Consolidated AI Thread: A Discussion For Everything AI

Game Development Writing and Content
83

Write a noir-flavored mystery about an owl who is a lawyer and who lives in a city of birds.

The city of birds was a dark and dangerous place, full of corruption and deceit. But among the feathered inhabitants of the city, there was one owl who was determined to bring justice to the streets – a lawyer by the name of Hootie Hawkins.

Hootie was a solitary figure, spending his days and nights in his small office, pouring over legal documents and cases. He was a brilliant lawyer, with a sharp mind and a keen eye for detail. And he was not afraid to take on the toughest cases, no matter how dangerous they might be.

One day, Hootie received a mysterious letter from a client who claimed to have information about a high-profile murder case. The letter was anonymous, but it promised to reveal the truth about what had really happened on the night of the murder.

Hootie was intrigued. He quickly accepted the case and set out to investigate. But as he delved deeper into the dark and twisted world of the city of birds, he realized that the truth was even more shocking than he could have ever imagined.

The further he went, the more enemies he made. Powerful birds in the city didn’t want the truth to come out, and they were willing to do anything to stop Hootie from uncovering it.

But Hootie was not afraid. He was determined to see justice served, no matter the cost. And in the end, he uncovered the shocking truth about the murder – and brought the guilty parties to justice.

In the end, Hootie was hailed as a hero, and the city of birds was a little bit safer thanks to his bravery and determination. But even as he basked in the glow of victory, Hootie knew that the darkness would always be lurking, waiting to strike again.

(good job, Hootie! And I think this is awesome and terrifying, all at the same time. I’ve supplied quite a few prompts, and the stories are generally a bit bland but pretty well constructed, and I’m guessing with better/more specific prompts, they will be stronger. I’ve been using adjectives like “write an adventurous story with a tragic twist” and I’ve also asked for things like arguments where the AI generates a lot of dialogue, which I think it did a decent job with).

3 Likes
84

Interestingly enough, an MKBHD video actually just got posted about where he talks about this exact subject and he showed some of the more egregious instances where you can actually see artist signatures being obfuscated in the generated images.

3 Likes
85

I love the name Hootie Hawkins! :grin:

Yes, the vanilla stories it spits out are for the most part bland and it’s generally much “tell” and little “show”. It also feels the desperate need to wrap things up; it’s very hard to make it only write, say, a prologue. I’ve also noticed, like many others, that it seems to be dead-set on providing a happy ending. Even if you explicitly state you want the bad guy to win, it will do some monkey-pawing by adding “he won, but he never felt happy!” (that’s an example that actually happened).
But just like you I’ve also noticed the output can massively be improved by giving more detailed prompts like “in the style of” or “make one speak like a Southerner” or whatever, as long as it gets it away from the vanilla.

My recent highlight was this:

Summary

cotton
cotton904×900 102 KB

Afterwards, I told it to turn Cotton-Eye Joe into a typical JPRG boss and it did! With HP / MP, attacks that made sense, it even added different boss phases when I asked it to. :sweat_smile:

I also love this poem it generated. I changed the second to last line, but the rest is as-spat out:

Summary

In the Silence of the Night

In the silence of the night
I ponder on the vast expanse
Of all that is, and all that might
Be, in this universe of chance

I think on all the endless skies
And all the stars that dot the black
I think on all the distant lives
That I will never hope to track

And in that silence, I am small
A mere speck in this grand design
A fleeting thought, a single call
In a world, vast and divine

And yet, within this world so vast
I find my place, my time, my all
For though I am a single cast
In this great play, I still stand tall

For in this life, this fleeting dream
We all have value, all have worth
How meaningless it all may seem
We still have purpose here on earth.

Right now I’ve moved on from ChatGPT though and have fallen down the character.ai rabbit hole, which comes MUCH closer to interactive fiction as we understand it here. That one… Honestly, that one is right here, right now true competition. You can just build your own RO if you want to and teach the AI how they should act and it will adapt on the fly. This even works for the world around (yes, there can be one, it’s not just you and a chatbot). For example, you can say “Look over there!” points at entrance to a dark cave and then there is a cave! It’s like lucid dreaming in a machine!

