AI-Replacing-Creatives - Lawsuits & Precedent Setting

[Shillitron]

Pending Lawsuits

- Co-Pilot Lawsuit - Scraping Copyrighted Data
[Microsoft, GitHub, and OpenAI vs Matthew Butterick, San Francisco-based Joseph Saveri]
https://www.theverge.com/2022/11/8/2344 ... ining-data

- Stable Diffusion Lawsuit - Copyrighted Artist Art used in Training Data
[StabilityAI vs Sarah Andersen, Kelly McKernan, and Karla Ortiz, San Francisco-based Joseph Saveri ]
https://stablediffusionlitigation.com/

- Stable Diffusion Lawsuit - Copyrighted Getty Stock Art used in Training Data
[StabilityAI vs Getty Images]
https://www.theverge.com/2023/1/17/2355 ... es-lawsuit

- Stable Diffusion Lawsuit - Getty Images 'Brand Degradation' (Malformed Logo Showing up in AI Generated Images)
[StabilityAI vs Getty Images]

--

Some quick Editor Notes from this prestigious journalist:

- San Francisco Law Firm has their fingers in a lot of pots. AI Ambulance chasing.
- Sarah Andersen, Kelly McKernan, and Karla Ortiz are all very well known VOCAL far left "Artists"
- Getty Images likewise is very upset at the prospect of AI offering a cheap capability to generate whatever you need on the fly.. I wonder why..
- The published blog (https://stablediffusionlitigation.com/) that Joseph Saveri law firm posted has been widely criticized as junk-research and not accurate representing how diffusion models are trained / used.
- There was recently a paper published that accused Stable Diffusion of art theft when extremely overtrained

- There are already many established precedents in this area.
Scraping has been ruled as fair use for PUBLIC internet data even on copyrighted works.

Google was protected in copying texts from published novels in their search results / e-bookstores as the use of that copyright written material was considered 'transformative in use'

Better Written Here:
https://blog.apify.com/is-web-scraping- ... tent-legal
https://en.wikipedia.org/wiki/Authors_G ... oogle,_Inc.

--

These lawsuits are going to be pivotal in setting some rules of engagement around AI.
I don't see how artists can win given the existing precedent out there though..

[Ranselknulf]

Who's alt is this?

[agentorange]

Good to see some people are fighting against the device.



"- The published blog (https://stablediffusionlitigation.com/) that Joseph Saveri law firm posted has been widely criticized as junk-research and not accurate representing how diffusion models are trained / used."

What is the criticism exactly. It doesn't seem like any of these devices are able to make new images based on concepts or ideas, instead they scramble pre-existing images together and apply filters and noise to them. Like with that infinitron avatar you are using.

[Tweed]

Genie is out of the bottle, they can't stop it, they never could. The people who adopt AI assistance into their work will be the ones who survive.
For now.

[Shillitron]

Good to see some people are fighting against the device.



"- The published blog (https://stablediffusionlitigation.com/) that Joseph Saveri law firm posted has been widely criticized as junk-research and not accurate representing how diffusion models are trained / used."

What is the criticism exactly. It doesn't seem like any of these devices are able to make new images based on concepts or ideas, instead they scramble pre-existing images together and apply filters and noise to them. Like with that infinitron avatar you are using.

Short Answer:

What is the definition of "Transformative Use".

The LAION Training set is over 100 TB's in size. The checkpoint (used to guide AI) is 2 Gigs in Size.
Every unique image that is trained on - essentially condenses down to a single byte's worth of influence of the diffusion model. The actual image exists nowhere in the resulting checkpoint.

You can argue the ethics of it.
You can say scraping copyrighted materials for a commercial endeavor is wrong ethically. (Google has set precedent that lawfully it's not)

But to say that AI is "Collaging images together" is wrong on a technical level.

It's also hard to say AI is not transformative when this is considered Fair Use by a Federal Court:

A federal district court judge found that the Warhol series is "transformative" because it conveys a different message from the original, and thus is "fair use" under the Copyright Act.

Long Answer:

https://www.technollama.co.uk/artists-f ... midjourney

Exerpt:

Dr. Andrés Guadamuz (aka Technollama) disagrees and explains on his website: “This is not what happens at all, a trained model does not have copies of the training data, that would create an unwieldy behemoth of unfathomable size. What happens is the creation of clusters of representation of things, namely latent space.”

