Apparently, stealing other people’s work to create product for money is now “fair use” as according to OpenAI because they are “innovating” (stealing). Yeah. Move fast and break things, huh?
“Because copyright today covers virtually every sort of human expression—including blogposts, photographs, forum posts, scraps of software code, and government documents—it would be impossible to train today’s leading AI models without using copyrighted materials,” wrote OpenAI in the House of Lords submission.
OpenAI claimed that the authors in that lawsuit “misconceive[d] the scope of copyright, failing to take into account the limitations and exceptions (including fair use) that properly leave room for innovations like the large language models now at the forefront of artificial intelligence.”
Or, or, or, hear me out:
Maybe their particular approach to making an AI is flawed.
Its like people do not know that there are many different kinds of ways that attempt to do AI.
Many of them do not rely on basically a training set that is the cumulative sum of all human generated content of every imaginable kind.
Well, off the top of my head:
Whole Brain Emulation, attempting to model a human brain as physically accurately as possible inside a computer.
Genetic Iteration (not the correct term for it but it escapes me at the moment), where you set up a simulated environment for digital actors, then simulate quasi-neurons, quasi-body parts dictated by quasi-dna, in a way that mimics actual biological natural selection and evolution, and then you run the simulation millions of times until your digital creature develops a stable survival strategy.
Similar approaches to this have been used to do things like teach an AI humanoid how to develop its own winning martial arts style via many many iterations, starting from not even being able to stand up, much less do anything to an opponent.
Both of these approaches obviously have drawbacks and strengths, and could possibly be successful at far more than what they have achieved to date, or maybe not, due to known or existing problems, but neither of them rely on a training set of essentially the entirety of all content on the internet.
That sounds like a great idea for making an intelligent agent inside a video game, where you control all aspects of it’s environment. But what about an AI that you want to be able to interact with our current shared reality. If I want to know something that involves synthesis of multiple modalities of knowledge how should that information be conveyed? Do humans grow up inside test tubes that only consume content that they themselves have created? Can you imagine the strange society we would have if people were unleashed upon the world without having any shared experiences until they were fully adults?
I think the OpenAI people have a point here, but I think where they go off the rails is that they expect all of this copyrighted information to be granted to them at zero cost and with zero responsibility to the creators of said content.
🤖 I’m a bot that provides automatic summaries for articles:
Click here to see the summary
Further, OpenAI writes that limiting training data to public domain books and drawings “created more than a century ago” would not provide AI systems that “meet the needs of today’s citizens.”
OpenAI responded to the lawsuit on its website on Monday, claiming that the suit lacks merit and affirming its support for journalism and partnerships with news organizations.
OpenAI’s defense largely rests on the legal principle of fair use, which permits limited use of copyrighted content without the owner’s permission under specific circumstances.
“Training AI models using publicly available internet materials is fair use, as supported by long-standing and widely accepted precedents,” OpenAI wrote in its Monday blog post.
In August, we reported on a similar situation in which OpenAI defended its use of publicly available materials as fair use in response to a copyright lawsuit involving comedian Sarah Silverman.
OpenAI claimed that the authors in that lawsuit “misconceive[d] the scope of copyright, failing to take into account the limitations and exceptions (including fair use) that properly leave room for innovations like the large language models now at the forefront of artificial intelligence.”
Saved 58% of original text.
I will repeat what I have proffered before:
If OpenAI stated that it is impossible to train leading AI models without using copyrighted material, then, unpopular as it may be, the preemptive pragmatic solution should be pretty obvious, enter into commercial arrangements for access to said copyrighted material.
Claiming a failure to do so in circumstances where the subsequent commercial product directly competes in a market seems disingenuous at best, given what I assume is the purpose of copyrighted material, that being to set the terms under which public facing material can be used. Particularly if regurgitation of copyrighted material seems to exist in products inadequately developed to prevent such a simple and foreseeable situation.
Yes I am aware of the USA concept of fair use, but the test of that should be manifestly reciprocal, for example would Meta allow what it did to MySpace, hack and allow easy user transfer, or Google with scraping Youtube.
To me it seems Big Tech wants its cake and to eat it, where investor $$$ are used to corrupt open markets and undermine both fundamental democratic State social institutions, manipulate legal processes, and undermine basic consumer rights.
Agreed.
There is nothing “fair” about the way Open AI steals other people’s work. ChatGPT is being monetized all over the world and the large number of people whose work has not been compensated will never see a cent of that money.
