that means much, if at all, since many aspects of phenotypes are not controlled by the difference in our protein coding regions.
I would disagree with that. Much of creating art is a feedback loop. An artist rarely, if ever, has a fully formed picture in their mind of what a finished piece will look like, and it's even more rare that the final piece conforms to that image. In the process of creating, the artist takes in what is in progress and has to adjust accordingly.
The final piece is influenced by the medium it's created in, the process of creating it, constraints both momentary and well-known, and the context of the time and whims of the artist.
The billions of images and its software determine its outputs right? Is there the same analogy for humans?
I don’t think so. It’s much harder to imagine a deterministic story of how we developed set theory for example. Nothing empirical forces us into set theory. It’s not needed by physics for example.
The billions of images and software plus the entered phrase fully determine the output right? What possibly similar story gave rise to things like set theory?
Artists should be thankfull for DALL-E. A century ago (more or less) another tech invention - camera - started to create realistic pictures. Thanks to the camera, artists were able to move away from painting realistic pictures into abstract art. They should really be thankfull to be able to live in the times when another tech invention forces them to move the art further forward.
Are we sure about that? The cave paintings at Lascaux are of recognizable animals. Egyptian and Maya art are naturalistic too, and highly refined toward realistic portrayals.
For most of Western art's history, content and style were both dictated by the patron. Even the ideas we hold around art and freedom of artistic license had to wait until the 1800s and the Salon de Refusé to come into existence.
Most buyers and makers of art today still gravitate toward figurative drawing and naturalistic landscapes. It seems like we have always used art primarily to reflect and comment on the real world.
Consumer-ready black and white photography is about 150 years old. Many “realistic” (whatever people mean by that) paintings were made after that point. Consumer-ready color photography is about 50 years old, and not a lot of artists followed any traditional style by that time.
In most simple terms, the oscillation between vaguely “metaphysical” (imaginary, romantic) and vaguely “physical” (realistic) approach has occurred for a long time.
On the plus side modern studio art is so execrable that AI laying waste to it will probably be a net positive. The human urge for expression will always find an outlet and perhaps this will channel it in new and better ways.
Definitely. There is a weird elision on the author's part that other forms of artistic creativity exist. FWIW there is plenty of room at the bottom for those artists that can pull together the tenuous threads between seemingly incompatible disciplines.
Decades ago the art world traded skill (lifetime of practice, apprenticeship, in relative poverty) in the reproduction of realistic forms on canvas and in marble, for novelty (Piss Christ, Duchamp’s Urinal, Basquiat’s graffiti). DALL-E will just precipitate the next stage I think.
I make no judgement as to the validity or quality of the art, either classical or modern. I think art is much too personal a thing to speak of in quite that way. Only that the art world has decided to make the trade I speak of. Someone who can produce photo realism today, will not set the art world on fire. Neither will someone who takes 6 years to produce a single large anatomically precise sculpture after 25 years training. However if you can inject novelty (I recently read of a statue made of the artists own congealed blood, and another made of their ear wax) then you have the attention of the patron.
You seem to be describing "technical skill", like "can he move his pencil in a straight line". Essentially, can he draw the things he intends to draw on the paper?
I'm talking about "artistic skill", which is the ability to effectively put ideas or emotions into a work of art.
An analogy which might explain my thinking: A software engineer who writes a 1,000 line program in a month is producing work beneath their "skill" because they could've written a 100,000 line program that does the same thing, in that same month. Typing is a technical skill for programmers, but I wouldn't say that typing skill makes you a better software engineer.
I'm talking about the same skill you are. You (and many others) think that the third picture requires a lot of artistic skill. As far as I see, the third picture is a prototype with a lot of technical debt.
I don't think we can meet on common terms because our definitions of art are different. I do not subscribe to the opinion that art is about challenging our thinking. Instead, I consider good art to be about immersion. You do not have to convince yourself to like it. You do not feel the need to like it ironically. Yes, you can get used to it and it loses some of it's power, but the greatest art is the art that you can keep looking at (or listening to etc.) without feeling the need to justify it. (I cannot remember which philosopher I borrowed that idea from.)
