A.I. : "ART"-ificial Intelligence?

[ThatGuy]

Here's something to discuss while we wait for WWIII...something to harken back to the good ol' days, when the biggest harbinger of doom in O-Land wasn't war, or viruses, or economic collapse, but battles over "proper" art and music! (Ah, the good ol' days...'twas a simpler time...)

Anyway...Here's an article from PJ Media about "art" created by artificial intelligence, and whether or not it is, in fact, art (spoiler alert: the writer of that article says "no".) There are different "engines" out there: Dall-E, Midjourney, and more. Basically, you type in a suggestion for an art work, and the program scours the web for related images, and comes back with what it thinks you want. Some are pretty uncanny, and can even do different styles, such as photo-realism, anime, impressionism, etc.

Now, for the controversy: As mentioned, the writer of this article doesn't think it is, in fact, art, because the A.I. is not aware, not actively creating.
B

Quote

 

ut DALL-E isn’t creative. It just has incredibly broad access to the creative work of human beings, and a rapid ability to kind of smudge it all together into something new.

As “intelligent” as DALL-E is, however, it still has no idea what it’s creating — it doesn’t know right or wrong. It knows only how to smudge.

Sometimes the result is breathtaking. Sometimes the result is too wrong to be salvageable.

But what it isn’t, is art.

 

 

We can compare that to the Objectivist theory of art, which would probably say the same, for similar reasons.

I'll start by throwing out a suggestion/counter-point: But then, it might be argued that it is not NOT art...if "art" is a "selective re-creation of reality according to an artist's metaphysical value-judgements", then the "Art" would come in via the person making the suggestions, and the A.I. is simply the "middle-man" in the creative/technical process. Now, whether or not it's "good" art is another topic...

Check it out, see what you think...I'll post some examples, as well. There's even a link to Dall-E, which will give you 5 free credits to try making your own "art"...If you’d like to give DALL-E a try, you can sign up here.

DALL-E: Inside the Brave New World of AI That Makes Images So Real They’ll Creep You Out

PJMEDIA.COM

---

DALL-E is an artificial intelligence that can create images of literally anything simply by typing in a brief description of what you’d like to...

 

[ThatGuy]

Some examples, and an article at the end talking about the sales of A.I.-generated "art":
 

The best examples of DALL-E 2's strange, beautiful AI art — Quartz

QZ.COM

---

Twitter and Instagram have been awash in sample images from DALL-E 2, the image generation model developed by OpenAI, which creates...

"Otter With a Pearl Earring"



DALL-E 2’s interpretation of “A photo of an astronaut riding a horse.”



Avacodo Chair:



Sunset:

You can now sell your DALL-E 2 artwork, but it feels grim - 71Bait - 71 Bait

71BAIT.COM

---

OpenAI has promised to give one million subscribers on the DALL-E 2 queue access to the amazing AI art generator, and now they can sell...

This sunset was created by DALL-E 2 (Image credit: Reid Hoffman)

"Hoffman took to Twitter (opens in new tab) At the end of July and wrote: 'A picture says more than a thousand words. But with DALL•E it’s the other way around: a single word says more than a thousand pictures.'"

[Michael Stuart Kelly]

11 hours ago, ThatGuy said:

Basically, you type in a suggestion for an art work, and the program scours the web for related images, and comes back with what it thinks you want.

TG,

That's not the way it works.

(btw - I'm glad you brought this up, though.)

This kind of program takes bits and pieces (and style approaches and other things) from an ever-growing database of images (which, granted, comes from scouring the Internet) and combines them into new images. Meanwhile, there is a "self-learning" algorithm embedded in it, too. So it can expand on its own without further human input. Also, I have no doubt these programs will be sharing their databases before long since all this stuff is on the cloud.

Usually, these AI "art from text prompt" programs give you 4 results from your prompt. Then you can keep one or all, or you can have it run one of the options as an image prompt for further images. And you can keep doing that over and over if you wish.

I think the real art here will be in people becoming expert at coming up with text prompts, and then in mashing the images together with other stuff. AI art works well in a post-modern kind of way.

It won't replace painters though. The public prizes authenticity. See an example from literature. Once in a while a fake memoir will turn into a bestseller. Once the public becomes aware that the memoir is fiction, not fact, the book tanks and the author is not taken seriously anymore. I think something similar will happen with AI art. The public will love it in its many uses that are still to come (like the public loves fiction), but if a person gets busted for passing off AI art as his own work, he will be discredited.

 

There is an interesting legal aspect. All AI art, by definition, is in the public domain. That guarantees that it will become widely used. But that will also cause the copyright and trademark people heartburn for a long time to come.  

 

btw - Rand's definition of art is narrow in scope. It applies to only one kind of art, the kind within those parameters. That art is valid, but it's not all-encompassing. In hierarchical terms, art is the main concept. Rand's definition of art is--in reality--a category of that concept even though Rand claims it to be the whole concept. It's like car is the main concept and sedan is a subcategory of car. Rand's definition is better represented metaphorically by the sedan idea than the car idea.

Think of it more like this. In Rand's terms, art is a specific activity to be done in a specific way. But the whole enchilada of art is more an overall process than a specific activity. The higher brain expresses itself in several methods or processes. Language is one process. Art is another. Gestures (body language) is another. And if I think on it, I am sure I can come up with others. That's a long discussion, though...

Michael

[ThatGuy]

7 hours ago, Michael Stuart Kelly said:

That's not the way it works.

(btw - I'm glad you brought this up, though.)

This kind of program takes bits and pieces (and style approaches and other things) from an ever-growing database of images (which, granted, comes from scouring the Internet) and combines them into new images. Meanwhile, there is a "self-learning" algorithm embedded in it, too. So it can expand on its own without further human input. Also, I have no doubt these programs will be sharing their databases before long since all this stuff is on the cloud.