At the same time the LLM will be used to replace (at least some of ) the people who created those works in the first place.
Tech bros are disgusting.
At the same time the LLM will be used to replace (at least some of ) the people who created those works in the first place.
This right here is the core of the moral issue when it comes down to it, as far as I’m concerned. These text and image models are already killing jobs and applying downward pressure on salaries. I’ve seen it happen multiple times now, not just anecdotally from some rando on an internet comment section.
These people losing jobs and getting pay cuts are who created the content these models are siphoning up. People are not going to like how this pans out.
The flip side of this is that many artists who simply copy very popular art styles are now functionally irrelevant, as it is now just literally proven that this kind of basically plagiarism AI is entirely capable of reproducing established styles to a high degree of basically fidelity.
While many aspects of this whole situation are very bad for very many reasons, I am actually glad that many artists will be pressured to actually be more creative than an algorithm, though I admit this comes from basically a personally petty standpoint of having known many, many, many mediocre artists who themselves and their fans treat like gods because they can emulate some other established style.
Any company replacing humans with AI is going to regret it. AI just isn’t that good and probably won’t ever be, at least in it’s current form. It’s all an illusion and is destined to go the way of Bitcoin, which is to say it will shoot up meteorically and seem like the answer to all kinds of problems, and then the reality will sink in and it will slowly fade to obscurity and irrelevance. That doesn’t help anyone affected today, of course.
Tech bros are disgusting.
That’s not even getting into the fraternity behavior at work, hyper-reactionary politics and, er, concerning age preferences.
Yup. I said it in another discussion before but think its relevant here.
Tech bros are more dangerous than Russian oligarchs. Oligarchs understand the people hate them so they mostly stay low and enjoy their money.
Tech bros think they are the savior of the world while destroying millions of people’s livelihood, as well as destroying democracy with their right wing libertarian politics.
I suspect the US government will allow OpenAI to continue doing as it please to keep their competitive advantage in AI over China (which don’t have problem with using copyrighted materials to train their models). They already limit selling AI-related hardware to keep their competitive advantage, so why stop there? Might as well allow OpenAI to continue using copyrighted materials to keep the competitive advantage.
Yep, completely agree.
Case in point: Steam has recently clarified their policies of using such Ai generated material that draws on essentially billions of both copyrighted and non copyrighted text and images.
To publish a game on Steam that uses AI gen content, you now have to verify that you as a developer are legally authorized to use all training material for the AI model for commercial purposes.
This also applies to code and code snippets generated by AI tools that function similarly, such as CoPilot.
So yeah, sorry, either gotta use MIT liscensed open source code or write your own, and you gotta do your own art.
I imagine this would also prevent you from using AI generated voice lines where you trained the model on basically anyone who did not explicitly consent to this as well, but voice gen software that doesnt use the ‘train the model on human speakers’ approach would probably be fine assuming you have the relevant legal rights to use such software commercially.
Not 100% sure this is Steam’s policy on voice gen stuff, they focused mainly on art dialogue and code in their latest policy update, but the logic seems to work out to this conclusion.
Do musicians not buy the music that they want to listen to? Should they be allowed to torrent any MP3 they want just because they say it’s for their instrument learning?
I mean I’d be all for it, but that’s not what these very same corporations (including Microsoft when it comes to software) wanted back during Napster times. Now they want a separate set of rules just for themselves. No! They get to follow the same laws they force down our throats.
Some relevant comments from Ars:
leighno5
The absolute hubris required for OpenAI here to come right out and say, ‘Yeah, we have no choice but to build our product off the exploitation of the work others have already performed’ is stunning. It’s about as perfect a representation of the tech bro mindset that there can ever be. They didn’t even try to approach content creators in order to do this, they just took what they needed because they wanted to. I really don’t think it’s hyperbolic to compare this to modern day colonization, or worker exploitation. ‘You’ve been working pretty hard for a very long time to create and host content, pay for the development of that content, and build your business off of that, but we need it to make money for this thing we’re building, so we’re just going to fucking take it and do what we need to do.’
The entitlement is just…it’s incredible.
4qu4rius
20 years ago, high school kids were sued for millions & years in jail for downloading a single Metalica album (if I remember correctly minimum damage in the US was something like 500k$ per song).
All of a sudden, just because they are the dominant ones doing the infringment, they should be allowed to scrap the entire (digital) human knowledge ? Funny (or not) how the law always benefits the rich.