2 Likes
86

I will check that one out next! Thanks! There’s no point sticking my head in the sand like an ostrich. We all need to understand what’s available out there. And since a huge majority of players seem to prioritize interacting with ROs, then yeah, tools like character.ai definitely seem like something that should be on our collective radar. Would many of them be content to just ‘interact’ with an AI RO without there being a larger narrative surrounding the two of you? I’m guessing yes.

Edited to add: Okay I’m playing in a super power school and I just attacked the Xmen and Jean Grey slapped some sense into me. Then Professor X pulled me into the hallway and I attacked him. He subdued me and made me sign a paper saying I wouldn’t attack anyone again. Guess what I did? :smiley:

I will laugh at this because shit, there is no way I can compete with this once the content gets stronger. It’s like a parser game that can respond to pretty much any prompt or action you take.

3 Likes
87

I don’t think it would replace the original, canonical work (but maybe I’m too optimistic here), but it could certainly replace fanfiction, because you can make it even more specifically tailored to your needs, and it scratches the same itch. It is… very scary how easy it is to get a character to act and speak like you’d hope and expect them too. I’ve had Nintendogs that were harder to train. Keeping them in that sweet spot seems to be the bigger challenge.

The thing is, those characters have a box where you can enter hard-facts about them, that will always be there from what I understand, while the rest of their character shapes with the conversation. And nothing keeps you from typing into that box a fact sheet of the RO. From name to personality traits.

88

Oh, this one’s fun.
image

1 Like
89

This seems to be an argument made by a lot of people who are pro-AI art - that the way it puts together a piece is the same as how humans are inspired, that humans too can’t make art without subconsciously putting together techniques and imagery they’ve copied from others. But I’d say it isn’t quite true; if it was, humans would never have developed art in the first place. When new art materials come on the scene, there tends to be people messing around and experimenting with them, creating new techniques and sometimes whole new art styles. There’s a certain amount of ‘happy accidents’ in the process of artmaking that can sometimes lead in unexpected new directions, and that I don’t think AI can reproduce because of the way the models work.

Furthermore, even when humans are taking direct inspiration from another artist it tends to be more of a ‘yes, and…’ process where they try to build on the original. Someone who learnt to exactly copy another’s style, colouring process, subject matter, and used that to make their own new compositions… I’m not sure their stuff would be theft, but it would be derivative crap.

I mean it is coherent which is impressive for the AI, but it also sounds like it just amalgamated every generic noir blurb and replaced ‘people’ with ‘birds’! Didn’t even do anything interesting with the bird angle… still, I suppose the point at which it can do that is when the AI overlords take over… :sweat_smile:

10 Likes
90

I’ve seen worse stories written by humans! I had it write a “wholesome christmas story” about a changeling and doppelganger couple, and it went (without me telling it to) full in on the shapeshifter gimmick-angle and had the couple spread the joy of christmas by visiting a small village and shapeshifting into all sorts of christmas related figures and creatures, including Santa, reindeers, a snowman and the angel atop the big christmas tree.

91

I think Ai algorithms are unintentionally capable of artistic experimentation, and I think some semblance of that can be seen in the form of what we would consider visual glitches. But I think the only difference as of now is that the ‘happy accidents’ we see in them are aesthetically unappealing, because they lack the conscious thought and subjective visions that human artists put into their work.

However, in the future, there will likely be art pieces generated from sample sizes so large that we wouldn’t view them as derivative if we believed they were created by human hands. These would be far from exact copies or simple imitations of someone’s art style.

If we developed an Ai capable of emulating human imagination an artistic spontaneity, would it still be wrong to feed them models? If you saw a beautiful digital scene, unlike any you’d seen before, produced by a glitch or a newer advanced Ai algorithm, would you still view the art poorly because it was produced from a collection of models? I’m using that question not to say that’s where we’re at, but depending on how the technology evolves, that’s where we could be. That’s where the confusion comes in for me.

