AI Image Generation

Mist

Eeyore Enthusiast
<Gold Donor>
30,476
22,325
Prompt: Give me a screencap from the dune movie trailer.

AI: gives you a screencap from the dune trailer.

shockedpikachu.jpg
If you ask me to get you a screencap from Dune, I'm going to go get a screencap from Dune.
You don't seem to get how this works. It is not browsing the web and grabbing a screencap from Dune for you. That screencap from Dune is stored inside the model from the training run; that math is part of the model's neural network and model weights, its knowledgebase. Therefore, they have incorporated copies of someone else's work (actually, millions of people's work) into the very foundational math that makes their product work. That is copyright infringement by any definition.
 

Denamian

Night Janitor
<Nazi Janitors>
7,204
19,002
You don't seem to get how this works. It is not browsing the web and grabbing a screencap from Dune for you. That screencap from Dune is stored inside the model from the training run; that math is part of the model's neural network and model weights, its knowledgebase. Therefore, they have incorporated copies of someone else's work (actually, millions of people's work) into the very foundational math that makes their product work. That is copyright infringement by any definition.

Time to lawyer up and sue all the human artists that do the same damn thing. The training data in their brains is infringing on someone's copyright.
 
  • 2Like
Reactions: 1 users

Bandwagon

Kolohe
<Silver Donator>
22,849
59,959
And thus we arrive at the entire crux of the argument that this new technology is at the center of.

"How is that any different than what human artists have been doing?"

What are we even arguing here?

If Mid journey is actually AI or not?

If it (or the video creation ones) are actually doing anything impressive?

If it's copyright infringement when a LLM does the same thing humans have been doing for years?


Or is this a wag the dog sort of conversation where Mist Mist just steers the discourse in whatever direction her cynicism takes her minute to minute?
 

Palum

what Suineg set it to
23,602
34,137
Time to lawyer up and sue all the human artists that do the same damn thing. The training data in their brains is infringing on someone's copyright.
What's the alternative? No IP law because you can put anything into a magic black box and it comes laundered IP free out the other end?
 

Palum

what Suineg set it to
23,602
34,137
And thus we arrive at the entire crux of the argument that this new technology is at the center of.

"How is that any different than what human artists have been doing?"

What are we even arguing here?

If Mid journey is actually AI or not?

If it (or the video creation ones) are actually doing anything impressive?

If it's copyright infringement when a LLM does the same thing humans have been doing for years?


Or is this a wag the dog sort of conversation where Mist Mist just steers the discourse in whatever direction her cynicism takes her minute to minute?

Well it isn't like those people aren't prosecuted when they can be, it simply isn't feasible to take down every person on Etsy who paints a Mario portrait or whatever and is trying to sell it.
 

Bandwagon

Kolohe
<Silver Donator>
22,849
59,959
What's the alternative? No IP law because you can put anything into a magic black box and it comes laundered IP free out the other end?
I'm not an expert in any of this shit, but I rub shoulders with a lot of people who are just because of what I do for work. I don't know anyone at the AI companies, but know more than a few academics and startup guys putting all their effort into it.

Anyways, every one of these people hates the concept of "intellectual property" and patents. All of them. They want no restrictions on usage once something pops into existence.

I don't agree for the most part, but after much debate we figured out it's just because I want the creator/inventor to be well rewarded for their "discovery". Beyond that, I also would support ending the concept of intellectual property and patents. I do want people rewarded for their contributions to technology that's beneficial to everyone, but I don't want them to have total control over it indefinitely.

That was a bit of a tangent, but I was just trying to give some backstory before saying - Fight it all you want, but I think it's inevitable and unstoppable. Yes, I think the future we're heading towards is essentially no intellectual property, except that which you protect yourself.
 

Denamian

Night Janitor
<Nazi Janitors>
7,204
19,002
What's the alternative? No IP law because you can put anything into a magic black box and it comes laundered IP free out the other end?

That I don't know. The AI image generators are doing nothing humans don't do already. They can just do it at scale.

Should a human artist be sued if you ask them to create an image of a still from Dune? Not screencapping it, but creating it from scratch.
 
  • 1Like
Reactions: 1 user

Cad

I'm With HER ♀
<Bronze Donator>
24,496
45,438
You have walked into my trap card.

Google hyperlinks to material. Google does not present those images as their own product.

Thank you for making my argument for me, Mr. Lawyer.
When I go to images.google.com and search batman, I get a bunch of batman images. They are not served by paramount, they are served by google. When I actually click on the thumbnail, that goes to wherever; the thumbnails and the original images are absolutely stored by Google (thus the page loads instantly). Furthermore, for the search to even WORK, google would have had to analyze and index all those photos (and keep that analysis online) in order for the search to work.

This is a stupid argument and nothing you said changes anything because your initial premise was wrong.
 
  • 1Like
Reactions: 1 user

Palum

what Suineg set it to
23,602
34,137
That I don't know. The AI image generators are doing nothing humans don't do already. They can just do it at scale.

Should a human artist be sued if you ask them to create an image of a still from Dune? Not screencapping it, but creating it from scratch.
Yea, they do get sued. Obviously they can't enforce it at scale (until AI judiciary process).
 

