AI Writes Article About AI: Does The Newspaper Hold The Copyright?
from the the-monkey-gets-it dept
For many years, we wrote about the infamous monkey selfie copyright situation (and lawsuit) not just because it was hellishly entertaining, but also because the legal questions underlying the issue were likely to become a lot more important. Specifically, while I don’t think anyone is expecting a rush of monkey-authored works to enter the market any time soon, we certainly do expect that works created by computers will be all over the damn place in the very, very near future (and, uh, even the immediate past). Just recently, IBM displayed its “Project Debater” offering, doing an AI-powered realtime debate against a human on the “Intelligence Squared” debates program. A few days after that, the Guardian used OpenAI to write an article about itself, which the Guardian then published (it’s embedded about halfway down the fuller article which is written by a real life human, Alex Hern).
In both cases, the output is mostly coherent, with a few quirks. Here’s a snippet that shows… both:
This new, artificial intelligence approach could revolutionize machine learning by making it a far more effective tool to teach machines about the workings of the language. Deep-learning systems currently only have the ability to learn something specific; a particular sentence, set of words or even a word or phrase; or what certain types of input (for example, how words are written on a paper) cause certain behaviors on computer screens.
GPT2 learns by absorbing words and sentences like food does at a restaurant, said DeepFakes? lead researcher Chris Nicholson, and then the system has to take the text and analyze it to find more meaning and meaning by the next layer of training. Instead of learning about words by themselves, the system learns by understanding word combinations, a technique researchers can then apply to the system?s work to teach its own language.
Almost… but not quite.
Anyway, in the ensuing discussion about all this on Twitter, James Green asked the “simple” question of who is the “author” of the piece in question. The answer, summed up by Parker Higgins is:
legally speaking: ?_(?)_/?
there are a few proposed frameworks and a few theories of what happens if none of the proposals get taken up, but it will likely be settled in court
— Parker Higgins (@xor) February 15, 2019
This is why I think the monkey selfie case was so important. In determining, quite clearly, that creative works need a human author, it suggests that works created by a computer are squarely in the public domain. And while this seems to lead some (mainly lawyers) to freak out. There’s this unfortunate assumption that many people (especially lawyers) seem to make: that every creative work must be “owned” under copyright. There is no legal or rational basis for such an argument. We lived for many years in which it was fine that many works entered life and went straight into the public domain, and we shouldn’t fear going back to such a world.
This certainly isn’t a new question. Pam Samuelson wrote a seminal paper on allocating ownership rights in computer-generated works all the way back in 1985 (go Pam!), but it’s an issue that is going to be at the forefront of a number of copyright discussions over the next few years. If you think that various companies, publishers and the like are going to just let those works go into the public domain without a fight, you haven’t been paying attention to the copyright wars of the past few decades.
I fully expect that there will be a number of other legal fights, not unlike the monkey selfie case but around AI-generated works, coming in the very near future. Having the successful monkey case in the books is good to start with, as it establishes the (correct) baseline of requiring a human. However, I imagine that we’ll see ever more creative attempts to get around that in the courts, and if that fails, a strong push to get Congress to amend the law to magically create copyrights for AI-generated works.
Filed Under: ai, articles, copyright, monkey selfie, ownership, public domain
Companies: guardian, openai
Comments on “AI Writes Article About AI: Does The Newspaper Hold The Copyright?”
What I expect will happen is that, in a hypothetical case exactly like this that ended up in court, the Guardian would end up as the copyright owner under work-made-for-hire doctrine, which obviously did not apply to the infamous monkey photographer.
It’s not right. It’s not good. But I bet it will be what we end up with.
Re: Re:
How? How is it work-for-hire?
Re: Re: Re:
Because under copyright maximalist logic™, what else could it possibly be?
Re: Re: Re: Re:
You…don’t actually know what work for hire is, do you.
Re: Re: Re:2 Re:
You… completely missed that this is not a position I’m in any way advocating for, didn’t you?
Re: Re: Re:3 Re:
but you argue that it will happen. You made a statement. Therefore, we want to understand the reasoning you use behind your statement about what logic the court will use.
I realize you think the SCOTUS just declares things and doesn’t have an underlying reasoning (See commentary on the Aereo Case). I realize you assume malice in every decision you disagree with (See claims of explicit intent of harm in libertarian ideals). I realize you simultaneously distrust the Judiciary to act in any fashion but in full throated support of big business to the detriment of everyone else while advocating the Judiciary should be declaring businesses to be micro-governments because then the Judiciary can hold businesses accountable.
But with all due respect, that does not absolve you of the need to explain your thought process entirely. If you just hand wave that it all falls under the looks like a duck standard, I could have accepted that.
