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.”
Try to train a human comedian to make jokes without ever allowing him to hear another comedian’s jokes, never watching a movie, never reading a book or magazine, never watching a TV show. I expect the jokes would be pretty weak.
A comedian isn’t forming a sentence based on what the most probable word is going to appear after the previous one. This is such a bullshit argument that reduces human competency to “monkey see thing to draw thing” and completely overlooks the craft and intent behind creative works. Do you know why ChatGPT uses certain words over others? Probability. It decided as a result of its training that one word would appear after the previous in certain contexts. It absolutely doesn’t take into account things like “maybe this word would be better here because the sound and syllables maintains the flow of the sentence”.
Baffling takes from people who don’t know what they’re talking about.
I wish I could upvote this more than once.
What people always seem to miss is that a human doesn’t need billions of examples to be able to produce something that’s kind of “eh, close enough”. Artists don’t look at billions of paintings. They look at a few, but do so deeply, absorbing not just the most likely distribution of brushstrokes, but why the painting looks the way it does. For a basis of comparison, I did an art and design course last year and looked at about 300 artworks in total (course requirement was 50-100). The research component on my design-related degree course is one page a week per module (so basically one example from the field the module is about, plus some analysis). The real bulk of the work humans do isn’t looking at billions of examples: it’s looking at a few, and then practicing the skill and developing a process that allows them to convey the thing they’re trying to express.
If the AI models were really doing exactly the same thing humans do, the models could be trained without any copyright infringement at all, because all of the public domain and creative commons content, plus maybe licencing a little more, would be more than enough.
Exactly! You can glean so much from a single work, not just about the work itself but who created it and what ideas were they trying to express and what does that tell us about the world they live in and how they see that world.
This doesn’t even touch the fact that I’m learning to draw not by looking at other drawings but what exactly I’m trying to draw. I know at a base level, a drawing is a series of shapes made by hand whether it’s through a digital medium or traditional pen/pencil and paper. But the skill isn’t being able replicate other drawings, it’s being able to convert something I can see into a drawing. If I’m drawing someone sitting in a wheelchair, then I’ll get the pose of them sitting in the wheelchair but I can add details I want to emphasise or remove details I don’t want. There’s so much that goes into creative work and I’m tired of arguing with people who have no idea what it takes to produce creative works.
It seems that most of the people who think what humans and AIs do is the same thing are not actually creatives themselves. Their level of understanding of what it takes to draw goes no further than “well anyone can draw, children do it all the time”. They have the same respect for writing, of course, equating the ability to string words together to write an email, with the process it takes to write a brilliant novel or script. They don’t get it, and to an extent, that’s fine - not everybody needs to understand everything. But they should at least have the decency to listen to the people that do get it.
Well, that’s not me. I’m a creative, and I see deep parallels between how LLMs work and how my own mind works.
Either you’re vastly overestimating the degree of understanding and insight AIs possess, or you’re vastly underestimating your own capabilities. :)
This whole AI craze has just shown me that people are losing faith in their own abilities and their ability to learn things. I’ve heard so many who use AI to generate “artwork” argue that they tried to do art “for years” without improving, and hence have come to conclusion that creativity is a talent that only some have, instead of a skill you can learn and hone. Just because they didn’t see results as fast as they’d have liked.
Alternatively, you might be vastly overestimating human “understanding and insight”, or how much of it is really needed to create stuff.
Children learn by watching others. We are trained from millions of examples starting from before birth.
When people say that the “model is learning from its training data”, it means just that, not that it is human, and not that it learns exactly humans. It doesn’t make sense to judge boats on how well they simulate human swimming patterns, just how well they perform their task.
Every human has the benefit of as a baby training on things around them and being trained by those around them, building a foundation for all later skills. Generative models rely on many text and image pairs to describe things to them because they lack the ability to poke, prod, rotate, and disassemble for themselves.
For example, when a model takes in a thousand images of circles, it doesn’t “learn” a thousand circles. It learns what circle GENERALLY is like, the concept of it. That representation, along with random noise, is how you create images with them. The same happens for every concept the model trains on. Everything from “cat” to more complex things like color relationships and reflections or lighting. Machines are not human, but they can learn despite that.
In general I agree with you, but AI doesn’t learn the concept of what a circle is. AI reproduces the most fitting representation of what we call a circle. But there is no understanding of the concept of a circle. This may sound nit picking, but I think it’s important to make the distinction.
That is why current models aren’t regarded as actual intelligence, although people already call them that…
I understand. I didn’t mean to imply any sort of understanding with the language I used.
It makes sense to judge how closely LLMs mimic human learning when people are using it as a defense to AI companies scraping copyrighted content, and making the claim that banning AI scraping is as nonsensical as banning human learning.
But when it’s pointed out that LLMs don’t learn very similarly to humans, and require scraping far more material than a human does, suddenly AIs shouldn’t be judged by human standards? I don’t know if it’s intentional on your part, but that’s a pretty classic example of a motte-and-bailey fallacy. You can’t have it both ways.
I don’t understand what you mean, can you elaborate?
What you count as “one” example is arbitrary. In terms of pixels, you’re looking at millions right now.
The ability to train faster using fewer examples in real time, similar to what an intelligent human brain can do, is definitely a goal of AI research. But right now, we may be seeing from AI what a below average human brain could accomplish with hundreds of lifetimes to study.
If the AI models were really doing exactly the same thing humans do, the models could be trained without any copyright infringement at all, because all of the public domain and creative commons content, plus maybe licencing a little more, would be more than enough.
I mean, no, if you only ever look at public domain stuff you literally wouldn’t know the state of the art, which is historically happening for profit. Even the most untrained artist “doing their own thing” watches Disney/Pixar movies and listens to copyrighted music.
If we’re going by the number of pixels being viewed, then you have to use the same measure for both humans and AIs - and because AIs have to look at billions of images while humans do not, the AI still requires far more pixels than a human does.
And humans don’t require the most modern art in order to learn to draw at all. Sure, if they want to compete with modern artists, they would need to look at modern artists (for which educational fair use exists, and again the quantity of art being used by the human for this purpose is massively lower than what an AI uses - a human does not need to consume billions of artworks from modern artists in order to learn what the current trends are). But a human could learn to draw, paint, sculpt, etc purely by only looking at public domain and creative commons works, because the process for drawing, say, the human figure (with the right number of fingers!) has not changed in hundreds of years. A human can also just… go outside and draw things they see themselves, because the sky above them and the tree across the street aren’t copyrighted. And in fact, I’d argue that a good artist should go out and find real things to draw.
OpenAI’s argument is literally that their AI cannot learn without using copyrighted materials in vast quantities - too vast for them to simply compensate all the creators. So it genuinely is not comparable to a human, because humans can, in fact, learn without using copyrighted material. If OpenAI’s argument is actually that their AI can’t compete commercially with modern art without using copyrighted works, then they should be honest about that - but then they’d be showing their hand, wouldn’t they?
Sure, if they want to compete with modern artists, they would need to look at modern artists
Which is the literal goal of Dall-E, SD, etc.
But a human could learn to draw, paint, sculpt, etc purely by only looking at public domain and creative commons works
They could definitely learn some amount of skill, I agree. I’d be very interested to see the best that an AI could achieve using only PD and CC content. It would be interesting. But you’d agree that it would look very different from modern art, just as an alien who has only been consuming earth media from 100+ years ago would be unable to relate to us.
the sky above them and the tree across the street aren’t copyrighted.
Yeah, I’d consider that PD/CC content that such an AI would easily have access to. But obviously the real sky is something entirely different from what is depicted in Starry Night, Star Wars, or H.P. Lovecraft’s description of the cosmos.
OpenAI’s argument is literally that their AI cannot learn without using copyrighted materials in vast quantities
Yeah, I’d consider that a strong claim on their part; what they really mean is, it’s the easiest way to make progress in AI, and we wouldn’t be anywhere close to where we are without it.
And you could argue “convenient that it both saves them money, and generates money for them to do it this way”, but I’d also point out that the alternative is they keep the trained models closed source, never using them publicly until they advance the tech far enough that they’ve literally figured out how to build/simulate a human brain that is able to learn as quickly and human-like as you’re describing. And then we find ourselves in a world where one or two corporations have this incredible proprietary ability that no one else has.
Personally, I’d rather live in the world where the information about how to do all of this isn’t kept for one or two corporations to profit from, I would rather live in the version where they publish their work publicly, early, and often, show that it works, and people are able to reproduce it, open source it, train their own models, and advance the technology in a space where anyone can use it.
