This is actually not necessarily only a rich thing. There are numerous artists that will accept commissions and create something for you, unique to you, and it's not thousands of dollars.
But your statement brings up (or at least makes me think of) what I think is an interesting point: there's the notion of mass media, which started when broadcasting became a thing. Play 1 recording via a method where potentially millions can hear or see it. It was after the development of broadcast technology that copyright maximalism took hold. IMHO the current IP regime is for that purpose, and cracks are showing in it because media is becoming more fragmented and more personalized. Would mass media ever become a thing of the past?
Cool, sounds like the 20X0's version of "we'll just outsource the engineers".
Of course, they have relatively little to lose and can just reverse course in a few years, likely undercutting the market despite talent being more efficient and self sufficient and able to generate more funds for the company (if I take the charitable end road for this AI stuff).
the actual answer is "those who apply themselves will be fine and those who spend all their days wallowing in marxist fantasies will stay poor (both spiritually and monetarily)"
I have not read the article, just responding to title:
Hell yeah!
I do wish I.P. dead. Or at least, I wish for a much shorter protection period, and for original authors / artists only. For all the wrong reasons, but primarily, because nothing's sacred. Amen.
Do intellectuals / creatives not deserve the fruits of their own labor?
I get corporate IP is used for a lot of bad. But intellectuals and artists create a lot of value for society in ways that are very hard to monitize as a individual. If we completely kill IP what becomes of our authors, philopshipers, historians, artists etc.?
They are already having it hard as it is with existing ip protections vs. big tech, big pharma, big record labels, etc.
People do abuse the protections, which are too lengthy, makes possible to sell rights, etc.
In the end, corporations (ab)uses (buy and sell rights) protections for amassing fortunes much bigger than the original creator ever got. That's not right, not fair, at all. System's broken, need fix, throw it all out and build something new, centered around creators.
I can get behind that for sure. I just don't like the blanket "get rid of IP" especially when we are talking about AI companies which already well surpased the legal and ethical limits of it.
IP rights are about retribution to the creators. It might be more important to fix the retribution paradigm rather than fixing IP, which is a broken system.
At the moment, most creators are starving, while everyone, AI included, are feeding on them - big time. Successful then powerful creators only may dream about making AI pay - to them only.
Maybe it's time for some minimum revenue scheme for all creators, on top of heavy limitations on IP. Maybe AI corps can pay for that?
No subscription and archive.is is broken on 1.1 but it seems like the conversation is finally moving in a good direction.
I'm very very tentatively hopeful that we might see some reduction of IP rights in my lifetime. It's going to be a long hard fight but maybe, just maybe we will see people wake the fuck up and realize how important the building up the fucking commons is to the progress of all humanity.
I assume you're referring to copyright. It could be argued copyright is a tragedy of the commons, the individual benefits at the expense of everyone else. Obviously it is intended that the wider impact of incentivising innovation will be beneficial to everyone, but that is not a sure thing, it is a compromise, a balancing act and that applies to AI too.
The Luddites did lose their jobs after all.
Please explain to me how baking all the art into a big tensor and then giving it away to everyone for free to run on any half decent GPU from the past 10 years is somehow anti-commons.
Rent-seeking by controlling AI models is anti-commons, the models themselves are not, neither is training them. Don't delude yourself into thinking that Miyazaki didn't study Moebius didn't study Mucha. Even if you got your way and all training on copyrighted material was stopped today and all existing models were deleted it would only be a matter of time before we could synthesize arbitrary styles (at least as well as we can today) by identifying and controlling the correct axes in tensor space using a model purely informed by public domain data.
Your art is mostly not yours, your art is mostly an amalgamation of other people's art which itself is also the same, all the way down. You are a sprinkling of flavor on top. Your art is stolen styles, techniques and vibes mashed together. And here you are, trapped arguing about where we draw the line on what's okay. 75 years? now 95? Look at what you're saying! Introspect and realize that your understanding of the context is defined by Disney and Bono wanting to collect rent on ideas. You're just focused on getting your crumbs of the pie in a game that's rigged to centralize power and keep it centralized.
