AI is going to be our biggest game changer
This discussion has been closed.
Daz 3D is part of 
Connect
DAZ Productions, Inc.
7533 S Center View Ct #4664
West Jordan, UT 84084
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2025 Daz Productions Inc. All Rights Reserved.
Comments
:sigh: Some people WILL be jerks. This is why we can't have nice things.
This technology can be fundamentally jerk all over. If you don't understand or neglect to adapt rules, "it" or the companies behind it will dictate the course of things. That may well end up not being the same as "having nice things".
Or maybe it forces "us" to have a fundamental change of copyright, loosening the straps on any sort of remixing in general. "Reinventing Bach"...
Yeah, the fact that AI is lacking intelligence is quite obvious. But it manages the artificial part nicely.
:snicker: It's got quite the lack of intelligence at times and here's the evidence to prove it: https://www.daz3d.com/gallery/user/6386104575983616#gallery=newest&page=1&image=1251877
Now I ask you, seriously, does THAT look like a cat?!
well, this is probably a picture from right after the witch's cat, while cleaning itself in the way cats usually do it, slipped from the witch's shoulder and fell down on the ground (and therefor out of sight). Had you defined the picture to be bigger... or rather longer... so that the full witch could been shown, the cat would in the picture. Looking suspiciously un-bothered by the fall, which, of course, would never ever can happened, as cats never ever slip or drop randomly from places they sit on.
LOL! There was never supposed to be a witch in the first place. She just popped in out of nowhere, nice as you please. Might have been wearing an invisibility cloak that the cat made off with and is currently hiding under, because I didn't hear her apparate and there's not a broomstick in sight.
Oh, you said "cat"! It thought you said "hat".
So far our impending AI artist overlords aren't worrying me.
The cat was supposed to be wearing the hat. Maybe that confused it. Dunno why it also gave me witches without hats, though. And no cat in those renders either. Unless maybe I has invisible kitteh?
Have you guys never heard of the Cat in the Hat? Shhh knows about that.
Just wanted to mention...
One of the first images I created with my Stable Diffusion free credits developed with what was quite obviously a checkered watermark layered in...meaning the AI pulled a stock photo from a site like Shutterstock as part of it's source material. Another couple developed with artist signatures in a corner. (I've also seen some folks mention the signature thing and they thought it was cute without realizing the implications.) Bad, bad juju I say. That means none of our own artwork is safe as they're likely out there being pulled in as source images as well. Seems like any image on the internet is fair game regardless of copyright. *sigh*
I get it...all artwork is eventually stolen...mine has been stolen more than once...but that doesn't mean I should be ok or complacent with it. AI is potentially another way of seeing your artwork end up being stolen or sampled and there's legit nothing you can do about it.
I've seen watermarks pop up a few times, text from promotional images, and even some images that were pulled from movies/games complete with their game UI. I had a few tavern pictures that added the UI elements from whatever game(s) it pulled the image(s) from. It didn't quite match the prompt, so the results were quite unexpected.
I mean, I knew the images had to come from somewhere, but still found it disconcerting the first time I saw someone's signature.
After reading and thinking about it, I can't find any positive about the tech - even though I like playing with it.
I hope there will be a lot of law suites against the AI companies. It can't be that they steal everything from everyone and then make money with it. Same for the open source stable diffusion models. They are also trained with stolen material.
Just thinking: They might have stolen all images from the DAZ gallery and are now selling your art knowledge as their AI tech. They take your work and not even ask you beforehand.
I feel these companies have lost all morality.
EDIT
:
After evaluation with my local stable diffusion install I now don't think, they stole much from the gallery.
Example prompts:
is always outputting a naked women with big breasts.
"DAZ studio genesis 8 clothed" outputs, who knew, a naked women with big breasts
Sorry, another one - i am sitting here with a fit of laughter at the moment:
This is what I get with "DAZ studio genesis wearing a dress".
(Image removed by mod for mature content)
Try reverse psychology. What do you get when you type in "NVIATWAS"?
The thing about watermarks and logos and whatnot is that ...
Imagine someone with no real self-awareness learning to paint. This person goes into various museums and copies various artists. Picks up styles and techniques particularly to specific artists.