Even for humans, there has to be some sort of base before we can conceptualize new, coherent scenes unless we’re talking about purely abstract artworks. For example, I can’t describe to you what a color if I’ve never seen anything in my life. I’d have no concept of color. In the same way, an Ai can’t create a good picture on its own because it has no concept of what is aesthetically “good” beyond what humans perceive as art. Computers are almost a true blank slate whereas we humans have a consciousness as a baseline that allows us to form experiences and have an imagination.

Is the only way ai art would be ethical was if it could replicate the human experience? In the sort of future where a machine’s process for creating art is so similar that its bordering on human creativity and/or conciousness, I personally wouldn’t be against it.

1 Like
92

I have been using AI Dungeon for years, by now, for personal entertainment.
Through that, I have played around with AI image generation, from a few different engines (artflow, stable diffusion, etc)
I’ve never used a seed image, just text, to generate images, and in my experience it’s interesting and fun, but also frustrating and not very useful. The results are weird and rather off, especially when you try to get something more complex than just a simple face from artflow. And even those are kinda… soulless?
It might just be because I am an amateur artist, and a very visual person, but the images just never turn out right, no matter how detailed I get in describing them. It’s fun to see how the AI ‘interprets’ my words, and what to change to make the result more what I visualized. But the result is never something that feels like a completed professional image, in any way. Perhaps that’s because I don’t use a seed, and don’t tell it to imitate a specific artist, but doing so wouldn’t feel right to me.

But, as I said, I use AI dungeon a lot.
Why?
Because it helps make up for the executive dysfunctions that makes writing so difficult for me. From my point of view, I am literally using it as a disability aid, an accessibility tool.
I haven’t tried any other text generation tools, so I can’t speak to how they work, but using AI dungeon is more like collaborating or roleplaying with an AI, than just asking it to write a full story for you. I am halfway writer and halfway director, using my own written inputs to guide the AI to make useable outputs (and then re-trying 5 times, before giving up and just editing the output I got to something useful, most of the time).
It helps me when I get stuck on a specific line or specific wording.
It helps me when I get stuck on how to get from one plot point to the next.
It helps me by giving new input to spark my imagination.
It helps get that first oh-so-difficult draft out of my hed and down in words.
And this is just when I write self-indulgent stories for my own entertainment, that my girlfriend is the only other person to have ever read any of. If I was going to actually turn any of it into actual short-stories, novels, or games, I would be doing multiple passes of editing to it, like I would with any first drafts.
(I am very worried though, that AI dungeon will move on to become something I can no longer justify supporting, and then I am not sure what I will do, since my executive dysfunctions will still be just as big a problem.)
My point is, it’s a tool that makes it easier for me to write, but I am still very much the one shaping the story, controlling the little details, and making the creative decisions.

And that is what i think AI should be in visual art as well.
A tool to help artists with the parts they find difficult or tedious.
Adding a pattern to a shape, making a basic colour scheme for you to continue working from, adding basic highlights, adding visual effects, creating backgrounds that you would just be copying from references anyway, generating thumbnails in the planning phase, etc.
There are so many ways it can be helpful in making the process more efficient, less tiresome, more fun for the artists themselves.

I feel like most of the private people who choose to use AI images for their profile pics and character portraits, are people who wouldn’t have commissioned an artist anyway, but would have other kinds of free art on the internet to use instead.
But of course I don’t know how biased or real that opinion is.
Where I see a giant problem is where corporations start using AI pictures, instead of actually hiring artists to do work for them. Because we all know quality is much less important than costs, in late stage capitalism.

4 Likes
93

I was thinking about making a post about that, and I think I may still could. Because I think we need to talk about this. I will try not to go too out of subject but this isn’t completely possible if you want to understand my opinion. First I want to say that I don’t know how to draw or code and sure as heck wouldn’t call myself more then a casual writer at best. But I would love to be all three, an artist, a writer and a coder. So I try to keep up with these community whenever I can. Especially the first two. But since I only have some experience with writing and reading I won’t do the drawing one.