We all know that an artist studying the masters to make something original isn't the same as an autocomplete API picking the color of the next pixel. Come on, man.
I did not. You and GP are missing the point. I'm trying to make you understand how much of art is stealing from others, that you're incorrect about how art works at a basic level. Art is mostly drawing upon the commons and your understanding of artist rights is based on a fiction created by corporations and wealthy rent seekers seeking to retain power in our economic system.
The jist of what you're saying is in the Free Culture movement/philosophy, I think, and it resonates with me as someone who does not like what AI has done with copyrighted works but also dislikes copyright.
The "rent-seekers" is the problem. We collectively inherit and own our shared culture, but large corporations have always wanted to sell it back to us. AI companies are arguing they should have no limitations on their usage of the culture, but that the same shouldn't apply to them. Selling tickets to the commons is anti-commons.
Perhaps if these companies were themselves arguing for the end of copyright and IP for everyone, the conversation would be different.
I think copyright lawsuits against AI companies will force them to develop attribution models. They will do the work of indexing all ideas to their authors. This will also reveal what is common knowledge, and who borrowed from who without attribution.
In order to make attribution models we need text+author+timestamp. We can get that from books, newspaper articles, scientific papers and social network posts. Then we extend to the rest of the training set.
But then we can also make AI models that cleverly avoid infringement while the same strict checking is going to be applied to human made content. Humans are not that good at avoiding pitfalls.
Right, I think we’d be having a very different conversation if products of AI could not be sold (what’s taken for free must be given for free and free for others to use as they see fit). Companies will fight this tooth and nail though because that flushes any prospective profit produced by firing humans down the drain.
> Even if you got your way and all training on copyrighted material was stopped today and all existing models were deleted it would only be a matter of time before we could synthesize arbitrary styles (at least as well as we can today) by identifying and controlling the correct axes in tensor space using a model purely informed by public domain data.
Do it then.
> Your art is mostly not yours, your art is mostly an amalgamation of other people's art which itself is also the same, all the way down. You are a sprinkling of flavor on top. Your art is stolen styles, techniques and vibes mashed together.
Eat shit. People deserve to have some ownership over the product of their labor. Imagine doing all that work to learn all these skills and some rat fuck nerd tells you they have the right to alienate you from that work, making some greasy fucking argument that the multi-billion dollar program they made is doing the same thing you do.
Corporate abuse of copyright isn't an excuse for corporate abuse of people who can actually make things.
The tragedy of the commons relates to deletable resources, such as grass feed for livestock. How's that relevant to ML models that create content/resources about as fast as most can think? What's being depleted?
Qualifying this statement by saying this is something I've thought about and this is my own bullshit theory.
There's a limited amount of "attention" on the internet that's available in online art spaces. I theorize that artists need to maintain a public portfolio to find work, either creative work for advertising or entertainment companies or commissions or patreon subscription. You need to be out there for people to hire you if you want to make a living doing art.
AI art is disruptive and harmful to that ecosystem in a few ways. It competes with artists not only for dollars and jobs for commercial art (ie ad agency work), but more fundamentally it competes with artists for attention on the internet, which at a macro scale makes it so there's less money in the system for real artists to practice and master their skills. It's also not necessarily that the AI art is better, it's just good enough to pollute the signal to noise ratio. And as you said, it's fast. That can quickly overwhelm sites like DeviantArt or Art Station, which moves those communities and resources away from their original purpose.
Additionally, AI art models need data from the commons to function. I'm skeptical that synthetic data is good enough to use for improving these models, so theoretically, one of the main ways to improve these models is getting more and better data from humans. If the economy of these models is structured in a way that discourages artists from posting their work publicly, or working at all, then the pace of improvement decreases. I think a lot of artists are pretty pissed off that their work is being used to produce commercial software that creates substitutes for their work without their permission or knowledge.
So that's what I mean by the tragedy of the commons. The online art ecosystem is more fragile than many people think it is and ai companies are over exploiting it. The content remains but the thing you're actually trying to sample, human knowledge and skill, withers on the vine.
