Art Made with Artificial Intelligence Wins at State Fair | Hacker News Reader

Art Made with Artificial Intelligence Wins at State Fair

smithsonianmag.com

52 pointsbrettcvz50 comments

Steven Pinker talks about art in one of his books. I believe it’s in “How the Mind Works” he splits art in to two rough categories:

“Museum Art” and “Motel Art”.

Motel Art might be a picture of a flower or a field.

Museum art is things like Picasso and Rembrandt - what can be in a museum changes over time. E.g, if I paint a Picasso that would be passable in a motel, but not a museum. Picasso has been done. Doing it again is not museum art.

That is, motel art is pretty to look at. Museum art is culturally and historically defined.

A less favorable interpretation is that museum art is an intellectual pissing contest, where curators and visitors can measure their dicks by how much they know about the art history that lead up to this work or how much that work challenges status quo.

Making pretty motel art is a solved problem.

Making museum art, the prettiness or the craftsmanship has for a long time been largely been irrelevant. It might help form the story around the piece, but it is the story around the piece that is important.

That story may be ruined by the revelation that a given piece is produced by AI or it may be the center piece of the story that propels a piece into art history.

We have already seen childrens paintings and monkeys paintings win museum curators favor. We’ve also seen toilets and feces been placed in museums and art history.

I don’t see anything here changing.

Art has no absolute value by itself. Curators and buyers give it value. So if curators will decide that some motel art or some AI art has value - we will see it displayed in a museum. Of course there are exceptions but this covers the vast majority of art pieces.

Everyone gives things value. I don't need a museum curator to tell me I like a particular piece, even if it's motel art. If enough people like some motel art, does that not give it value?

Nothing has absolute value by itself, not even gold. If there weren't people to want gold, it wouldn't be worth something by itself. Value is a people thing.

And money laundering

Some big accusation of nowhere. Do you have a bunch of recent examples?

Money launderers buy art, and contribute to price signaling. Arguably inflating art, whether human or AI generated.

https://www.nytimes.com/2021/06/19/arts/design/money-launder...

It's just the natural progression of anything really. Nobody would argue that you shouldn't drive a car because modern machines were used to build it. This kind of art will be inflationary and that's the end of it - the skill required is getting outsourced to machines. What it means to be an artist will be defined by how you use these new creative tools.

Which museum is the first documented piece of AI art held in?

I don’t know who the first was but the digital artist Refik Anadol has pieces in a current exhibit at MoMA: https://www.moma.org/magazine/articles/658

The model was trained on pieces from MoMA’s collection.

There is a lot of art that doesn’t fit into that categorization: underground, street and outsider art have all existed in limbo before being processed by the museum and the motel.

Are you saying that the artists in that contest made "Motel Art"?

"Motel Art" just took $300 from an artist who is naturally creative and talented and who might have went on to become the next Van Gogh.

What bothers me is the lack of middle ground. Art that is aesthetically compelling but also has meaning.

It feels like most of what I see is meaningless "motel art" or unaesthetic museum art (think of the artists who do things like bolting a urinal to a wall or abstract art that's basically a blank canvas with some lines).

I'm not saying that there isn't a place for avant garde art. I just think that the art community has lost its way in focusing on it almost exclusively.

Art is in how people interact with the work.

Bliss.jpg was originally motel art. It was a stock photo of a random pretty place, and that's what Microsoft bought it as. The success of XP made it popular motel art.

Finally, years later, people managed to track down the original location of the photo and found the rolling meadows replaced by vineyards. That in my mind is the moment when bliss.jpg became 'real' art.

> Motel Art might be a picture of a flower or a field.

> Museum art is things like Picasso and Rembrandt

I dont know Pinker, but I think he might be pulling the readers leg there. Either that or he just doesn’t know the subject he’s commenting on. Flowers by Picasso: https://images.app.goo.gl/U6gZiidQfQpcbFjSA and a field by Rembrandt: https://images.app.goo.gl/qXEpg3urfPokFzQj6

On this debate people are confusing artistic illustration skills with creating art. An artist created this artwork at the state fair, not the AI. The artist had a vision, directed the AI towards that outcome and selected the result they wanted.