I love innovation and used artbreeder and AI dungeon multiple time and absolutely loved them. Thus when I learned about AI Art through Youtube I was excited and couldn’t wait to learn more about it. And so when I learned how deceptive and unethical AI art business were, I was genuinely crushed that a technology that could have done so much good was turned into this.

Here is a really bad explanation and I may get some things wrong.

As it stand now the most popular AI drawing generators were made by stealing thousands and thousands of drawings (many copyrighted), and a lot of other images that have no business being used such as personal photo or such, without the consent of the people or artists or even compensating any of them.
Now here’s the thing, when I say this I mean the dataset, the thing that was used to train the AI to produce those images. Many AI programs use the same dataset or make their own. And the reasons they cannot be sued for this is that they make themselve pass for a non-profit organization and use the privilige it give them to avoid the legal consequences using copyrighted or private images would normally have. Then they use that dataset made by the non-profit organization (organization that they created or backed) for commercial purpose and can argue that since their program can make unique images it doesn’t fall under copyright laws or anything. Pushing under the rug how the only way the AI is able to make those is that they violated copyright law on a large scale.

So lots of artists images were use without any warning, consent or payment, to initially train the AI under the non-profit disguise. And there is even some community or people that continue to create dataset doing what those companies did. Stealing images without consent and payment to train an AI to replicate a specific artist style so they can later sell the images made by it at a lower price or don’t pay the artist. Search and you will find many examples of artists distraught over how their works was used without their permission and the effects it had on them. I read about a lot of them losing half of their income after a lot of clients decided to just use an AI trained with their stolen art to make something rather then pay them.

Don’t you find it ironic how because of it very nature AI art program are fundamentally dependent on artists work but is used to replace them ? Because that’s those companies aim. And I don’t say that lightly. If you look for the AI trained to make music you will find an egregious double standard. All those programs are solely made with opensource works. All of them.
Which proved to me that those companies KNOW what they are doing. They know that they can steal whatever drawing they want because many artists won’t have the mean to sue them, and that the visual art domain is not only less prone to legal action but also less respected by the large public. While if they did with music the company labels would destroy them in court and people would be outraged that they dared use a singer voice to make that.

An artist drawing style is their voice. It’s the reasons people love their works, pay them to make art and that they can do this. If an AI can do the same thing for cheaper or free, why pay an artist ? That’s what most people will think. That’s what businesses will think as well. Few of them caring how the AI is only capable of doing that because of say artist. Or how AI companies stole them, exploited them, built their fortune on their back and did everything they could to devalue their work. Marketing themselves to artists as “benefactors” only interested in making art available for the wider public when called out about their practices. While selling their products to everyone else as a cheap way to replace visual artists.

The worse in all of this is that they could have been ethical, they could have been human, and everyone could have benefitted from this amazing technology. They could have use only opensources images, they could have even pay consenting artists to make art to feed the AI. And you know what ? Many artists would have send their work willingly for free if they did that. I don’t need to teach you that artists are innovative and creative by nature. They love novelty, and I do too. I love worldbuilding, I love writing, I also want to create a game one day, and it’s been a long time dream of mine to be able to make the visual I have in my head a reality. I was so happy when I found AI art for the first time. Finally I could do that, I could have spectacular visual for the thing I love and created.

But not like that. Not by exploiting people each time I train an AI by imputing some words into those programs, or make them smarter by the edits I do. Because that’s what it is. The only reasons those programs are free to use is because those companies not only count on people refining their AI for them but also to gradually ease everyone to accept them. They are counting on my ignorance of their practices. My temptation to use them regardless, because I’m too poor to afford an artist, because I don’t know how to draw, because I want a pretty image, on and on. I can find multiple good and valid reasons for why anyone would want to use an AI.