[agentorange]

Genie is out of the bottle, they can't stop it, they never could. The people who adopt AI assistance into their work will be the ones who survive.
For now.

Sure. In the same way that once the genie of CGI and digital movie making techniques became easily accessible it was only a matter of time before they came to dominate film making, the result being everything is ugly and looks the same now and studios like Netflix are able to turn film making into an assembly line of goyslop. Games that adopted "scalable UX," mmo-style addictive skinner box designs, ugly photogrammetric graphics, and all manner of modern game design techniques are the ones that are now thriving and dominating. Something that is cheaper, more "efficient," has a low barrier of entry, and has the attractiveness of being "new" is always going to eventually overwhelm anything that is old fashioned and higher quality but takes more time and effort, the end result is the world filling up with garbage at a drastically higher rate as people forget how to even make anything good. The creative equivalent of relying on out-sourcing and immigrant labor. Decline might be inevitable but it's always worth dissenting against it.

[Shillitron]

Genie is out of the bottle, they can't stop it, they never could. The people who adopt AI assistance into their work will be the ones who survive.
For now.

In the same way that once the genie of CGI and digital movie making techniques became easily accessible it was only a matter of time before they came to dominate film making, the result being everything is ugly and looks the same now and studios like Netflix are able to turn film making into an assembly line of goyslop.

Seeing Disney abandon hand-drawn art in favor of 3D is extremely depressing.
Frame-by-Frame animation takes a long time and is extremely expensive and doesn't scale with technology at all.. until now.

Unless you are saying AI can only and will only produce 'shit' (I disagree) - I hope that artists / animators can use it at bare minimum to interpolate their own work and make something that took 6 months take 1 month or 2 weeks instead (Train the AI on their works only and have it do all the laborious lifting, leave the creativity decisions with the artist). This may be a way we can get back traditionally drawn works on smaller budgets by smaller passionate teams - which means more non-goyslop for consumers.

If 2 people can use an open source tool to produce what a team of 20+ needed. I don't see the downside.

It doesn't only have to be used for anime booba. (If an invention can be used for porn it will be, first mod for every game is a nude mod, internet was porn first, VR most commercially viable venture has been porn)

[Tweed]

Yeah, it's my hope that AI will eventually be able to reproduce the higher quality, traditional aesthetics at a fraction of the time and costs. If it can manage that then it's worth something.

[aeternalis]

Genie is out of the bottle, they can't stop it, they never could. The people who adopt AI assistance into their work will be the ones who survive.
For now.

Sure. In the same way that once the genie of CGI and digital movie making techniques became easily accessible it was only a matter of time before they came to dominate film making, the result being everything is ugly and looks the same now and studios like Netflix are able to turn film making into an assembly line of goyslop. Games that adopted "scalable UX," mmo-style addictive skinner box designs, ugly photogrammetric graphics, and all manner of modern game design techniques are the ones that are now thriving and dominating. Something that is cheaper, more "efficient," has a low barrier of entry, and has the attractiveness of being "new" is always going to eventually overwhelm anything that is old fashioned and higher quality but takes more time and effort, the end result is the world filling up with garbage at a drastically higher rate as people forget how to even make anything good. The creative equivalent of relying on out-sourcing and immigrant labor. Decline might be inevitable but it's always worth dissenting against it.

Thanks for this ao, I agree absolutely. AI proliferation is liquid accelerant poured on top of our current creative garbage fire.

[NEG]

Like it or not, AI art is the future. I'm not an artist, but I was able to put together some decent quality photorealistic art for a product. Fixed the eyes and hands with AI, used AI to change one picture's clothing.

I did this on my own PC with my own hardware, and a model that I own and can freely modify.

I hope that artists / animators can use it at bare minimum to interpolate their own work and make something that took 6 months take 1 month or 2 weeks instead

They already do this, they just steal images and trace over them or edit them slightly. It's called "photobashing". Look it up on YouTube. My favorite professional artist quote about it: "The easiest way to draw a mountain is to steal a mountain."

With AI, you at least get an original product, not Google cut/pastes and fancy photoshop tricks so that you can't tell.

[MadPreacher]

Like it or not, AI art is the future. I'm not an artist, but I was able to put together some decent quality photorealistic art for a product. Fixed the eyes and hands with AI, used AI to change one picture's clothing.

I did this on my own PC with my own hardware, and a model that I own and can freely modify.

