Comment by jrockway
2 years ago
I was perusing some Simpsons clips this afternoon and came across a story to the effect of "So and so didn't want to play himself, so Dan Castellaneta did the voice." It's a good impression and people didn't seem very upset about that. I am not sure how this is different. (Apparently this particular "impression" predates the Her character, so it's even easier to not be mad about. It's just a coincidence. They weren't even trying to sound like her!)
I read a lot of C&D letters from celebrities here and on Reddit, and a lot of them are in the form of "I am important so I am requesting that you do not take advantage of your legal rights." I am not a fan. (If you don't want someone to track how often you fly your private jet, buy a new one for each trip. That is the legal option that is available to you. But I digress...)
Surely there’s some kind of difference between “voice impression for a two-line cameo in one episode of an animated sitcom” and “reproducing your voice as the primary interface for a machine that could be used by billions of people and is worth hundreds of billions of dollars.”
Is there a name for this AI fallacy? The one where programmers make an inductive leap like, for example, if a human can read one book to learn something, then it’s ok to scan millions of books into a computer system because it’s just another kind of learning.
If famous actors could sue over the use of a less-famous actor that sounds just like them, what's to stop less-famous actors from suing over the use of a famous actor who sounds just like them in big-budget movies? ... and that's when you discover that "unique voice" is a one-in-a-million thing and thousands of people have the same voice, all asking for their payout.
> what's to stop less-famous actors from suing over the use of a famous actor who sounds just like them in big-budget movies?
Not having idiots (or ChatGPT) for judges.
12 replies →
Nobody stops anyone from suing, but the less-famous actor would have to make a plausible case that the big-budget movie intended to copy the voice of the less-famous actor.
> for example, if a human can read one book to learn something, then it’s ok to scan millions of books into a computer system because it’s just another kind of learning.
Since this comes up all the time, I ask: What exactly is the number of books a human can ingest before it becomes illegal?
This is a bit like someone saying they don't want cars traveling down the sidewalk because they're too big and heavy, and then having someone ask how big and heavy a person needs to get before it becomes illegal for them to travel down the sidewalk.
It misses the point, which is that cars aren't people. Arguments like "well a car uses friction to travel along the ground and fuel to create kinetic energy, just like humans do", aren't convincing to me. An algorithm is not a human, and we should stop pretending the same rules apply to each.
8 replies →
Depends on similarities between existing data and generative outputs, so minimum is zero. Humans are caught plagiarizing all the time.
4 replies →
For a human? Whatever they can consume within their natural life.
3 replies →
100 per second.
> Surely there’s some kind of difference between “voice impression for a two-line cameo in one episode of an animated sitcom” and “reproducing your voice as the primary interface for a machine that could be used by billions of people and is worth hundreds of billions of dollars.”
There are too many differences to understand what you're saying. Is the problem too much money is in the company doing it? Fox is also pretty wealthy.
I think the pertinent question is: does having it sound like Scarlett Johansenn mean they get to access billions of people? If not, then while she might get paid out a few million, it'll be from OpenAI's marketing budget and not because of actual value added.
How unique is a voice? I'm sure there's enough people out ther who sounds like Johansson. There's probably some argument for voice + personality + face + mannerisms, some gestalt that's more comparable to copying the likeness "person". But openAI is copying a fictional character played by Johansson, it's not her. Do actor/esses get to monopolize their depiction of fictional characters? Especially when it's not tied to physical represenation. What if OpenAI associate it with an avatar that looks nothing like her. I'm sure hollywood and/or actors union is figuring this out.
> “Do actor/esses get to monopolize their depiction of fictional characters? Especially when it's not tied to physical represenation.”
If Annapurna Pictures (the production company that owns the rights to “Her”) made a sequel where the voice AI is played by someone else than Johansson but sounded the same and was marketed as a direct continuation, I think there would be a lawsuit.
She didn’t write the script or develop the character, but I think there’s enough creative authorship in her voice portrayal that it would be risky for the production company.
4 replies →
"How unique is X" is something we can start to get quantitative answers to with strong machine learning models, and for most things people care about, it seems like the answer is "not very unique at all".
1 reply →
Fair use[1]
[1]: https://en.wikipedia.org/wiki/Fair_use
It's not a fallacy. Behind the AI are 180M users inputting their own problems and giving their guidance. Those millions of books only teach language skills they are not memorized verbatim except rare instances of duplicated text in the training set. There is not enough space to store 10 trillion tokens in a model.