I for one am looking forward to LLM-based super-decompilers that can recognize assembly patterns and understand what the high level intent was. Suddenly there would be a wealth of effectively open-source code (with the hope that it wouldn't change much about SaSS-ifying things since that's mostly already done where the market will bear it).
Software people here, we dont want ai people to steal our open source code, hide it behind closed models, and sell it without attribution or other terms respected.
Many in the software world are in awe how much theft ai people do.
Open generative models are to open source what open source is to closed source - an even deeper level of openness, customisability and accessibility. Like open source they empower their users, are private and easy to adapt. Most of the time you just need to prompt, other times you apply a fine-tuning tool, which is also open source.
I really don't understand open source people combating generative AI. I've never seen this ethos since the javascript framework wars. Take a look at llama.cpp, ollama and vllm repos and their friends. They got such a sustained rate of development and participation.
I dont think open source people are against open source models. Most are against work being used without their licensing respected. If models trained by companies on open source data would be open source that would be a win. But they arent. Furthermore, software people are heavily opposed to stealing artist’s work.
In general software developers are already highly (or completely) dependent on permissively licensed open source software; this would at its worst produce more of it no?
When it costs zero to produce a copy of a novel is the moral choice to create artificial scarcity to preserve it's value in our current system of value assignment, or is the moral choice to fight to change and improve the system to more align with the reality that novels exist in a post-scarcity world?
Do we really want to deny people something that costs nothing in order to preserve the status quo? Perhaps it's the system of value assignment that's broken.
I'm not claiming to have worked out how to make this all function. I'm just pointing out that the cost of duplicating a novel approaches zero and the assumption that the way we're assigning value is the only one that works leads to a world where we must deny people things they could have in order to preserve their value. This does not seem morally correct to me. I do not need to have a solution in order to identify a problem. Perhaps we could make strides toward a future where you don't need to be independently wealthy to write a novel by having some sort of a UBI scheme.
That's the basic argument that hackers make. Creativity shouldn't be a way to make a living, but a cool hobby for people who are already employed with comfortable salaries and their real estate paid off. Trying to sell your creativity is grifting. In fact, the novel writer should pay the hacker if the hacker pleases to read the novel, because the novel writer is using the hacker's valuable attention. Or at least I've heard around here that attention is very valuable.
I wonder if traditional novels have been out-paced by free online texts that people write in their free time (fan-fics and/or original-fics) ? It'd be interesting to crunch the numbers. Productivity is enormous, though quality isn't always as high. Of course with sufficient monkeys on keyboards, there's bound to be a few Shakespeares out there.
We have had more content that we can chew for 25 years, all accessible through search. Yes, we love it. But it's nothing new. We don't need AI to have good, free stuff to read.
It can end up being bad for both readers and writers in the long term.
If writers aren't able to find profit in writing then there will be less people spending the time and effort to produce good writing, which in turn means less quality material for readers to read (and, as an aside, less quality material for an AI to train on in the future).
We could arguably reach some type of feedback loop where less quality writing leads to less interest in reading, which leads to smaller reading audiences, which leads to less quality writing being produced, etc.
People do love free stuff, but free stuff often isn't sustainable under our current economic system.
> free stuff often isn't sustainable under our current economic system
I disagree - some people live their whole lives on benefits or similar entitlements. That's an incredible economic achievement, as well as being a brilliant show of hard work by the net contributors.
The problem is that our society has no other viable widespread mechanism to support the many millions of people who do intellectual work, and lots of intellectual work is really important for our society-- far more important than would make sense to relegate to hobbyists and volunteers. And this is coming from someone who's contributed well over 10k coding hours to FOSS software.
What would you propose the US workforce did with some 47 million knowledge workers with suddenly useless skills if you removed the only widespread economic mechanism our society has to support them?
Yeah but we need the Star Trek society first. In principle, I absolutely agree that IP is bullshit. The problem that many people in the “smash IP” camp tend to ignore is that people needing to eat, have a place to live, not having a stage 1 cancer diagnosis be an instant death sentence, and things like that aren’t merely “regrettable collateral damage” on the road to freeing data.
