Chaosium should allow work that uses generative models for illustration in the BRP design challenge

[hix]

The exclusion of such material is based on the idea that AI art "steals" or "uses" work from human illustrators. This is an untenable idea based on a common misconception, but don't take my word for it. The Electronic Frontier Foundation is a group of people that specialize in the intersection of computer science, copyright law, and artistic material. Generative models do not trace. They do not compile images to average them. They do not "use" reference images, they literally learn from them and use that learning to create something new. This is the most exciting advance in artistic technology since the invention of digital composition, and creators in the BRP Design Challenge should be able to use it. Please reconsider your position, as it is based on faulty reasoning and incomplete understanding of the technology involved.

https://www.eff.org/deeplinks/2023/04/how-we-think-about-copyright-and-ai-art-0

https://www.eff.org/deeplinks/2023/04/ai-art-generators-and-online-image-market

https://www.eff.org/document/eff-two-pager-ai

Edited March 30 by hix

[soltakss]

I agree, but one of the issues with AI is that it has been trained on images without the explicit consent of copyright owners. Sure, it is OK to train AI on images which are out of copyright, so the paintings of long-dead artists are fair game, but when an artist is alive and says "I do not consent to my images being used to train AI", then there is a difficulty.

RPG supplements feed a fair amount of income to artists, I know that I have paid artists thousands of dollars, that income might be lost if we used generative AI instead of commissioning from artists. I know that there is a counter-argument of many professions being made redundant by advances in technology, for example we don't see typing pools or rooms of clerks filling in ledgers nowadays. 

Personally, I would like a balance where I can use generative art but prefer to commission and pay artists.

[hix]

@soltakss Fair use does not require the consent of the creator. The articles posted make an incredibly good case that ai learning is Fair Use, which is its current legal status.

[MelronPeredor]

And I for one will simply stop buying Chaosium products if they start allowing this.

[Atgxtg]

1 hour ago, soltakss said:

I agree, but one of the issues with AI is that it has been trained on images without the explicit consent of copyright owners. Sure, it is OK to train AI on images which are out of copyright, so the paintings of long-dead artists are fair game, but when an artist is alive and says "I do not consent to my images being used to train AI", then there is a difficulty.

Uh, why?

When people study art they look at and train on the artwork of various artists who came before them. This is true in all creative fields not just images. No one comes out and tells a kid that they can't learn how to draw from their Hulk comic books. Yes, they can't directly copy/trace another person's work, but they certainly can (and should) look at any art they have available to them.  But every artist learns from the work of those who came before them, and copywrite and consent never stopped someone from studying a piece of art. The same holds true for RPGs and RPG adventures. No one says you can't look at an old D&D or Cthulhu rulebook or adventure and learn from it. Quite the opposite, we all told to look at what has come before and learn from it. I don't think a single person submitting to the new Chasoium contest hasn't looked at and been "trained" on some previous BRP product.

So why is it different with A.I.? 

1 hour ago, soltakss said:

RPG supplements feed a fair amount of income to artists, I know that I have paid artists thousands of dollars, that income might be lost if we used generative AI instead of commissioning from artists. I know that there is a counter-argument of many professions being made redundant by advances in technology, for example we don't see typing pools or rooms of clerks filling in ledgers nowadays. 

Yeah, I don't see anyone bemoaning desktop publishing and print-on-demand services, despite their impact of the printing industry. The whole reason why anyone can publish an RPG these days is because they don't have to go to a processional print shop and pay the various up front costs associated with getting a book printed. No one is complaining how Lulu is jeopardizing the jobs of the printers who print the game, the distributors and delivery drivers who distribute the game, or the people at "friendly local gaming stores" who sell the games. In fact most RPG companies are more than willing to cut out the middle men and sell directly to the consumer. But somehow artits are different.

If "RPG supplements feed a fair amount of income to artists," that nice for them, especially since they seem to have been paid better than most RPG writers, but why should artists income be protected when no one else's income is protected? 

 

1 hour ago, soltakss said:

Personally, I would like a balance where I can use generative art but prefer to commission and pay artists.

Personally I would like to see a free and open market when the people working on a project get to decide for themselves. Just like how we can choose to buy from local farmers or shop at a supermarket. Our choice. It better for us, and it's better for the artists. Nothing will put the artists out of business faster than some rule that forces people to hire them for their art. 