You could hypothesize of a middle ground where they do the research, but aren’t allowed to profit from it without licensing every bit of data they train on. But the reality of AI research is that it only happens to the extent that it generates revenue. It’s been that way for the entire history of AI. Douglas Hofstadter has been asking deep important questions about AI as it relates to consciousness for like 60 years (ex. GEB, I am a Strange Loop), but there’s a reason he didn’t discover LLMs and tech companies did. That’s not to say his writings are meaningless, in fact I think they’re more important than ever before, but he just wasn’t ever going to get to this point with a small team of grad students, a research grant, and some public domain datasets.
So, it’s hard to disagree with OpenAI there, AI definitely wouldn’t be where it is without them doing what they’ve done. And I’m a firm believer that unless we figure our shit out with energy generation soon, the earth will be an uninhabitable wasteland. We’re playing a game of climb the Kardashev scale, we opted for the “burn all the fossil fuels as fast as possible” strategy, and now we’re a the point where either spent enough energy fast enough to figure out the tech needed to survive this, or we suffocate on the fumes. The clock is ticking, and AI may be our best bet at saving the human race that doesn’t involve an inordinate number of people dying.
OpenAI are not going to make the source code for their model accessible to all to learn from. This is 100% about profiting from it themselves. And using copyrighted data to create open source models would seem to violate the very principles the open source community stands for - namely that everybody contributes what they agree to, and everything is published under a licence. If the basis of an open source model is a vast quantity of training data from a vast quantity of extremely pissed off artists, at least some of the people working on that model are going to have a “are we the baddies?” moment.
The AI models are also never going to produce a solution to climate change that humans will accept. We already know what the solution is, but nobody wants to hear it, and expecting anyone to listen to ChatGPT and suddenly change their minds about using fossil fuels is ludicrous. And an AI that is trained specifically on knowledge about the climate and technologies that can improve it, with the purpose of innovating some hypothetical technology that will fix everything without humans changing any of their behaviour, categorically does not need the entire contents of ArtStation in its training data. AIs that are trained to do specific tasks, like the ones trained to identify new antibiotics, are trained on a very limited set of data, most of which is not protected by copyright and any that is can be easily licenced because the quantity is so small - and you don’t see anybody complaining about those models!
OpenAI are not going to make the source code for their model accessible to all to learn from
OpenAI isn’t the only company doing this, nor is their specific model the knowledge that I’m referring to.
The AI models are also never going to produce a solution to climate change that humans will accept.
We already know what the solution is, but nobody wants to hear it
Then it’s not a solution. That’s like telling your therapist, “I know how to fix my relationship, my partner just won’t do it!”
expecting anyone to listen to ChatGPT and suddenly change their minds about using fossil fuels is ludicrous
Lol. Yeah, I agree, that’s never going to work.
categorically does not need the entire contents of ArtStation in its training data.
That’s a strong claim to make. Regardless of the ethics involved, or the problems the AI can solve today, the fact is we seeing rapid advances in AI research as a direct result of these ethically dubious models.
In general, I’m all for the capitalist method of artists being paid their fair share for the work they do, but on the flip side, I see a very possible mass extinction event on the horizon, which could cause suffering the likes of which humanity has never seen. If we assume that is the case, and we assume AI has a chance of preventing it, then I would prioritize that over people’s profits today. And I think it’s perfectly reasonable to say I’m wrong.
And then there’s the problem of actually enforcing any sort of regulation, which would be so much more difficult than people here are willing to admit. There’s basically nothing you can do even if you wanted to. Your Carlin example is exactly the defense a company would use: “I guess our AI just happened to create a movie that sounds just like Paul Blart, but we swear it’s never seen the film. Great minds think alike, I guess, and we sell only the greatest of minds”.
It isn’t wrong to use copyrighted works for training. Let me quote an article by the EFF here:
and
What you want would swing the doors open for corporate interference like hindering competition, stifling unwanted speech, and monopolization like nothing we’ve seen before. There are very good reasons people have these rights, and we shouldn’t be trying to change this. Ultimately, it’s apparent to me, you are in favor of these things. That you believe artists deserve a monopoly on ideas and non-specific expression, to the detriment of anyone else. If I’m wrong, please explain to me how.
If we’re going by the number of pixels being viewed, then you have to use the same measure for both humans and AIs - and because AIs have to look at billions of images while humans do not, the AI still requires far more pixels than a human does.
Humans benefit from years of evolutionary development and corporeal bodies to explore and interact with their world before they’re ever expected to produce complex art. AI need huge datasets to understand patterns to make up for this disadvantage. Nobody pops out of the womb with fully formed fine motor skills, pattern recognition, understanding of cause and effect, shapes, comparison, counting, vocabulary related to art, and spatial reasoning. Datasets are huge and filled with image-caption pairs to teach models all of this from scratch. AI isn’t human, and we shouldn’t judge it against them, just like we don’t judge boats on their rowing ability.
And humans don’t require the most modern art in order to learn to draw at all. Sure, if they want to compete with modern artists, they would need to look at modern artists (for which educational fair use exists, and again the quantity of art being used by the human for this purpose is massively lower than what an AI uses - a human does not need to consume billions of artworks from modern artists in order to learn what the current trends are). But a human could learn to draw, paint, sculpt, etc purely by only looking at public domain and creative commons works, because the process for drawing, say, the human figure (with the right number of fingers!) has not changed in hundreds of years. A human can also just… go outside and draw things they see themselves, because the sky above them and the tree across the street aren’t copyrighted. And in fact, I’d argue that a good artist should go out and find real things to draw.
AI don’t require most modern art in order to learn to make images either, but the range of expression would be limited, just like a human’s in this situation. You can see this in cave paintings and early sculptures. They wouldn’t be limited to this same degree, but you would still be limited.
It took us 100,000 years to get from cave drawings to Leonard Da Vinci. This is just another step for artists, like Camera Obscura was in the past. It’s important to remember that early man was as smart as we are, they just lacked the interconnectivity to exchange ideas that we have.
I think the difference in artistic expression between modern humans and humans in the past comes down to the material available (like the actual material to draw with).
Humans can draw without seeing any image ever. Blind people can create art and draw things because we have a different understanding of the world around us than AI has. No human artist needs to look at a thousand or even at 1 picture of a banana to draw one.
The way AI sees and “understands” the world and how it generates an image is fundamentally different from how the human brain conveys the object banana into an image of a banana.
I think the difference in artistic expression between modern humans and humans in the past comes down to the material available (like the actual material to draw with).
That is definitely a difference, but even that is a kind of information shared between people, and information itself is what gives everyone something to build on. That gives them a basis on which to advance understanding, instead of wasting time coming up with the same things themselves every time.
Humans can draw without seeing any image ever. Blind people can create art and draw things because we have a different understanding of the world around us than AI has. No human artist needs to look at a thousand or even at 1 picture of a banana to draw one.
Humans don’t need representations of things in images because they have the opportunity to interact with the genuine article, and in situations when that is impractical, they can still fall back on images to learn. Someone without sight from birth can’t create art the same way a sighted person can.
The way AI sees and “understands” the world and how it generates an image is fundamentally different from how the human brain conveys the object banana into an image of a banana.
That’s the beauty of it all, despite that, these models can still output bananas.
Humans learn mostly from real life. Go touch some grass
When you look at one painting, is that the equivalent of one instance of the painting in the training data? There is an infinite amount of information in the painting, and each time you look you process more of that information.
I’d say any given painting you look at in a museum, you process at least a hundred mental images of aspects of it. A painting on your wall could be seen ten thousand times easily.
That’s what humans do, though. Maybe not probability directly, but we all know that some words should be put in a certain order. We still operate within standard norms that apply to aparte group of people. LLM’s just go about it in a different way, but they achieve the same general result. If I’m drawing a human, that means there’s a ‘hand’ here, and a ‘head’ there. ‘Head’ is a weird combination of pixels that mostly look like this, ‘hand’ looks kinda like that. All depends on how the model is structured, but tell me that’s not very similar to a simplified version of how humans operate.
Yeah but the difference is we still choose our words. We can still alter sentences on the fly. I can think of a sentence and understand verbs go after the subject but I still have the cognition to alter the sentence to have the effect I want. The thing lacking in LLMs is intent and I’m yet to see anyone tell me why a generative model decides to have more than 6 fingers. As humans we know hands generally have five fingers and there’s a group of people who don’t so unless we wanted to draw a person with a different number of fingers, we could. A generative art model can’t help itself from drawing multiple fingers because all it understands is that “finger + finger = hand” but it has no concept on when to stop.