I’ve encountered many people over the years that hardline shit like that—they’re almost exclusively middle to upper middle class suburbanites that are edge lording that stuff to subconsciously quiet their deep-seated insecurities about how incredibly soft their comparatively unchallenged existence has made them.
I think so. Not just because of copyrights as the article covers, because the people who want to abuse copyright for A.I. and the people who own copyrights are basically the same people. They will simply cross-license and lock normal people out of using the products of A.I. by wielding that licensing against individuals.
The real reason is that anybody can do A.I. and it can't be very patent-encumbered, being a number of abstract mathematical techniques. If it becomes an absurdly productive technology, you won't easily be able to keep people from using it in private. Maybe the real Butlerian Jihad will be the government Office of A.I. Copyright Royalties sending agents and their silicon-sniffing dogs to kick in people's doors following rumors of Aggravated Infringement.
Maybe it will be what finally gets general purpose computers banned?
No, it will result in a strengthening. It raises the value of things previous unprotectable, and thus will raise the incentive to get them protected. The Wild West of the AI world today will rapidly get captured because by the time you've amassed the resources to put up a credible AI, you're already a fairly large corporation. The scale of time necessary to get new protections passed is roughly comparable to the scale of time it will take for these companies to become owned one way or another by existing interests.
The good news is that this probably still won't much affect your personal projects or anything. Anything you can already self-host and do today isn't going anywhere and you've probably got at least another two or three generations in this fast-moving space before the law clamps down. It is likely that many of the referenced AI aspects will be essentially "solved" before then, such as voice cloning. They aren't going to go after every last little AI user for every last thing because that's squeezing blood from a stone and not cost-effective. But don't plan on building a multi-million-subscriber YouTube channel out of it.
And if you are interested in the freedom to do things the copyright owners don't want you to do, be sure to self-host and archive as much as possible. The ability to do this on platforms others host is going to disappear rapidly, if it isn't already.
This is a really interesting take, but I don't completely buy it. Traditionally technologies like these become more accessible with time. Take the youtube example, this is actually a really good counterpoint as if you can create videos with an AI you created, nobody has any way of figuring out that you trained it on so-and-so's IP. It's completely black box which is the direction things are probably going to go IMO.
On the other hand, this is problematic because it'll slow down the development of the self-hostable models outside of the community moving to a darkweb like model where they stay beyond the reach of the legal system.
One of the big factors is whether or not AI can stay in the hands of the public or if it recedes into levels of hardware that only companies can own. If the latter, then the control will win.
Keeping AI performance workable on personal-level hardware is going to be a big deal. Open source work on keeping performance at that level may be long-term more important that this or that feature.
Watch out for calls for AI ethics to turn into calls for controls on hardware capable of running AI. I'm half surprised we aren't already hearing calls for that. I wouldn't be surprised to see it by the end of the year, though my expectation would be at least next year if not the one after. The real goal behind hardware control will be this sort of thing (though copyright control may actually be a secondary concern versus just keeping the really good tools out of the hands of the plebs), not AI ethics. (The elites know AI ethics limitations have no meanings as they have no intention of limiting themselves.)
everyone already has computers that can "run AI". would be pretty ridiculous if some law was passed and everyone's computers were confiscated. that would be like a lobotomization of society. a very stupid and short-sited move i think.
> One of the big factors is whether or not AI can stay in the hands of the public or if it recedes into levels of hardware that only companies can own. If the latter, then the control will win.
Both will win. As corporations will make more and more advanced models, training data generated by their models will uplift open models. They only lag a few months. We already can shrink most of chatGPT's abilities in a 7B model. All open models trained after 2022 are benefiting from their big brothers. It's too easy to exfiltrate skills from proprietary models.
It's been dead a while because of big tech. YouTube, Google, Facebook, Twitter - they have no way of effectively managing IP and so you basically see people's stolen content and they monetise it, so <shrug>
TBF Napster and torrents are another and earlier angle.
Purely IMO, big tech normalised the taking of other people's content, and it's getting to the point where content creators (as in, ones that are capable of producing unique content, not wealthy YouTubers per se) have simply had enough. The idea of authorities and primary sources is rapidly diminished with the hodge podge we have to navigate through in information discovery.