Bandwagon

Kolohe
<Silver Donator>
22,849
59,959
When I go to images.google.com and search batman, I get a bunch of batman images. They are not served by paramount, they are served by google. When I actually click on the thumbnail, that goes to wherever; the thumbnails and the original images are absolutely stored by Google (thus the page loads instantly). Furthermore, for the search to even WORK, google would have had to analyze and index all those photos (and keep that analysis online) in order for the search to work.
So Google isn't creating images on demand, but theyre indexing other peoples images. Right? Indexing other people's "work"?

Are they profiting off of this?
 

Cad

I'm With HER ♀
<Bronze Donator>
24,496
45,438
So Google isn't creating images on demand, but theyre indexing other peoples images. Right? Indexing other people's "work"?

Are they profiting off of this?
I do believe Google is a commercial operation, yes. How could they get away with this rampant use of other people's property?
 

Mist

Eeyore Enthusiast
<Gold Donor>
30,476
22,325
When I go to images.google.com and search batman, I get a bunch of batman images. They are not served by paramount, they are served by google. When I actually click on the thumbnail, that goes to wherever; the thumbnails and the original images are absolutely stored by Google (thus the page loads instantly). Furthermore, for the search to even WORK, google would have had to analyze and index all those photos (and keep that analysis online) in order for the search to work.
You are very obviously way out of your element here on a technical level, and your argument sucks on top of it. When you do a Google Image Search and are presented with the indexed images, you get this:

1703803739958.png


So the image you're presented in the index is in fact pulled from the original source. There's some complex caching at the CDN level to speed this up, which is what you see in the img src tag, but the content is pulled from the original source and links to the original source. Further, Google is not presenting this as their own work, unlike an LLM or Diffusion model.

Additionally, you completely skipped over the fact that Google provides countless ways for content owners to control how their content is used by Google. If Paramount suddenly wanted all of their content removed from Google, they would access their Google Search Console, delist their websites from Google, and send C&D orders and DMCA letters and it would be gone within days. This cannot happen the way LLMs or Diffusion models work, there is no way to snip the content out of the model.

How do you make your living as a lawyer with arguments like these?
 

Cad

I'm With HER ♀
<Bronze Donator>
24,496
45,438
You are very obviously way out of your element here on a technical level, and your argument sucks on top of it. When you do a Google Image Search and are presented with the indexed images, you get this:

View attachment 506222

So the image you're presented in the index is in fact pulled from the original source. There's some complex caching at the CDN level to speed this up, which is what you see in the img src tag, but the content is pulled from the original source and links to the original source. Further, Google is not presenting this as their own work, unlike an LLM or Diffusion model.

Additionally, you completely skipped over the fact that Google provides countless ways for content owners to control how their content is used by Google. If Paramount suddenly wanted all of their content removed from Google, they would access their Google Search Console, delist their websites from Google, and send C&D orders and DMCA letters and it would be gone within days. This cannot happen the way LLMs or Diffusion models work, there is no way to snip the content out of the model.

How do you make your living as a lawyer with arguments like these?
It must frustrate you to be so wrong all the time but think you're right. If you think when you load the images.google.com page that that is loading from hundreds of different sites but all loads instantly I have a very large bridge to sell you. I'm not digging into the tagging of a webpage (and what you've done where is highlight the link (href is the link the picture goes to, not where the image was loaded from, but you should know that as a tech person shouldn't you?) to find exactly how they loaded the pictures, but I've been alive long enough to know if you tried to load from 30 different sites they wouldn't all load instantly and at the same time.

But you carry on being wrong.
 

Cad

I'm With HER ♀
<Bronze Donator>
24,496
45,438
There is no valid legal argument that compressing the original content itself and storing it inside the model, aka OpenAI's product, qualifies as either Fair Use or Transformative Use.

Authors Guild, Inc. v. Google, Inc.

 

Mist

Eeyore Enthusiast
<Gold Donor>
30,476
22,325
It must frustrate you to be so wrong all the time but think you're right. If you think when you load the images.google.com page that that is loading from hundreds of different sites but all loads instantly I have a very large bridge to sell you. I'm not digging into the tagging of a webpage (and what you've done where is highlight the link (href is the link the picture goes to, not where the image was loaded from, but you should know that as a tech person shouldn't you?) to find exactly how they loaded the pictures, but I've been alive long enough to know if you tried to load from 30 different sites they wouldn't all load instantly and at the same time.

But you carry on being wrong.
You're 100% demonstrably wrong.

Our own forum threads do this all the time. Because of the storage limits on this website, much of the content of any of the picture and gif threads are linked from elsewhere, either to Imgur's backend, or Discord's CDNs, etc, or many other websites. That is why images on older pages frequently break.

Further, any website you visit with ads is generally pulling content from a dozen different domains that serve the ad content.

This isn't AOL. The internet got very fast over the past 15 years. Your browser can load content from dozens of different domains very quickly without you even noticing.

And you're still ignoring the point about the fact that if Paramount wanted all of their served content removed from Google, or removed from the caches of various CDNs, they could make it happen. There are mechanisms to allow this to happen.
 

Cad

I'm With HER ♀
<Bronze Donator>
24,496
45,438
And you're still ignoring the point about the fact that if Paramount wanted all of their served content removed from Google, or removed from the caches of various CDNs, they could make it happen. There are mechanisms to allow this to happen.
Did anybody ask any of the AI companies to remove their content and they refused?