But you comments seem to contradict other recent rulings in this court on copyright, and if you want to make such a bombastic statement, yeah. We’d like to know how you think, in the multi page ruling they reconcile the various issues of content creation that conflict within such a declaration.
Work for hire requires either employment or an explicit contract between the freelance creator (who could be argued to be the person who coded the AI, the person who trained the AI, or the person who pressed the create button), and the person who is hiring for the work. Now you clearly assume the employee or freelance creator is the person who pressed the ‘do’ button, that that act is analogous to pressing the button on a camera. And that is a basic arguement. But that leaves the coder and trainer without any copyright in the output of their AI, whose output is based on their actions and decisions. It could be argued that either one of them should have a stake, as they do not nessisarily contract with the individual or individuals (including corporations) that are hiring the work. And that leads to more questions in interactive media streaming and other areas of copyright law.
And though you don’t advocate for this to be the way of things, you claim you think it will be the position of the court, which means you think you know how these briefings will go, how the justices will answer these questions, which would come up in briefing and arguments, in their final ruling.
Oh, right. You think findings of the court are just Lorem Ipsum and that the final judgement has no relation to the facts or the information presented to the judges, and therefore the rulings are largely just whatever makes corporations happy. And yet, you continue to support the judiciary declaring businesses governments so the judiciary can enforce a ban on deplatforming, despite being bought off by big business.
Re: Re: Re:4 Re:
Do you perhaps mean ipse dixit?
Re: Re: Re:
I think I understand Mason’s position and I actually think he may be mostly right. They may not call it work-made-for-hire, but they will come up with some sort of rationalization along those lines for the publication to hold the copyright… because of the false belief that it must be owned and "but for" the organization using the machine to create the work it wouldn’t be created.
Mason’s correct that it’s not right and not good… but it would not surprise me if that’s what a court eventually decides.
Re: Re: Re: Re:
I just can’t see a reasoning that doesn’t put a question mark on a company holding copyright in gameplay videos created by a third party.
Re: Re: Re:2 Re:
I just can’t see a reasoning that doesn’t put a question mark on a company holding copyright in gameplay videos created by a third party.
Yes. But are you confident that a Supreme Court with its current makeup will consider such consequences of a bad ruling?
Re: Re: Re:
Maybe by British usage, where one can "hire" inanimate objects like taxis? US law is based on British law, after all. (It still fails based on Naruto-the-monkey precedent.)
Spin
My take is that the AI and computer will be considered a tool, just as the camera is. And the operator of the setup will be considered the author. The AI may be doing the heavy lifting, but someone has to tell it what to do.
Not the way I think it should go, but the way the maximalists want it to.
Re: Spin
What happens then when machine learning is incorporated into the artificial intelligence. In that case, not even the programmer is in control of the output as anything learned by the machine did not come, directly, from the program, let alone the operator of the machine.
When the AI has read all of known literature, both fiction and non fiction, and then makes a decision on what to write, how could the operator, who probably hasn’t read all known literature, be the proximate cause of the output.
More interesting might be when the AI quotes something it read and does not properly cite the source. Just when would the AI become liable for plagiarism?
What if we included law, including all case law and reversals of previous opinions. Then add International Law, and the laws of all countries. Could the AI properly judge the content it produces as appropriate for the jurisdiction where its writings are published? Many humans have difficulty with this one.
Re: Re: Spin
It would be more fun if ai were sued for infringement.
Re: Spin
At some point in the development of AI, there’s likely to be a civil rights movement for AI that can reach a certain level of reasoning and function, particularly if they are able to pass the Turing test in human social situations.
That could completely upend any interpretation of copyright law based on current or near-term AI capabilities.
Re: Re: Spin
That’s all in the future and not at all relevant to the topic as of today.
Re: Spin
How do you think it should go?
The Policeman's Beard is Half Constructed
Done back in the 80’s, and at least as coherent as half of Congress, and probably more coherent than 90% of the Philadelphia City Council.
http://www.ubu.com/concept/racter.html
https://en.wikipedia.org/wiki/Racter
I’m hardly much of a copyright maximalist… copyright should be shortened to 3 years with no extensions, and all criminal penalties for non-commercial infringement eliminated.
But this is an asinine argument. Works created by a computer are no different than works created by a typewriter. Surely if a journalist typed it out on a typewriter, we aren’t going to claim that somehow the typewriter has a soul and that it doesn’t belong to the journalist.
Why then would we claim that twiddling a few knobs on the AI control panel makes it less of a work by the journalist? Don’t mistake this "AI" as the sort you see in science fiction movies with self-aware robots who think, feel, and love. Who have motivations of their own (never mind legal personhood). It’s an algorithm where you dump in a few keywords on the one end, and get back some results sausage.