If this person is told 'paint an image of a robot roller skating as if painted by Claude Monet,' they will take what they know and apply it.
But this person has no real self-awareness. This person might add in squiggles in the bottom, because every time they studied a Monet, there was a signature. They lack an understanding of what a signature is, beyond 'this set of swirls that Monet paintings have.'
This applies to watermarks and other things; unless the AI has special processes to recognize what a watermark or signature is and when to avoid using it, these symbols and shapes will appear.
It doesn't mean specific images are being copied, but rather that the AI thought that's the kind of thing you wanted to see for a particular style/artist.
It isn't clear to me that using an image to train the AI is not a breach of copyright, regardless of whether the output with the signature is otherwise a composite/creation drawing on open source content or content with IP owned by the developers
It shows very well what this so called intelligence really does though. It's just a mix and match copying of existing work done by humans. It's fascinating but also kinda questionable when you think about it. And yet, how much copying and duplicating has been going on in art among humans throughout history? Maybe the only difference is now you don't need any art skills anymore. But then at some point nobody will care about art at all.
Continue down this road, full games could be made by AI pretty much destroying the games industry. Full movies could be made destroying Hollywood, songs could be composed at the push of a button destroying the music industry. Who needs DJ's in clubs when you can put an AI on stage with real time generating electronic music and real time controlling of lights and effects into a spectacular show that no human could ever match.
So then we will probably be an enitrely hedonstic society that doesn't need to work or really do anything at all, just shower in infinite auto-generated entertainment. I will then tell my personal AI manager that I would like to experience one more time how life felt in 2022 and I will promptly find myself in a virtual world typing this very post.
Cue Zager and Evans. :-)
We won't see how this plays out until AI generated art gets a certain amount of littigation behind it. I'm sure AI generated art will be a similar kind of game changer as people using digital assets and rendering software, or the same as photography. The printing press was a game changer. And let's not forget online publishing. A game changer indeed. But people will decide what that change looks like.
While alot of the output makes me laugh I have seen some interesting stuff from midjourney. But it would be more interesting to me if they'd invented an AI interface to all the museum and private collections of art and knowledge of technique, etc. An AI that would customize instruction on art history and techniques driven by a person's questions and interests. Sure, for degrees you want specific things taught to set expectations. But an AI that could teach about art to anyone who has access to a computer (or tablet) could make art history and technique more accessible to those with a casual interest. Some of the museum web sites are quite nice, but they're limited to presenting their collections. There are lots of nice books about national musuem collections or works of various artists, but they're quite expensive. I'd like to learn more but bouncing around search results chews through a lot of time. Cue the holodeck rendition of da Vinci and be an apprentice (without the nagging and unuseful criticism I hope; useful criticism would be welcome). I'd settle for storage technology that let me have my own digital copy of the complete and unabridged works in the Library of Congress and the various National museums.
Well, here's the question... are you allowed to look at artwork, study it, and try to learn to make similar stuff?
In my example, I didn't use a specific artist or style insomuch of an actual style...I just used "unreal render" for the style. But even that got me the checkered watermark in one instance, and a few different cases of signatures on the bottom. Subject matter was something to the effect of "a portrait of a cottontail with flowers".
Came across this video not only are we seeing AI artwork and the like, but now they are working on AI generated animation.
Art as old as Monet's lacks any valid copyright. Van Gogh & others too. That's why you see every kind of kitsche product with those artists' artwork printed on the product in some form or another.
Yes, I am. Because it is me - one person. I learn and invest my human being to be able, after years of training, to copy that art and then - maybe - find my own style. If I only make shameless copies, I get sued, if the original artist notice me. If I find my own style, even if it has similarities to existing art (which is inevitably because everything has been done already) everybody will be fine. But with those AI systems no individual effort is required anymore. A company is just mass copying everything. There is no human evolvement anymore, just copy and paste (and writing a prompt with "make it like artist xyz" is not creating art, it is copying art).
It is simple. If I make something and somebody else wants to use it for whatever purpose, he/she has to ask me and in case of pay me. I see a copyright being the patent of art. If I invent a new machine and get a patent, nobody is allowed to copy it as a whole or in parts. Why should that be allowed with art? I will applaude if this art theft gets banned by law - but, until then I will continue to play with it
so, I admit, I am a two face in this case.