Some very famous artist exposed in museums today don’t physically hold the tools that « make » the artwork, they direct a team of assistants who act as their hands. Some like to hold the paintbrush themselves but they don’t have to and that’s why it’s not a criteria for whether they are the artist or not.

Much like a movie director directing a team to create a movie doesn’t hold the mic boom or deal with the photography themselves. They orchestrate other humans into creating the piece of art they have envisioned.

This is exactly the same. Instead of a known visual artist directing art students to fabricate the artwork they’ve envisioned, the artist directs the AI.

I guess by this logic, Elon Musk really is the real life Iron Man.

I’m talking about concrete examples of well-known artists, let’s not confuse this with the delusions of sociopathic CEOs taking credit for other people’s work and constructing a fictional vision a posteriori for the media. There’s nothing artistic there.

Getting good AI art requires finding the right prompt and sorting through potentially hundreds of pictures. Picking out the one that embodies your idea (or sometimes the one that stands out among the rest that you would never have expected from that input prompt) is the way you turn AI output into art.

> The artist had a vision, directed the AI towards that outcome and selected the result they wanted.

And a monkey randomly pressing keys on a typewriter will eventually type the entire Bible

The point is the barrier to entry and the skills required dropped, it dropped so far that you now need no skills besides writing a sentence and clicking on an image. We've became the proverbial typewriter monkey

See also "generative art" which has been around for a few decades. There's still a human curating + tweaking the end result.

https://en.wikipedia.org/wiki/Generative_art

Who would have ever imagined that one of the first groups to be unemployed by AI would be artists?

I can imagine a new chain of mall stores where you come in and a counselor creates a series of prototypes after chatting with you. Pick the one you want and they will 'paint' and frame it while you wait.

With a looser definition of Artificial "Intelligence": Calculators used be be a human profession before being replaced by Computerized Calulators

With All sorts of Artificial "Intelligence" rapidly becoming commonplace, it's easy to discount past strides in "Intelligence" - Even though they are so pathetic in comparison to the supercomputer in your pocket...

>Who would have ever imagined that one of the first groups to be unemployed by AI would be artists?

>Jason Allen, a video game designer in Pueblo, Colorado, spent roughly 80 hours working on his entry to the Colorado State Fair’s digital arts competition. Judges awarded him first place, which came with a $300 prize.

I don't know, 80 hours sounds like a lot of human labor for something that's described as a replacement for human labor.

But he could probably make the next piece in 80 minutes, if not 80 seconds.

Perhaps, if he completely omits quality control and tweaking. Most of everything ever artist makes is by their own standards garbage or rough work to be discarded.

I had the same idea for a business, but it would just be a photo booth at an arcade.

The vast majority of artists are unemployed.

I also love that your vision of the future of this happens in a mall. I genuinely find it charming.

A great photograph can take just a few minutes to get right, it's still great art.

There used to be a debate whether photography really was an art form, but that was long ago. Pretty much everyone now agrees that photography is a tool, and a great artist can produce great art with it. There is nothing about text-to-image learning models trained on huge corpuses that makes it fundamentally different from that.

Art is constantly evolving. That's what makes contemporary art relevant. The judges, who one may expect to have seen a lot of highs and lows, seem to find the tool used here less controversial than some of the commentators.

> A great photograph can take just a few minutes to get right

As a professionally selling photographer whose work appeared in magazines I can say with great authority that you are wrong.

> There used to be a debate whether photography really was an art form

Yes, and we are just beginning that debate with AI.

In an artists medium there are constraints. For the painter, the paint, for the photographer, the subject. With AI there are no constraints, so it is not art.

In Stable Diffusion constraints are

- the imagination of the prompter, or ability to find good inspiration somewhere

- the concepts learned by the model, or the training set used in creating the model

But you're right, it might not be art. It's something else. It's one-time-use-art because most of the images only get one glance and are forever deleted. It's an imagination-bicycle for the mind. Art does not follow your thoughts like the models do. It's a cognitive enhancement.