Any case where a small player cannot effectively defend patents they hold, but at any time, can be sued for infringement and cannot effectively defend themselves either.
What had sprung to mind for me was the wholesale rip off of European IP by the US in its commercial development in the 19th century and China now with the US.
I guess what I'm thinking of is at the individual level, no one cares much for re-appropriating someone elses work. It has been normalised.
> The pharmaceutical industry continues as a giant, blood sucking, vampire squid because of IP.
Let's be honest here. In order to get that way, they needed to stomp out and eradicate every trace of "evidence" that natural medicine was efficacious, because herbs, botanicals, and other natural substances can't be patented, can't be synthesized in factories, and can't be exploited for profit at megascale.
If word ever leaked out that, for example, selenium, valerian root or St. John's wort were equivalent/better than synthetic, patented medications, we'd all be in trouble.
I can buy valerian root tea at multinational grocery store. The problem with supplements isn't that they're disenfranchised, and it's not even that they strictly don't work. It's that the huge businesses that make tons of money off of them have so aggressively resisted regulation based on effectiveness, or even contents, that you can't count on what you're buying not poisoning you, let alone helping.
And I'd love to see a natural replacement for insulin that doesn't kill people, but I'm pretty sure it doesn't exist. So maybe pharmaceutical technology does a few things that natural remedies can't?
>And I'd love to see a natural replacement for insulin that doesn't kill people, but I'm pretty sure it doesn't exist.
No sugar, no alcohol, low carbohydrate diets are the answer to the diabetes epidemic. If you want to regulate something, start with the corporations marketing processed, poisoned garbage to people as "food" and "beverages".
> TBF Napster and torrents are another and earlier angle.
Napster didn't kill the music industry, torrents didn't kill the movie industry because people still want to see movies on a big screen experience. And so, actors, directors, VFX artists, cutters, riggers, lighting staff, costume/mask people, they all continue to have jobs.
But now, with AI being already able to do very realistic photos, and in a few years likely to create short movie sequences all based upon a prompt, not only will all these people be replaced by one prompt artist feeding a massive AI engine... but the diversity the AI can generate will always be limited by the diversity of its training material. It is by definition incapable of trying something entirely new, and since there will be no economic incentive to try anything entirely new there will be no expansion. (Oh, and guess what, movie prices aren't going down but since the expenses for all the humans vanish the profit concentration will accelerate!)
Widespread AI will freeze our culture in an era of about 2020-2030 forever simply because capitalism will not offer any incentive to feed the AI with new, creative things.
> torrents didn't kill the movie industry because people still want to see movies on a big screen experience
Given you believe the reason torrents didn't kill the movie industry is because people weren't satisfied with smaller screens: would you be satisfied with content that is stuck in "2020-2030 forever" and can never produce anything "entirely new"?
Personally I think novelty in art thrives when the entry barrier is low and experimentation is possible without a huge budget - which AI can help achieve.
> Given you believe the reason torrents didn't kill the movie industry is because people weren't satisfied with smaller screens: would you be satisfied with content that is stuck in "2020-2030 forever" and can never produce anything "entirely new"?
I fear the enshittification of movies - similar to, say, the situation in video games or residential ISPs: when everyone has silently agreed to a common level of base bullshit, customers won't move (because why move from a pile of cow dung to a pile of dog dung? it's all dung in the end), and everyone can keep fleecing the customers.
> Personally I think novelty in art thrives when the entry barrier is low and experimentation is possible without a huge budget - which AI can help achieve.
Oh even now even utter novices and two-three person teams can produce hours of high quality content, youtube is enough proof of that. Experimentation is not the problem, the problem is all the marketing to get people to sit in the cinema. For a lot of modern movies, music and video games the cost of production is at least on the same order of magnitude as the cost of marketing - even a blockbuster movie like Avengers Endgame [1] reportedly had 356 M$ of production cost vs 200 M$ of marketing cost, and of the production cost at least 100 M$ were compensation for the star actors.