By the way, just startet to fiddle with a DS scene again because I need a brake from Stable Diffusion
(this post sounds more serious that it is meant).
Short version first: An artist usually knows when they're copying other people's works 1:1, and they nowadays do have to be somewhat weary of what they do on that account. So for the quick version, making everyone such an artist by "AI", just without the worrying part, basically will remove copyright.
Concerning the "train on them all"-part, a software like stablediffusion may put in large parts of your work into the results, and even if not, i'd ad-hoc compare it to stealing software code and incorporating it into a cloud-based software, violating the license of the original software - "nobody is ever going to notice, right?". Distantly similar, though not really reflecting the "learning from" and the "experience" parts. Yet, does a machine learning system "learn from" and has "experience" in some similar kind of sense which we might assume humans to? They're trained "hard" and then run off that mill (for now), for one view, despite impressive results, whatever they can be made to do. These systems excel at problems with logic rules, and it remains to be proven, if "art" or "creating images from text" has logic enough rules to be beaten by such a system.
Allowing some "AI" to be used for just anything, using literally everything for training, could/would change the fundamentals of copyright. Artists who publish would just participate in their own oblivion, so this would at some point fundamentally change "everything", and probably deserves a discussion, before letting it happen or rolling it out on a global scale. These systems not only could be used literally everywhere, they also could be trained on amounts of data from everywhere, that hardly any "experienced being" will have had access to throughout their entire life. In my opinion, a clean way forward would be to let the artist/rights-holder decide, while not allowing implicit consent via TOS from social networks, for instance, and further make it transparent in the internet via meta-data, and also define what kind of results may be copyrightable or in which exact way, uh and tackle the scraping question somehow, maybe make it non-profit and research only? So, even Scraping a website for specific purposes may be against their TOS and could be illegal, if you just head off straight forward. And again, the big tech companies are sitting on a pile of data, who would have guessed, which may be turned into money. This kind of concentration of power is another question to keep in mind. And it's not too simple to solve in any lenient ways, because allowing "everybody" to scrape all the internet for instance, would not be doable traffic-wise. That's in a way similar to discussions about free speech and democratic principles, concerning social networks.
On yet another side of things, even if only using contents on a base of consent, one or another one of these kinds of systems, perhaps one of the next generations ;), will become so good and interesting, that it wouldn't even matter, if all contemporary artists of this world declined to participate, and it'll become the moblie device camera of those times.
I'd say it remains a pretty heavy topic, and whatever way lawmakers will decide, some people won't be happy with it. Still, de facto not being able to publish, if you don't want your stuff to be used by "AI" would be a pretty drastic consequence for "not playing along as an artist". It would be near-equivalent to removing copyright as we know it, at frist glance. Of course results from "AI" may be put "not copyrightable" and even be "not allowed for commercial exploitation" by law, which would again twist things elsewhere, in terms of watering copyright on the one hand, but allowing some artists to live on.
Edit: For a final plot-twist on the words, machine learning-based systems tend to be more like efficient killer-machines, which is why so many misconceptions about actual strong ai are present, or are associated with machine learning-based systems, because those are labelled "ai" too. A real (still artificial) intelligence, might be so complicated that we can't even copy it, can't halt it without destroying it, or it may become a permanent Kaspar Hauser thing, because we want it to be trained in 3 years, while it can't be, it won't do what we want it to do, it may just grow up like humans do, or not grow up at all, like humans do too, or overtake us all on a single day with, say, broadband internet access. Restraining such a system on copyright, would probably not make sense anymore, as this will strongly depend on the capabilities that such a thing/being in existence will have then. I think the question of copyright also basically is due to the capabilities and the easy-ish scaling of current machine learning-based systems, given the system's inability to judge questions of copyright, concerning the output.
Another thought:
A company hires an Artist to make an advertizement image. They pay him to get the exclusive rights. From now on only the company is allowed to use the image. AI comes along copies the image from the website. The next day another artist writes a prompt "make it like companie's x advertizement". He gets a new image which is similar to the one of company x. And now this artist sells it to somebody else. What is with the exclusive right of company x now?