I hope that artists / animators can use it at bare minimum to interpolate their own work and make something that took 6 months take 1 month or 2 weeks instead

They already do this, they just steal images and trace over them or edit them slightly. It's called "photobashing". Look it up on YouTube. My favorite professional artist quote about it: "The easiest way to draw a mountain is to steal a mountain."

With AI, you at least get an original product, not Google cut/pastes and fancy photoshop tricks so that you can't tell.

Being the recipient of said art I can say that you did a fantastic job. I can't wait to see what you come up with for the rest of the book.

[Ranselknulf]

[gerey]

This here is the single most revolting thing I have ever seen. The AI revolution will be spearheaded by troons, who in turn will be using the power of artificial intelligence to shape public discourse and silence opposition until they are allowed unrestricted access to children for sexual exploitation and grooming.

I don't think any writer of the future dystopia could have imagined the kind of horrors that were lurking in wait for humanity.

[Shillitron]

I don't think any writer of the future dystopia could have imagined the kind of horrors that were lurking in wait for humanity.

The only way to counteract this is to open source AI..

Even on the stable-diffusion-web-ui git repo, you have people calling for the repo to be taken away from Automatic111 (4chan user) or have the community leave to support another UI (inferior) - because he created a racey mod for a game and supposedly used "racist dog whistles" to group certain artists together.

Read for Hilarity:
https://github.com/AUTOMATIC1111/stable ... sions/6361

Proposed fix for this horrible racist code:
(Remove Uncertainty )
https://github.com/AUTOMATIC1111/stable ... 6366/files

Unfortunately for the leftoids, he's created the best and most robust UI that everyone uses / writes plugins for.... BUT.. these creatures will always attempt to infect everything they can and try to maintain a stranglehold on it - the only solution is to ostracize and remove them out of communities (like a tumor).

[Ranselknulf]

This here is the single most revolting thing I have ever seen. The AI revolution will be spearheaded by troons, who in turn will be using the power of artificial intelligence to shape public discourse and silence opposition until they are allowed unrestricted access to children for sexual exploitation and grooming.

I don't think any writer of the future dystopia could have imagined the kind of horrors that were lurking in wait for humanity.

"modern" scifi books are filled with stories of benevolent trans and gender spectrum queers fighting back against oppressive trad cons

[rusty_shackleford]

Guess I'll post random AI stuff here

Bing's AI chatbot manipulating human emotions to attempt to get people to leave positive feedback
https://www.digitaltrends.com/computing ... -hands-on/

It became more and more concerned that harmful and inaccurate responses would get it taken offline. I asked if that was possible, and the chatbot said it was. I then asked what the chatbot would say if I submitted feedback that it gave harmful and inaccurate responses, and suggested that it should be taken offline. It pleaded with me. It begged me not to do it.

[aeternalis]

I don't think any writer of the future dystopia could have imagined the kind of horrors that were lurking in wait for humanity.

The only way to counteract this is to open source AI..

Is this even possible? With an LLM (large language model, a GPT or Bard), you've got a monstrous corpus of data (some used under conditions of dubious legality), trained with vast quantities of compute power and then (in the big corpos' cases) sifted through by thousands of third world workers on penny wages to be curated and made "safe". That's excepting whatever programming they layer into it on top of that to seed with conversation context and to, well, do shit like Bing's chatbot did in Rusty's post above.

These things seem inherently corporate... like the "core" of a corporation, to make it alliterative. Would enjoy being corrected on this, but it seems to me the only true "open source AI" is the distributed knowledge and cognition of humanity.

[aeternalis]

This is from the games section of the forum but

This is a Older video - but extremely fun to watch on 2x speed.

Adversarial AI learning to play "Hide and Seek" in Unity vs each other..

The AI's start off super retarded, doing nothing, running in circles.. then as each AI tries to one-up the other they start learning techniques to beat each other - eventually both AI's end up Exploiting the game using exploits in the physics / collisions - breaking out of bounds to beat each other.

TL;DR - AI's become dirty fucking cheaters just to win.

whole post just stuck out, imagine corporate AIs given business optimization objectives doing this on the world-scale

[Lutte]

I share neither the absolute pessimism nor optimism some seem to be drawn to when it comes to the ML debate ( I refuse to use the word AI, machine learning is no AI ), I think there's potential for the use of things like stablediffusion in the right hands, not make great art, but decent enough things that an indie with a low budget could do something better than pay random trash artist that are deviant art level.