> the imagination of the prompter, or ability to find good inspiration from somewhere

Imagination is unconstrained. You can literally imagine anything.

> the skills learned by the model, or the training set used in creating the model

This is not a constraint of the artist, it is a constraint of the AI if is even a constraint.

It's the recognition of perceived merit based on effort. If you believe the human artist did nothing more than you could have done yourself with your current skills, it's natural to devalue the work. Sometimes that effort is intellectual, sometimes it's just brute force, and most often it's reflective of years of training and honing of skills that might be commonplace. At some point you might value the artwork, but not the human artist because you don't think much effort was expended to achieve that work.

In this case the article says:

  "Allen created Théâtre D’opéra Spatial by entering various words and phrases 
  into Midjourney, which then produced more than 900 renderings for him to 
  choose from. He selected his three favorites, then continued adjusting them 
  in Photoshop until he was satisfied. He boosted their resolution using a tool
  called Gigapixel and printed the works on canvas."

The artist claims 80 hours. Is that enough effort? It's certainly not turnkey. That's the question.

Yes. It took time, not effort.

I'm on the fence about calling myself an "AI artist". I think a better term is "muse to an AI"

I couldn't possibly create the images that I get out of Stable Diffusion, so I consider it the real artist. But it can't possibly come up with the ideas of what to draw, so I am the muse.

What's more, I feel that the ideas that I come up with are uniquely my own. By which I mean, it is unlikely for another human to have the same idea.

For me, the point at which it becomes art is when it expresses something. It has to evokes thought in the viewer.

Why not just call yourself a programmer? You tell it what to do, it does the work. I'm not sure there's a difference

Maybe you shouldn't consider Stable Diffusion the "artist" but actually the "work of art". Several artist have been experimenting with 'machines' that can make art for a long time, but it is always the machine that is considered "work of art" not its output.

One such artist is Jean Tinguely. You can go to the Jean Tinguely museum and make a free "painting" using one of his machines, and that picture has no value and isn't very good. The machine itself however is very valuable and is what is everybody (including Tinguely) considers the actual work of art.

The main differences now is that now the machine can produce aesthetically pleasing results.

Sing, O prompt engineer, of the artstation anime waifus

I'm not a muse to a CPU, I'm a code monkey and I don't pretend to have 1% of the skills of the people who actually put in the work to make it possible.

Same with AI "art", you're an input and a filter, at most. Everybody has ideas of what to type in a prompt and everybody has an idea about what looks visually appealing to them

In Tantric philosophy (at least according to Christopher Wallis in Tantra Illuminated), the concept of rasa (usually translated as aesthetics) describes the experience of someone experiencing art. It’s not the modernist idea of “this is art because I say it is art”. Rather, something has rasa because the consciousness experiencing the art, in that moment, remembers profound spiritual truths. It’s more similar to what psychonauts might realize during visionary experiences. And if something doesn’t evoke rasa, it isn’t art.

I think everyone more or less has a connection to some source of inspiration and creativity. But not all of us have the craft or skill to execute that vision. These AIs can help bring that out. I am already seeing indie fiction authors creating sketches of characters they have created. I wouldn’t be surprised when someone hooks up GPT-X driven interactive fictions with generative art to illustrate them. If properly curated, little kids take naturally to this stuff — it’s all Make Believe.

Generative music is probably coming next.

This debate on what is art, and artists who do have the craft feeling threatened by the technology is just a surface debate. These models are only as good as the examples we feed them, and there is a place for people to execute on new art styles and variants.

Rather, this ability for anyone to execute on their personal source of inspiration means we will be surfacing up content deep within our collective consciousness. There will be the sublime, and there will be the horrific. Any and all fantasies, both wholesome and beyond the pale, will emerges. What we do about this as a civilization, is I think, will be a much bigger debate than what is art.

One of my favorite artists Grimshaw [0] used to use a camera in his paintings, at the time it wasn't considered 'real art' [1] much like the use of AI now isn't considered real art.

