Sorry if this comes off as offensive, but this isn’t new news, and I think we all need a reality check here. I’ll also be forward, the artifice of doubt cast on this makes me pretty angry and is a threat to every one of us.
So, OK. Speaking as a thirty-year programmer, a neurobiology minor, and a hobby animator, you know what these objections sound like? They sound like somebody who just learned to program, with JavaScript, at the top of that initial peak. You know the one, when you feel like you can do anything and present as such. You’ve never had to worry about stack overflows or memory exceptions or, quite possibly, even fundamental networking utilities, but you got a web page to do something which is, in fact, legitimately cool, and you’re proud of yourself—as you friggin’ should be. However, you don’t know how much you don’t know, and are way further from the top than you suspect you are.
Until you learn something like C, and really walk away from it with a sore ass and a sense of perspective, you don’t know what you don’t know. This isn’t because we all aren’t rooting for you, it’s just an expected rite of passage; when you really begin to learn. We’re totally rooting for you, we used to be you! And this is also true of visual art and programming; they have nothing to do with each other, and as far as I can see this case is open and shut, and we’re still mulling over the details.
Artists don’t copy, they analyze. It’s difference between reading all of the answers on a Stack Exchange site, considering them, discussing them, putting them in the context of your own life, and applying them in an organized and personal fashion; versus simply copying the code verbatim and jamming things in until it arguably works, which is literally what an LLM neural network is designed to do. We’ve all seen this with extremely questionable code output by GPT-3 (and, yes, GPT-4, it still happens, just not as frequently).
The art neural networks are the same—an artist charges for that inspiration, because it was a lot of physical and emotional toil for them. It was a lot of feedback and self-critique. “Prompt engineering” is neither art nor, if we’re honest with ourselves, engineering.
I maintain that this is basically a slightly obfuscated recap of the Napster trials from twenty years ago, who, ironically, fell back on almost the same “fair use” defense. It’s going to falter just as hard, as there’s a massive difference between showing smaller, low-resolution images on a search page; and using meaningful elements of those images to produce brand new, and competing, works; which happen to be flawed, but look the same to someone who browses JPEGs on Google like a channel surfer.
The last thing I’m going to say is that we need to stop describing LLMs as “AI”. “AI” doesn’t mean anything. It could refer to a neural network, machine learning, A* path finding, collective intelligence, and any number of other things, the colloquial definition being technology that “performs a function which was previously only possible for a human”, and come on, once upon a time you could describe a hammer that way. AI is not a formal industry term, it’s marketing flak. LLMs are effectively a database of connections browsed with the use of a neural network.
You want to keep using Midjourney and Stable Diffusion? Great! Go for it. You want to use ChatGPT to help you understand some multivariate calculus? Go nuts, I do it all the time, most mathematicians are terrible at articulating. However, they should, without question, have to pay for the art that they used, or cease using it if the sale won’t be completed. Any other outcome is absolutely going to lead to an economic collapse.
However, they should, without question, have to pay for the art that they used, or cease using it if the sale won’t be completed. Any other outcome is absolutely going to lead to an economic collapse.
This is the part that drives me crazy, the conclusion doesn’t follow from the premise. Just because machine learning poses an economic threat to artists/writers/ect, does not mean it somehow makes how they are trained unethical. It undoubtedly does pose a threat, but not because they’re “stealing” work, but because our economic system is fucked. I would challenge anyone to try applying the same copywrite legal argument to the more common training sets using reddit, Twitter, and other online forum text data, which does not have the same copywrite protections, and isn’t identifiable in the ML outputs. Machine learning applications have the potential to liberate humanity from millions of hours of meaningless work, why should we be whining about it just because “they took our jobs!”?
Just like the Napster trials, I think our economic system and industry ought to adapt to the new technology, not cling to legal precedent to protect it from changing. Employment should not be a prerequisite to a standard of living, full stop. If some new technology comes along and replaces the labor of a couple million people, our reaction shouldn’t be to stop the progress, but to ensure those people put out of work can still afford to live without creating more meaningless work to fill their time .
