Your thoughts on A.I. art creation

[lapsking]

(Warning for...ooer...saucy language.)

Sums up the situation rather eloquently.

If Artificial Intelligence can steal your job, better find a job that doesn't need artificial intelligence, or just like me stop working altogether Hunting and gathering? Better than whining, hey? Nobody wants to be challenged in his cozy bed, but that's not what life is about. Although I don't deny evil aspects of AI. Even knife can be Gnostic, chopping potatoes for soup or stabbing an innocent child. AI will be with us for the rest of our lives, just like macrowave and washing machine, we have no choice but to suck it up and figure out how to deal with it.

Yes, I realized I'm getting a bit verbose.

[Babar]

While I'm not against generative AI in and of itself, it can't really be denied that the way it has currently been built up and implemented is definitely immoral, and probably illegal as well. Reducing that to "Quit whining about losing your jobs and find better jobs" is about as meaningful as someone stealing someone else's house and then everyone else telling them "Quit whining and find another house".

The dataset that was used to build up generative AI that is used for images was made up of work that was stolen (probably the majority of it, in fact).
The dataset that was used to build up generative AI that is used for text was made up of work that was stolen (probably the majority of it, in fact).

The moral (and legal) way to have done this would have been to use artwork in the public domain. We have thousands of years of art to draw from, and thousands of years of the written word, never mind stuff that people have given their permission for today.
The moral (and legal) way to move forward from here would be to remove access to these current datasets, and reset the AIs trained on them, and make use of tools that can identify any art/text that made use of those AIs. Obviously, this is only going to happen if someone pushes legal action.

As it is right now, I am feeling morally (although I guess not legally) ok with pirating software and media from services that have made use of this current stolen data, and enabling ad-blockers of services that make use of it- if they are blatantly making use of stolen data, I don't quite see how they can justify whining about others doing it to them. Lets hope one of these big corporations get hurt enough to push legal action that will result in the removal of these datasets.

[lapsking]

While I'm not against generative AI in and of itself, it can't really be denied that the way it has currently been built up and implemented is definitely immoral, and probably illegal as well. Reducing that to "Quit whining about losing your jobs and find better jobs" is about as meaningful as someone stealing someone else's house and then everyone else telling them "Quit whining and find another house".

The dataset that was used to build up generative AI that is used for images was made up of work that was stolen (probably the majority of it, in fact).
The dataset that was used to build up generative AI that is used for text was made up of work that was stolen (probably the majority of it, in fact).

The moral (and legal) way to have done this would have been to use artwork in the public domain. We have thousands of years of art to draw from, and thousands of years of the written word, never mind stuff that people have given their permission for today.
The moral (and legal) way to move forward from here would be to remove access to these current datasets, and reset the AIs trained on them, and make use of tools that can identify any art/text that made use of those AIs. Obviously, this is only going to happen if someone pushes legal action.

As it is right now, I am feeling morally (although I guess not legally) ok with pirating software and media from services that have made use of this current stolen data, and enabling ad-blockers of services that make use of it- if they are blatantly making use of stolen data, I don't quite see how they can justify whining about others doing it to them. Lets hope one of these big corporations get hurt enough to push legal action that will result in the removal of these datasets.

Well, I remember the same argument when machines were invented after industrial revolution, though I wasn't born. "They are going to steal our jobs". Though it turned out to be a little more than a fuss. Also the same story with robots. Nobody is stealing nobody's job. It's competition, just because you can do a better job cheaper than me, doesn't mean you are stealing my job. The problem occurs with copyright though, which needs laws and regulations, although I personally have nothing copyrighted, so personally I'm not worried about that either, but I can understand. By the way cryptocurrency is not any dirtier than governmental paper currency money which is backed by nothing but bubble and politics. Good old days, money was backed by gold at least. Paper money is scam. And about AI giving wrong answers, you just need to God forbidden deal with cancer, to see how "real" doctors and big pharma can already be whether ignorant or corrupt anyway.

