AI Art Generation: Beneficial or Problematic?

[GloomCookie]

I think AI has the potential to disrupt major industries in ways that I'm sure not many people are thinking about. For example, if they can make deep-fakes of real people... why not make deep-fakes of fake people? Think about the future in which AI actors and actresses are used to produce movies, with everything from their voices to their actions being directed by AI. The models are built using an amalgamation of different people until they produce the perfect archetype for that genre, and then they have the AI build essentially any kind of movie they want. No longer are the movie studios in months or years long negotiations with the Screen Actors Guild, but instead they can produce a movie whenever they want using an actor or actress that never, ever grows old unless they need them to, and can still produce a movie at any age they need the protagonist to be a hundred years from now.

Generations of producers could be involved with a particular 'faux actor' since everything is done via computers. People watch these because unlike a movie requiring some expensive actor, they can devote the special effects necessary for a fraction of the cost. You think I'm kidding, but... what happens if Pixar decides to do this with any of their animated films? Toy Story was 100% computer generated, and it was only because of voice actors that the film wasn't 100% artificial. Now what happens if, a hundred years from now, the copyright on Tom Hanks' voice runs out and suddenly we get the next Toy Story that uses an AI fake voice to bring Woody back to life?

It's not too far off even with the technology of today. It's shocking how far programs like Blendr have come due to people animating countless porn parodies of films, so much so that you could almost watch films made with Blendr and chart what year it came out by the graphical fidelity. Couple it with AI voice, and suddenly you have yourself an exceptionally powerful tool in the hands of the average person.

This will be both good and bad. Imagine the backlash the first time someone like AOC, already on a lot of Republican's shit list for a laundry list of reasons, suddenly having porn parodies of her made (and there are already some out there) and released to the public prior to her next campaign season. Imagine if Sarah Palin, who already had people like Tina Fey impersonating her on TV, having deep fake nudes released by some angry Democrat because how dare she be hot and run as a Republican? These are the things that will absolutely ruin lives.

But there's also incredible potential for good. Imagine some teenager sitting in their bedroom putting together a short animated film using AI tools and voice bots to produce a film that gets them some recognition, and they keep at it. In their twenties, they go to film school, learning the structure of movies and figuring out what works and what doesn't, learning formal AI tools and such in class and applying it to their work, and by the age of 30 has produced his first full length film and released it as a self-published work similar to self-publishing books today. It would be incredible seeing the potential that could come out of this as entirely new genres we can't even imagine come from this.

So yeah, there will definitely be good and bad, and it'll take a lot of people putting their heads together to figure out how we're going to navigate this road going forward, but I do think AI has the potential to do a lot more good than harm, even with the bad actors out there doing horrible things.

[Zaer Darkwail]

Overall, any tool, technology, or discovery can be used good or bad all based on people using it. Bad actors have turned many beneficial things into harmful things over the existence of humanity overall. AI is one such milestone and tool introduced in our era and we are in front seats to watch it be born and evolve in its early stages.

Besides art AI affects majorly labor-driven industries and workplaces; already Amazon has robots doing inventory tasks in storage houses. Soon robots driven by AI can be flipping burgers in fast food places and long run even restaurant chefs are replaced by AI. Ofc these things are not yet possible but the rate and speed how human civilization now advances in tech it could be seen in our generation.

With this comes the huge big question; what works humans can do that AI cannot do? Not that many, and there is lots and lots of people on Earth who need jobs and income to get money to get by. This is the reason why all major AI developers press on governments to develop and already establish universal basic income. This works similarly to how some countries in the EU have basic income for those who are unemployed but not retired yet. But it's working without the caveat that you need be actively search for work; you get the basic income just by registering and cash flows in to ensure you can buy food and pay bills. Basically, you can live without ever working in your entire life. It would also lead education to be free as well (basic one anyways, albeit in Finland example higher education is free as well so long you can buy books/supplies yourself and get a discount for those as a student).

As an alternative to that, it will not be pretty at all what the consequences will be. Starvation, medical needs, and survival mode kicking in for lots of people can turn them very desperate and violent.

[GloomCookie]

Since we're discussing UBI, I thought I'd share this video.

[Zaer Darkwail]

I say sorry in advance for this going a bit deep to a side subject (albeit it relates to AI Art as we are discussing AI in general now), the problem with UBI is who is going to pay for it. The tax route definitely would not work, nor the government printing extra money cause it would cause inflation or even hyperinflation which would be catastrophic to the economy.