Thanks, Michael, for explaining it better (mine was obviously very over-simplified, as I was a bit a rush.)
 

 

7 hours ago, Michael Stuart Kelly said:

It won't replace painters though. The public prizes authenticity. See an example from literature. Once in a while a fake memoir will turn into a bestseller. Once the public becomes aware that the memoir is fiction, not fact, the book tanks and the author is not taken seriously anymore. I think something similar will happen with AI art. The public will love it in its many uses that are still to come (like the public loves fiction), but if a person gets busted for passing off AI art as his own work, he will be discredited.

 

There is an interesting legal aspect. All AI art, by definition, is in the public domain. That guarantees that it will become widely used. But that will also cause the copyright and trademark people heartburn for a long time to come.  

I did not know about the public domain aspect, that's interesting. I have seen people call for art samples on Twitter, with the warning not to include AI generated art (which is how I learned about this, to begin with). (Mostly for comic book illustrations, which much of that AI art looking suitable for that medium.) I saw what you're talking about first-hand (the part about the demand for authentic art vs. A.I. , and the transparency issues that go with that), but the public domain part adds another aspect to that (people not just wanting something "authentic", but PROFITABLE.) Harder to profit off of something in the public domain, competition-wise, if it's non-exclusive...

 

7 hours ago, Michael Stuart Kelly said:

btw - Rand's definition of art is narrow in scope. It applies to only one kind of art, the kind within those parameters. That art is valid, but it's not all-encompassing. In hierarchical terms, art is the main concept. Rand's definition of art is--in reality--a category of that concept even though Rand claims it to be the whole concept. It's like car is the main concept and sedan is a subcategory of car. Rand's definition is better represented metaphorically by the sedan idea than the car idea.

Yup. I've seen the past debates here between Jonathan and Newberry, and your scope arguments, which is why I brought up the O'ist approach, to compare/contrast against the objection of the author in that article to calling it "art". It's interesting to see how the definition of art, and the accusations of "that's NOT art!" still continue, not just in O-land, but outside as well, whenever a new technology comes along (like it did with photography...)
 

[Michael Stuart Kelly]

1 hour ago, ThatGuy said:

Harder to profit off of something in the public domain, competition-wise, if it's non-exclusive...

TG,

That's a premise worth checking.

Ask Walt Disney.

As for public domain art in books, that is used constantly. There is nothing like free to keep costs down.

I have a lot of courses on how to make money with public domain. There's a guy who made his claim to fame on using public domain stuff for new works, and now he is all over AI text to art--Tony Laidig. Here is his YouTube channel so you can get a good idea of what he does Tony Laidig.

The reason I believe AI art will be used in mashups more than by itself is that you can copyright the part you contribute and you can copyright composite works.

Disney obviously does not hold a copyright on, say, the fairy tale Sleeping Beauty. But it does have a copyright on the movie it made. And it even has copyrights and trademarks on the cartoon characters from Sleeping Beauty.

In other words, you are free to make a new depiction of Sleeping Beauty and the characters in the story, just not Disney's depiction. 

I think AI art in comics would be awesome if you have good storylines and a lot of original art to go along with it.

I even see famous painters painting over parts of an AI painting to add their name to a work (or give their work more value than it has  ). Of course, all that would come with respective intellectual property protection.

 

Here is an article from The Smithsonian showing how a pure AI work was denied copyright protection, even though it was created from prompts describing a near-death experience.

 

U.S. Copyright Office Rules A.I. Art Can't Be Copyrighted

WWW.SMITHSONIANMAG.COM

---

An image generated through artificial intelligence lacked the "human authorship" necessary for protection

 

But here is a lady who registered a copyright for her graphic novel where she openly said it contained AI art in it. That's because a lot of human effort goes into a graphic novel.

Artist receives first known US copyright registration for latent diffusion AI art

ARSTECHNICA.COM

---

Registration of AI-assisted comic comes amid fierce online debate about AI art ethics.

 

In fact, here is an entire site devoted to AI comics.

AI Comic Books

 

 

Back to public domain.

You might be interested in knowing there are tons of comics and even comic book characters in the public domain. Laidig covers them in his courses on PD. 

Here's a taste, a site that lists public domain comic book characters in alphabetical order. This list includes super heroes and villains.

Public Domain Characters Characters - Comic Vine

COMICVINE.GAMESPOT.COM

---

Today there are a wealth of Golden and Silver Age comic book characters who have fallen into Public Domain. This is a list of such characters.

 

If you can come up with a great story, I see no problem in using any of these things. And mixing AI characters, AI art, and original stuff.

Just think of the monsters alone.

 

 

Michael

[ThatGuy]

57 minutes ago, Michael Stuart Kelly said:

TG,

That's a premise worth checking.

Ask Walt Disney.

As for public domain art in books, that is used constantly. There is nothing like free to keep costs down.

It's all ripe for the checkin'. Check away!

 

[tmj]

I saw an article last week about an AI art generator and some such demon artifact , as with most things the article was more click bait than not , but boy was it spooky, lol. Some demon image of horrendous looking old woman had some how infiltrated the data base or ‘net’ space the AI(s ?) pull the ‘stuff’ to produce the images via prompts. Here is another article about it that speaks more to the technical side of how to manipulate prompts , interesting stuff.

https://newsrnd.com/tech/2022-09-21-who-is-the-disturbing-character-"loeb"-that-appears-uninvited-in-ai-images----walla!-technology.rJ3aPxKZo.html

[ThatGuy]

Picture of an A.I. generated macaroni and cheese smoothie...ART!

(in Homer Simpson voice): "mmmm...macaroni smoothie..."