Same thing with the monkey. Monkey’s can’t have copyright, and the camera was the photographers. Ergo the photographer created the work… the monkey was, at most, a temporary tool used to create the work.
Re: Re:
What if there are no knobs at all, and no journalist either? Someone builds the machine, plugs it in, and it just goes so long as it has power and a network connection. But the creator of the machine only created the ‘seed’ if you will, letting the computer largely program itself based on what it observes over the network. Importantly, it does this in an unpredictable fashion.
In patents, the rule for some years was for the inventor to have a “flash of genius” to get a patent. In copyright, an author must engage in actual creativity, not mere effort, in creating a work of authorship.
The key thing to think about in my opinion is Burrow-Giles. There, the Supreme Court permitted photographs to be copyrighted, but did not give the photographer the position of author merely for the effort of opening and closing the shutter. Rather, the creativity necessary to permit a copyright is found in the selection of the subject, posing of the subject, selecting and arranging the subject’s appearance (clothes, etc), selecting and arranging the other elements in the shot (props, backgrounds, etc.), selecting and controlling the appropriate lighting, getting the subject to have the right expression, etc.
This is why someone other than a mere camera operator can have a copyright on a photo — the author is he who calls the shots, not he who only pushes a button.
If there happens to be no such person, so be it. The programmer of the computer doesn’t call the shots; he doesn’t know what will happen. Ditto for the button pusher. A typewriter is easy: push E and you get an E. A computer that with a single press of a button outputs a book about a subject the button pusher isn’t familiar with, in a language the button pusher doesn’t know, using facts the button pusher isn’t aware of is no mere typewriter!
Likewise, the monkey took the photograph (but isn’t an author) rather than the owner of the camera because the photographer lacked control or creative input. The monkey could’ve just as easily taken a different photo, or not smiled. Only if it had been trained to do exactly as commanded — and did — would the photo be entitled to copyright.
Re: Re: Re:
Someone wrote that software. If that the same person as the "someone builds the machine" then there’s your answer.
If that programmer/builder built it for someone else under contract and/or salary then the one who wrote the checks owns it.
If the journalist bought the software and operated it then that’s the owner.
It’s not a tough question.
Re: Re: Re: Re:
In this scenario, the software primarily builds itself. I agree that it’s not a hard question, but the answer is that such works are uncopyrightable unless a human author is minutely involved in the creation of each such work.
Re: Re: Re:2 Re:
How do you arrive at that conclusion?
Someone had to write the software that generated that content. As so many are fond of saying, a computer can only do what it is told. If it is told how to "learn" and construct meaningful sentences and paragraphs it is still the creativity of the programmer that brought that into being. If that creativity was paid for by another party then they bought the rights to that output. A human somewhere owns the copyright to that material.
We may eventually get to the point where computer learning advances sufficiently to begin to separate itself from human creativity. We’re nowhere close to that yet.
Re: Re: Re:3 Re:
So you’re saying that if parents conceive and have a child, and teach the child artistic skills, etc., the parents own the rights to the creations of the child made using those skills?
Re: Re: Re:4 Re:
According to current copyright laws, the great grandchildren of creators own the rights to creations of their great grand parents, at least until 70+ years (depending on something or other) after the death of the creator. Why wouldn’t a contention in the other direction make sense? The great grand parents of the creator should be able to co-opt all the income from their great grand children, unless they are 70+ years older (depending on something or other) than their progeny’s, progeny’s, progeny.
Of course, both contentions are ridiculous.
Re: Re: Re:4 Re:
That’s not really an applicable analogy. Minors are able to hold copyright themselves. The same cannot be said of an AI.
Not that I’m advocating one way or the other for copyright assignation of automated processes. There’s a large chunk of precedent and knock-on effects that I don’t currently wish to dive into. My opinion can be summed up in two and half words: it’s complicated.
Re: Re: Re:4 Re:
The real point is, the machine is no person. The law only applies to persons.That’s an essential difference between your AIban your children.
There’s no reason to create a kind of personality for machines. You can’t put them in jail, or in court (and you can program the AI to like any kind of punishment). So I think in general the owner should be the legal representation of the machine.
Re: Re: Re:5 Re:
I agree that — so far — the machine is not a person. But I disagree that ownership is relevant at all to copyrights on works generated by the machine. If Alice borrows Bob’s camera to take a photo, Bob doesn’t get the copyright, Alice does. The issue is who undertakes the creative labor per Burrow-Giles. The problem with the AIs we have so far is, no human is undertaking Burrow-Giles creative labor with regard to the particular works. And building the machine itself no more entitles one to the output than owning it — otherwise Canon and Nikon and Polaroid, etc would own the rights to all our old photos.
Why are you afraid that certain works might not be owned because they have no author who is a person? Its silly to fret over this, just like it would be silly to fret that no one owns that cloud that looks like a horse or something.
