6 ms·

1) Legality and morality are obviously different and unrelated concepts. More people should understand that. 2) Copyright was the wrong mechanism to use for co

1) Legality and morality are obviously different and unrelated concepts. More people should understand that. 2) Copyright was the wrong mechanism to use for code from the start, LLMs just exposed the issue. The thing to protect shouldn't be creativity, it should be human work - any kind of work. The hard part of programming isn't creativity, it's making correct decisions. It's getting the information you need to make them. Figuring out and understanding the problem you're trying to solve, whether it's a complex mathematical problem or a customer's need. And then evaluating solutions until you find the right one. (One constrains being how much time you can spend on it.) All that work is incredibly valuable but once the solution exists, it's each easier to copy without replicating or even understanding the thought process which led to it. But that thought process took time and effort. The person who did the work deserved credit and compensation. And he deserves it transitively, if his work is used to build other works - proportional to his contribution. The hard part is quantifying it, of course. But a lot of people these days benefit from throwing their hands up and saying we can't quantify it exactly so let's make it finders keepers. That's exploitation. 3) Both LLM training and inference are derivative works by any reasonable meaning of those words. If LLMs are not derivative works of the training data then why is so much training data needed? Why don't they just build AI from scratch? Because they can't. They just claim they found a legal loophole to exploit other people's work without consent. I am still hoping the legal people take time to understand how LLMs work, how other algorithms, such as synonym replacement or c2rust work, decide that calling it "AI" doesn't magically remove copyright and the huge AI companies will be forced to destroy their existing models and train new ones which respect the licenses.

wvenable 8mo ago
> If LLMs are not derivative works of the training data then why is so much training data needed? If you went to school for 12-16 years, that's a lot of training. Does that mean anything you produce is a derivative work?
- martin-t 8mo ago
  I see this argument sometimes and it's annoying because: 1) People phrase it as a question even when they've already made up their mind (whether that's your case or not). 2) It implicitly assumes that humans and algorithms are the same. They are not - humans have rights and free will, algorithms don't. Humans cannot be bought or sold, etc. To your question: a) If you're asking whether teachers should get compensated according to how good a job they do, I think so. They are very often undervalued, especially the good ones - but of course that means the job attracts people who do it because they enjoy it (and are therefore more likely to be good at it) rather than those who chose jobs according to money and then do the bare minimum. b) There's a critical difference - consent. Teachers consented to their knowledge being used by those they taught. I did not consent to my code being used for training LLMs. In fact I purposefully chose a licence (AGPL) which in any common sence interpretation prohibits this used unless the resulting model is licensed under the same license - you can use my work only if you give back. Maybe there's a hole in the law - then it should be closed. I am now gonna pose a question to you in turn. Do you think people should be compensated for the full transitive value of their work?
  - wvenable 8mo ago
    > It implicitly assumes that humans and algorithms are the same. They are not - humans have rights and free will, algorithms don't. Humans cannot be bought or sold, etc. I don't think that's a necessary condition for that argument. You're making the implicit assumption that humans are special snowflakes and anything that we do cannot be replicated by computers, in any form. That's a very strong position to make without evidence. Is an LLM even an algorithm in the traditional sense? Is human cognition not an algorithm of some sort? I studied cogitative science decades ago and these questions weren't clear then, they're certainly even less clear now. It's also somewhat begs the question; this isn't even relevant to what we are talking about. Whether something is a derivative work or not does not require this discussion. Teachers are not relevant to conversation. You can learn by reading books, watching TV, using and reading software. Basically all of copyrighted and non-copyrighted human expression is available for you to consume and then creatively produce your own works using that knowledge. > Do you think people should be compensated for the full transitive value of their work? The short answer is no. Not everything that someone simply dreams up can or should be monetized forever when sampled by other people. That sounds like a radical position but actually the current state of "intellectual property" has only existed for an extremely brief bit of human history. What has most greatly shaped our culture and knowledge has been effectively free for anyone to use, modify, and reproduce for hundreds of years. That's not to say I don't support copyright as a means to support creative works but I would argue that it's an imperfect system. We're starving human minds of modern culture and knowledge often not even for someone's monetary gain but simply because the system demands it. It's ironic that artificial intelligence might actually free us from these constraints. I purposefully choose a license (Apache) for my open source work to make it widely and freely available.
flavionm 7mo ago
> The person who did the work deserved credit and compensation. That's the part of the argument in favor of copyright that is inherently flawed. Doing some amount of work doesn't entitle you to anything besides whatever you've agreed to get for that work, or possession of the output, in case you did it for yourself. But that's all you're entitled to get. Work itself doesn't have any intrinsic value, only output does. The scarcity of output is what dictates what is actually valuable. Creative work has the characteristic of its marginal cost being very high for the first copy, but nearly zero for additional copies. That's true simply because of the nature of such work, it isn't something that is unfairly imposed upon creative workers. Whenever you choose to engage in creative work, you know that, or at least you should. And if you choose to give away the first copy for free, or very cheap, that's your prerrogative, but it doesn't inherently entitle you to anything else besides the value of that first copy. Yes, there are laws such as copyright laws that exist to artificially inflate the value of additional copies, but they go against how things work naturally, so you shouldn't rely on them, and you certainly shouldn't base your moral compass on them. Now, I do still prefer copyleft licenses over permissive ones for the work I choose to give away for free, but only to stop corporations from taking that work and then using copyright laws to keep it exclusive to them. Once copyright is no longer an issue, they won't be necessary anymore.
- martin-t 7mo ago
  > it doesn't inherently entitle you to anything else besides the value of that first copy What if many people want to buy a copy but nobody wants to be the first because he'd pay the full cost and the others could get the next copy for free? What if they instead agree to share the cost equally? But then this group doesn't wanna be the first because the next group could get it for free? So what if they get the whole country together and agree to share the cost? How much? They agree to determine that by supply and demand. Congratulations, we're just reinvented copyright laws. Anarchists and especially ancaps never game things out in their heads because they'd arrive at some variation of the current system. Sad because they usually could come up with an improved version. But they choose to throw out the baby with the bathwater.