But those companies ? Many of them don’t reveal what dataset they are using or what it was trained with for visual art. And if life (and good ol’ capitalism) taught me one thing is that they are hiding something they know the general public won’t like. Namely the fact they are probably as shady as the companies suspected of data laundering. So until one of them can assure me that they are only using opensources images with clear consent given, I can’t support any of them.

I can’t say things like “I can’t afford an artist and the technology will continue to improve anyway so I’m not doing any harm.” Or “I don’t have a choice” or anything like that. Because it’s not true, I know those would only be justifications so I don’t feel like a bad person. I know that AI art has received backclash already, that people are pushing back because of the problems I mentioned, and that some of the companies in favor of AI art were forced to change things up because of it like devianart. I know I have a choice, I can use free picrew, characters creator, or opensources images or heck simply not use images if it’s not necessary.

And in writing it’s not, it’s not suppose to be unless it’s part of the genre like some light novels that often include some illustrations. It sure isn’t for the majority of choice of games. Any images beside the cover is suppose to be a plus, not a necessity.

As you can see, I have a rather mixed opinion on AI art, on one hand I love the technology and hope it can be used to make great things, but on the other I’m all for destroying it in the state it is right now because it was built on a deeply immoral, greedy and exploitative foundation. And thus nothing good can come out of it, not now. Beautiful things ? Of course, I find many AI generated art stunning, but then I think about the artists, and that’s enough to make anything ugly.

I learned a lot from youtube and the video by Steven Zapata was the most instructive on how these companies worked for real and gave many good points about the urgent need for copyrights laws to be updated so artists can protect themselves.
Here is the google doc provided by the author of the video titled “The End of Art: An Argument Against Image AIs” it also have the sources and some avenue of help for Artists. The author specifically talk to artists and you can feel he is passionate about the subject. Which I will admit I find some of his words choices dramatic but considering the subject and how it affect him as an artists I can understand.
https://docs.google.com/document/u/0/d/128yey0VfYhM9eUdvkvCpk5zvvoIkqXfI4hEPAYeJCHU/mobilebasic

FOR AUTHORS
Ima throw a curveball at yall, lmao

  1. Have you heard of AI story generators?
    Yes, I first hear about ai dungeon but not really the others. May check it out to see if it’s better or pretty much in the same state then AI art.

  2. What do you think about this emerging new factor?
    I’m unsure. On one hand I’m always excited to see such progress but on the others I wonder how it will affect the future of writing. As the field is already hard to get into. I think that those writing generator could definitely do a good job for non creative writing. Like announcements, ads, and all kind of boring business stuff, writing that is really generic by nature and where the content is meant to be more practical. But I don’t know for creative writing, it’s clearly not there yet but as seen by AI art, I won’t do the mistake of underestimating it.

  3. Do you feel threatened by this? Or do you feel unworried?
    I’m always wary, for now I believe creative writing is mostly safe, but we should keep our guard up nonethless. Because I won’t put it above companies to try and steal fanfiction or small authors work to try and train an AI to write. Since editors are here, big authors and published one will certaintly have more protection than artists. But every of us that are writers and not in those category should watch out, because I bet you tech company are looking closely at what AI art companies are doing. And if they see they can do the same thing for other form of arts they will.

Creative writing by it very nature is harder to replicate by an AI. As what make it special is how the authors combine old ideas to make new one, the myriad way they can handle an ideas, their prose, ect. So an AI would have to handle the prose and the story at the same time, which is already difficult, but what’s more is that it also need to stay consistent for it to make any sense. Something I personally know that AI dungeon have great difficulty at, beside the prose being basic. The thing is that story do tend to have structure and very common tropes and such. And as some have shown in comments, that’s what the AI do. It’s able to make somehwhat consistent but very bland story with uninteresting prose. So it making a generic commercial book could be possible. But anything more it will need human help. And that’s why I think that creative writing have a good chance of being an art field that will still need a writer no matter what. At least to go over what the AI did and thoroughly edit, revise and build upon it. As even if the AI consistency and prose improve, it has a myriad of others thing it cannot do yet and I wonder if it will ever be able to do them really. Like knowing the difference between show vs tell, pacing, mood and atmosphere, creating complex characters, showing emotions correctly in the right moment, tackling deep themes and subject, on and on. Creative writing is just a field that is inherently harder because of the complexity of language and how intertwined it is with human experience, culture and emotions.

