Should the creator of Photoshop be part of every discussion about digital art? Should the inventor of the camera be a part of every discussion about photography?
When I took philosophy, we always had to agree as a group to an “operational definition” of whatever term or phrase we were discussing before the discussion began. I think this thread is a great example of why. I have my own concept/belief of what constitutes an artist, and you have a different concept/belief of what constitutes an artist. That prevents us from having any common ground for discussion because, in your mind, you are right and, in my mind, I am right.
I think Jasmine is referring to the AI prompter instead of the people who actually produced Midjourney (or other AI art programs) when she uses the term AI creator.
Hence it’s the work of the guy who made Théâtre D’opéra Spatial and submitted it for copyright protection that’s driving the current conversation versus Midjourney founder David Holz.
I agree, but that stems from a complete misunderstanding of what AI is or how it works, because they also say that the “AI Creator” is a “Great programmer”. You need to be a programmer to create your own AI tools, but not to use them.
It’s like if you heard that someone is a digital artist, so you assumed they coded Microsoft Paint from scratch and then doodled in it.
If an academic wants to publish a paper including a table from a previously published work that he wrote, he still has to get permission from the publisher and pay a licensing fee. He made it but it’s not his. Copyright law isn’t really interested in protecting artistic integrity so much as protecting profitability.
Effort and skill are also completely independent from art itself except as bullshit means of gatekeeping (and there’s lots of bullshit gatekeeping in the art world).
An awful lot of negative sentiment about generative art is that it’s too easy, which is exactly what people wrung their hands about when photography was popularized and was 100% going to destroy painting. Narrator: “it didn’t.”
What happened instead is that people recognized that producing good photography required a different skillset and that appreciating good photography required a different viewpoint and language. That’s where we’ll eventually get with generative art.
This is something I’ve been arguing ever since I first saw a talk on using genetic algorithms to design things. I came at it from a patenting angle, but it’s the same issue: if a computer is doing the work, who is the inventor/author/creator? All the statutes are written such that the inventor/author/creator creates the intellectual property, even if the ownership of that IP is immediately transferred to some other entity, like a corporation. If there’s no human in the creative process, then who is it who created the work?
For patents, this is less of an issue, because they only exist for a fixed period of time. But copyright duration is defined based on the lifetime of the author. Who determines when a computer dies? With cloud computing, can you even tell which computer(s) did the processing?
Whomever has intent. DallE doesn’t make anything on its own. It still requires human intention, and intention is art.
DallE isn’t “doing the work” any more than a camera is “doing the work” of a painter.
Someone above made the analogy to hiring an artist to paint something for you. this kind of “work for hire” is well-established in the law and jurisprudence. The person who hired the artist owns the copyright, as soon as it is created. That is not in dispute.
What is in dispute is, who created the copyright in the first place? Under work for hire rules, it was the artist under contract, who is assumed to be human. In this AI system, it would be the computer that created the copyright.
“Intent”, as you call it, in both cases comes from someone other than the entity that actually laid down the images. “Paint me a cow.” “Now, make it blue.” I end up with a painting of a blue cow, I had the “intent”, and under work for hire, I own the copyright. But who created the copyright? I can’t own what doesn’t exist.
The Monkey Selfie Issue is relevant here.
I think I see what you’re saying. Your argument is that DallE is incapable of handing the copyright to the user because a computer can’t create the copyright in the first place? Why is it not OpenAI granting the copyright?
No, it wouldn’t, because a computer is not a person that can create copyright. It lacks intent. The human generating the AI art creates the copyright, just like the human who pushes the clicker on a camera or the human who pushes the buttons in Photoshop or MSPaint.
Yes, pretty much. This is a question that the courts have (mostly) ruled on. If a monkey can’t create a copyright, I can’t see how a computer can.
But that’s under the existing law. What we need to decide is, “Do we, as a society, want AI generated art/inventions to be protected under the law?” Answer that, and amend the law accordingly.
For the same reason Kodak doesn’t grant the copyright on a photograph. They build equipment, not art. Otherwise, why not allow paint brush companies to own copyrights on every painting?
Frankly, I think the monkey decision was beyond stupid.
If I set up a camera on a 10 second timer, I own the copyright. If I set up a motion activated camera in the forest and take pictures of animals, I own the copyright. And if I take a camera to a clearing where a troop of monkeys is, and I play with them, and one of them bumps the button on the camera - I should own the copyright for that, too.
You managed to neatly miss my point. The computer in this case is doing far more than a camera does when you click the shutter. When the photographer clicks the shutter, they have a generally good idea of what the end result will be. Is that the case with these AI systems? Isn’t part of the point that they produce things a human would not expect?
But the computer still doesn’t have intent.
And that’s the argument I think we should be having. Write your congressmonkey and amend the law!
Isn’t there an implication of a contract between artist and person who hires that artist which makes the copyright be owned by the hirer? Computers can’t enter into contracts, so work for hire might not be relevant.
Here’s another angle. Say the AI produces an exact copy of what it produced for someone else - very possible. If the user can get a copyright for the AI’s work, are they liable for copyright infringement if some other user created the work first? Ordinarily you might be able to prove that the accused infringer had no access to the first work, but here the AI uses the same data in both cases.
AI Artists might want to be careful about this.
If I put a GoPro on my dog’s collar (and start the video myself to avoid any monkey business), then let my dog run around outside - let’s say I own a few acres out in the country rather than a house in the suburbs so there’s more to explore than just my back yard - isn’t the entire point of doing this (rather than taking a walk with a camcorder myself) to get video footage from a vantage point that a human would not expect?
But then we have the history of slavery to fall back on. Slaves could not enter into contracts either, and were barred from patenting their inventions.
And these are the kinds of questions that make lawyers rich. This is why I started warning people about this problem years ago, but no one listened!
The level of randomness is actually a discrete setting called temperature. Some models enable it for users and some don’t.
Sure, that makes sense to me. Does copyright make a distinction between a product (a local copy of Stable Diffusion) and a service (a subscription to OpenAI)? Doesn’t OpenAI have a right to claim copyright on anything made using its tools as part of its service agreement?
Again, good questions that we should ask the people in charge to answer.
Video graphics card have filters that use AI to improve the appearance of the scenes they’re displaying. Does this mean a screenshot of a game can no longer be protected by copyright?
Or, what about an artist who manually draws a sketch of a scene on a sheet of paper, scans it into a computer, and then uses AI to add details. Can that final image be copyrighted? Does it matter how rough the sketch is, or if it started on paper?