And that’s the reason why LLM generated content isn’t considered creative.
I do believe that the person using the device has a right to copyright the unique method they used to generate the content, but the content itself isn’t anything worth protecting.
You say that yet I initially responded to someone who was comparing an LLM to what a comedian does.
There is no unique method because there’s hardly anything unique you can do. Two people using Stable Diffusion to produce an image are putting in the same amount of work. One might put more time into crafting the right prompt but that’s not work you’re doing.
If 90% of the work is handled by the model, and you just layer on whatever extra thing you wanted, that doesn’t mean you created the thing. That also implies you have much control over the output. You’re effectively negotiating with this machine to produce what you want.
Wouldn’t that lead to the same argument as originally brought against photography, though?
A photographer is effectively negotiating with the sun, the sky and everything else to hopefully get the result they are looking for on their device.
One difference is that the photographer has to go the places they’re taking pictures of.
Another is that photography isn’t comparable to paintings and it never has been. I’m willing to bet photography and paintings have never coexisted in a contest. Except, when people say their generative art is comparable to what artists have been producing by hand, they are admitting that generative art has more in common with photography than it does with hand-crafted art but they want the prestige and recognition those artists get for their work.
more time into crafting the right prompt
Thats not work to you? My company pays me to spend time to do the right thing, even though most of the work does the computer.
I see where you are going at, but your argument also invalidates other forms of human interaction and creating.
In my country copyright can only be granted if a certain amount of (human) work went into something. Any work.
The difficult part is finding out whats enough and what kind of work qualify to lead to some kind of protection, even if partial.
The difficult part was not to create something, but to prove someone did or didnt put enough work into it.
I think we can hold generated or assisted goods to the same standard.Putting a simple prompt together should probably not be granted protection as no significant work went into it. But refining it, editing the result… maybe thats enough, thats really up to the society to decide.
At the same time we have to balance the power of machines against human work, so the human work doesnt get totally invalidated, but rather shifted and treated as sub-type.
Machines already replaced alot of work, also creative ones. Book-printing, forging, producing food… the scary part about generative AI is mainly the speed of them spreading.So as a data analyst a lot of my work is done through a computer but I can apply my same skills if someone hands me a piece of paper with data printed on it and told me to come up with solutions to the problems with it. I don’t need the computer to do what I need to do, it makes it easier to manipulate data but the degree of problem solving required needs to be done by a human and that’s why it’s my job. If a machine could do it, then they would be doing it but they aren’t because contrary to what people believe about data analysis, you have to be somewhat creative to do it well.
Crafting a prompt is an exercise in trial and error. It’s work but it’s not skilled work. It doesn’t take talent or practice to do. Despite the prompt, you are still at the mercy of the machine.
Even by the case you’ve presented, I have to ask, at what point of a human editing the output of a generative model constitutes it being your own work and not the machine’s? How much do you have to change? Can you give me a %?
Machines were intended to automate the tedious tasks that we all have to suffer to free up our brains for more engaging things which might include creative pursuits. Automation exists to make your life easier, not to rob you of life’s pursuits or your livelihood. It never should’ve been used to produce creative work and I find the attempts to equate this abomination’s outputs to what artists have been doing for years, utterly deplorable.
I don’t choose my words man. I get a vague sense of the meaning I want to convey and the words just form themselves.
As an artist you draw with an understanding of the human body, though. An understanding current models don’t have because they aren’t actually intelligent.
Maybe when a human is an absolute beginner in drawing they will think about the different lines and replicate even how other people draw stuff that then looks like a hand.
But eventually they will realise (hopefully, otherwise they may get frustrated and stop drawing) that you need to understand the hand to draw one. It’s mass, it’s concept or the idea of what a hand is.
This may sound very abstract and strange but creative expression is more complex than replicating what we have seen a million times. It’s a complex function unique to the human brain, an organ we don’t even scientifically understand yet.
That’s not the point though. The point is that the human comedian and the AI both benefit from consuming creative works covered by copyright.
Yeah except a machine is owned by a company and doesn’t consume the same way. It breaks down copyrighted works into data points so it can find the best way of putting those data points together again. If you understand anything at all about how these models work, they do not consume media the same way we do. It is not an entity with a thought process or consciousness (despite the misleading marketing of “AI” would have you believe), it’s an optimisation algorithm.
It’s a glorified autocomplete.
It’s so funny that this is something new. This was Grammarly’s whole schtick since before ChatGPT so how different is Grammarly AI?
Here is the bigger picture: The vast majority of tech illiterate people think something is AI because duh its called AI.
Its literally just the power of branding and marketing on the minds of poorly informed humans.
Unfortunately this is essentially a reverse Turing Test.
The vast majority of humans do not know anything about AI, and also a huge majority of them can also barely tell the difference between, currently in some but not all forms, output from what is basically a brute force total internet plagiarism and synthesis software, from many actual human created content in many cases.
To me this basically just means that about 99% of the time, most humans are actually literally NPCs, and they only do actual creative and unpredictable things very very rarely.
I call it AI because it’s artificial and it’s intelligent. It’s not that complicated.
The thing we have to remember is how scary and disruptive AI is. Given that fear, it is scary to acknowledge that we have AI emerging into our world. Because it is scary, that pushes us to want to ignore it.
It’s called denial, and it’s the best explanation for why people aren’t willing to acknowledge that LLMs are AI.
And human comedians regularly get called out when they outright steal others material and present it as their own.
The word for this is plagiarism.
And in OpenAIs framework, when used in a relevant commercial context, they are functionally operating and profiting off of the worlds most comprehensive plagiarism software.
They get called out when they use others work as a template, not as training data.
A comedian isn’t forming a sentence based on what the most probable word is going to appear after the previous one.
Neither is an LLM. What you’re describing is a primitive Markov chain.
You may not like it, but brains really are just glorified pattern recognition and generation machines. So yes, “monkey see thing to draw thing”, except a really complicated version of that.
Think of it this way: if your brain wasn’t a reorganization and regurgitation of the things you have observed before, it would just generate random noise. There’s no such thing as “truly original” art or it would be random noise. Every single word either of us is typing is the direct result of everything you and I have observed before this moment.
Baffling takes from people who don’t know what they’re talking about.
Ironic, to say the least.
The point you should be making, is that a corporation will make this above argument up to, but not including the point where they have to treat AIs ethically. So that’s the way to beat them. If they’re going to argue that they have created something that learns and creates content like a human brain, then they should need to treat it like a human, ensure it is well compensated, ensure it isn’t being overworked or enslaved, ensure it is being treated “humanely”. If they don’t want to do that, if they want it to just be a well built machine, then they need to license all the proprietary data they used to build it. Make them pick a lane.
Neither is an LLM. What you’re describing is a primitive Markov chain.
My description might’ve been indicative of a Markov chain but the actual framework uses matrices because you need to be able to store and compute a huge amount of information at once which is what matrices are good for. Used in animation if you didn’t know.
What it actually uses is irrelevant, how it uses those things is the same as a regression model, the difference is scale. A regression model looks at how related variables are in giving an outcome and computing weights to give you the best outcome. This was the machine learning boom a couple of years ago and TensorFlow became really popular.
LLMs are an evolution of the same idea. I’m not saying it’s not impressive because it’s very cool what they were able to do. What I take issue with is the branding, the marketing and the plagiarism. I happen to be in the intersection of working in the same field, an avid fan of classic Sci-Fi and a writer.
It’s easy to look at what people have created throughout history and think “this looks like that” and on a point by point basis you’d be correct but the creation of that thing is shaped by the lens of the person creating it. Someone might make a George Carlin joke that we’ve heard recently but we’ll read about it in newspapers from 200 years ago. Did George Carlin steal the idea? No. Was he aware of that information? I don’t know. But Carlin regularly calls upon his own experiences so it’s likely that he’s referencing a event from his past that is similar to that of 200 years ago. He might’ve subconsciously absorbed the information.
