I was in the "Alte Pinoteke" in Munich, Germany, on Sunday. It is a fabulous gallery full of huge and old master paintings.
There was a certain 4 frame series with each frame about 5m long by 3m high. Each was an architectural study of Rome with each being from a different view direction. What struck me was the text written below the paintings which indicated that the artist, after *3 years of work*, had been quoted that he was sick of the job and wished that the process would be now over. 3 years to produce 4 images so that some aristocrat could hang it on his wall and be reminded of his holiday adventure.
The distance between this state of the world, where to make a representation of reality (unreality) at the highest levels of fidelity would take years to achieve, versus what is now possible with digital cameras and AI image generation is staggering. Now anybody can create a Mona Lisa and have it hung in their own toilet to admire. Is this a bad thing that ordinary people have access to the tools to create content vs the old way where only the very very rich could afford to pay a master for three years to produce a few images?
I can only assume that the same arguments and complaints were raised at every stage a new technology was introduced that lowered the cost and time to produce "content".
> I can only assume that the same arguments and complaints were raised at every stage a new technology was introduced that lowered the cost and time to produce "content".
Which of those technologies were based on billions of copyright violations? And of course this copyright still exists but is only enforced in one direction, just like on YouTube where random corporations can claim copyright on any video and hurt the income of actual creators without negative consequences. AI will likely not improve this situation.
I'll think differently about that comparison if it is possible for random people to legally train models & create images from Disney's intellectual property on affordable hardware. But I don't have any false hope that equivalent enforcement of copyright will be a reality in the foreseeable future.
What's the infringement? You can't copyright style, no? If artists can look at each others copyright-in-full-force works and then imitate it, why can't AI?
Corporations having too much power is different from technology encroaching on previously uniquely human faculties.
Of course both are phenomena that ought to be addressed by society. At which point we see that it's a political problem, and of course politics is broken.
> I'll think differently about that comparison if it is possible for random people to legally train models & create images from Disney's intellectual property on affordable hardware.
... that's a very strange set of criteria. that said, last year one of my designer friends tried Midjourney, one of the pictures was so and so in Star Wars style.
Of course generating pictures is different from using them, let's say in ads.
> If artists can look at each others copyright-in-full-force works and then imitate it, why can't AI?
AI can reconstruct many training images in surprising detail. Show me an artist who can draw you an image they practiced on 10 years ago including the rough shape of the watermark.
> Of course generating pictures is different from using them, let's say in ads.
Microsoft's Copilot is a paid service, same with OpenAI's upcoming ChatGPT Plus subscription. Nobody here claims that it's illegal to draw Mickey Mouse and hang it on your fridge, but those AI models aren't trained just for fun. If Microsoft is allowed to make money from ignoring countless license terms I should be allowed to make my own short film trained with Disney cartoons and sell it. But we all know how well that would turn out.
> Show me an artist who can draw you an image they practiced on 10 years ago including the rough shape of the watermark.
I can't but that doesn't mean there are none. For music we have a lot of examples of artists who can copy perfectly. For visual arts we have people who take a helicopter tour of a city and then draw it in amazing detail.
Yes, time discounts everything for humans. That doesn't mean human artists don't cargo-cult watermark-like things. (One example is putting text/symbols on clothes that the designer doesn't bother to understand.)
> Copilot and MickeyMouse
This thread started by someone commenting about paintings in the Alte Pinakotheke, then about Disney.
Now it's language models.
Obviously they are different, but what's more important is that copying is more clear-cut in code, less degrees of freedom compared to visuals.
That said, at this point it's important to discuss how copyright works. You can look at someone's code. You can copy the style. Let's say you never saw functional code, you see a lot of it, then you learn functional style. Neat. But you can't copy the Haskell standard library 1:1 without infringing.
If Copilot produces Haskell stdlib, the copyright holders can sue.
And of course that's again where politics comes in. Since this is a new phenomena that disadvantages many small copyright holders this is where society ought to step in and help them to enforce their copyrights.
(So it's important to note that copyright and the license terms don't kick in if it's not a derived work. And that's the discussion we should be having. Maybe as a quick hotfix society should tweak copyright law and make it more broad in terms of what it considers a derivative. We can also put in the specifier that "except for humans, if you swear you did not use an artificial neural net" or whatever. I don't think this makes much of a difference, unfortunately. AI - and other kinds of technological changes - makes a lot of existing inequality problems worse. The general solution is to fix the inequalities. Of course if politics were actually representative of people's needs the inequalities were also likely inconsequential.)
I think the point here is that now you can. Your "AI" can draw it for you with extreme accuracy and then... is it yours? That's the real question here. Pretty simple.