[Danvzare]

As it is right now, I am feeling morally (although I guess not legally) ok with pirating software and media from services that have made use of this current stolen data

I've heard a lot of moral justifications for pirating. So many in fact, that I'm fairly sure every pirate who does justify it, is just in denial, and doesn't actually believe it's morally right to pirate. I mean as soon as you're try to justify it, that means you're on the defensive.

Personally, I think if you're going to pirate, then just pirate. There's no need to justify it anymore than you need to justify drinking water. Enjoy and remember to use a VPN.

But I also think that the people who make and use AI, are basically just pirates. And not the good kind either. They're the scummy kind who claim ownership of what they've stolen, and charge others for it.
But yeah, piracy and AI is basically in the same ballpark for me.

EDIT:
Also, don't forget that Sterling is a huge hypocrite that didn't pay their editor properly (it take a bit of looking up, but you can find Nick's response to the whole ordeal if you look around). I wouldn't listen to anything that grifter has to say if I was any of you. Although I suppose even a broken clock is right twice a day.

[lapsking]

Slaughtering animals in millions on daily basis and eating their flesh doesn't sound moral either, but here we go, who cares. People suddenly get so sensitive over AI morality while calling vegetarians cultist. I suppose if an "intelligent being" can't beat Aritificial Intelligence, should call it stupid or whatever or how would one cope with his unintelligent behavior?

EDIT: I don't use old-fashion grandpa clocks anymore, my phone is enough, though any intelligent person knows time is an illusion.

[LimpingFish]

Also, don't forget that Sterling is a huge hypocrite that didn't pay their editor properly (it take a bit of looking up, but you can find Nick's response to the whole ordeal if you look around). I wouldn't listen to anything that grifter has to say if I was any of you. Although I suppose even a broken clock is right twice a day.

Well, I'm not going to get into that, only to say that a lot of info I found on the topic (apart from the editors original statement) seems come from center-right to right-wing youtubers; the same kind of people who claim similar ill of Hasan Piker, Lindsay Ellis, or any number of "leftist/liberal" commenters.

I've been watching, listening, and reading JS Sterling for a very long time, and rarely have I found myself with opposing opinions. But nobody is perfect.

And we all have our biases; if someone linked to a Destiny rant, or such, I'd also object.

Regardless, she's dead right in this instance.

But back to AI...

While I'm not against generative AI in and of itself, it can't really be denied that the way it has currently been built up and implemented is definitely immoral, and probably illegal as well.

Quite, but you also have to realize that it's also working exactly as intended. This was always the plan, and there is no going back. You'll find that these corporations are willing to spend billions of dollars, if the end goal is (in their eyes) unlimited profit, to circumvent current or future laws.

...

Also, piracy is fine.

[lapsking]

I think I had to sign up on TikTok or Insta to understand what are these anti-vaccine, anti-AI and flatearth conspiracy trends and a long list of hilly billy names are about, before commenting. Excuse my ignorance!

[Babar]

I think I had to sign up on TikTok or Insta to understand what are these anti-vaccine, anti-AI and flatearth conspiracy trends and a long list of hilly billy names are about, before commenting. Excuse my ignorance!

I've usually just been ignoring the extra bits you add to your conversations, but I am very much confused now, so I might as well just ask...
What does cryptocurrency, paper money, vaccines, flat-earth have to do with anything?

PS: And while we are off-topic, people seem to have gotten very confused when I mentioned piracy, so just to note, I don't pirate, I have no need to, most of the software I use is free and FOSS, the music I listen to are either things I already own, or indie bands that are available and I want to support on bandcamp, the games I play are stuff I buy discounted from steam and gog or free stuff here, and movies and tv shows I watch are on streaming services. I was speaking of piracy from a moral position: i.e. if a company makes use of theft of someone else's property to make their products, it is hypocritical of them to complain of theft when someone pirates their products.