If this is considered to be "fair use" then I don't see a reason anymore why a fake handbag should be a problem. Hey, I made pictures of the handbag, recreated it, changed it a bit and now sell it as my own creation. Sorry GUCCI I am just learning from you and it's fair use of your IP. And don't say it's a copy - the logo says GUCI not GUCCI, so it's a creative interpretation.
I remember the times on trade fairs where we introduced new machines. They were on display and the engineers were very proud of them. Immediately crowds of businessmen entered the booth and tried to make as many pictures of the inner workings as they could. Of course they were always told "no photos allowed" (of course the real secrets were anyway hidden behind covers). They smiled and left after some friendly small talk. But what if we call that fair use? We can send AI robots to any fair in the world to copy any technology available - no patents anymore, no IPs.
And you story writers out there: No need to think about your story telling art anymore. Go to https://novelai.net/ - they have the AI to make your text like it came from Sir Arthur Conan Doyle etc. No need to ask the IP holders for permission. We are an AI company and can take anything from anyone - it's all fair use. But, of course, if you copy our software code we will sue you.
Well, the "AI" doesn't reach THAT far (yet). Best not expect magic, even if it is "oh so great". Because it'll still be dwelling in the PAST, by the nature of the thing, not a smart intelligence which adapts in real time and learns from everything and at the same time always fulfills the payroll-related requirements (...), not such. They may update based on input by people and imitate stuff, but in the end society will dry out and lose culture and actual abilities, while not gaining any new ones, certainly no compensation, if we marched on with such. Instead there will just be a few more extremely rich people with too much power.
That's not meant to cover all, it's the extreme scenario of relying on the technology we have now, ceasing to have brains, basically. Problem being, the tech is no brain.
Of course this kind of technology may be part of bigger things, even crucial, who knows, and it may bring forth great tools, though "even better" will likely still have similar limitations. I won't tell anyone to not touch it or so, i'd just suggest not to assume magic here. On the other hand, with how publishing used to work, we also had strong irrigation effects, not always and everywhere for the good. The difference would probably be transparency, especially in a world, where information you see and stories you're told become personalized for you.
This all may be terribly wrong. Like "Kaspar Hauser", the absence of communication in the beginning rather is a joke on Alpha Go/Zero, which of course start with logic rules instead of a childhood. Though there is a twist, i just didn't want to write "child soldiers" ;). Because what happens, if you have a working concept for actual intelligence? Fit for a purpose, just much much better? That's highly uncertain, and it could happen, that they'll learn and nurture an actual AI for as long as it takes for it to develop some skills and get interested in learning, and then suddenly pull the chords and change all that it percieves to something that is crafted towards whatever it is intended to be used for. If the technology, even if just in early stages of "growing", or with a smaller than possible "brain", allows copying, they'll go for evolutionary approaches from there, and keep developing on with the ones best for a purpose. "Better rats" may be the name of the game. A little bit of chaos control, inception, deception, and so on. In fact that's probably for the movies. The generic approach would be to learn an actual ai within a system of further subsystems that then consist of data gathering, playing games to perferction and then let them steer drones and the likes. The perception needs an abstraction layer then, but the ai would have a real life with necessities and rewards, pitfalls and set-backs perhaps, though i'd still be sceptical if anyone ever makes it that "far". Perhaps they'll try the inception part to the point, they can influence human brains enough to make "it" work. Ouch.
This happens already though - no AI required. Marketing firms always push the boundaries on what's considered plagiarism vs. derivative works.
Yes it is a problem already and the AI crawlers now take it a step further with fully automatic copies and usage of millions if not billions of images. And in their terms they simply write "if you get sued because you used our service, it's your problem not ours. Some even write: "if we get sued because you used an image from our service, we will put everything that happens on you, we will find you, we will sue you, we will put our attorney costs on you" and so on. That's really what they have in their terms. Not sure if it will work like that in the end, but so far it is what we agree to when using those services. It is a legal and ethic nightmare.
There's also the ones that claim they retain copyright on any generated artwork...all while providing an option to upload your own artwork to "enhance" the AI...and then claim copyright on the output...that used your own artwork as a piece of the formula. Massive side-eye to Wombo Dream.