What if we had a copyright law that did give like a percent of detail that has to be different, so you make a new drawing that has the required amount of difference, and it's yours right? Now your "AI" can be trained to make differences in drawings that specifically meet the requirement of however much difference between the original and yours will negate the ability of the original author to make a copyright claim. Pretty straightforward. It's not the case now, but when we have enough content that has to be judged by humans on whether or not it falls under copyright... Anyhoo... I'm not selling content, so I shouldn't care, right?</snark>
Are we equating the capabilities of humans and AI now? AI isn't "imitimating" it's synthesizing an output, in seconds, based on a giant database of art scraped from the internet by human researchers, without permission from those artists, to be crystallized in a form that will outlive every artist born and yet born till the heat death of the universe.
If copyright doesn't distinguish between that then copyright needs updated. It's a tool, not a person, and one that people want to use commercially so they can create the billion pieces of art to train it on themselves.
One one hand I support authors' rights to place restrictions on their art for a limited time. (Not the current forever + 75 years.) So let them draw whatever lines they want in the sand.
Because otherwise there's no general good way to delineate generative AI from other tools. It's a very fast magic brush. (You might have seen timelapse videos of artists do a digital painting. It's already pretty magical. Especially to me, because I can't even draw to eggs to match.)
Any copyright currently is about the not-for-free right to _copy_, imitate-right on the other end, still has to be invented. I don't see how that would work in practice.
AI isn't "imitating" anything. Imitation is a conscious effort a human makes. Any artist could try copying someone's style by looking at a piece of art.
AI can't look at an image and say "I'll imitate that". A human "trains" an AI on images to create a model of how X looks with different examples and piecemeals together something that looks new but is in fact a million tiny data copies from all the images it was trained on so the original source cant be detected. The AI can't do anything it hasn't been trained on.
As I understand it, while it may not copy the exact image it has copied data associated with the image in a way that copyright law could never have anticipated.
So the solution is simple, train your models on your own stuff if you want to claim copyright to the work commercially or pay artists for your training data.
It always boils down to the same question when discussing a.i. Is there a difference between people's imagination (producing art that is considered new, but is unmistakably influenced by existing pieces of art) and copying/transforming and stitching together pieces of existing art? In a few years from now, probably nobody can tell the difference anymore.
Same question holds for the reasoning and intelligence capabilities of a.i. Soon we can't tell the difference in at least several knowledge/skill domains. Is there any difference, will become a hypothetical question.
Even if there was no difference at all in the process of image generation between AI and humans I would argue humans and AI should not be held to the same standard of law (and we're not even talking about true AI).
A human that copies someone will die. An AI can live, multiply, and be improved forever. Humans will never compete against anything they wind up being capable of. Society is built for humans not AI.
>Which of those technologies were based on billions of copyright violations?
All of them, no one is reinventing the wheel every time something new is made, every invention is based on the inventions that proceeded it, it's how human society advances and learns.
I don't know how it is in other countries but in general in France you cannot just take a picture of a building and put it on Instagram without asking the architect for rights :
> the image of a building cannot be reproduced without the authorization of the architect who holds the copyright or of his successors in title. Otherwise, the reproduction of a protected work without the authorization of its author constitutes, according to the provisions of the article L122-4 of the Code of the intellectual property an act of counterfeit.
I read some year ago that this kind of thing applies to New York as well, that was in the context of an open-world Spider-Man game featuring NY skylines.
France must get a lot of money from all the photos of the Eiffel Tower on the internet. Has anyone been sued/convicted of breaking that and paid reparations?
Fun fact, the Eiffel Tower architecture is not protected since 1990, but the lighting is. So it's illegal to use commercialy a night time photo of the tower[1][2].
I don't think anybody has ever been sued but the SETE (Société d'exploitation de la tour Eiffel) do make money with the image.
This is a separate (but important) issue. The development and effect that tech like this will have on the place of the artist or designer in the future is going to be a similarly seismic shift for the industry, just as the camera and printing press were in their time. This is going to happen more-or-less regardless of what the outcome of the copyright issue is.
There is a reason the art you were looking at was in a museum, and images generated by AI tools is not.
There is something impressive and inspiring about the process those artists went through, embedded in the physical work itself. Like we are impressed watching someone running a marathon, even though they could cover the same distance much more quickly in a car.
But you aren't saying that "Google trying to build a self-driving car is the biggest triple-E (Embrace Extend Extinguish) bait-and-switch in its history and completely unfair on marathon runners who, if it had any morals, it would be striving to pay".
There was a certain 4 frame series with each frame about 5m long by 3m high. Each was an architectural study of Rome with each being from a different view direction. What struck me was the text written below the paintings which indicated that the artist, after *3 years of work*, had been quoted that he was sick of the job and wished that the process would be now over. 3 years to produce 4 images so that some aristocrat could hang it on his wall and be reminded of his holiday adventure.
The distance between this state of the world, where to make a representation of reality (unreality) at the highest levels of fidelity would take years to achieve, versus what is now possible with digital cameras and AI image generation is staggering. Now anybody can create a Mona Lisa and have it hung in their own toilet to admire. Is this a bad thing that ordinary people have access to the tools to create content vs the old way where only the very very rich could afford to pay a master for three years to produce a few images?
I can only assume that the same arguments and complaints were raised at every stage a new technology was introduced that lowered the cost and time to produce "content".