 

Now, that said, it's Chasoium's contest so they can (and should) get to make the rules. So I see "no A.I. art" as a fair restriction here, the same as any other rule for submission. Just like they could say "no superheroes" in a Call of Cthulhu adventure contest.  Their contest, their rules.

 

 

 

 

[g33k]

For purposes of the "Design Challenge" contest, it's too late.  Shifting the goalposts mid-contest will only generate distrust, discontent, and complaints (particularly around such a contentious issue).  Chaosium has already made their overall position on AI very clear, and it's unsurprising that they have the same position on the contest they are sponsoring.

The balance of the topic remains, though, regardless of the "BRP Design Contest" issue:  is AI-generated art "fair use"?

Even if the AI training data scrapes the work of living artists, who have not consented to such use of their work?

Even if they have explicitly asked to be removed?
 

Here, I think, is the crux of EFF's argument:

Quote

Machine Learning is a Fair Use
The process of machine learning for generative AI art is like how humans learn - studying other works – it is just done at a massive scale. Huge swaths of data (images, videos, and other copyrighted works) are analyzed and broken into their factual elements where billions of images for example could be distilled into billions of bytes, sometimes as small as less than one byte of information per image. In many instances the process cannot be reversed because too little information is kept to faithfully recreate a copy of the original work.

The analysis work underlying the creation and use of training sets is likethe process to create search engines. Where the search engine process is fair use,it is very likely that processes formachine learning are too. While theact of analysis may potentiallyimplicate copyright, when that act is a necessary step to enabling a non-infringing use,it regularly qualifies as fair use.

-- https://www.eff.org/document/eff-two-pager-ai

 

While I generally like EFF & their mission, I find several weaknesses in this line of thought.  To begin with, I don't think "done at a massive scale" gets a free pass.  "Massive Scale" is not just a quantitative but a qualitative difference... it's kids' sandcastles vs. the Hoover Dam.  EFF even says as much in the next sentence:  "... where billions of images ... could be distilled ..."  This is something no human could do, qualitatively different.

And this is the crux of my own objection:  the EFF wants to grant a legal right of "fair use" to something that's humanly-impossible.  It wants to grant the machinery (and (until true "general" AI is produced (and granted freedom)) the corporations/nations/etc owning the machinery) a right that is previously unheard-of and even unimaginable... a "right" that no human can possess.  This seems like a problem, to me.

And I'll re-post the NYT article, where an AI created a clearly-copyright-infringing work:
https://www.nytimes.com/interactive/2024/01/25/business/ai-image-generators-openai-microsoft-midjourney-copyright.html

Given that the generative AI model cannot even tell when it IS creating a (c)-infringement, I'm disinclined to say that the method is presumptively fair-use.  Generative AI is brand-new, and exciting, and society is still coming to terms with it.  Pre-emptively enshrining a previously-unknown phenomenon as "fair use" seems extremely premature.
 

Edited March 30 by g33k

[Atgxtg]

On a related issue, what about A.I. generated stories and adventures? RPGs are headed that way, just look at WOTC's plans for D&D. I mean if the community is going to ban A.I. generated art on ethical grounds,  then how can it allow other A.I. generated content? What if that content ends up better than human produced content? Do we use it or run something inferior out of spite?

And worst still, just where do we draw the line? I for one, use random generators at times for ideas, treasure, etc. Should those be allowed or are they too much A.I.? Think of all those random tables for things in the past and how they can be combined together for use by a computer. There are many such generators available for use on-line already, and they are only going to get better and more expansive.

And that's before mentioing computer GMing. 

 

I hate to say it but I think we are at a "Battleship" moment. Despite what people might wish, the airplane has made the old battleship navy obsolete, and I think we are going to have to adapt to the new reality no matter our personal wishes on the matter. 

[hix]

@g33k Fair use is not granted. It is up to a claimant to prove that something is not fair use. These attempts have all failed. This is not a wild west undecided issue. Adobe and Microsoft make generative tools, but I assume Chaosium's conviction is not so strong as to ban the use of these company's products.
https://www.computerworld.com/article/3709691/artists-lose-first-copyright-battle-in-the-fight-against-ai-generated-images.html

The decision

https://copyrightlately.com/pdfviewer/andersen-v-stability-ai-order-on-motion-to-dismiss/?auto_viewer=true#page=&zoom=auto&pagemode=none

Also, your NYT article is behind a paywall, got something that everyone can see?