The point is that the way these models have been trained is unethical. They used material they had no license to use and they’ve admitted that it couldn’t work as well as it does without stealing other people’s work. I don’t think they’re taking the position that it’s intelligent because from the beginning that was a marketing ploy. They’re taking the position that they should be allowed to use the data they stole because there was no other way.
branding
okay
the marketing
yup
the plagiarism
woah there! that’s where we disagree… your position is based on the fact that you believe that this is plagiarism - inherently negative
perhaps its best not use loaded language. if we want to have a good faith discussion, it’s best to avoid emotive arguments and language that’s designed to evoke negativity simply by their use, rather than the argument being presented
I happen to be in the intersection of working in the same field, an avid fan of classic Sci-Fi and a writer
its understandable that it’s frustrating, but just because a machine is now able to do a similar job to a human doesn’t make it inherently wrong. it might be useful for you to reframe these developments - it’s not taking away from humans, it’s enabling humans… the less a human has to have skill to get what’s in their head into an expressive medium for someone to consume the better imo! art and creativity shouldn’t be about having an ability - the closer we get to pure expression the better imo!
the less you have to worry about the technicalities of writing, the more you can focus on pure creativity
The point is that the way these models have been trained is unethical. They used material they had no license to use and they’ve admitted that it couldn’t work as well as it does without stealing other people’s work
i’d question why it’s unethical, and also suggest that “stolen” is another emotive term here not meant to further the discussion by rational argument
so, why is it unethical for a machine but not a human to absorb information and create something based on its “experiences”?
First of all, we’re not having a debate and this isn’t a courtroom so avoid the patronising language.
Second of all, my “belief” on the models’ plagiarism is based on technical knowledge of how the models work and not how I think they work.
a machine is now able to do a similar job to a human
This would be impressive if it was true. An LLM is not intelligent simply through its appearance of intelligence.
It’s enabling humans
It’s a chat bot that’s automated Google searches, let’s be clear about what this can do. It’s taken natural language processing and applied it through an optimisation algorithm to produce human-like responses.
No, I disagree at a fundamental level. Humans need to compete against each other and ourselves to improve. Just because an LLM can write a book for you, doesn’t mean you’ve written a book. You’re just lazy. You don’t want to put in the work any other writer in existence has done, to mull over their work and consider the emotions and effect they want to have on the reader. To what extent can an LLM replicate the way George RR Martin describes his world without entirely ripping off his work?
i’d question why it’s unethical, and also suggest that “stolen” is another emotive term here not meant to further the discussion by rational argument
If I take a book you wrote from you without buying it or paying you for it, what would you call that?
you know how the neurons in our brain work, right?
because if not, well, it’s pretty similar… unless you say there’s a soul (in which case we can’t really have a conversation based on fact alone), we’re just big ol’ probability machines with tuned weights based on past experiences too
You are spitting out basic points and attempting to draw similarities because our brains are capable of something similar. The difference between what you’ve said and what LLMs do is that we have experiences that we are able to glean a variety of information from. An LLM sees text and all it’s designed to do is say “x is more likely to appear before y than z”. If you fed it nonsense, it would regurgitate nonsense. If you feed it text from racist sites, it will regurgitate that same language because that’s all it has seen.
You’ll read this and think “that’s what humans do too, right?” Wrong. A human can be fed these things and still reject them. Someone else in this thread has made some good points regarding this but I’ll state them here as well. An LLM will tell you information but it has no cognition on what it’s telling you. It has no idea that it’s right or wrong, it’s job is to convince you that it’s right because that’s the success state. If you tell it it’s wrong, that’s a failure state. The more you speak with it, the more fail states it accumulates and the more likely it is to cutoff communication because it’s not reaching a success, it’s not giving you what you want. The longer the conversation goes on, the more crazy LLMs get as well because it’s too much to process at once, holding those contexts in its memory while trying to predict the next one. Our brains do this easily and so much more. To claim an LLM is intelligent is incredibly misguided, it is merely the imitation of intelligence.
but that’s just a matter of complexity, not fundamental difference. the way our brains work and the way an artificial neural network work aren’t that different; just that our brains are beyond many orders of magnitude bigger
there’s no particular reason why we can’t feed artificial neural networks an enormous amount of … let’s say tangentially related experiential information … as well, but in order to be efficient and make them specialise in the things we want, we only feed them information that’s directly related to the specialty we want them to perform
there’s some… “pre training” or “pre-existing state” that exists with humans too that comes from genetics, but i’d argue that’s as relevant to the actual task of learning, comprehension, and creating as a BIOS is to running an operating system (that is, a necessary precondition to ensure the correct functioning of our body with our brain, but not actually what you’d call the main function)
i’m also not claiming that an LLM is intelligent (or rather i’d prefer to use the term self aware because intelligent is pretty nebulous); just that the structure it has isn’t that much different to our brains just on a level that’s so much smaller and so much more generic that you can’t expect it to perform as well as a human - you wouldn’t expect to cut out 99% of a humans brain and have them be able to continue to function at the same level either
i guess the core of what i’m getting at is that the self awareness that humans have is definitely not present in an LLM, however i don’t think that self-awareness is necessarily a pre-requisite for most things that we call creativity. i think that’s it’s entirely possible for an artificial neural net that’s fundamentally the same technology that we use today to be able to ingest the same data that a human would from birth, and to have very similar outcomes… given that belief (and i’m very aware that it certainly is just a belief - we aren’t close to understanding our brains, but i don’t fundamentally thing there’s anything other then neurons firing that results in the human condition), just because you simplify and specialise the input data doesn’t mean that the process is different. you could argue that it’s lesser, for sure, but to rule out that it can create a legitimately new work is definitely premature
“Soul” is the word we use for something we don’t scientifically understand yet. Unless you did discover how human brains work, in that case I congratulate you on your Nobel prize.
You can abstract a complex concept so much it becomes wrong. And abstracting how the brain works to “it’s a probability machine” definitely is a wrong description. Especially when you want to use it as an argument of similarity to other probability machines.
“Soul” is the word we use for something we don’t scientifically understand yet
that’s far from definitive. another definition is
A part of humans regarded as immaterial, immortal, separable from the body at death
but since we aren’t arguing semantics, it doesn’t really matter exactly, other than the fact that it’s important to remember that just because you have an experience, belief, or view doesn’t make it the only truth
of course i didn’t discover categorically how the human brain works in its entirety, however most scientists i’m sure would agree that the method by which the brain performs its functions is by neurons firing. if you disagree with that statement, the burden of proof is on you. the part we don’t understand is how it all connects up - the emergent behaviour. we understand the basics; that’s not in question, and you seem to be questioning it
You can abstract a complex concept so much it becomes wrong
it’s not abstracted; it’s simplified… if what you’re saying were true, then simplifying complex organisms down to a petri dish for research would be “abstracted” so much it “becomes wrong”, which is categorically untrue… it’s an incomplete picture, but that doesn’t make it either wrong or abstract
*edit: sorry, it was another comment where i specifically said belief; the comment you replied to didn’t state that, however most of this still applies regardless
i laid out an a leads to b leads to c and stated that it’s simply a belief, however it’s a belief that’s based in logic and simplified concepts. if you want to disagree that’s fine but don’t act like you have some “evidence” or “proof” to back up your claims… all we’re talking about here is belief, because we simply don’t know - neither you nor i
and given that all of this is based on belief rather than proof, the only thing that matters is what we as individuals believe about the input and output data (because the bit in the middle has no definitive proof either way)
if a human consumes media and writes something and it looks different, that’s not a violation
if a machine consumes media and writes something and it looks different, you’re arguing that is a violation
the only difference here is your belief that a human brain somehow has something “more” than a probabilistic model going on… but again, that’s far from certain
You do know that comedians are copying each others material all the time though? Either making the same joke, or slightly adapting it.
So in the context of copyright vs. model training i fail to see how the exact process of the model is relevant? At the end copyrighted material goes in and material based on that copyrighted material goes out.
Text prediction seems to be sufficient to explain all verbal communication to me. Until someone comes up with a use case that humans can do that LLMs cannot, and I mean a specific use case not general high level concepts, I’m going to assume human verbal cognition works the same was as an LLM.
We are absolutely basing our responses on what words are likely to follow which other ones. It’s literally how a baby learns language from those around them.
If you ask an LLM to help you with a legal brief, it’ll come up with a bunch of stuff for you, and some of it might even be right. But it’ll very likely do things like make up a case that doesn’t exist, or misrepresent a real case, and as has happened multiple times now, if you submit that work to a judge without a real lawyer checking it first, you’re going to have a bad time.
There’s a reason LLMs make stuff up like that, and it’s because they have been very, very narrowly trained when compared to a human. The training process is almost entirely getting good at predicting what words follow what other words, but humans get that and so much more. Babies aren’t just associating the sounds they hear, they’re also associating the things they see, the things they feel, and the signals their body is sending them. Babies are highly motivated to learn and predict the behavior of the humans around them, and as they get older and more advanced, they get rewarded for creating accurate models of the mental state of others, mastering abstract concepts, and doing things like make art or sing songs. Their brains are many times bigger than even the biggest LLM, their initial state has been primed for success by millions of years of evolution, and the training set is every moment of human life.