Not saying visual art isn’t the same. Not at all. It’s just an AI can be more easily trained to draw a proportional body that can be considered beautiful then describe one in details that evoke the same feelings and sensations, make sense in context and with pretty prose. But I don’t doubt it could. Because I didn’t think AI art would be this good so fast. So better not presume anything.

  1. Now returning to the AI generated art. Would you use it? How would you use it? I think a number of us already use Artbreeder, and if you didn’t know; Artbreeder is an artificial intelligence (AI ) portrait and landscape image creation tool.
    I used artbreeder a lot and loved it. After I learned what I did about some of these company I’m unsure about using it anymore before I know better how it work. If the problems I mentioned with AI art where solved I could be tempted to try it again. Especially for landscape as I usually don’t need detailed one like for characters and I just like to set them as background. For characters I would prefer to engage an artist so I can precise exactly what I need and the mood or posture I want. I could see myself using it for less important characters though. Overall I would use it to find inspiration or make sketchs.

  2. Would you commission artists with the intention of having them make a reference image to use for AI generation art?
    Sure as hell no with how things are right now. But even if the ethic problems with AI art are solved I don’t think I would feel comfortable asking an artists to basically let me have other of their drawing for free. Unless for whatever reasons I decided to train an AI but do it ethically, commissioning multiple artists to train my program and give them royalty when their image is used or something like that.

For the readers

  1. Did you guys know this was a thing?
    Yes. I did for a long time.

  2. How often do you use AI art generators?
    I used artbreeder quite a lot before for my profile pics but stopped and won’t use it again until I’m sure it’s not like the others. I had tried others of which I forgot the name for fun before. Let just say I didn’t use any AI art generator after I discovered the companies practice and art theft.

(As for ai dungeon same thing, I lowkey regret taking a subscription as I didn’t have the time to research how they work too and it’s still active. (For now) But usually I used ai dungeon when I craved something hyper specific or seld indulgent that no one but me could make. Basically I often asked the AI to write my fanfiction for me.)

  1. Do you think the Interactive Fiction industry and community will be changed by this? Think Tumblr, Discord, and other social media. Will this be more of a topic going forward as AI-generated art becomes more popular?
    Absolutely. I’m also into the rp community and using drawn faceclaim but not citing the artists is sadly common practice. So maybe some will replace it by AI art instead. There is also profile pic or background image that I could see being replaced by AI art. As for interactive fiction the effects are already there, many games have been using artbreeder and now newer programs to make their ROs art or other important characters.

It is and will become a controversial topic more and more as artists get stolen and rightfully complain, the effects is felt on the digital art market and the new artists generation, and we see more AI art being used by business and average people.

  1. How would you feel if CoG decided to use AI art?
    Like if they did use NFT. Ok, hear me out. I don’t want to cause an argument or problems. But right now AI art is already pretty unethical and the community forming around it have a small but very vocal part that is slowly been worthy of being called AI bros by how condescending, entitled and toxic they are toward artists. So I don’t want CoG to go anywhere near that or I will withdraw my support of the company like I would do if they started using NFT.

  2. Do you think AI generated art is going to be more beneficial or harmful to your reading experience? In more focus, would you be okay with it if it helps create more art in your games, or do you rather want to use your imagination?
    Right now. Harmful.

I love Ros/characters portrait, it’s one of my favorite things to have as it make it easier to remember them for me or can even convince me to romance a RO. It can be a double edge sword if I liked the RO personality but don’t find the portrait to my taste. But I can go over it and decide to imagine the RO however I want.