Most models are trained unethically, relying on weird statements about how humans learn the “same way” (looking at a few references when drawing a specific thing, you need to know how it looks to draw it lol) as large models (more or less averaging and weighting billions of images stolen from internet with no regards to the licenses)
I don’t think i said “humans learn the same way”, but I do think it helps to understand how ML algorithms work in comparison with existing examples of copyright infringement (i.e. photocopies, duplicated files on a hard drive, word for word or pixel for pixel duplication’s, ect.). ML’s don’t duplicate or photocopy training data, they “weight” (or to use your word choice, “average”) the data against a node structure. Other, more subjective copyright infringements are decided on a case-by-case basis, where an artist or entity has produced an “original” work that leans too heavily on a copyrighted work. It is clear that ML’s aren’t a straight-forward duplication. If you asked an MLA to reproduce an existing image, it wouldn’t be able to recreate it exactly, because that data isn’t stored in its model, only the approximate instructions on how to reproduce it. It might be able to get close, especially if that example is well represented in the data set, but the image would be fundamentally “new” in the sense that it has not been copied pixel by pixel from an original, only recreated through averaging.
If our concern is that AI could literally reproduce existing creative work and pass it off as original, then we should pursue legal action against those uses. But to claim that the model itself is an illegal duplication of copyrighted work is ridiculous. If our true concern (as I think it is) that the use of MLAs may supplant the need for paid artists or writers, then I would suggest we re-think how we structure compensation for labor and not simply place barriers to AI deployment. Even if we were to reach some compensation agreement for the use of copyrighted material in the training of AI data, that wouldn’t prevent the elimination of artistic labor, it would only solidify AI as an elite, expensive tool owned only by a handful of companies that can afford the cost. It would consolidate our economy further, not democratize it.
In my opinion, copyright law is already just a band-aid to a broader issue of labor relations, and the issue of AI training data is just a drastic expansion of that same wound.
My concern is that billions of works are being used for training with no consent and no regard to the license, and that the model “learns” is not an excuse. If someone saved some of my content for personal use, sure, I don’t mind that at all, but huge scale scraping for-profit operation downloading all content they physically can? Fuck off. I just blocked all the crawlers from ever accesing my websites (well, google and bing literally refuse to index my stuff properly anyway, so fuck them too, none of them even managed to read the sitemap properly, and it was definitely valid)
With a universal basic income artists would be free to choose to make art for fun instead of survival. Given enough job destruction in transport a UBI-like solution may be manditory as there are not enough jobs for humans generally.
Sorry if this comes off as offensive, but this isn’t new news, and I think we all need a reality check here. I’ll also be forward, the artifice of doubt cast on this makes me pretty angry and is a threat to every one of us.
So, OK. Speaking as a thirty-year programmer, a neurobiology minor, and a hobby animator, you know what these objections sound like? They sound like somebody who just learned to program, with JavaScript, at the top of that initial peak. You know the one, when you feel like you can do anything and present as such. You’ve never had to worry about stack overflows or memory exceptions or, quite possibly, even fundamental networking utilities, but you got a web page to do something which is, in fact, legitimately cool, and you’re proud of yourself—as you friggin’ should be. However, you don’t know how much you don’t know, and are way further from the top than you suspect you are.
Until you learn something like C, and really walk away from it with a sore ass and a sense of perspective, you don’t know what you don’t know. This isn’t because we all aren’t rooting for you, it’s just an expected rite of passage; when you really begin to learn. We’re totally rooting for you, we used to be you! And this is also true of visual art and programming; they have nothing to do with each other, and as far as I can see this case is open and shut, and we’re still mulling over the details.
Artists don’t copy, they analyze. It’s difference between reading all of the answers on a Stack Exchange site, considering them, discussing them, putting them in the context of your own life, and applying them in an organized and personal fashion; versus simply copying the code verbatim and jamming things in until it arguably works, which is literally what an LLM neural network is designed to do. We’ve all seen this with extremely questionable code output by GPT-3 (and, yes, GPT-4, it still happens, just not as frequently).
The art neural networks are the same—an artist charges for that inspiration, because it was a lot of physical and emotional toil for them. It was a lot of feedback and self-critique. “Prompt engineering” is neither art nor, if we’re honest with ourselves, engineering.
I maintain that this is basically a slightly obfuscated recap of the Napster trials from twenty years ago, who, ironically, fell back on almost the same “fair use” defense. It’s going to falter just as hard, as there’s a massive difference between showing smaller, low-resolution images on a search page; and using meaningful elements of those images to produce brand new, and competing, works; which happen to be flawed, but look the same to someone who browses JPEGs on Google like a channel surfer.