LLMs aren’t nearly at that level. That’s not to say what they do isn’t impressive, because it really is. They can also synthesize unrelated concepts together in a stunningly human way, even things that they’ve never been trained on specifically. They’ve picked up a lot of surprising nuance just from the text they’ve been fed, and it’s convincing enough to think that something magical is going on. But ultimately, they’ve been optimized to predict words, and that’s what they’re good at, and although they’ve clearly developed some impressive skills to accomplish that task, it’s not even close to human level. They spit out a bunch of nonsense when what they should be saying is “I have no idea how to write a legal document, you need a lawyer for that”, but that would require them to have a sense of their own capabilities, a sense of what they know and why they know it and where it all came from, knowledge of the consequences of their actions and a desire to avoid causing harm, and they don’t have that. And how could they? Their training didn’t include any of that, it was mostly about words.
One of the reasons LLMs seem so impressive is that human words are a reflection of the rich inner life of the person you’re talking to. You say something to a person, and your ideas are broken down and manipulated in an abstract manner in their head, then turned back into words forming a response which they say back to you. LLMs are piggybacking off of that a bit, by getting good at mimicking language they are able to hide that their heads are relatively empty. Spitting out a statistically likely answer to the question “as an AI, do you want to take over the world?” is very different from considering the ideas, forming an opinion about them, and responding with that opinion. LLMs aren’t just doing statistics, but you don’t have to go too far down that spectrum before the answers start seeming thoughtful.
Am I a moron? How do you have more upvotes than the parent comment, is it because you’re being more aggressive with your statement? I feel like you didn’t quite refute what the parent comment said. You’re just explaining how Chat GPT works, but you’re not really saying how it shouldn’t use our established media (copyrighted material) as a reference.
I don’t control the upvotes so I don’t know why that’s directed at me.
The refutation was based on around a misunderstanding of how LLMs generate their outputs and how the training data assists the LLM in doing what it does. The article itself tells you ChatGPT was trained off of copyrighted material they were not licensed for. The person I responded to suggested that comedians do this with their work but that’s equating the process an LLM uses when producing an output to a comedian writing jokes.
Edit: Apologies if I do come across aggressive. Since the plagiarism machine has been in full swing, the whole discourse around it has gotten on my nerves. I’m a creative person, I’ve written poems and short stories, I’m writing a novel and I also do programming and a whole host of hobbies so when LLMs are used to put people like me out of a job using my own work, why wouldn’t that make me angry? What makes it worse is that I’m having to explain concepts to people regarding LLMs that they continue to defend. I can’t stand it so yes, I will come off aggressive.
Sorry, I was essentially emphasizing on my initial point “am I a moron?”, lol, because I legitimately didn’t get your point at first like others do in this thread.
I get what you mean now after reading it couple more times
There’s this linguistic problem where one word is used for two different things, it becomes difficult to tell them apart. “Training” or “learning” is a very poor choice of word to describe the calibration of a neural network. The actor and action are both fundamentally different from the accepted meaning. To start with, human learning is active whereas machining learning is strictly passive: it’s something done by someone with the machine as a tool. Teachers know very well that’s not how it happens with humans.
When I compare training a neural network with how I trained to play clarinet, I fail to see any parallel. The two are about as close as a horse and a seahorse.
Not sure what you mean by passive. It takes a hell of a lot of electricity to train one of these LLMs so something is happening actively.
I often interact with ChatGPT 4 as if it were a child. I guide it through different kinds of mental problems, having it take notes and evaluate its own output, because I know our conversations become part of its training data.
It feels very much like teaching a kid to me.
I mean passive in terms of will. Computers want and do nothing. They’re machines that function according to commands.
The way you feel like teaching a child when you feed input in natural language to a LLM until you’re satisfied with the output is known as the ELIZA effect. To quote Wikipedia:
In computer science, the ELIZA effect is the tendency to project human traits — such as experience, semantic comprehension or empathy — into computer programs that have a textual interface. The effect is a category mistake that arises when the program’s symbolic computations are described through terms such as “think”, “know” or “understand.”
A comedian walks on stage and says, “Why is there a mic here?”
AIs are not humans. Humans cannot read millions of texts in seconds and cannot split out millions of output at the same time.
Try to train a human comedian to make jokes without ever allowing him to hear another comedian’s jokes, never watching a movie, never reading a book or magazine, never watching a TV show. I expect the jokes would be pretty weak.
deleted by creator
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.
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.
I mostly disagree (especially on the long term), but hope you’re right
It’s garbage for programming. A useful tool but not one that can be used by a non-expert. And I’ve already had to have a conversation with one of my coworkers when they tried to submit absolutely garbage code.
This isn’t even the first attempt at a smart system that enables non-programmers to write code. They’ve all been garbage. So, too, will the next one be but every generation has to try it for themselves. AGI might have some potential some day, but that’s a long long way off. Might as well be science fiction.
Other disciplines are affected differently, but I constantly play with image and text generation and they are all some flavor of garbage. There are some areas where AI can excel but they are mostly professional tools and not profession replacements.
OpenAi, please generate your own source code but optimized and improved in all possible ways.
not how programming works, but tech illiterate people seem to think so
It was of no use whatsoever to programming or image generation or writing a few years ago. This thing has developed very quickly and will continue to. Give it 5 years and I think things will look very differently.
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.
Literally every artist copies, it’s how we all learn. The difference is that every artist out there does not have an enterprise-class-data-center-powerd-super-human ability to absorb \ and then be able to spit out anything instantly. It still takes time and hard work and dedication. And through the years of hard work people put into learning how their heroes do X, Y, and Z, they develop a style of their own.
It’s how artists cut their teeth and work their way into the profession. What you’re welcoming in is a situation where nobody can find any success whatsoever until they are absolutely original and of course that is an impossible moving target when every original ideal and design and image can just be instantly siphoned back up into the AI model.
Nobody could survive that way. Nobody can break into the artistic industry that way. Except for the wealthy. All the low level work people get earlier in their careers that helps keep them afloat while they learn is gone now. You have to be independently wealthy to become a high level artist capable of creating truly original work. Because there’s no other way to subsidize the time and dedication that takes when all the work for people honing their craft has been hoovered up by machines.
No, I am not welcoming an artist apocalypse, that would obviously be bad.
I am noting that I find it amusing to me on a level I already acknowledged was petty and personal that many, many mediocre artists who are absolutely awful to other people socially would have their little cults of fandom dampened by the fact that a machine can more or less to what they do, and their cult leader status is utterly unwarranted.
I do not have a nice and neat solution to the problem you bring up.
I do believe you are being somewhat hyperbolic, but, so was I.
Yep, being an artist in a capitalist hellscape world with modern AI algorithms is not a very reliable way to earn a good living and you are not likely to be have such a society produce many artists who do not have either a lot of free time or money, or you get really lucky.
At this point we are talking about completely reorganizing society in fairly large and comprehensive ways to achieve significant change on this front.
Also this problem applies to far, far more people than just artists. One friend of mine wanted her dream job as running a little bakery! Had to set her prices too high, couldn’t afford a good location, supply chain problems, taxes, didn’t work out.
Maybe someone’s passion is teaching! Welp, that situation is all fucked too.
My point here is: Ok, does anyone have an actual plan that can actually transform the world into somewhere that allow the average person to be far more likely to be able to live the life they want?
Would that plan have more to do with the minutiae of regulating a specific kind of ever advancing and ever changing technology in some kind of way that will be irrelevant when the next disruptive tech proliferates in a few years, or maybe more like an actual total overhaul of our entire society from the ground up?
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.
deleted by creator
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.
deleted by creator
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.
So why is so much information (data) freely available on the internet? How do you expect a human artist to learn drawing, if not looking at tutorials and improving their skills through emulating what they see?
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.
The problem is not the use of copyrighted material. The problem is doing so without permission and without paying for it.
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.
What’s stopping AI companies from paying royalties to artists they ripped off?
Also, lol at accounts created within few hours just to reply in this thread.
The moment their works are the one that got stolen by big companies and driven out of business, watch their tune change.
Edit: I remember when Reddit did that shitshow, and all the sudden a lot of sock / bot accounts appeared. I wasn’t expecting it to happen here, but I guess election cycle is near.