But if not by tax, the government needs to pay for it but how? Simply cutting much of expenses or bloating in their own internal cost in running their systems and investing their existing tax income differently. Ofc what they need to cut is their own kind of hydra to deal with, more so some countries aren't as wealthy as some others (and there is ofc local economy where need to take consideration of local prices in goods to determine how much UBI value is there compared to US). Or they run things differently; for example US has schooling and prisons run mostly as businesses run by entrepreneurs (there are some government funded but they are a minority). In Finland example schooling and prisons are paid for and run by the government itself.

US gained a bit over 5 trillion in taxes, so UBI would take half of that with present tax rates. Tax revenues go for government activity (salaries and infrastructure maintenance) and for goods and services for US citizens. Now salary cut could work but likely not be enough. Infrastructure maintenance is public facilities and transport, which need improved a lot to cut any extra expenses from those to gain more revenue to spend to UBI.

So, overall making UBI itself a reality needs major improvements in how governments are run. Regardless, even if UBI cannot made into reality it does not erase the problem which is avoidable but rather an eventuality in the future and it's better it gets addressed now before the crisis is already happening.

[MightyMaiden]

We're going well wide of the original topic here, but UBI has already worked in innumerable test cases in a wide swath of countries including local-to-me tests in Canada where people I know had their lives directly improved by it until parties in power changed and the programs were scrapped. (yep I'm bitter)

One of the major benefits of UBI is minimal management overhead compared to existing systems. Huge numbers of expensive-to-manage welfare and low-income support systems can be eliminated and condensed into the UBI program which while perhaps not a 1 to 1 cost replacement, can certainly offset UBI costs dramatically rather than existing on top as an additional cost to the economy.

Of course, if a country already lacks any form of welfare or economic protection for its most vulnerable, then UBI is equally a non-starter, but wealthier nations need to lead by example.

Of course, this hardly relates to art AI or otherwise since a skilled artist who's able to earn a living income for their work is relying as much on their business acumen as their technical ability. Artists, as with many fields make money by being known, by skilled marketing, customer management, and value adds, etc. The most qualitatively skilled artists aren't necessarily the most successful, the ones who can sell themselves and their products are.

In fact since AI art has been for the time being determined to be uncopyrightable, selling it is highly problematic as there is nothing to sell, given that when selling artwork in the modern age, outside of physical artwork, the rights are really what are being negotiated and currently AI art has no rights available to it to negotiate.

[Zaer Darkwail]

Indeed, a pure AI product has no copyright protection and copyright is where money comes in (and protects your income) in the artistic field. Only if AI creates something and is then edited by an artist or skilled image editor, thus including 'human effort', it then gains copyright protection.

In my vision, AI Art can be useful in a corporation setting in producing raw ideas and rough work sketches faster. Or help multiple artists working on the same project and brush all their work with AI program to 'unify' the look of the work to be nearly similar to each other despite different styles the artists possess (like in animation or comic, as usually, artists of different styles need to have period training themselves to specific style so workflow and quality are same; with AI it can be skipped entirely possibly the training period and jump right into production phase).

When comes to freelancers; I view AI Art generator to be a useful tool that can help create quick works that artists can then edit and redraw to meet their own quality standards for the final product.

When comes to non-artists; I have already seen already trend in tabletop gamers using AI art generators to create their characters or NPC's or images for settings. They do not aim focus to on generating income or money; just have fun with it. As I myself use AI Art generator for fun, I find actual joy when after several prompts and tests I achieve something that looks beautiful, albeit it's a random chance but still over time I have through experimenting grown more competent in getting desired results faster.

It may not be true art nor I would call myself an artist, but it's a skill regardless which can improve and I find enjoyment result of hard work in using the program which can sometimes go in whacky directions. But this does not diminish for joy of seeing real artists working in art programs and it has inspired me to try to spark my own creativity using digital tools to create art, but continue still play around AI Art generators as maybe in a year or two I use digital tools to use my earlier AI experiments to be drawn properly.

I do understand that the theft of art; basically taking an artist's public work to train a digital program and then make a profit from said software is wrong and should addressed for sure (as you cannot take artist's public work online and place into book cover, I view it same deal as AI uses the art as reference to function and provide paid service). But I do not understand (nor accept) otherwise demonizing AI Art generator users and treating them as thrash for using the software.

[MightyMaiden]

None of the generative AI models I use are for profit. None of the training I have done using images is for profit. I do all my AI generation on my own home PC(s) using freely available open-source products. 