The last thing I’m going to say is that we need to stop describing LLMs as “AI”. “AI” doesn’t mean anything. It could refer to a neural network, machine learning, A* path finding, collective intelligence, and any number of other things, the colloquial definition being technology that “performs a function which was previously only possible for a human”, and come on, once upon a time you could describe a hammer that way. AI is not a formal industry term, it’s marketing flak. LLMs are effectively a database of connections browsed with the use of a neural network.
You want to keep using Midjourney and Stable Diffusion? Great! Go for it. You want to use ChatGPT to help you understand some multivariate calculus? Go nuts, I do it all the time, most mathematicians are terrible at articulating. However, they should, without question, have to pay for the art that they used, or cease using it if the sale won’t be completed. Any other outcome is absolutely going to lead to an economic collapse.
This is the part that drives me crazy, the conclusion doesn’t follow from the premise. Just because machine learning poses an economic threat to artists/writers/ect, does not mean it somehow makes how they are trained unethical. It undoubtedly does pose a threat, but not because they’re “stealing” work, but because our economic system is fucked. I would challenge anyone to try applying the same copywrite legal argument to the more common training sets using reddit, Twitter, and other online forum text data, which does not have the same copywrite protections, and isn’t identifiable in the ML outputs. Machine learning applications have the potential to liberate humanity from millions of hours of meaningless work, why should we be whining about it just because “they took our jobs!”?
Just like the Napster trials, I think our economic system and industry ought to adapt to the new technology, not cling to legal precedent to protect it from changing. Employment should not be a prerequisite to a standard of living, full stop. If some new technology comes along and replaces the labor of a couple million people, our reaction shouldn’t be to stop the progress, but to ensure those people put out of work can still afford to live without creating more meaningless work to fill their time .
Most models are trained unethically, relying on weird statements about how humans learn the “same way” (looking at a few references when drawing a specific thing, you need to know how it looks to draw it lol) as large models (more or less averaging and weighting billions of images stolen from internet with no regards to the licenses)
I don’t think i said “humans learn the same way”, but I do think it helps to understand how ML algorithms work in comparison with existing examples of copyright infringement (i.e. photocopies, duplicated files on a hard drive, word for word or pixel for pixel duplication’s, ect.). ML’s don’t duplicate or photocopy training data, they “weight” (or to use your word choice, “average”) the data against a node structure. Other, more subjective copyright infringements are decided on a case-by-case basis, where an artist or entity has produced an “original” work that leans too heavily on a copyrighted work. It is clear that ML’s aren’t a straight-forward duplication. If you asked an MLA to reproduce an existing image, it wouldn’t be able to recreate it exactly, because that data isn’t stored in its model, only the approximate instructions on how to reproduce it. It might be able to get close, especially if that example is well represented in the data set, but the image would be fundamentally “new” in the sense that it has not been copied pixel by pixel from an original, only recreated through averaging.
If our concern is that AI could literally reproduce existing creative work and pass it off as original, then we should pursue legal action against those uses. But to claim that the model itself is an illegal duplication of copyrighted work is ridiculous. If our true concern (as I think it is) that the use of MLAs may supplant the need for paid artists or writers, then I would suggest we re-think how we structure compensation for labor and not simply place barriers to AI deployment. Even if we were to reach some compensation agreement for the use of copyrighted material in the training of AI data, that wouldn’t prevent the elimination of artistic labor, it would only solidify AI as an elite, expensive tool owned only by a handful of companies that can afford the cost. It would consolidate our economy further, not democratize it.
In my opinion, copyright law is already just a band-aid to a broader issue of labor relations, and the issue of AI training data is just a drastic expansion of that same wound.
My concern is that billions of works are being used for training with no consent and no regard to the license, and that the model “learns” is not an excuse. If someone saved some of my content for personal use, sure, I don’t mind that at all, but huge scale scraping for-profit operation downloading all content they physically can? Fuck off. I just blocked all the crawlers from ever accesing my websites (well, google and bing literally refuse to index my stuff properly anyway, so fuck them too, none of them even managed to read the sitemap properly, and it was definitely valid)
With a universal basic income artists would be free to choose to make art for fun instead of survival. Given enough job destruction in transport a UBI-like solution may be manditory as there are not enough jobs for humans generally.