 

[soltakss]

31 minutes ago, Atgxtg said:

Uh, why?

When people study art they look at and train on the artwork of various artists who came before them. This is true in all creative fields not just images. No one comes out and tells a kid that they can't learn how to draw from their Hulk comic books. Yes, they can't directly copy/trace another person's work, but they certainly can (and should) look at any art they have available to them.  But every artist learns from the work of those who came before them, and copywrite and consent never stopped someone from studying a piece of art. The same holds true for RPGs and RPG adventures. No one says you can't look at an old D&D or Cthulhu rulebook or adventure and learn from it. Quite the opposite, we all told to look at what has come before and learn from it. I don't think a single person submitting to the new Chasoium contest hasn't looked at and been "trained" on some previous BRP product.

As I said ...

 

2 hours ago, soltakss said:

I agree

I can understand why companies have taken a position against generative AI, but I would like to see more generative AI being used.

18 minutes ago, Atgxtg said:

On a related issue, what about A.I. generated stories and adventures? RPGs are headed that way, just look at WOTC's plans for D&D. I mean if the community is going to ban A.I. generated art on ethical grounds,  then how can it allow other A.I. generated content? What if that content ends up better than human produced content? Do we use it or run something inferior out of spite?

I am happy for AI to train on any of my published supplements. It would be great to see what an AI could write for Dorastor, for example.

 

[g33k]

42 minutes ago, hix said:

... Also, your NYT article is behind a paywall, got something that everyone can see?

Yeah, not sure what happened. I had previously read that article; I found it again to post in  the other thread  (I skimmed it to be sure it was what I recalled).

But when I searched again -- to post in this thread -- it was paywalled...??!

Not a NYT subscriber... I had presumed they had a few articles outside their paywall, and/or had a 1-article-per-day or three-per-week or what-have-you.
 
Maybe they have an AI which detects popularity, and monetizes their older articles behind a paywall after too many hits ???   🤣

Google for:
  nyt joker ai
Possibly click for just images.
The generated image shows up alongside the (c) original (at least on my search; I know Google filters their results, so YGMV -- Your Google May Vary).

I'm pretty sure the (c)-violations are self-evident, without the text; it's only "Fair Use" here as a news-item (that such an extremely (c)-violating use is so trivially-easy to achieve).

Edited March 30 by g33k

[hix]

@g33k You just Fumbled your Idea roll. I dropped articles from disinterested parties and legal decisions, you're going to have to do better than "trust me bro, it's self evident, check out this pic of the joker behind a paywall"

[g33k]

47 minutes ago, hix said:

@g33k Fair use is not granted. It is up to a claimant to prove that something is not fair use.

I believe you are mistaken:  the owner of the (c) does need to file a claim, but then:

Quote

 ... infringer has the burden of proving their use was a fair use ...

--  https://www.jdsupra.com/legalnews/artificial-intelligence-and-copyright-6563561/
 

 

1 hour ago, hix said:

... to prove that something is not fair use. These attempts have all failed.  This is not a wild west undecided issue ...

While it's true that "these attempts have all failed," it's not like there's ample case law to cite!
It very much is "wild west" (that is, one without much relevant precedent).

 

58 minutes ago, hix said:

 ...
https://www.computerworld.com/article/3709691/artists-lose-first-copyright-battle-in-the-fight-against-ai-generated-images.html

The decision

https://copyrightlately.com/pdfviewer/andersen-v-stability-ai-order-on-motion-to-dismiss/?auto_viewer=true#page=&zoom=auto&pagemode=none

 

 That is an extremely-limited ruling, and based largely upon technical legal points (with the artists invited to resubmit amended suits).

[g33k]

8 minutes ago, hix said:

@g33k You just Fumbled your Idea roll. I dropped articles from disinterested parties and legal decisions, you're going to have to do better than "trust me bro, it's self evident, check out this pic of the joker behind a paywall"

Were you able to see the two images, side by side?
One Generative-AI, one (c)?

"Worth a thousand words" as they say...

But here, have this:
https://www.giantfreakinrobot.com/ent/ai-joker-copyright-infringement.html
 

Edited March 30 by g33k
typo

[hix]

"... infringer has the burden of proving their use was a fair use ... "

OK, this is a worthwhile distinction. Youtube is notorious for allowing copyright claimants to go wild.