[lapsking]

The crypto currency was mentioned as something like vampire corporation sucking blood thing like AI in the video that was sent. Anti-vaccine were refereing to this paranoia about anything new that pops up and trying to TOTALLY diss it instead of trying to manage it. Flat earth was again refering to this idea that everything is conspiracy and they are constantly trying to make naughty plans. There are already lots of naughty plans. Always have been, what can we do? Did the devil make corporations while God was sleeping? I personally don't use AI much, but once drilled a too big hole in the wall and gave me good tips how to fit the screw in. Can't totally diss it. More relevant than Hasan Piker.

EDIT: vegetarianism was an ugly sarcastic tickle for Jesus Christ Reincarnations and their hypocritical morality preachings.

[Khris]

Here's another great facet of image generating algorithms:

Where Facebook's AI Slop Comes From

I'm also really looking forward to this bubble of diarrhea finally bursting, which can't take much longer. The energy cost is astronomical, meaning these fucking "AI" companies are destroying the planet at a much more rapid pace than ever before and the worst thing about this BS is that there's not even a payday at the end of this. Not even for the psychopathic CEOs.

[Snarky]

This is basically what I predicted a couple of years ago, @Khris:

I had a thought that isn't particularly new or original, but that suddenly struck me forcefully.

We've discussed how GPT-3 and other machine learning systems can produce ever more convincing texts (as well as images and voice). And we know that bots are used extensively by spammers and scammers, including to post spam on these forums, and for example on dating apps, because for most of their schemes, the (cost of the) human effort involved is the limiting factor.

It just occurred to me that it's inevitable that as these ML systems become better and better at impersonating a real person, they will be widely adopted by crooks, who will flood every digital channel and forum with deceitful material meant to scam people out of their money (or for other nefarious, monetizable purposes). There will be so much AI-generated bullshit online that it will drown out any genuine human attempts at communication.

[...]

Capitalism + lightweight identity + AI = end of human connection

I think you're fooling yourself if you believe AI will go away because "the bubble will burst" (any more than the Internet went away when the Web 1.0 bubble burst). It's like plastics: way too useful for way too many things for people to give it up, regardless of the negative consequences. For better and worse, we'll need to figure out how to live with it and how to mitigate its problems.

[Khris]

@Snarky The bubble will burst in the NFT sense. Those are still around, but the public doesn't care.

Sure, text and image generation is here to stay*, but OpenAI isn't heading for a self-aware computer program.
Everybody will realize that the people hyping AI are simply stupid and something else will be all the rage, like 8K monitors or whatever.


*It's (illegal) plagiarism on a massive scale, and the quality is getting worse and worse because it's already taking in its own slop. Companies also will have to start charging for image generation and the like, so MAYBE it will actually go away in the sense that it simply stops being useful.

[cat]

It's like plastics: way too useful for way too many things for people to give it up, regardless of the negative consequences. For better and worse, we'll need to figure out how to live with it and how to mitigate its problems.

I love this analogy.

[Snarky]

Snarky The bubble will burst in the NFT sense. Those are still around, but the public doesn't care.

Sure, text and image generation is here to stay*, but OpenAI isn't heading for a self-aware computer program.
Everybody will realize that the people hyping AI are simply stupid and something else will be all the rage, like 8K monitors or whatever.

The difference between NFTs and generative AI is that generative AI (like plastics or online shopping or social media and other developments that have been associated with massive hype and have massive downsides) is actually extremely useful for lots of things people want to do. I'm sure at some point the "hype" will die down just because it becomes a fact of life, and there will be business models that pass the costs on to customers, but it's not going to become less of a thing. That's just wishful thinking.

Nor do I think you are correct in claiming that generative AI is getting worse. "Training data inbreeding" is a (potential) problem, but so far the systems are still getting better and better (harder and harder to distinguish from reality or from human work). But at least it's a good counterargument to the "intelligence explosion"/"singularity" predictions: it shows one reason why it's unlikely that an AI will be able to build a better AI that will be able to build an even better AI and so on.