Money is not always the issue. FOSS software for example. Who wants their FOSS software gobbled up by a commercial AI regardless. So there are a variety of issues.
I don’t care if any of my FOSS software is gobbled up by a commercial AI. Someone reading my code isn’t a problem to me. If it were, I wouldn’t publish it openly.
I do, especially when someone’s profiting from it, while my license is strictly for non commercial.
Same. I didn’t write it for them. I wrote it for folks who don’t necessarily have a lot of money but want something useful.
Well, for $20/mo I get a super-educated virtual assistant/tutor. It’s pretty awesome.
I’d say that’s some good value for people without much money. All of my open source libs are published under the MIT license if I recall correctly. I’ve made so much money using open source software, I don’t mind giving back, even to people who are going to make money with my code.
It makes me feel good to think my code could be involved in money changing hands. It’s evidence to me that I built something valuable.
$20/mo
good value for people without much money
The absolute majority of people can not afford that. This is especially true for huge part of the art that was used to train various models on.
AI currently is a tool for rich people by rich people which uses the work of poor people who themselves won’t be able to benefit from it.
And yet it is orders of magnitude less than it cost a year ago to hire someone to do research, write reports, and tutor me in any subject I want.
If an artist can’t afford $20/mo they need a job to support that hobby.
What’s stopping AI companies from paying royalties to artists they ripped off?
profit. AI is not even a profitable business now. They exist because of the huge amount of investment being poured into it. If they have to pay their fair share they would not exist as a business.
what OpenAI says is actually true. The issue IMHO is the idea that we should give them a pass to do it.
Uber wasn’t making profit anyway, despite all the VCs money behind it.
I guess they have reasons not to pay drivers properly. Give Uber a free pass for it too
When you think about it, all companies would make so much more money if they didn’t have to pay their staff, or pay for materials they use! This whole economy and capitalism business, which relies on money being exchanged for goods and services, is clearly holding back profits. Clearly the solution here is obvious: everybody should embrace OpenAI’s methods and simply grab whatever they want without paying for it. Profit for everyone!
- This is not REALLY about copyright - this is an attack on free and open AI models, which would be IMPOSSIBLE if copyright was extended to cover the case of using the works for training.
- It’s not stealing. There is literally no resemblance between the training works and the model. IP rights have been continuously strengthened due to lobbying over the last century and are already absurdly strong, I don’t understand why people on here want so much to strengthen them ever further.
There is literally no resemblance between the training works and the model.
This is way too strong a statement when some LLMs can spit out copyrighted works verbatim.
https://www.404media.co/google-researchers-attack-convinces-chatgpt-to-reveal-its-training-data/
A team of researchers primarily from Google’s DeepMind systematically convinced ChatGPT to reveal snippets of the data it was trained on using a new type of attack prompt which asked a production model of the chatbot to repeat specific words forever.
Often, that “random content” is long passages of text scraped directly from the internet. I was able to find verbatim passages the researchers published from ChatGPT on the open internet: Notably, even the number of times it repeats the word “book” shows up in a Google Books search for a children’s book of math problems. Some of the specific content published by these researchers is scraped directly from CNN, Goodreads, WordPress blogs, on fandom wikis, and which contain verbatim passages from Terms of Service agreements, Stack Overflow source code, copyrighted legal disclaimers, Wikipedia pages, a casino wholesaling website, news blogs, and random internet comments.
Beyond that, copyright law was designed under the circumstances where creative works are only ever produced by humans, with all the inherent limitations of time, scale, and ability that come with that. Those circumstances have now fundamentally changed, and while I won’t be so bold as to pretend to know what the ideal legal framework is going forward, I think it’s also a much bolder statement than people think to say that fair use as currently applied to humans should apply equally to AI and that this should be accepted without question.
I’m gonna say those circumstances changed when digital copies and the Internet became a thing, but at least we’re having the conversation now, I suppose.
I agree that ML image and text generation can create something that breaks copyright. You for sure can duplicate images or use copyrighted characterrs. This is also true of Youtube videos and Tiktoks and a lot of human-created art. I think it’s a fascinated question to ponder whether the infraction is in what the tool generates (i.e. did it make a picture of Spider-Man and sell it to you for money, whcih is under copyright and thus can’t be used that way) or is the infraction in the ingest that enables it to do that (i.e. it learned on pictures of Spider-Man available on the Internet, and thus all output is tainted because the images are copyrighted).
The first option makes more sense to me than the second, but if I’m being honest I don’t know if the entire framework makes sense at this point at all.
The infraction should be in what’s generated. Because the interest by itself also enables many legitimate, non-infracting uses: uses, which don’t involve generating creative work at all, or where the creative input comes from the user.
I don’t disagree on principle, but I do think it requires some thought.
Also, that’s still a pretty significant backstop. You basically would need models to have a way to check generated content for copyright, in the way Youtube does, for instance. And that is already a big debate, whether enforcing that requirement is affordable to anybody but the big companies.
But hey, maybe we can solve both issues the same way. We sure as hell need a better way to handle mass human-produced content and its interactions with IP. The current system does not work and it grandfathers in the big players in UGC, so whatever we come up with should work for both human and computer-generated content.
I can spit out copyrighted work verbatim.
“No Lieutenant, your men are already dead”
See?
But AI isn’t all about generating creative works. It’s a store of information that I can query - a bit like searching Google; but understands semantics, and is interactive. It can translate my own text for me - in which case all the creativity comes from me, and I use it just for its knowledge of language. Many people use it to generate boilerplate code, which is pretty generic and wouldn’t usually be subject to copyright.
This is how I use the AI: I learn from it. Honestly I just never got the bug on wanting it to generate creative works I can sell. I guess I’d rather sell my own creative output, you know? It’s more fun than ordering a robot to be creative for me.
I have used it as a collaborator when doing creative work. It’s a great brainstorming buddy, and I use it to generate rough drafts of stuff. Usually I use it while developing roleplaying scenarios for TTRPGs I run for my friends. Generative AI is great for illustrating those scenarios, too.
I know it inherently seems like a bad idea to fix an AI problem with more AI, but it seems applicable to me here. I believe it should be technically feasible to incorporate into the model something which checks if the result is too similar to source content as part of the regression.
My gut would be that this would, at least in the short term, make responses worse on the whole, so would probably require legal action or pressure to have it implemented.
The key element here is that an LLM does not actually have access to its training data, and at least as of now, I’m skeptical that it’s technologically feasible to search through the entire training corpus, which is an absolutely enormous amount of data, for every query, in order to determine potential copyright violations, especially when you don’t know exactly which portions of the response you need to use in your search. Even then, that only catches verbatim (or near verbatim) violations, and plenty of copyright questions are a lot fuzzier.
For instance, say you tell GPT to generate a fan fiction story involving a romance between Draco Malfoy and Harry Potter. This would unquestionably violate JK Rowling’s copyright on the characters if you published the output for commercial gain, but you might be okay if you just plop it on a fan fic site for free. You’re unquestionably okay if you never publish it at all and just keep it to yourself (well, a lawyer might still argue that this harms JK Rowling by damaging her profit if she were to publish a Malfoy-Harry romance, since people can just generate their own instead of buying hers, but that’s a messier question). But, it’s also possible that, in the process of generating this story, GPT might unwittingly directly copy chunks of renowned fan fiction masterpiece My Immortal. Should GPT allow this, or would the copyright-management AI strike it? Legally, it’s something of a murky question.
For yet another angle, there is of course a whole host of public domain text out there. GPT probably knows the text of the Lord’s Prayer, for instance, and so even though that output would perfectly match some training material, it’s legally perfectly okay. So, a copyright police AI would need to know the copyright status of all its training material, which is not something you can super easily determine by just ingesting the broad internet.
skeptical that it’s technologically feasible to search through the entire training corpus, which is an absolutely enormous amount of data
Google, DuckDuckGo, Bing, etc. do it all the time.
I don’t see why it wouldn’t be able to. That’s a Big Data problem, but we’ve gotten very very good at searches. Bing, for instance, conducts a web search on each prompt in order to give you a citation for what it says, which is pretty close to what I’m suggesting.
As far as comparing to see if the text is too similar, I’m not suggesting a simple comparison or even an Expert Machine; I believe that’s something that can be trained. GANs already have a discriminator that’s essentially measuring how close to generated content is to “truth.” This is extremely similar to that.
I completely agree that categorizing input training data by whether or not it is copyrighted is not easy, but it is possible, and I think something that could be legislated. The AI you would have as a result would inherently not be as good as it is in the current unregulated form, but that’s not necessarily a worse situation given the controversies.