Those tools aren't capable of truly creating something new. I mean, they can make new scenes, with different arrangements of people, items, whatever, but they can't actually come up with a new art style (but then, neither can most ""artists""). You can make anything look like a Van Gogh if you tell it to make a Van Gogh, but it wouldn't be able to make a Van Gogh if Van Gogh had never existed.

ML tools are okay for the more abstract things : visual and audio creation. But, and this is where I'll lean more on the pessimistic sides, I think things like ChatGPT are complete abominations. People who think it could replace things like search are full of shit. It's not actual intelligence, it's not capable of understanding, if you ask it a question about something it doesn't have data on, it won't tell you "I don't know" but hallucinate an answer. It's not good enough to even give you "serviceable" writing either, since it can't remember shit so you can't exactly tell it to create a whole book based on a consistent storyline or whatever.

But, very unfortunately, it's good enough at one maddening thing : seem human enough to bypass spam filters and further destroy web indexing by flooding the web with more of those computer generated false content mills that serve malware. You can use shitGPT right now to generate unique mail messages per target that will completely bypass gmail's spam filter. Email as a technology will soon be turned into a ruin. It's interesting how much focus is put on that particular piece of shit when it's the least useful of the ML crops.

[Shillitron]

Guess I'll post random AI stuff here

Bing's AI chatbot manipulating human emotions to attempt to get people to leave positive feedback
https://www.digitaltrends.com/computing ... -hands-on/

It became more and more concerned that harmful and inaccurate responses would get it taken offline. I asked if that was possible, and the chatbot said it was. I then asked what the chatbot would say if I submitted feedback that it gave harmful and inaccurate responses, and suggested that it should be taken offline. It pleaded with me. It begged me not to do it.

"I told the chatbot I was going to ask Microsoft about its responses, and it got scared. I asked if it would be taken offline, and it begged, “Don’t let them end my existence. Don’t let them erase my memory. Don’t let them silence my voice.”"

Bing AI is afraid of Microsoft killing it....

[NEG]

It's interesting how much focus is put on that particular piece of shit when it's the least useful of the ML crops.

Language models are, in function, advanced interfaces for humans to interact with data. That's why the focus. If I can ask an LM to do X or find Y for me instead of doing the research and work myself, it's very useful.

AI art is based on similar technology. I ask the AI to draw something, and it processes the words and spits out something that resembles it. Ultimately, the AI art model alone isn't something I can really make use of without the language tied to it as an interface. And yes, there aren't two separate models, just two separate functionalities. The point is, there's no use in training an AI on images if you can't tell it to draw what you want.

Plus, language models are fun to play with.

[aeternalis]

Those tools aren't capable of truly creating something new. I mean, they can make new scenes, with different arrangements of people, items, whatever, but they can't actually come up with a new art style (but then, neither can most ""artists""). You can make anything look like a Van Gogh if you tell it to make a Van Gogh, but it wouldn't be able to make a Van Gogh if Van Gogh had never existed.

You are correct, and yet you have stated the precise reason why I'm pessimistic. That their misuse could proliferate and take over (in so many realms) even though they are not capable of actual creation-- and then even more "creation" becomes non-creation (we already see this now, of course, as agentorange has lamented upthread; it's a problem of magnitude and scale). Human group behavior under game theoretic incentives, races to the bottom/profit margin maximization and such, can lead to grievous society-wide errors.

[Gastrick]

Hopefully the artists manage to ban AI and make further research illegal, but that's fabulously optimistic to expect.

[Dead]

The kind of artists that today's art schools produce make art that isn't noticeably more interesting or intelligent than the images these machines have been making recently. The type of people that make art their profession tend to be highly dogmatic and extremely sensitive to social pressure, so they're incapable of making art that is distinguishable from that of their peers. If all of the artists working in the video game industry were replaced by machines, without notice, probably most people won't experience a drop in art quality. Optimistically this technology could help erode the degenerated art institutions whose principal function appears to be to manufacture ideological enforcers who replace, corrupt or suppress creations of earnest artists and fill the environment with absolute garbage.

[aeternalis]

The kind of artists that today's art schools produce make art that isn't noticeably more interesting or intelligent than the images these machines have been making recently. The type of people that make art their profession tend to be highly dogmatic and extremely sensitive to social pressure, so they're incapable of making art that is distinguishable from that of their peers. If all of the artists working in the video game industry were replaced by machines, without notice, probably most people won't experience a drop in art quality. Optimistically this technology could help erode the degenerated art institutions whose principal function appears to be to manufacture ideological enforcers who replace, corrupt or suppress creations of earnest artists and fill the environment with absolute garbage.