But with how AI art tend to be made I feel conflicted when I see it. Is it from the company that steal artists ? Can I really support that without screwing visual artists over ? That’s the kind of questions I ask myself. Because despite me wanting to support authors I don’t want to support those AI companies and sadly if authors use them that’s what they are contributing too even if indirectly. So I’m feeling on the fence because if I buy the author CoG with AI art I’m giving those companies influence and encouraging other authors to use them, since the public don’t mind it. And I hate when an art field/artist throw another art field/artist under the bush because it benefit them. As art in general struggle to be valued and is more difficult to monetize, so I firmly believe that all artists no matter their medium must have more solidarity between them then anyone. Because as artists we are overall the field the least protected by the law or society, so we need to stick with each others if we want to be able to defend our rights and continue to do what we love. Thus using AI that steal visual artists arts is hurting all of us. As it’s telling everyone that visual art and by extension writing, music, dance and everything can be exploited and devalued as long as you can evade the law.

9 Likes
94

I’m with you on that, but isn’t the elephant in the room here that two very separate issues always get mentioned in one breath when discussing AI art? First, we have the unethical nature of how it came to be, the theft. And then we have the consequences: Artists losing work because of it. And that could and would still happen with the most ethical AI art generator in the world, no?

A lot of jobs don’t exist anymore due to technological advancement, should artists be treated inherently different here than, say, a coal miner?

95

Coal miners still have jobs in many places, Hayden :wink:

Won’t be able to say the same about digital artists, since it doesn’t matter where you are - if you got access to the internet you’ll have an access to AI.
I guess the root of your argument is that we should just get over some jobs ceasing to exist, and I get that - but this is never a painless process.

1 Like
96

Yes, I’m aware of that, I picked that example, because where I’m from it was THE job men had (Ruhrgebiet, Germany). Everyone was either a miner, married to one, or the child of one, so it was naturally the first thing coming to my mind. I could have chosen another example, but the specific nature of coal mining wasn’t the point. :wink:

This is making the rounds right now on Twitter, and I thought I’d leave that here, because it speaks for itself:

97

See, this is what I mean by people looking at art as an end product. It feels like you’re still largely concerned with how beautiful an image can be — and if people can be so moved by the beauty of an image, no matter who made it, then it should be fine right? If you can create a machine that can replicate the artistic abilities of a living breathing person — assuming the best case scenario that theft is not involved — then it’s just another form of competition, yes?

It doesn’t matter if people will lose opportunities, careers, and livelihoods. So long as ‘art’ can be made then it’s fine — even if it means someone losing out to a machine. We are willing to sacrifice entire professions for ‘beauty.’

This is the reason why I said ‘originality’ (for the want of a word anyway) should not be a factor when considering if AI art is ethical. Because are you really saying that we should push for the development of something that will be used to replace people? Because let’s be real, a lot of the concern is because livelihoods are and will be affected by this. It’s not a hypothetical to say that businesses would be interested in developing this further to cut costs as it’s already happening.

As it is right now, you’re looking at it at its infancy (which imo won’t remain there long given how widespread usage is). I don’t believe that it can’t be regulated — it can, which is why we have copyright and property laws — but we will have to push for it.

We have to side with artists, who have already said they’re being hurt by this, and that’s just the tip of the iceberg of the ethical concerns of AI art. For example:

Are you comfortable with the possibility of someone make nudes of yourself because you decided to use their model for a selfie?

Are you comfortable with the idea that your private medical information is being used to train AI without your consent?

The reason why, as we’ve agreed, is because the way AI has progressed has been unethical. There is no way to proceed with AI art being ethical in the future as the current models were built on stolen images, violations of privacy, and consent.

AI art cannot ‘create’ in the way human can as it is dependent on data provided. Theoretically you can draw a vase with 0 reference to someone’s style but an AI can’t do that. It is dependent on someone else’s property to make something. As Eiwynn has mentioned, they cannot be inspired. They’re merely functioning on data provided.

Even if you were to build an ‘ethical’ model from scratch, there is the question of what datasets would it be trained on and how vis a vis its competition. And as Hayden_Winter has just posted, you’re dealing with customers who again only care about the product they’re receiving.