But time passes and now not only is it considered 'real art' but it's also (subjectively to me) some of the best art ever created.

Although it's hard to envisage I imagine the same is going to happen with AI art, it's hard to see now but in the future it will be considered 'normal'.

[0] https://en.wikipedia.org/wiki/John_Atkinson_Grimshaw

[1] https://www.trinityhousepaintings.com/looking-back-artist-jo...

I would argue you can clearly see this piece has been created by Midjourney. It just has that look to it.

> The winner, Théâtre D’opéra Spatial, “depicts a strange scene that looks like it could be from a space opera, and it looks like a masterfully done painting,” Matthew Gault writes for Vice. *“Classical figures in a Baroque hall stare through a circular viewport into a sun-drenched and radiant landscape.”*

Curious, has someone with access to Midjourney tried using this exact description to generate a new image? I'd be fascinated to see the AI-generated result of a human's description of an AI-generated image.

listed them for sale for $750

Regardless how you feel about AI-generated art, I feel like this piece will be worth a heck of a lot more than that soon, given the controversy / infamy it's generating!

In an artists chosen medium, there are constraints. For the painter, the paint; for the photographer, the subject.

It is the constraints that determine creativity.

With AI there are no constraints, so it is not art.

How are there no constraints with AI? I'm not arguing that this piece is good art, but rather that "AI" is merely a tool, with limitations like any other tool. In fact, considering how aesthetically uniform (centerless, fragmented, "dead") works using Midjourney and similar programs are, these programs strike me as extremely limited, little more than artistic crutches for producing novelties that will almost immediately become dated.

> "AI" is merely a tool, with limitations like any other tool.

What are the limitations?

The "deadness" you are feeling is not from AI's limitations, but from a lack of constraints. You have to think about this for it to sink in. I know it sounds un-intuative, but it is true. AI is nothing but a human given no constraints. The same thing is true in digital photography. Have you ever noticed how it all looks the same? There was a time this was much less true.

> What are the limitations?

Inability to procedurally generate highly stochastic forms or textures (for instance what a haboku painter, or any kindergartner, can get by splattering paint); reliance on a finite bank of source or "training" images; no way for the artist to turn an artistic vision into a meaningful input for an I/O program; no demonstrated ability or proposed algorithm for producing on its own (without human inputs) anything resembling any component of art aside from formulaic patterns.

> AI us nothing but a human given no constraints.

I think you have a key insight that productive constraints lie at the living core of the arts, and I agree that if some hypothetical entity that was a human with no constraints produced imaginative works, Kipling's devil[0] would finally get to make a point. I had a similar objection to the hype when AlphaGo beat Lee Sedol – Lee was playing under the constraints of human cognition, which the "AI" with its brute computing power did not even pretend to approximate, yet still overall held his own. That's where the beautiful or "living" quality of the game emerges.

But no existing or proposed AI is anything like "a human without constraints."

There are productive constraints, and their are counterproductive constraints. The 5-7-5 haiku form is a productive constraint; complete exclusion of the letter "e" from an novel[1] turned out to be an unproductive one. Having to include anything generated by Midjourney in an extremely unproductive constraint.

[0] https://www.kiplingsociety.co.uk/poem/poems_conundrum.htm [1] https://en.wikipedia.org/wiki/Gadsby_(novel)

Paint isn’t the constraint of a painter. Imagination is. The same imagination that constrains AI prompting.

The value a person prescribed to a painting isn’t in how the painter was able to manipulate the paint, it’s in what the painting as a whole means to them.

> Paint isn’t the constraint of a painter. Imagination is.

What can you paint, right now, at this moment, this second. Probably very little

Now what, at this same moment, can you imagine. Anything. You can imagine absolutely anything. Imagination is unlimited. Whether what is imagined is art or not is besides the point.

> The value a person prescribed to a painting isn’t in how the painter was able to manipulate the paint, it’s in what the painting as a whole means to them.

The craft does not matter? Only the content? How do you separate the craft from the content?

Ok. Computer.

Reply on news.ycombinator.com