Interestingly while we complain that these companies are using our data to generate new products without paying us (a value which would be with fractions of a penny based on any individual work's value to the greater whole), the entire business model of search engines is the same. They ingest our generated data, more or less in the same manner using robots crawling the web, then return their own, monetized product, based on that gathered information we generated. (News companies removing themselves from Google's monetized platforms, for this reason, have almost universally discovered their mistake and come back. In Canada, the ridiculous solution was to put media companies onto a UBI system, but that's a different rant)

Why is it that few complain about this? Is it simply because individuals are already making money off it to have their livelihood stolen from them? I think companies like Ask Jeeves and AltaVista would argue otherwise.

Now I can't speak for LLM models like chat GPT, but if you think you can exactly reproduce a work by any artist whose work has been used as training material, I say, good luck, and more power to you as you have figured out how to reverse engineer not only the PRNG system, but the entire maths involved in creating diffusion models. Sure you can complain that they're mimicking your style, but then so are folks I can hire from fiverr.com to rip off other people's work.

Legislation should be used to stop bad actors, things like deep fake porn and other deliberately malicious works, are abhorrent REGARDLESS of the technology used to implement them. Regulating technology just means your regulations are out of date before they're implemented. The AI process I use to create my character art today is already wildly different from that which I used 6 months ago.

[Oniya]

So, I had to laugh about AI recently.  There's a company called 'Fanatics', that produces licensed merchandise for Major League Baseball (and other sports franchises).  Twice, now, they've released products that should have never made it past a design review.

In 2024, they released a line called 'Team Shadow', which featured the team logo superimposed over a series of fading repeats of the logo.  What's the harm in that?

It didn't work out well for the Oakland A's.  (Although, since they were leaving Oakland at the time, it did drive sales among pissed-off fans.)

You'd think that this would prompt the company to take more care in their filter setups - at least run them past a few human eyes, but no.  This year, they superimposed the logo over the team's name, obscuring the center letters (the 'Overlap' collection).  This has lead to the Philadelpia PhiPies, the Washington NaWals (I think Washington should lobby for a mascot change.  Narwhals are pretty epic), and

The Texas TeTas. 

In a state where at least 75% of the population has some rudimentary Spanish comprehension.  (I'm also partial to the Houston AsHos.)

[SausageDoctor]

Personally, I'm a firm believer that AI should be used to ASSIST, not replace. Writers should be using it to help them with things like character names or fixing minor grammatical errors and the like or asking for ideas, but never to have it do 99% of the work. Visual artists are bit more complicated since you can't nor would want generative AI to "touch up" your drawing. Moreover, AI is absolutely garbage in making straight coherent writing in the first place, it's like trying to have a person in their 80s with severe ADHD and an English degree try to write a story.

People saying that "AI will replace X thing" or "Company replaces employees with AI" are just a bunch of techbros trying to make a quick buck from selling their product (An AI LLM). It will never work out the same way they thought it would and they're just over-hyping it.

[Zaer Darkwail]

Indeed, present AI in writing is toddler stage and not really useful besides rough frameworks. In art, you can use it to get a concept or reference image, which you then use as a base when you draw yourself. Already, some artists who were good before have started tracing over AI art (drawing over them and fixing mistakes). I agree AI should be used to assist in tasks, helping to make them quicker than replacing humans entirely human in the workflow. It would take decades or more before AI gets good enough to replace humans.

But then there is a group that is adamant about not even using AI at all to 'train to replace' the people, even if it happens in a decade or two. I myself view it not as something you can realistically fight; as technology progresses, inventions that replace human labor or work are inevitable, as it always has been since industrialization started to progress rapidly. So society as a whole needs to adapt to it; that there is simply too many people and humans and number of jobs which employ people will be somepoint always be less than the human population.

So a universal income that ensures minimum acceptable living conditions (or provides it without needing to pay for anything) to everyone, despite whether they work or not should be established.

I myself I am still against how AI is trained, in sense that art which is taken online to train AI program, artists are not paid for the commercial license to use their art to train AI. Some AI companies have taken steps to train AI 'inside the house' meaning using their own artists or hired artists to train AI but not all of them. At least law (I think this was in US somepoint) which dictates that purely AI created work is not protected by copyright is one solid step stop profiting from AI entirely but it leads then developing group of workers which specialize on 'touching up' the AI work and thus bypass the law cause once human interacts and fixes AI work it then gets copyright protection.

[Cosmo_ac]

I admit, I have very mixed views on this.