The art world has always been about novelty, but what was considered novelty centuries ago is now considered as "classic". There are countless examples of artists that didn’t get much recognition because what they did was too "modern" for the times they lived in (see Caravaggio for example). Most of the things you see in museums are notorious because they were extraordinarily novel for their time, not because they would be hard to reproduce.
I've always pondered if code could dream up idiosyncratic styles of art. Instead of using 'training' material, it procedurally generates art using random noise as the base input, and uses a different algorithm each time it generates art, instead of a predictable algorithm (to avoid everything looking the same). You could, in effect, generate serendipity on demand, and use algorithms to create algorithms, a very meta thing to do, but would be an interesting project to undertake.
It won't work because the resonant part of art doesn't come from random noise. It comes from an inflection point on a particular emotional introspection or connection. When the viewer or listener shares this connection, art happens. Without this emotional exchange, art is lifeless and uninteresting.
Interestingly, music works the same way. Random, unpitched noise sounds to us like wind or rain. In music synthesis, randomness has a use (like in DALL-E) but it is limited. What humans want hear to "get into" a song emotionally comes from specific arrangements of pitches - not random ones. The artist chooses them because they "feel" a certain way, an exclusively human concept. When the listener has the same emotional reaction, art succeeds. No DALL-E painting will ever achieve this because it has no feelings to begin with.
> DALL·E 2 is a research project which we currently do not make available in our API. As part of our effort to develop and deploy AI responsibly, we are studying DALL·E’s limitations and capabilities with a select group of users.
> AI art is about to lay utter waste to traditional visual art forms. This will be so much more destructive than what the Internet did to music. It will be a technological conquest of one of the great human avenues of spiritual transformation.
AI now challenges what we've always thought was unique about us, "soul" and crap. Some people who can't do anything else obviously get desperate. And don't get me started on the fact that "art" is not defined or explained nowhere in this essay. Let me define "art". Art is not Malevich's Black Square - that's 50%/50% crap/PR.
Art is ~90% work and ~10% ingenuity. AI cannot reproduce real art because no one will see any work behind it. People go to galleries not to just appreciate ideas, they go their to appreciate work. Real artists are safe. Sleep tight.
Okay, but then we shouldn't call something "art" due to the creator's style or brush technique either.
What I like about AI and algorithmic art is that it can produce near infinite varieties around a style or technique, and often produces more interesting results than an artist leaning on a style with very little meaning or reason behind it.
I think that many spectators of human-made art also fall for something close to Pareidolia, Apophenia, or the sensing of meaning that is not (necessarily) there. Even expert critics can't infer exactly what authors or creators had in mind when creating a work of art or literature and without a statement from the artist it's just a heuristic guess based on culture and personal taste. People have wondered for hundreds of years what the Mona Lisa's expression means, making assumptions about Da Vinci's technique and intentions for her smile, but without definitive evidence from the artist himself it's all guesswork. Likewise in this article the author assumes that art is about communication but in fact art is a very lossy and ambiguous communication medium. Human artists choose to put concepts in juxtaposition with each other but what that means to people differs with each individual, and so what an artist intends to communicate may not come across even if people strongly feel that they are right about their own interpretation. Dall-E2 is almost as capable at the same kind of juxtaposition (astronaut riding a horse) that makes people think about what it might mean.
Ultimately the discussion falls to the notion of whether a simulacrum is experience.
If you see a painting of a horse accompanied with a description of why the artist painted it, does it matter if those molecules were arranged that way by human experience?
In either case, you are now being prompted by a simulacrum. And should you then artistically be inspired, your readers are viewing a simulacrum of a simulacrum. Does this compound the "error" or damp it?
I may come across as dismissive but most people don't care for abstract or portrait-y art. I know I don't.
What I appreciate is comic books. With the amount of computer tools available today, comic books are way more beautiful than two or three decades ago. Some of the most gorgeous art I've seen is in comic books.
DALL-E can't replicate that and I doubt any AI can (or will be able to) draw exactly what some writer is envisioning in their brain.
So artists aren't going anywhere.
Why not? DALL-E can already emulate comic book style. With more tweaks and training I imagine it could be made pretty good at it. DALL-E2 is much more precise and coherent than DALL-E1 was, and I see no reason why we wouldn't get an improved DALL-E3/4/5.