I'm not going to get into the copyright/plagiarism/ethics arguments other than to say that I don't see a substantial ethical dimension to the question of where models get their training data; only to what they produce. As Jesus would say: "It is not what goes into the training that defiles an AI model; but what comes out of it, this defiles an AI model."

[LimpingFish]

Look, just let me say this. If artists feel the need to distance their work from Generative AI, going so far as to highlight a lack of AI in their work as a selling point, then Generative AI is not in a good place to influence a massive paradigm shift such as those seen with the examples @Snarky listed. I'm not saying it's impossible, or even unlikely, just not as it currently stands. Corporations will tell us otherwise, because of the unending profits they envisage (visions which are purely responsible for the head-spinning speed of AI deployment), but until someone, someone not currently invested in the con job, shows us a usage beyond "It can write a book/make a movie/compose a symphony for you!...Hell, it can write a card to your daughter on her birthday!"... it won't happen.

That is, until it makes society ask how they ever lived without it, Generative AI will remain something of a niche product; useful to a few, but more of a weird toy to the public at large, with so much negative baggage that only those looking to exploit it will openly admit to using it for artistic purposes.

At least that's how I see it.

...

Ps. Fuck Generative AI.

EDIT: Unless you've been living in a cave with your eyes closed and your fingers in your ears, you'll have no doubt experienced one of the many adverts extolling the inclusion of AI-based tools specific to photo-editing on your cellphone (from Samsung, Google, etc.), that will make everybody in your photos smile, or delete some creepy dude in the background of a photo of your dog skateboarding, and such. But, beyond being antithetical to the truth (imagined or otherwise) of documenting a single moment in time, I find that I kinda don't have that big of a problem with it.

Go figure.

[Snarky]

I don't follow your logic, @LimpingFish. Isn't that like saying (25 years ago) that there's no way Amazon is going to be a big deal given the opposition from other book sellers? The question is what audiences think.

And while generative AI is certainly controversial among artists, it's not as universally condemned as you seem to think. There are a number of established artists who have embraced it and started to use it in their process.

Having just visited an Arts & Crafts museum during Mittens, I think there are some clear parallels. The Arts and Crafts movement was one attempt to push back against commodification, mass production and the loss of handicraft traditions (as well as the proletarianization and impoverishment of the workers engaged in production). It had some successes, but ultimately became something of a niche. (You see a lot of handmade jewelry and knickknacks, but how many people today have handmade cutlery or even a lot of handmade furniture? I venture to claim: mainly people with too much money.) And several of the more successful of its adherents ended up adopting aspects of industrial production.

It would have been a serious mistake to have taken the criticisms from the Arts & Crafts movement (valid as they were) as evidence that industrial mass production was "not in a good place to influence a massive paradigm shift." There wouldn't be such a passionate backlash if the paradigm shift wasn't happening.

[LameNick]

...
I'm not going to get into the copyright/plagiarism/ethics arguments other than to say that I don't see a substantial ethical dimension to the question of where models get their training data; only to what they produce. As Jesus would say: "It is not what goes into the training that defiles an AI model; but what comes out of it, this defiles an AI model."

I'm not sure if, what you're indicating, is that copyright owners (largely creators) shouldn't have a say in whether their work is used in training AIs, but if yes I just wanted to respond to that.
I'm not completely clear on why some people think the owner shouldn't have control over the use of their work in this regard, but I'll try to respond to what seems as some commonly held notions that are misguided in my opinion.
Prohibiting a person from directly learning from your work that is publicly available is absurdly restrictive to their most basic personal liberties in a way, that asking someone not to use your work to create a machine that will churn out stuff derived from it and make money of it, while not alleviating some serious suffering with it, is not.
The damage another person can cause to your professional life with learning from your work and then deriving work from this knowledge is quite restricted in a way, that again a machine that can, churn out tons of stuff based on features characteristic of your expression, is not.
Putting to use the mathematical description of the process of "learning" doesn't automatically provide some divine immunity and moral high-ground beating the ownership of mental property.