"see the images"

That wouldn't prove anything. We'd need to see the workflow to judge, as well as examine the dataset and any presets it might have been using. It is entirely possible to use img2img processing to change the color of the cover of RQ:RiG. I could then put it out and completely violate copyright. I could also do this with Photoshop. This is a user issue, not a tech issue.

To be fair, the ruling was about output, and Chaosium's argument is about the training. I won't argue that point, as EFF does a much better job in the first article i linked, and they are actually qualified to discuss it on that level.

[Atgxtg]

2 hours ago, soltakss said:

As I said ...

 

I can understand why companies have taken a position against generative AI, but I would like to see more generative AI being used.

I can't understand why companies would take a position against A.I. I can understand why people would.

But I also think this is a case of shuttle the barn door after the horses have left. The technology is here, it will be used. The artists taking a stand now won't matter since A.I. can replace them. It's kinda like broadcast TV taking a stand now. 

2 hours ago, soltakss said:

I am happy for AI to train on any of my published supplements. It would be great to see what an AI could write for Dorastor, for example.

Well, you might just get to see that. 

Considering that 30 years ago we went through something like thing with computer generated graphics, I suspect it won't be all that long before we have A.I. bots writing adventures. 

[Atgxtg]

 

2 hours ago, g33k said:

And this is the crux of my own objection:  the EFF wants to grant a legal right of "fair use" to something that's humanly-impossible.  It wants to grant the machinery (and (until true "general" AI is produced (and granted freedom)) the corporations/nations/etc owning the machinery) a right that is previously unheard-of and even unimaginable... a "right" that no human can possess.  This seems like a problem, to me.

It will be the the corporations/nations/etc owning the machinery that will end up using A.I. In the long run it won't be someone with a PC who will be making A.I. Joker stuff for sale, but the movie studios and comic book companies. THey will do it to replace the various people they currently need to make these things. THat's where the real debate is.

For instance, if Warner/DC fire all their actors, production crew, artist and writer and replace them with A.I for thier next Joker film. what is the legal complaint? Warner owns the character. 

 

 

 

2 hours ago, g33k said:

And I'll re-post the NYT article, where an AI created a clearly-copyright-infringing work:
https://www.nytimes.com/interactive/2024/01/25/business/ai-image-generators-openai-microsoft-midjourney-copyright.html

Given that the generative AI model cannot even tell when it IS creating a (c)-infringement, I'm disinclined to say that the method is presumptively fair-use. 

Creating it isn't an infringement. Distributing it  might be.

That's important. If you, sitting at home draw a picture of the Joker or any other character you can do so. If you try to sell it, you could be in trouble.  Please, don't offset an existing problem with a worse one. 

 

 

[Jeff]

1 hour ago, hix said:

@g33k You just Fumbled your Idea roll. I dropped articles from disinterested parties and legal decisions, you're going to have to do better than "trust me bro, it's self evident, check out this pic of the joker behind a paywall"

You are aware that the judge did not dismiss the copyright infringement claim against Stability AI? To me that is the crux of the matter here. Remember, these programs are NOT people and, at least as far as the USPTO is concerned, cannot create transformative work - they are just "dumb machines". There's no "creative work" in the way that phrase is used in intellectual property law. The end user of the program has cut up the works of others, reassembled them, and is passing it off an unauthorised derivative work. You might dig the result, might not care about the impact on the artist's ability to continue to get paid - but in the end, I cannot see how one gets away from the copyright violation unless you "train the machine" (again I put that in quotations because the very terminology we use can actually be very misleading) on a walled garden of materials that you own the rights to make derivative use.

But until the appellate courts come to some consensus (or the SCOTUS hears a case), this will likely trundle on like Napster until it doesn't.

[Bilharzia]

Quote

Generative AI Has a Visual Plagiarism Problem

Both OpenAI and Midjourney are fully capable of producing materials that appear to infringe on copyright and trademarks. These systems do not inform users when they do so.

Quote

Following up on the New York Times lawsuit, our results suggest that generative AI systems may regularly produce plagiaristic outputs, both written and visual, without transparency or compensation, in ways that put undue burdens on users and content creators. We believe that the potential for litigation may be vast, and that the foundations of the entire enterprise may be built on ethically shaky ground.

https://spectrum.ieee.org/midjourney-copyright

Edited March 30 by Bilharzia

[hix]

@Jeff Thank you for an honest and well considered reply.