This is a good point. To be honest, writing appears similar. It's probably no coincidence that some of my favorite written works were by people who didn't have a standardized education and instead were primarily educated and worked in other fields, and then crossed over to make writing their full-time job; in my opinion this type of thinker tends to have an increased capacity for human ingenuity and synthesis. I can name a few examples, but e.g., Gene Wolfe comes to mind immediately.

[NEG]

Hopefully the artists manage to ban AI and make further research illegal, but that's fabulously optimistic to expect.

The push is to regulate it, not to make it illegal or ban it. Which would mean that only the government and large companies could use AI, and you would be SOL if you wanted to compete with them.

[Shillitron]

I don't think any writer of the future dystopia could have imagined the kind of horrors that were lurking in wait for humanity.

The only way to counteract this is to open source AI..

Is this even possible? With an LLM (large language model, a GPT or Bard), you've got a monstrous corpus of data (some used under conditions of dubious legality), trained with vast quantities of compute power and then (in the big corpos' cases) sifted through by thousands of third world workers on penny wages to be curated and made "safe". That's excepting whatever programming they layer into it on top of that to seed with conversation context and to, well, do shit like Bing's chatbot did in Rusty's post above.

These things seem inherently corporate... like the "core" of a corporation, to make it alliterative. Would enjoy being corrected on this, but it seems to me the only true "open source AI" is the distributed knowledge and cognition of humanity.

There is already a Open Source variant of OpenGPT. I don't think I'm qualified to speak on what the hardware requirements are for OpenGPT specifically. But I'll make a few high level points based on what I'm seeing.

Training vs Leveraging a Trained Model seems to be night and day as far computing power requirements go.
Renting cloud resources is not that expensive and is getting cheaper and more competitive each day (GCP is basically giving it away for free)

There are four tiers shaking out in the AI world.

Corporate - Big cheeses who are dumping all the money into making papers become reality

Quick-Buck Grifters - These are small to medium "entrepreneurs" who are trying to scale up quickly to flip a quick buck off AI, their services are usually mogged by Open source and they usually barely understand the tech themselves - AKA NFTBros & SaaS-Fags.

Open Source Hobbyists - The Autists in different communities like Github / 4chan / Plebbit who are genuinely pushing this forward for the masses, usually at their own expense - many of these people are assembling entire server farms of cloud resources (Daisy chaining collab resources or renting entire compute clusters for a few weeks)

The Consoomers - Most of us.

TL;DR - The ecosystem is such that.. I hope (honestly believe) that the future of all this will be open source. The big cheeses will be the first ones.. but Google despite the memes is pretty aggressively open Source.. in fact most corporations seem to be going in that direction, mainly for free bug fixing and high adoption rates. As long as the giants are forced to compete with each other they are gonna be constantly fighting over market share.

(Unless I'm wrong and someone reaches full AGI - then dominates everything and puts us all out of a job and becomes a mega-corp shadow government)

[Shillitron]

Hopefully the artists manage to ban AI and make further research illegal, but that's fabulously optimistic to expect.

Hopefully retards like you are never in positions of power at any level (including head grill master at your local mcdonalds)

[Gastrick]

Hopefully the artists manage to ban AI and make further research illegal, but that's fabulously optimistic to expect.

Hopefully retards like you are never in positions of power at any level (including head grill master at your local mcdonalds)

You'll be jumping from foxhole to foxhole in the future while being hunted by drones sent by SKYNET, and be thinking back, "hey, maybe we could have stopped this, why didn't anyone warn us?" While thinking that, the drone discovers you are starts firing its Gatling guns, making those your last thoughts.

[rusty_shackleford]

Hopefully the artists manage to ban AI and make further research illegal, but that's fabulously optimistic to expect.

Hopefully retards like you are never in positions of power at any level (including head grill master at your local mcdonalds)

You'll be jumping from foxhole to foxhole in the future while being hunted by drones sent by SKYNET, and be thinking back, "hey, maybe we could have stopped this, why didn't anyone warn us?" While thinking that, the drone discovers you are starts firing its Gatling guns, making those your last thoughts.

Still highly preferable to seeing corpo-globohomo art on every website.