On top of that, one of the common defenses for AI is that it is learning from material just as humans do, but humans also can differentiate between copyrighted and public works. For the defense to be properly analogous, it would make sense to me that it would need some notion of that as well.
It’s actually the other way around, Bing does websearches based on what you’ve asked it and then the answer it generates can incorporate information that was returned by the websearching. This is why you can ask it about current events that weren’t in its training data, for example - it looks the information up, puts it into its context, and then generates the response that you see. Sort of like if I asked you to write a paragraph about something that you didn’t know about, you’d go look the information up first.
but humans also can differentiate between copyrighted and public works
Not really. Here’s a short paragraph about sailboats. Is it copyrighted?
Sailboats, those graceful dancers of the open seas, epitomize the harmonious marriage of nature and human ingenuity. Their billowing sails, like ethereal wings, catch the breath of the wind, propelling them across the endless expanse of the ocean. Each vessel bears the scars of countless journeys, a testament to the resilience of both sailor and ship.
Thank you for your thoroughly analytical take on the subject. Solid points all around.
deleted by creator
Sorry AIs are not humans. Also executives like Altman are literally being paid millions to steal creator’s work.
I didn’t say anything about AIs being humans.
They’re also not vegetables 😡
I don’t understand why people on here want so much to strengthen them ever further.
It is about a lawless company doing lawless things. Some of us want companies to follow the spirit, or at least the letter, of the law. We can change the law, but we need to discuss that.
IANAL, why isn’t it fair use?
The two big arguments are:
- Substantial reproduction of the original work, you can get back substantial portions of the original work from an AI model’s output.
- The AI model replaces the use of the original work. In short, a work that uses copyrighted material under fair use can’t be a replacement for the initial work.
you can get back substantial portions of the original work from an AI model’s output
Have you confirmed this yourself?
In its complaint, The New York Times alleges that because the AI tools have been trained on its content, they sometimes provide verbatim copies of sections of Times reports.
OpenAI said in its response Monday that so-called “regurgitation” is a “rare bug,” the occurrence of which it is working to reduce.
“We also expect our users to act responsibly; intentionally manipulating our models to regurgitate is not an appropriate use of our technology and is against our terms of use,” OpenAI said.
The tech company also accused The Times of “intentionally” manipulating ChatGPT or cherry-picking the copycat examples it detailed in its complaint.
https://www.cnn.com/2024/01/08/tech/openai-responds-new-york-times-copyright-lawsuit/index.html
The thing is, it doesn’t really matter if you have to “manipulate” ChatGPT into spitting out training material word-for-word, the fact that it’s possible at all is proof that, intentionally or not, that material has been encoded into the model itself. That might still be fair use, but it’s a lot weaker than the original argument, which was that nothing of the original material really remains after training, it’s all synthesized and blended with everything else to create something entirely new that doesn’t replicate the original.
You said:
Substantial reproduction of the original work, you can get back substantial portions of the original work from an AI model’s output.
If an AI is trained on a huge number of NYT articles and you’re only able to get it to regurgitate one of them, that’s not a “substantial portion of the original work.” That’s a minuscule portion of the original work.
Agreed on both counts… Except Microsoft sings a different tune when their software is being “stolen” in the exact same way. They want to have it both ways - calling us pirates when we copy their software, but it’s “without merit” when they do it. Fuck’em! Let them play by the same rules they want everyone else to play.
That sounds bad. Do you have evidence for MS behaving this way?
Literally first hit on google (after the NYT links).
Any reasonable person can reach the conclusion that something is wrong here.
What I’m not seeing a lot of acknowledgement of is who really gets hurt by copyright infringement under the current U.S. scheme. (The quote is obviously directed toward the UK, but I’m reasonably certain a similar situation exists there.)
Hint: It’s rarely the creators, who usually get paid once while their work continues to make money for others.
Let’s say the New York Times wins its lawsuit. Do you really think the reporters who wrote the infringed-upon material will be getting royalty checks to be made whole?
This is not OpenAI vs creatives. OK, on a basic level it is, but expecting no one to scrape blogs and forum posts rather goes against the idea of the open internet in the first place. We’ve all learned by now that what goes on the internet stays there, with attribution totally optional unless you have a legal department. What’s novel here is the scale of scraping, but I see some merit to the “transformational” fair-use defense given that the ingested content is not being reposted verbatim.
This is corporations vs corporations. Framing it as millions of people missing out on what they’d have otherwise rightfully gotten is disingenuous.
This isn’t about scraping the internet. The internet is full of crap and the LLMs will add even more crap to it. It will shortly become exponentially harder to find the meaningful content on the internet.
No, this is about dipping into high quality, curated content. OpenAI wants to be able to use all existing human artwork without paying anything for it, and then flood the world with cheap knockoff copies. It’s that simple.
Shortly? It’s happening already. I notice it when using Google and Duckduckgo. There are always a few hits that are AI written blog spam word soup
Unfortunately you haven’t seen the full impact of LLMs yet. What you’re seeing now is stuff that’s already been going on for a decade. SEO content generators have been a thing for many years and used by everybody from small business owners to site chains pinching ad pennies.
When the LLM crap will kick in you won’t see anything except their links. I wouldn’t be surprised if we’ll have to go back to 90s tech and use human-curated webrings and directories.
I wonder how many comments in this thread are ai generated. I wonder how many comments on Lemmy will be in 5 years time.
It’s especially amusing when you consider that it’s not even fully autonomous yet; we’re actively doing this to ourselves.
Yep. The effect of this as currently framed is that you get data ownership clauses in EULAs forever and only major data brokers like Google or Meta can afford to use this tech at all. It’s not even a new scenario, it already happened when those exact companies were pushing facial recognition and other big data tools.
I agree that the basics of modern copyright don’t work great with ML in the mix (or with the Internet in the mix, while we’re at it), but people are leaning on the viral negativity to slip by very unwanted consequences before anybody can make a case for good use of the tech.
it’s so baffling to me that some people think this is a clear cut problem of “you stole the work just the same as if you sold a copy without paying me!”
it ain’t the same folks… that’s not how models work… the outcome is unfortunate, for sure, but to just straight out argue that it’s the same is ludicrous… it’s a new problem and ML isn’t going away, so we’re going to have to deal with it as a new problem
Having read through these comments, I wonder if we’ve reached the logical conclusion of copyright itself.
copyright has become a tool of oppression. Individual author’s copyright is constantly being violated with little resources for them to fight while big tech abuses others work and big media uses theirs to the point of it being censorship.
Perhaps a fair compromise would be doing away with copyright in its entirety, from the tiny artists trying to protect their artwork all the way up to Disney, no exceptions. Basically, either every creator has to be protected, or none of them should be.
IMO the right compromise is to return copyright to its original 14 year term. OpenAI can freely train on anything up to 2009 which is still a gigantic amount of material while artists continue to be protected and incentivized.
I’m increasingly convinced of that myself, yeah (although I’d favour 15 or 20 years personally, just because they’re neater numbers than 14). The original purpose of copyright was to promote innovation by ensuring a creator gets a good length of time in which to benefit from their creation, which a 14-20 year term achieves. Both extremes - a complete lack of copyright and the exceedingly long terms we have now - suppress innovation.
I’d favour 15 or 20 years personally, just because they’re neater numbers than 14
Another neat number is: 4.
That’s it, if you don’t make money on your creation in 4 years, then it’s likely trash anyway.
I’ve said it before and I’ll say it again! (My apologies if it happens to be to the same person, lol)
Early access developers in shambles!
that would mean governments prosecuting all offences, which is not going to happen. I doubt any country would have enough resources for doing that
deleted by creator
Apparently they’re going to just make only the little guy’s copyrights effectively meaningless, so yeah.
It’s crazy how everyone is suddenly in favour of IP law.
IP law used to stop corporations from profiting off of creators’ labor without compensation? Yeah, absolutely.
IP law used to stop individuals from consuming media where purchases wouldn’t even go to the creators, but some megacorp? Fuck that.
I’m against downloading movies by indie filmmakers without compensating them. I’m not against downloading films from Universal and Sony.
I’m against stealing food from someone’s garden. I’m not against stealing food from Safeway.
If you stop looking at corporations as being the same as individuals, it’s a very simple and consistent viewpoint.
IP law shouldn’t exist, but if it does it should only exist to protect individuals from corporations. When that’s how it’s being used, like here, I accept it as a necessary evil.
Word.
IP law used to compensate creators “until their death + 70 years”… you can spin it however you want, that’s just plain wrong.