I wonder if one of the reasons this is so...alarming for people is because unlike cases where simple manual labor or mathmatical calculations are done, this focuses more into the realm of creativity, something which for a very long time, people have claimed is more or less the sole demain of humans.  We can create art that resonates with people better, because it comes from people.  But, what happens if in twenty years from now, a machine is able to write the next Harry Potter, make the next Forest Gump, the next Red Dead Redemption 2, or the next Thriller?  What will it say about humanity if robots can create just as well as we can?    


I imagine in a lot of ways this in kind'a how a lot of people have felt over time, as machines, and then computers, replaced their jobs. It is unfortunate, but sadly is the way of progress.  I certainly dislike the AI learning off of other peoples are to replace them, especially for free, but I'm not sure how that can be reigned in.  I say this as somebody who enjoys writing and has at least some aspiration of publishing and making money in the future, and who has to a very limited extent, done so in the past.


That said, I will also say this.  There are many people out there who wish they had the ability to create art.  Some in written form, some in visual form.  I've always wanted to draw, but never put in the time.  That's not to say I didn't try at times, but the results were always poor.  Hell, I can't even cut a pizza evenly.  It takes time and effort to hone a skill, but I also understand that not everybody has the time, or in some cases like those with physical disabilities, the ability to make art.  AI gives people a tool to do this.

One of my favourite games is an RPG maker game.  I would love to make games like that, but I would also love them to have art.  However, I know I can't draw the art, so I would have the choice of either hiring an artist and spending hundreds, if not more, to make my game art, or possibly use an AI program to do it.  From a practical sense, the choice is easy.

Anyway, I'm just rambling here.  I see the benefits to AI, but I also see problems.  I have little doubt it will go forward, as the genie is out of the bottle.  While some countries might make laws, other places like China would probably do their own thing, and eventually it would spread to the US, either from major companies, or other means. 

[Zaer Darkwail]

Indeed, I believe the AI Art/Writing prop generators are genies out of the bottle situation. Artists and writers can fight, go on strike or do all they can to stop or propagandize against the AI products, but at the end of the day, the industry giants will use it, and the average consumer who does not care about artists nor writers; they care only about the end product.

If end product by AI is crappy and bad; that itself will kill the company. But once AI makes reliably good stuff in art and writing, then consumers do not care. That is the doom scenario that will inevitably happen unless some global law bans using AI at all. Which unlikely does, cause big corporations own the politicians, and politicians are ones who address and create new laws to be passed or denied.

Not saying consumers do not have power cause if large enough population if they do cares about the product, if they vote with their wallets and do not buy AI products, eventually corporations then give up (or try to trick folks). But overall, I think most present consumers do not actually care about whether the product is AI done or not if it gets good enough.

One which can slow down the progress if globally there is reinforced law in with generative AI that it must done inside house training, needing hire actual artists and writers to train the AI instead rip everything off the internet. As any product otherwise is considered breaking copyright violations. Have a specific digital stamp/seal that proves AI had not even glimpsed at the internet at all and instead all data it has has been taught by humans rather than by taking from the internet.

Also, there is already a law that says anything done by AI has no copyright protection, at the very least, forces some human interaction with the AI product (do final touches, fixes, and repairs before finalizing it). Albeit there is usual doom in saying 'history repeating itself' in this case I do hope it repeats itself; cause past methods of artistic expression still excist today and they can be profitable still in modern days.

Also, hopefully AI reaches a point where it can work as an assistant to artists instead of being a digital thief as it is now.

[LostInTheMist]

There are also currently pending lawsuits against AI for copyright infringement, since they look online, take what's online (which is most everything) and regurgitate it in combination with other works, or with slight modifications (or both) without asking for permission.

It's hard to predict how an emergent technology that has such broad-reaching implications will work out. For now, I'm going to focus on writing with people who, I assume, are real.

[TheGlyphstone]

The true Turing Test - making an AI that can write smut so convincingly you think it was made by a human.

[Silvered Sutures]

I personally voted "It Depends" and will elaborate on the why. Much like MightyMaiden;  None of the training I have done using images is for profit. I do all my AI generation on my own home PC(s), using freely available open-source products, images, and models. Certain model sources, like CivitAI, give you the option to remove a model from their platform if you can prove that the art is your own, regardless of any copyright. At which point, they ask you if you wish to appropriate the model, monetize it to your benefit, or simply delete it from their platform. 

Profiting from someone else's work without their consent, even if done "indirectly" by using their art without their consent in your model and then profiting from said model, is something I personally frown upon and believe to be art theft which should be criminalized. 

However, as someone who was an avid artist and martial artist in my youth, who destroyed my hands to the point that—while they have now regained their function for the most part—I am no longer capable of drawing a straight line due to the shakes that have put my dominant hand under a permanent vibrate setting... AI has been an absolute godsend, and I cried human tears the day I successfully generated something using StableDiffusion for the first time. 