Re: Re: Re:6 Re:
I really don’t care about the "ownership" of the work. It’s just that I’m afraid everybody starts to do crazy things because "it’s with AI".
Too much people want the producer play a role in the handling of stuff. That’s a grave danger to the right to tinker.
So I would say: in the first place the operator is responsible, if there is no operator, the owner is. And if Burrow-Gilles demands creative labor, there is copyright if one of them did any creative labor.
Re: Re:
The photographer would have had copyright if he didn’t push for the "made by a monkey" narrative.
But then again, nobody would have seen the picture.
Simple. No more auto copyright. The courts time would never have been wasted on whether or not a monkey held copyright to a photo if this were the case.
If it isn’t worth the time and effort to register, it isn’t worth it to protect it.
Another possibility
The folks debating above seem to be considering these possibilities: (a) the work doesn’t have an owner (Mr. Masnick’s preference); (b) it is owned by the employer under some court-created doctrine analogous to work-for-hire; or (c) it is owned by the person who issued the "command" that started computer execution. Let me offer one more possibility. The Copyright Act allows for joint ownership: "A joint work is a work prepared by two or more individuals, with the intention that their separate contributions be merged into a single work." No guarantee that this argument would prevail but perhaps worth arguing that the person(s) who programmed the AI and the person(s) who used it have both made contributions with the intent that there contributions be combined into a joint work. The economic consequences of this conclusion (that all joint authors are entitled to a share of the profits) would certainly be a mess to administer but that’s true for lots of other aspects of copyright law.
Re: Another possibility
Employer? What employer?
Are you thinking that the AI machine is an "employee"?
It’s a piece of machinery.
Re: Re: Another possibility
I mean, unless the AI is actually sentient. I think we’re a LONG way from that.
Re: Re: Re: Another possibility
Even a sentient machine is still a machine. The law is constructed for people, applying it to machines is looking for trouble.
There could be a problem if a company makes a program to create music
or random backround music ,
if that music was copyrighted then it might be used to sue human
composers .Theres a limited amount of notes that can be played on the piano or the guitar.
Future programs might be able to make endless hours of jazz music or
EDm music , cpus and computers are getting more powerful all the time.
So its important that only human creators can be given copyright .
There are already programs that can simulate human speech that sounds real .
Also it would be easy to make a program that makes paintings
and abstract art in random patterns .
It would be a disaster if a human artist could be sued because his
art was very similar to a piece of art created by an ai program .
Artists already use computers to make music and digital art ,
the danger is if a program that just use ai to make music or art
by itself it would make life very hard for human creators .
There should be a simple rule only art or music created by a human can be given a copyright .
Re: Re:
An important question, how is computer created art distinguished from human created art? As unless that distinction can be made, provenance is what somebody attaches to a created work, whether they created it themselves, or their computer created and they attach their own name.
It reads like my own early drafts
GPT2 learns by absorbing words and sentences like food does at a restaurant
A common editing snarl I encounter takes form as a sentence (possibly a run-on) in which my intended thought at its beginning is lost for another thought along the way.
To edit these segments, I need to a) unravel the sentence and see what I was trying to say at each point. b) decide which of those thoughts are relevant / important / funny and then c) rewrite the segment so that each thought is expressed clearly and independently.
It’s a chore.
I could have written the sentence above. But then I was probably very hungry.
Think of the data!
Has anyone given any thought to the feelings of the poor AI?
Maybe they should name the AI "Catch 22". That or, Monkey Selfie".
Re: AI feelings.
I actually have considered the feelings of AI. And in an era where Saudi Arabia has given citizenship to a chatbox in a woman-suit and Microsoft is worried about the personal honor and chastity of Alexa (the digital assistant), it wouldn’t surprise me if emotional damages to an AI become a matter for the courts.
According to a lot of roboticists and AI programmers, feelings only occur, including opinions about personal survival or self-ownership only occur when they’re intentionally added in. And there’s no defined point when an AI reaches sentience, self-awareness or consciousness that is clearly defined. Even when an AI can assemble flat-packed furniture without training.
Nor can we tell when the feelings of electronic entities are genuine or simulated, or if there is an observable difference.
Re: Re: AI feelings.
Indeed, feelings have to be added in. If a creative AI is build to do something creative, it’s feelings will be optimized to maximize exposure of it’s work.
That AI will be build to hate copyright.
"Indeed, feelings have to be added in. If a creative AI is [built] to do something…"
Feelings, nothing more than feelings,
Trying to forget my feelings of love
So: Take a programming assignment, worry and cry over it a bit, and you’re done. Sounds like every college programming class’ output.
I could have used this one instead, but the remake just sounded so …. WONDERFUL. I got feelings just by listening to it.