If you stop looking at corporations as being the same as individuals
That’s a separate bonkers legislation. Two wrongs don’t make one right.
I never said I like IP law. I explicitly said it shouldn’t exist. I wish they’d strip out any post-humous ownership, absolutely. But I’m fine beating OpenAI over the head with that or any other law. Whether I advocate for or against copyright law will ultimately have no impact on its existence, so I may as well cheer it on when it’s used to hurt corporations, and condemn it when it’s used to protect corporations over individuals.
That’s a separate bonkers legislation
I’m not talking about the legislation, I’m talking about the mindset, which is very prevalent in the pro-AI tech spaces. Go to HackerNews and see just how hard the AI-bros there will fellate each other over “corporate rights”.
My whole point is that there is nothing logically inconsistent with being against IP law, but also understanding that since its existence is reality, leveraging it as best as possible (i.e. to hurt corporations).
I’m not so much in favor of IP law as I am in favor of informed consent in every aspect of the word.
when posting photos, art and text content years ago, I was not able to imagine it might be trained off by an AI. As such I was not able to make a decision based on informed consent if I agreed to that or not.
Even though quotes such as “once you post it, its on the internet forever” were around, I was not aware the extend to which this reached and that had my art been vacuumed by a generative AI model (it hasnt luckily) people could create art that pretends to be created by me. Thus I could not consent
I think this goes for a lot of artists actually, especially those who exist far more publicly than I do, who are in those databases and who are a keyword to be used in prompts. There is no possible way they could have given informed consent to that at the time they posted art/at the time they started that social media profile/youtube channel etc.
To me, this is the real problem. I could care less about corporations.
I still think IP needs to eat shit and die. Always has, always will.
I recently found out we could have had 3d printing 20 years earlier but patents stopped that. Cocks !
It’s almost like most people are idiots who don’t understand the thing they’re against and are just parroting what they hear/read.
I’m the detractor here, I couldn’t give less of a shit about anything to do with intellectual property and think all copyright is bad.
Well in that case maybe chat gpt should just fuck off it doesn’t seem to be doing anything particularly useful, and now it’s creator has admitted it doesn’t work without stealing things to feed it. Un fucking believable. Hacks gonna hack I guess.
ChatGPT has been enormously useful to me over the last six months. No idea where you’re getting this notion it isn’t useful.
People pretending it’s not useful and/or not improving all the time are living in their own worlds. I think you can argue the legality and the ethics, but any anti-ai position based on low quality output (“it can’t even do hands!”) has a short shelf-life.
As with many things, the golden rule applies. They who have the gold, make the rules.
…so stop doing it!
This explains what Valve was until recently not so cavalier about AI: They didn’t want to hold the bag on copyright matters outside of their domain.
Then shutdown your goddamn company until you find a better way.
It’s also “impossible” to have multiple terabytes of media on my homeserver without copyright infringement, so piracy is ok, right!?
O no, wait it actually is possible, it’s just more expensive and more work to do it legally (and leaves a lot of plastic trash in form of Blurays and DVDs), just like with AI. But laws are just for poor people, I guess.
Even if it was impossible, would that make it okay?
I think viral outrage aside, there is a very open question about what constitutes fair use in this application. And I think the viral outrage misunderstands the consequences of enforcing the notion that you can’t use openly scrapable online data to build ML models.
Effectively what the copyright argument does here is make it so that ML models are only legally allowed to make by Meta, Google, Microsoft and maybe a couple of other companies. OpenAI can say whatever, I’m not concerned about them, but I am concerned about open source alternatives getting priced out of that market. I am also concerned about what it does to previously available APIs, as we’ve seen with Twitter and Reddit.
I get that it’s fashionable to hate on these things, and it’s fashionable to repeat the bit of misinformation about models being a copy or a collage of training data, but there are ramifications here people aren’t talking about and I fear we’re going to the worst possible future on this, where AI models are effectively ubiquitous but legally limited to major data brokers who added clauses to own AI training rights from their billions of users.
People hate them not because it is fashionable, but because they can see what is coming.
Tech companies want to create tools that would replace million of jobs without compensating the very people that created these works in the first place.
That’s not “coming”, it’s an ongoing process that has been going on for a couple hundred years, and it absolutely does not require ChatGPT.
People genuinely underestimate how many of these things have been an ongoing concern. A lot like crypto isn’t that different to what you can do with a server, “AI” isn’t a magic key that unlocks automation. I don’t even know how this mental model works. Is the idea that companies who are currently hiring millions of copywriters will just rely on automated tools? I get that yeah, a bunch of call center people may get removed (again, a process that has been ongoing for decades), but how is compensating Facebook for scrubbing their social media posts for text data going to make that happen less?
Again, I think people don’t understand the parameters of the problem, which is different from saying that there is no problem here. If anything the conversation is a net positive in that we should have been having it in 2010 when Amazon and Facebook and Google were all-in on this process already through both ML tools and other forms of data analysis.
Tech companies will create those tools no matter what. Then they will charge everyone through the nose for using them.
The question is whether:
- ONLY tech companies capable of paying scraps during 70 years after the author’s death are allowed to create those tools
- EVERYONE is allowed to train their own tool, without having to raise a few billion in seed capital
In this case, OpenAI is acting as “the devil’s advocate”… and it’s working to fool people into supporting the opposite position.
It is an open question. As others have pointed out, a human taking inspiration from the work of others is totally fine. My issue is that AI are not human.
A human’s production of work is limited. A human can only produce so fast for so long. An AI could theoretically be scaled infinitely and produce indefinitely. I don’t want to live in a world where FAANGCORP’s OmniAI is responsible for 90% of all art, media, and music because humans can’t keep pace with it.
A lot of this can be traced back to the invention of photography, which is a fun point of reference, if one goes to dig up the debate at the time.
In any case, the idea that humans can only produce so fast for so long and somehow that cleans the channel just doesn’t track. We are flooded by low quality content enabled by social media. There’s seven billion of us two or three billion of those are on social platforms and a whole bunch of the content being shared in channels is created by using corporate tools to make stuff by pointing phones at it. I guarantee that people will still go to museums to look at art regardless of how much cookie cutter AI stuff gets shared.
However, I absolutely wouldn’t want a handful of corporations to have the ability to empower their employed artists with tools to run 10x faster than freelance artists. That is a horrifying proposition. Art is art. The difficulty isn’t in making the thing technically (say hello, Marcel Duchamp, I bet you thought you had already litgated this). Artists are gonna art, but it’s important that nobody has a monopoly on the tools to make art.
“It’s too fast” is a really really dumb argument against AI
It’s like the classic “objection!” “On what grounds?” “It’s devastating to my case!” Scenario.
Throughout history technology has repeatedly been developed that lets people do things faster than the people currently doing it. That’s usually the point of technological progress. Of course the people left behind by that will complain, but that alone is no reason to limit the rest of us who would benefit from the advance.
who benefit from automating creative process
Anyone who enjoys creative things, since they now have access to a lot more of it a lot more easily.
is low quality
Then it’s not a threat to professional creatives, is it?
Mass produced garbage is still mass produced garbage. As you point out AIs aren’t human and while that removes the limitations of the flesh (including limitations that we might want there - no human ever says oops, I made a child porn), it imposes limitations of the machine. AI output isn’t that good at anything practical. It writes garbage code that even if you manage to get it working, the business manager or whoever isn’t capable of seeing the flaws in it. The art is devoid of any sort of soul and almost always has glaring flaws that require actual humans to identify and fix.
We are about to be inundated with AI produced garbage, sure, but that only proves the lie that shady internet sites and social media have always been a cesspool of shitty, unreliable content, and connecting with hundreds of thousands of faceless strangers was never a good idea. Hopefully we’ll come up with (or go back to) solutions that don’t treat the problem as simply one of volume.
It’s not right to say that ML output isn’t good at practical tasks. It is and it’s already in use and has been for ages. The conversation about these is guided by the relatively anecdotal fact that chatbots and image generation got good so this stuff went viral, but ML models are being used for a bunch of practical uses, from speeding up repetitive, time consuming tasks (e.g. cleaning up motion capture, facial modelling or lip animation in games and movies) or specialized tasks (so much science research is using ML tools these days).
Now, a lot of those are done using fully owned datasets, but not all, and the ramifications there are also important. People dramatically overestimate the impact of trash product flooding channels (which is already the case, as you say) and dramatically underestimate the applications of the underlying tech beyond the couple of viral apps they only got access to recently.