Now, to be fair; I do not profit from my use of AI, and it's purpose under my use is limited to two things; 
1. Generating face claims because I am personally uncomfortable with the use of a real person's picture without their consent, am unwilling to use my own face for RP or story-writing purposes, and feel likewise uncomfortable using someone else's character without their consent. (This is also why I simply do not RP canon characters). 
2. Making lewd art of the above characters for my personal enjoyment. 

I make 0 money from this, of course. 

So I personally think that anyone who gets in my face specifically over the use of AI art can suck it, but I can completely understand why someone would be frustrated if THEIR art was used without their consent and for public, profit-based, or both, uses. Hence my "It depends".

[Callie Del Noire]

I have a friend who makes a living doing Star Wars art. I get why he despises the AI training methodology. Scooping up buckets of original art to train the AI and claiming free use is utter BS. Particularly since most AI creators charge you for use.

[SINless]

I personally voted "It Depends" and will elaborate on the why. Much like MightyMaiden;  None of the training I have done using images is for profit. I do all my AI generation on my own home PC(s), using freely available open-source products, images, and models. Certain model sources, like CivitAI, give you the option to remove a model from their platform if you can prove that the art is your own, regardless of any copyright. At which point, they ask you if you wish to appropriate the model, monetize it to your benefit, or simply delete it from their platform.

Profiting from someone else's work without their consent, even if done "indirectly" by using their art without their consent in your model and then profiting from said model, is something I personally frown upon and believe to be art theft which should be criminalized.

However, as someone who was an avid artist and martial artist in my youth, who destroyed my hands to the point that—while they have now regained their function for the most part—I am no longer capable of drawing a straight line due to the shakes that have put my dominant hand under a permanent vibrate setting... AI has been an absolute godsend, and I cried human tears the day I successfully generated something using StableDiffusion for the first time.

Now, to be fair; I do not profit from my use of AI, and it's purpose under my use is limited to two things;
1. Generating face claims because I am personally uncomfortable with the use of a real person's picture without their consent, am unwilling to use my own face for RP or story-writing purposes, and feel likewise uncomfortable using someone else's character without their consent. (This is also why I simply do not RP canon characters).
2. Making lewd art of the above characters for my personal enjoyment.

I make 0 money from this, of course.

So I personally think that anyone who gets in my face specifically over the use of AI art can suck it, but I can completely understand why someone would be frustrated if THEIR art was used without their consent and for public, profit-based, or both, uses. Hence my "It depends".

I just want to point out that an AI scraper 'removing' art from their model is like a thief giving you back your empty wallet after having spent your cash and maxed out your credit cards. The machine has used it, and the resulting efforts are in there, even if the original is removed, and that's a big if to begin with.

[Silvered Sutures]

I just want to point out that an AI scraper 'removing' art from their model is like a thief giving you back your empty wallet after having spent your cash and maxed out your credit cards. The machine has used it, and the resulting efforts are in there, even if the original is removed, and that's a big if to begin with.

Agreed, but to be fair to the civitAI platform; it is a repository for people who wish to share their models, hence the need for this reporting system. 

[Zaer Darkwail]

Yeah, this entire AI argument would not have happened if AI developers had opened an open website where they advertise the engine but also ask people to donate an image, which they need to confirm they have copyright for first before the engine takes it and learns from it. The image donation would have had clear, worded terms of use, explaining what it means and how the image will be used (meaning the donor grants full commercial rights to the model donated, for free).

Instead of just opening a web browser and vacuuming all freely shown (but not freely used to train AI) images on the internet, and downloading them into the engine.

But the first ideal scenario would not have had much success cause artists would have known ahead if they trained AI they would get replaced (in corporate setting) and sink their careers (more so doing commercial free models is bonkers, better success if give some money/payment to artists but even then iffy), but then again, if some artists did donate models, then only they have to blame mainly is other artists, than AI developers.

In my case, I still continue to use AI and make 0 profit from it. Generate images for char references, and if starting to draw; get concept art from the generator and then draw your own way based on the image.

[Oniya]

https://gizmodo.com/chicago-paper-summer-reading-list-fake-books-ai-2000604708?utm_source=firefox-newtab-en-us

Ten of the fifteen books don't exist.  A couple of them (especially the Andy Weir book) look interesting.  I didn't check to see if all of the authors were real, but if any of those were also slop, that's an opportunity for some speed-writer to adopt a pseudonym and let the tail wag the dog.