wvenable 8mo ago

> If LLMs are not derivative works of the training data then why is so much training data needed? If you went to school for 12-16 years, that's a lot of training. Does that mean anything you produce is a derivative work?

martin-t 8mo ago
I see this argument sometimes and it's annoying because: 1) People phrase it as a question even when they've already made up their mind (whether that's your case or not). 2) It implicitly assumes that humans and algorithms are the same. They are not - humans have rights and free will, algorithms don't. Humans cannot be bought or sold, etc. To your question: a) If you're asking whether teachers should get compensated according to how good a job they do, I think so. They are very often undervalued, especially the good ones - but of course that means the job attracts people who do it because they enjoy it (and are therefore more likely to be good at it) rather than those who chose jobs according to money and then do the bare minimum. b) There's a critical difference - consent. Teachers consented to their knowledge being used by those they taught. I did not consent to my code being used for training LLMs. In fact I purposefully chose a licence (AGPL) which in any common sence interpretation prohibits this used unless the resulting model is licensed under the same license - you can use my work only if you give back. Maybe there's a hole in the law - then it should be closed. I am now gonna pose a question to you in turn. Do you think people should be compensated for the full transitive value of their work?
- wvenable 8mo ago
  > It implicitly assumes that humans and algorithms are the same. They are not - humans have rights and free will, algorithms don't. Humans cannot be bought or sold, etc. I don't think that's a necessary condition for that argument. You're making the implicit assumption that humans are special snowflakes and anything that we do cannot be replicated by computers, in any form. That's a very strong position to make without evidence. Is an LLM even an algorithm in the traditional sense? Is human cognition not an algorithm of some sort? I studied cogitative science decades ago and these questions weren't clear then, they're certainly even less clear now. It's also somewhat begs the question; this isn't even relevant to what we are talking about. Whether something is a derivative work or not does not require this discussion. Teachers are not relevant to conversation. You can learn by reading books, watching TV, using and reading software. Basically all of copyrighted and non-copyrighted human expression is available for you to consume and then creatively produce your own works using that knowledge. > Do you think people should be compensated for the full transitive value of their work? The short answer is no. Not everything that someone simply dreams up can or should be monetized forever when sampled by other people. That sounds like a radical position but actually the current state of "intellectual property" has only existed for an extremely brief bit of human history. What has most greatly shaped our culture and knowledge has been effectively free for anyone to use, modify, and reproduce for hundreds of years. That's not to say I don't support copyright as a means to support creative works but I would argue that it's an imperfect system. We're starving human minds of modern culture and knowledge often not even for someone's monetary gain but simply because the system demands it. It's ironic that artificial intelligence might actually free us from these constraints. I purposefully choose a license (Apache) for my open source work to make it widely and freely available.

flavionm 7mo ago

> The person who did the work deserved credit and compensation. That's the part of the argument in favor of copyright that is inherently flawed. Doing some amount of work doesn't entitle you to anything besides whatever you've agreed to get for that work, or possession of the output, in case you did it for yourself. But that's all you're entitled to get. Work itself doesn't have any intrinsic value, only output does. The scarcity of output is what dictates what is actually valuable. Creative work has the characteristic of its marginal cost being very high for the first copy, but nearly zero for additional copies. That's true simply because of the nature of such work, it isn't something that is unfairly imposed upon creative workers. Whenever you choose to engage in creative work, you know that, or at least you should. And if you choose to give away the first copy for free, or very cheap, that's your prerrogative, but it doesn't inherently entitle you to anything else besides the value of that first copy. Yes, there are laws such as copyright laws that exist to artificially inflate the value of additional copies, but they go against how things work naturally, so you shouldn't rely on them, and you certainly shouldn't base your moral compass on them. Now, I do still prefer copyleft licenses over permissive ones for the work I choose to give away for free, but only to stop corporations from taking that work and then using copyright laws to keep it exclusive to them. Once copyright is no longer an issue, they won't be necessary anymore.

martin-t 7mo ago
> it doesn't inherently entitle you to anything else besides the value of that first copy What if many people want to buy a copy but nobody wants to be the first because he'd pay the full cost and the others could get the next copy for free? What if they instead agree to share the cost equally? But then this group doesn't wanna be the first because the next group could get it for free? So what if they get the whole country together and agree to share the cost? How much? They agree to determine that by supply and demand. Congratulations, we're just reinvented copyright laws. Anarchists and especially ancaps never game things out in their heads because they'd arrive at some variation of the current system. Sad because they usually could come up with an improved version. But they choose to throw out the baby with the bathwater.