I bet it's only a matter of time when it can create an ok-quality full comic book based on a textual script. And perhaps the script could even be generated by GPT-like AI too.
I expect we will see lots of indie comic books made using these tools, and perhaps this kind of comic will ironically be seen as less corporate and more authentic?
XKCD is doing fine with minimal art, because it’s about the writing. Other writers will make comics very different in style, without having to settle for minimalism. Very long, story-driven comics will be more feasible for amateurs to make in a shorter time. Instead of drawing a few pages a week (which is a lot of work), they could write and illustrate a short chapter.
More labor-intensive comics that are more about gorgeous artwork than story will probably still be around, but a smaller niche than before.
I was thinking of Marvel/Star Wars/DC comics instead of webcomics like XKCD but yeah youre right.
These tools will lead to a boom in webcomics and memes where precision isnt important. The superheroesque traditional comics will be harder to replicate with this.
Comic book computer generation is not here yet, but if you want my opinion, it is coming soon. I don't see any reason why we cannot create a cartoon octopus call it Octofred, use it as a character and embed it in a story.
Dall E is a tool. Just like every other tech ever made. A paintbrush doesn't make the Mona Lisa, but it sure CAN. that ability to do something with a tool is the real art IMHO. Prompts are the paintbrush of DallE. Can i have random prompts? Yes. Can a strap a paintbrush to a double pendulum? Also yes.
Yeah the art here is the prompt, the input to the machine. I usually say, that any act of human like work, or exercise or singing, or doctors, or engineers of any kind, will become information-analyst and engineer, information-analyst and doctor, information-analyst and painter.
That information analysis does not necessarily need to be performed by the same person. One is the information analyst, another one one is the painter, and they both in collaboration create some art paintings. I don't see any problem with that. Which one is the artist in that case? The information analyst or the painter, or maybe both? I would argue they are both artists.
I have personally prompted and generated hundreds of pieces of art, real art and awesome art, and anyone who tells me that i don't know what a Picasso is, doesn't know what he is talking about. Picasso was my favorite band for a long time!
Let the market be the judge of what is art, if you didn't know this is AI would you buy it? as simple as that.
The rest is an opinion nobody needs.
When it comes to art nobody has a damn clue, so i got bad news for you mate.
145 comments
[ 0.21 ms ] story [ 241 ms ] threadSays who?
The final piece is influenced by the medium it's created in, the process of creating it, constraints both momentary and well-known, and the context of the time and whims of the artist.
Also the history of pigments is fascinating. There’s a whole lot lot more to painting than brush strokes.
True, but the line between art and great art is 99% thought and 1% skill.
I don’t think so. It’s much harder to imagine a deterministic story of how we developed set theory for example. Nothing empirical forces us into set theory. It’s not needed by physics for example.
The billions of images and software plus the entered phrase fully determine the output right? What possibly similar story gave rise to things like set theory?
https://www.youtube.com/watch?v=yk5EJd7S2uA
Look at this example, it's just beginning.
Art was never primarily about realistic mimicking of nature, even if some subset of it in some cultures may have temporarily appeared to be.
For most of Western art's history, content and style were both dictated by the patron. Even the ideas we hold around art and freedom of artistic license had to wait until the 1800s and the Salon de Refusé to come into existence.
Most buyers and makers of art today still gravitate toward figurative drawing and naturalistic landscapes. It seems like we have always used art primarily to reflect and comment on the real world.
In most simple terms, the oscillation between vaguely “metaphysical” (imaginary, romantic) and vaguely “physical” (realistic) approach has occurred for a long time.
> Now literally anything could be art
> This post is art
Edited for typo.
https://mymodernmet.com/wp/wp-content/uploads/2016/12/1-pica...
Than at 35:
https://mymodernmet.com/wp/wp-content/uploads/2016/12/6-pica...
And that he was an even worse artist at 90:
https://mymodernmet.com/wp/wp-content/uploads/2016/12/13-pic...
I'm talking about "artistic skill", which is the ability to effectively put ideas or emotions into a work of art.