" The end user of the program has cut up the works of others, reassembled them, and is passing it off an unauthorised derivative work"

This is exactly the misconception I am addressing. GAI do not do this. They do not combine previously existing work, they learn from it, and the amount learned from any potentially copyrighted work is clearly de minimis. But maybe that doesn't count since it's not a person. Let me get specific about my own workflow on the cover for Smoke and Aces, and I'll leave it to the audience to make up their own mind. While the amount of effort involved does not determine copyright status, this will illustrate the fact that my cover is a transformative work, and that ai tools can be used well.

-Plan out each figure for gen, choose a style (comic book, strong outlines, white background)

-gen each prompt dozens of times looking for something that has some motion and movement in it

-edit each image in photoshop to remove ai slop, ensure consistent style, and make everything make visual sense. These are minor edits like removing an extra finger, a weird squiggle on a face, closing a line to change the color of an eye, and so on.

-adjust the color of each image using photoshop to add bold color and visual appeal, add the branding and the text

-use layout software to collage all these images, cropping and editing along the way

Edited March 30 by hix

[Jeff]

29 minutes ago, hix said:

@Jeff Thank you for an honest and well considered reply.

" The end user of the program has cut up the works of others, reassembled them, and is passing it off an unauthorised derivative work"

This is exactly the misconception I am addressing. GAI do not do this. They do not combine previously existing work, they learn from it, and the amount learned from any potentially copyrighted work is clearly de minimis. But maybe that doesn't count since it's not a person. Let me get specific about my own workflow on the cover for Smoke and Aces, and I'll leave it to the audience to make up their own mind. While the amount of effort involved does not determine copyright status, this will illustrate the fact that my cover is a transformative work, and that ai tools can be used well.

-Plan out each figure for gen, choose a style (comic book, strong outlines, white background)

-gen each prompt dozens of times looking for something that has some motion and movement in it

-edit each image in photoshop to remove ai slop, ensure consistent style, and make everything make visual sense. These are minor edits like removing an extra finger, a weird squiggle on a face, closing a line to change the color of an eye, and so on.

-adjust the color of each image using photoshop to add bold color and visual appeal, add the branding and the text

-use layout software to collage all these images, cropping and editing along the way

Remember, as far as the legal system is concerned the GAI does not perform ANY transformable or copyrightable activity. It cannot. It is not a person and is not an intermediary between the copyright image or text and the user. So from the perspective of copyright law we have loads of copy-written works that end up being cut up and reassembled. You shouldn't take someone else's work and draw over it and expect to survive a copyright challenge - any more than George Harrison should have let "He's So Fine" get stuck in his head as he created "My Sweet Lord". 

[Atgxtg]

29 minutes ago, Jeff said:

. Remember, these programs are NOT people

Neither are corporations, yet they have the rights of people. And even if the A.I. don't have rights, the coprorations that own them or use them do. Nor do the A.I. get sued.

Ultimately this is people vs. other peopl, the A.I. is just a tool. And it's not the average person in the street who owns and operates an A.I. It's not even individuals, it's large corporations. Large corporations whose view of A.I. differernt depending on who'se A.I. it is. Google owns Youtube and restricts A.I. art, but Google also owns thier own A.I. which is capable of generating art. 

29 minutes ago, Jeff said:

and, at least as far as the USPTO is concerned, cannot create transformative work - they are just "dumb machines". There's no "creative work" in the way that phrase is used in intellectual property law. The end user of the program has cut up the works of others, reassembled them, and is passing it off an unauthorised derivative work.

And there are only 36 plots too.  The same argument could be made for things that people create. All stories break down to a handful of infinitely reused concepts. The devil is in the details. Should we go after George Lucas for the stuff he  "borrowed" for Star Wars? The Hidden Kingdom, The Heroe of A Thousand Faces, the story of Lugh Lamfada, the story of King Arthur. 

For art, how many people have drawn portraits? Or bowls of fruit. Nothing transformative there. Which is probably why the camera cut down on the nimder of portait painters. 