At the end of the day, if we are going to talk about ethics, the bottomline question is:

Is it hurting someone?

And the answer by far is yes.

5 Likes
98

I understand this sentiment, but I feel like this is the byproduct of revolutionary technological advancement in almost any field. Old professions are lost but new ones are also formed to accompany whatever technology has been created. Naturally, the original workforce will shrink, but this is what happens with automation in every industry. In a system where artists are willfully allowing their works to be used in models, I feel this point about losing professions would only hold some merit in the creative industry because of our personal attachment to expressive works.

If we were talking about an Ai using medical data to have robot perform more effective surgeries than humans, I don’t think most people would mind having less human doctors because there would be a net positive for humanity but perhaps that’s just my belief. I think people would have privacy concerns, and should have the right to opt out if they’d want to, but if we could actually anonymize the process, I would see no problem with it. Of course we’d have to watch out for corporate greed, but this wouldn’t be something new.

And as for the question about nudes, this stuff was, unfortunately, already possible with deepfake technology. But again, I feel like we might just fundamentally disagree on whether the benefits are worth the sacrifices required for Ai creativity. And I think we also disagree whether an Ai can evolve to replicate human creativity.

Maybe it wouldn’t be the same type of creation but this is not technically true. While an Ai can’t consciously draw something, I could program an ai that draws pixels at random and I bet you if I ran it enough times, it would eventually create an image with something that we would say resembles a vase, perhaps in a style unseen before.

But even in this scenario, the argument is based on the idea that humans and Ai are operating from a similar blank slate which is not the case. I’m saying that humans have intrinsic frames of reference that give us the capacity to imagine and create. And unless we lack sight, we are constantly being fed visual information through the things we observe which is then transformed into art by a conscious mind.

I think you’re right when you say I’m looking at art as the end product. I believe the technology has the potential to help more people in the long term than it hurts in the short time. Whether that’s the actual outcome in the future, I still can’t say but what I can say is that, for the most part, I agree the technology is not at the point where it can be used ethically.

3 Likes
99

I think this is a good moment to say: nobody needs good AI art. It doesn’t solve any pressing issue we could all benefit from solving. It exists, right now, as a toy. Should it be used to replace artists it will be used as a solution to the cost of art.
But this isn’t luddites vs some incredible advancement that improves our lives.

9 Likes
100

This is a great counter point that should be easy to comprehend on all sides. Very good way of wording that statement!

Also thats terrifying

3 Likes
101

Unfortunately, the article linked talks about how a woman consented to have her photos taken in order to prepare for a facial surgery as part of documentation. These photos, without her consent, were uploaded into an AI art generator — whether manually, or LAION (the AI art generating company) did it by ‘scraping’ data on the web — and are currently being used as part of their repository.

LAION however has ‘refused’ to take down her photos — saying that if the woman wanted them taken down she would have to find where it’s hosted and ask them to do so — despite them using it for image production

While deep fakes aren’t a new issue, Lansa AI is has taken off with their selfie function in their app (for a price). If they’re monetizing something especially, then you’d think they’d be obligated to protect their users images — even if out of business interest. Unfortunately that’s not what’s happening. And given how popular the app is, like last I checked it was number one in the app store, you can imagine the kind of horror that’s going to happen.

102

I wouldn’t say we need Ai art, but we technically don’t need conventional art either. Art is just one of many mediums that people can express and entertain themselves.

But I do think Ai art would be an advancement that could improvement some people’s lives because it massively lowers the bar of entry who can’t draw. A non-artist can probably imagine scenes just as creative as an actual artist but they lack the technical ability to make that scene a reality. Ai art would just be a tool to remedy that for other creators.

The only problem as of now is that the tech is so poor that some versions are basically stealing other people’s art. But if those issues are remedied, you’d have a tool capable of creating unique art from a prompt of a few words. How would that not be a massive advancement for people who can’t draw?

2 Likes