An analogy which might explain my thinking: A software engineer who writes a 1,000 line program in a month is producing work beneath their "skill" because they could've written a 100,000 line program that does the same thing, in that same month. Typing is a technical skill for programmers, but I wouldn't say that typing skill makes you a better software engineer.
I don't think we can meet on common terms because our definitions of art are different. I do not subscribe to the opinion that art is about challenging our thinking. Instead, I consider good art to be about immersion. You do not have to convince yourself to like it. You do not feel the need to like it ironically. Yes, you can get used to it and it loses some of it's power, but the greatest art is the art that you can keep looking at (or listening to etc.) without feeling the need to justify it. (I cannot remember which philosopher I borrowed that idea from.)
...but what about DALL-E 2?! I mean, that's an artist!
It is more subjectively art like.
like everything before it, it's just another tool that some will show a 100x mastery of than the average person
Have you tried having DALL-E draw a blue cube, red cube, and yellow sphere in the same image?
https://twitter.com/david_madras/status/1512573390896480267
Interestingly, music works the same way. Random, unpitched noise sounds to us like wind or rain. In music synthesis, randomness has a use (like in DALL-E) but it is limited. What humans want hear to "get into" a song emotionally comes from specific arrangements of pitches - not random ones. The artist chooses them because they "feel" a certain way, an exclusively human concept. When the listener has the same emotional reaction, art succeeds. No DALL-E painting will ever achieve this because it has no feelings to begin with.
We already have legal rulings to this effect with banning an AI as the inventor in a patent https://www.epo.org/law-practice/case-law-appeals/communicat... .
> DALL·E 2 is a research project which we currently do not make available in our API. As part of our effort to develop and deploy AI responsibly, we are studying DALL·E’s limitations and capabilities with a select group of users.
https://openai.com/dall-e-2/
AI now challenges what we've always thought was unique about us, "soul" and crap. Some people who can't do anything else obviously get desperate. And don't get me started on the fact that "art" is not defined or explained nowhere in this essay. Let me define "art". Art is not Malevich's Black Square - that's 50%/50% crap/PR.
Art is ~90% work and ~10% ingenuity. AI cannot reproduce real art because no one will see any work behind it. People go to galleries not to just appreciate ideas, they go their to appreciate work. Real artists are safe. Sleep tight.
What I like about AI and algorithmic art is that it can produce near infinite varieties around a style or technique, and often produces more interesting results than an artist leaning on a style with very little meaning or reason behind it.
If you see a painting of a horse accompanied with a description of why the artist painted it, does it matter if those molecules were arranged that way by human experience?
In either case, you are now being prompted by a simulacrum. And should you then artistically be inspired, your readers are viewing a simulacrum of a simulacrum. Does this compound the "error" or damp it?
What I appreciate is comic books. With the amount of computer tools available today, comic books are way more beautiful than two or three decades ago. Some of the most gorgeous art I've seen is in comic books.
DALL-E can't replicate that and I doubt any AI can (or will be able to) draw exactly what some writer is envisioning in their brain. So artists aren't going anywhere.
I bet it's only a matter of time when it can create an ok-quality full comic book based on a textual script. And perhaps the script could even be generated by GPT-like AI too.
XKCD is doing fine with minimal art, because it’s about the writing. Other writers will make comics very different in style, without having to settle for minimalism. Very long, story-driven comics will be more feasible for amateurs to make in a shorter time. Instead of drawing a few pages a week (which is a lot of work), they could write and illustrate a short chapter.
More labor-intensive comics that are more about gorgeous artwork than story will probably still be around, but a smaller niche than before.
These tools will lead to a boom in webcomics and memes where precision isnt important. The superheroesque traditional comics will be harder to replicate with this.
That information analysis does not necessarily need to be performed by the same person. One is the information analyst, another one one is the painter, and they both in collaboration create some art paintings. I don't see any problem with that. Which one is the artist in that case? The information analyst or the painter, or maybe both? I would argue they are both artists.
I have personally prompted and generated hundreds of pieces of art, real art and awesome art, and anyone who tells me that i don't know what a Picasso is, doesn't know what he is talking about. Picasso was my favorite band for a long time!