And if we wanted to replicate copyrighted text with a LLM, it would still be a bad idea, better to just find a copy online, faster and more precise, and usually free. We here are often posting paywalled articles in the comments, it's so easy to circumvent the paywalls we don't even blink twice at it.
Using LLMs to infringe is not even the intended purpose, and it only happens when the user makes a special effort to prompt the model with the first paragraph.
What I find offensive is restricting the circulation of ideas under the guise of copyright. In fact copyright should only protect expression not the underlying ideas and styles, those are free to learn, and AIs are just an extension of their human users.
I know there is some exceptions in US law for use of parody ???.
Sure, but what does that have to do with this?
> I was perusing some Simpsons clips this afternoon and came across a story to the effect of "So and so didn't want to play himself, so Dan Castellaneta did the voice."
IANAL, but parody and criticism are covered under Fair Use doctrine for Copyright law in the United States [1]. The Simpsons generally falls into that category, which is why they rarely get into trouble.
[1] https://en.wikipedia.org/wiki/Fair_use
The current system where you're allowed to "exploit" other people's image, but only if it's parody seems like a bit of an absurd loophole. Arnold as president in the Simpsons is okay, but Arnold as AI generated president in an action movie - suddenly not okay
Both arguably contributing the same minuscule amount to the "public discourse"..
That example isn't really pertinent, because in the case of the Simpsons it's fairly certain that the actors and actresses sign away the rights to their likeness to the company, otherwise there'd be major issues if one ever quit, became unable to work, just wanted a bunch of money, or whatever. There's probably some poor analogy with how if you write software, your company [generally] owns it.
For something more general look at Midler vs Ford [1], and lots of other similar cases. Ford wanted to use get Midler to sing some of her songs (that Ford owned the copyright to) for a commercial. She refused, so they hired an impersonator. They never stated it was Midler in the commercial, but nonetheless were sued and lost for abuse of 'rights of personality' even for content they owned the copyright to! Uncopyrightable characteristics highly associated with a person are still legally protected. Similar stuff with fight refs. Various trademark lines like 'Let's get it on!' or 'Let's get readddy to rumble.' are literally trademarked, but it's probably not even strictly necessary since it would be implicitly protected by rights of personality.
[1] https://en.wikipedia.org/wiki/Midler_v._Ford_Motor_Co.
I know it’s pendantic, but Ford did not own the copyright to either the original Bette Midler performance recording nor the lyrics/melody of the song. The marketing company that prepared the ‘Yuppie Campaign’ for Ford did negotiate a license for the lyrics/melody from the copyright holder. It doesn’t make a substantial difference, but commenters have been using wide ranging analogies in this thread and I wanted to make sure nobody jumped on a flawed foundation when arguing about the precedent case.
This sort of thing happens a lot, and is of course legal even if it isn't polite. I remember a decade or so ago when having "celebrity" voices for your GPS was a thing and there was an interview by the actor Michael Caine about how some company wanted him to do a GPS voice but he declined and later he found out that they then used an impersonator to make a voice that obviously was supposed to be his.
Just to clarify for people who don't read it, the article isn't claiming this was trained on the voice of someone doing a Scarlett Johannson impression. It says it was trained on the natural voice of someone who sounds similar to Johannson's, hired months before Altman reached out to her.
Who cares about this nothingburger
Lots of people, apparently: One of the biggest threads I've seen on here in the last weeks (months?). So not a "nothingburger".
I had similar thoughts based on a podcast I listened to once about voice actors hired for film spin off merchandise and whatnot. It's very common to look for voices that approximate a fictional character's voice, that was originally done by a different actor or actress.
Thinking about that episode, I imagine the legal risk is less in trying to sound like Scarlett Johansson, and more in trying to sound like Samantha, the AI character in Her. Warner Brothers or Spike Jonze probably has some legal rights to the character, and an argument could be made that OpenAI was infringing on that. The viability of that argument probably depends on how much people conflate the two or believe that the one was meant to represent the other.
Parody is protected in the US. The Simpsons can get away with a lot of stuff because of it
"The agent, who spoke on the condition of anonymity to assure the safety of her client, said the actress confirmed that neither Johansson nor the movie “Her” were ever mentioned by OpenAI."
But will either be mentioned by ChatGPT?