It feels quite ridiculous, that one can not post their work online, which is often necessary to reach customers, and have any control to keep this expression of their hard-earned skills and inner being, from getting artificially deconstructed and utilized against their interests.

Clearly there is some blurred line between doing something for the purpose of saving lives with no commercial expectations and for the purpose of deluding someone into believing that they actually created something, or just acquiring loads of assets for little expense and no effort, with the goal of making lots of money. But most of the generative AIs related to topics close to this thread and forum seem to me very much at the latter end.


As far as the future of generative AI in the field of art and entertainment goes, I can only say that I wish (though I'm not very hopeful) there will be at least bunch of people left, who enjoy and appreciate the fact, that there has been a conscious living being that can be related to, who wanted to convey something to another conscious being and who invested themselves into the work that they want to experience.

[Snarky]

The damage another person can cause to your professional life with learning from your work and then deriving work from this knowledge is quite restricted in a way, that again a machine that can, churn out tons of stuff based on features characteristic of your expression, is not.

That's my point: the issue is not the input, but the output. I think it is only (or, at least, mainly) problematic if the output closely resembles "features characteristic of your expression."

Many AI-critics are fond of claiming that they are nothing but plagiarism machines, but I don't agree. They can be used to produce plagiarist work (including outputs that are effectively copies of specific works they were trained on), but not everything they output can meaningfully be called that. In other cases those outputs are a mix of so many separate pieces of training data, recombined in novel ways based on the prompts, that they are clearly novel works. I don't think that using someone's work to train a model that produces such novel works amounts to copyright infringement, or any other infringement on the author/artist's legitimate rights. It is only when the output is plagiarism that the original creators' rights are infringed.

And while writers and artists may claim a generalized harm from the existence of the technology, I don't think they should have the right to withhold their work from such use, any more than movie studios should be able to deny reviewers the right to use clips from the movie in their video reviews, or bands should be able to deny people whose political opinions they don't like the right to play their music, etc. Sharing a creative work with the world does to some degree mean sharing it; it is no longer solely yours.

Of course, this is largely an academic argument, since existing systems do not in any (effective) way stop users from producing plagiarism, deliberately or inadvertently, so creators whose works have been used in training the models have plenty of good reason to object on those grounds.

[Creamy]

That's my point: the issue is not the input, but the output. I think it is only (or, at least, mainly) problematic if the output closely resembles "features characteristic of your expression."

The input can be problematic too if the results are too close to some sources.

Outputs from generative AI would probably fall under "fair use" in most cases but you can't assess that if the source pictures are not listed.

Since they probably use a very big number of pictures for each output, it could be in their interest to list the most significant sources with statistical weight.

[LameNick]

@Snarky
The distinction I tried to make there, is that I simply should have control over what my artwork is used for, unless it infringes on your more basic rights.
To force you not to speak your mind about my artwork(etc. write a review), to force you not to learn yourself from my artwork, to force you not to look at my artwork once its publicly available are some of those instances.
Perhaps using my artwork for testing, purely for research or for some other strictly non-commercial purposes.

Of course public good is another principle that should play a role, meaning if using my work for training should in some obvious way elevate suffering and can't be easily replaced, that should beat my right to stop it, but not if all it does, is on a push of a button, produce something pretty.

I believe the degree* to which my artwork(my labour) contributed to the NN's weights being adjusted, rather than other artwork or photo is the fact that should entitle me to have a say in whether its used that way, even if it didn't produce artwork that would pass under current laws as obvious plagiarism. Its turning my labour into my own competition in a systematic and undeniable manner.
My guess is, the laws now might more or less reflect what is fair, considering the damage a person can make with creating artwork similar to someone else rather than a machine spewing tons of it, combined with the fact that a person creating artwork on their own shouldn't need to struggle riddled with uncertainty just to avoid copyright infringement, as opposed to a AI model where you have clear choice in what you include in a data set.

*And the fact that proportionally to that degree the unapproved use of my work has contributed to me loosing future work, while someone else gets richer of it, its just hard for me not to see that as a sort of insult to an injury.