 

29 minutes ago, Jeff said:

 

 

You might dig the result, might not care about the impact on the artist's ability to continue to get paid - but in the end, I cannot see how one gets away from the copyright violation unless you "train the machine" (again I put that in quotations because the very terminology we use can actually be very misleading) on a walled garden of materials that you own the rights to make derivative use.

But until the appellate courts come to some consensus (or the SCOTUS hears a case), this will likely trundle on like Napster until it doesn't.

A.I. will ultimately win. Reason being that right or wrong, the money will win out in the end, and large corporations have spent tons on money on A.I. and will use it. Once the geneie is out of the bottle it's too late.

[hix]

1 hour ago, Jeff said:

Remember, as far as the legal system is concerned the GAI does not perform ANY transformable or copyrightable activity. It cannot. It is not a person and is not an intermediary between the copyright image or text and the user.

Now you're arguing against the output, which is a different beast than arguing against the training.

"You shouldn't take someone else's work and draw over it and expect to survive a copyright challenge"
Artists have survived copyright challenges doing exactly this  (pic related). It helps if you're a millionaire artist like Robert Prince and can afford good lawyers. I still don't think you could call my cover, or the images that comprise it, a derivative work, without an original to compare it to and claim derivation.

So we're left with the training issue, Chaosium's original claim. Which is indexing, AI tools grew out of the same programming as google's search engine. We're now not talking about generation, we're talking about how the machine learned and whether that is fair use. I'll refer again to the EFF article, in particular this quote.

Edited March 30 by hix

[AndreJarosch]

I am uncertain where i am standing regarding AI images. 

I can see why some people see it as a tool that learns some ideas about drawing, rather than copy an artists work. 
I can also see why some people see this as a copyright violation. 
I also understand that the AI is merging different styles and images, like a human artist would do when he pays homage to an artist he admires, or pays homage to a specific picture in another style (how many artists have drawn their version of Frank Franzettas Death Dealer?). Where does a homage end, and a copyright lawsuite begin?

I have two concerns about AI: 

1. AI learns from images it finds on the internet. The pictures it finds define the norm for it. AI doesn´t has a clue what the real norm is. (Try creating an "ugly woman", and you will still get a Miss Universe girl, with a little dirt and maybe a scar.)

2. AI creates images, but as a machine it doesn´t has any copyright on the created image. As far as i understand the person requesting the AI to create a specific picture also has no copyright to that picture, because he isn´t the artist. As far as i understand, this means that anybody can use an AI generated picture for any means he pleases; AI images are free to grab and alter (Photoshop, etc.) and do whatever you want with it. 
Why should a company, like Chaosium, let AI create Cthulhuoid or Gloranthan pictures, which anyone else also can use?
 

[el_octogono]

5 hours ago, hix said:

@Jeff Thank you for an honest and well considered reply.

" The end user of the program has cut up the works of others, reassembled them, and is passing it off an unauthorised derivative work"

This is exactly the misconception I am addressing. GAI do not do this. They do not combine previously existing work, they learn from it, and the amount learned from any potentially copyrighted work is clearly de minimis. But maybe that doesn't count since it's not a person. Let me get specific about my own workflow on the cover for Smoke and Aces, and I'll leave it to the audience to make up their own mind. While the amount of effort involved does not determine copyright status, this will illustrate the fact that my cover is a transformative work, and that ai tools can be used well.

And yet proof says otherwise. I tried modjourney for Elric of Melnibone images and I got images with Henry Cavil-like faces and some others were there was a clear source image.. (specifically the cover of a book where Elric is standing between two black/brown borders) probably because there are not many Elric images around so the AI doesn't have enough sources of "inspiration"

[svensson]

I'm a bit of a Luddite when it comes to AI art.

Firstly, it devalues an artist's skill and effort. It takes the humanity out of the experience of art.

Secondly, there is the significant issue of plagiarism. I've said on this board that 'The last truly original piece of writing was probably Gilgamesh. Everything else is plagiarism to one degree or another'. I still believe that to be true, but the threat AI poses to original artists and writers cannot and should not be discounted.

Thirdly, what will happen to actual artists when companies can just run some parameters in a software program and the computer will gin up a very pretty image, but one with no thought, no controversy, no technique, and no individuality?

Where I'm at right now with AI is this, "It's awfully pretty, but is it 'art'?" Or to be much more blunt with it, "AI artwork is the artistic equivalent of an HOA President... everything is bland, controlled, and various shades of beige."