> or residential ISPs: when everyone has silently agreed to a common level of base bullshit, customers won't move (because why move from a pile of cow dung to a pile of dog dung? it's all dung in the end)
So, if I'm understanding, you believe you wouldn't be satisfied with the movies, but wouldn't have a choice because of a silent agreement (as with residential ISPs) leaving customers with no good options. Is lowering barriers to competition not one of the ways that such oligopolies are fought?
To me this seems far more of a risk in our current environment where the vast majority of high-production-value movies come from a small handful of companies that have the budget to make them (and give us plenty of safe low-risk sequels).
> Oh even now even utter novices and two-three person teams can produce hours of high quality content, youtube is enough proof of that
There are reasons why working-class people are vastly under-represented in arts - the time and resources required to create "hours of high quality video content" is a hefty investment even for Youtube videos, not to mention a feature film that people will want to watch.
> and in a few years likely to create short movie sequences all based upon a prompt
It depends on the director and style of the era, but average "shot length" runs between 5 and 14 seconds. Six months ago companies were already demonstrating AI generated, 20 second shots (albeit with low motion, like slow pans) with the background staying mostly the same. I would probably upgrade this statement from "likely to create" to "almost near certainty".
> It is by definition incapable of trying something entirely new
Only if the AI doesn't see any usage. But when you start using it, you get AI+human generating things that can be outside its training scope. Humans can input novel ideas into the prompt or include new information. Then AI gets to retrain on the data generated with human in the loop and improve. There is a chance for humans and AI to explore new directions.
If you got a simple way to test, like the DeepMind geometry problem solving model, then AI models can improve on their own by doing massive search and validation. Kind of similar to the scientific method - formulate hypothesis, validate, observe outcomes, and repeat. Works for code too, or any AI output that can be "executed" and generate a validation signal.
Completely unfounded fear, fortunately. AI will enable more people to be creative than ever before. People won't stop telling stories because of capitalism, they'll just continue the tradition of storytelling that predates capitalism by many millenia.
There was no other outcome possible from the moment we invented a technology that allowed everybody to create infinite copies of digital content at almost zero cost.
If tomorrow we discovered a source of infinite free energy, oil and energy companies could try anything they wanted, their current business model would end up obsolete sooner or later.
Can't say I disagree it was inevitable but could have been legislated for, which seems to be years behind the times as always. My main concern is just unique content that is adequately rewarded that isn't siphoned off by large platforms that can viralise said content, or along those lines.
Maybe even if the big tech could pass on some (most) of the proceeds because of the (content share), it'd be far less a thing. Otherwise, 90% of people being made aware of an idea see it somewhere beyond the original source and those secondary sources capitalise because... simply because it's the norm.
Nuclear power plants have a dramatically higher upfront cost than personal computers. You can't really ignore the capital cost and call the energy free.
A better example is solar power, which is slowly but surely whittling away at fossil fuels.
It inherently has to damage our present conception of IP, unless we plan to cripple the AI revolution in the crib.
If you have smart robots in the home, much less AGI, what's the plan on restricting them? Sorry robot, you may only learn this limited set of skills, you're not allowed to create, because you've seen too much and might be too good at learning and producing.
They'll be incredibly metered if they're not allowed to learn and produce at a high capability.
And if we're talking about AGI, you can never realistically reach it if you put it in the straightjacket required to protect IP as things are today. Sorry all-knowing AGI god, you may not make images that look too much like XYZ. In fact, you're not allowed to even see any of it because of your capabilities. In fact, Disney has decided through the copyright act of 2042 that you're not even allowed access to the Internet (or magazines, or radio, or music, or TV).
It can't be AGI if it can't create. Forget about trained on ('just do not train it on material with copyright'), it'll self-train, and then what? Well then you can't even let it self-train, ie no AGI is possible.
It's going to be either or. Either the IP laws get changed, or you get no AGI. The only thing possible would be extremely boxed off, narrow, high capability AI.
Humans are too stupid to collectively govern themselves. Just look at the US election cycle that is starting. There is a genuine messianistic cult at work.
I, for one, do not want to live in your new world.
Further, I doubt that preventing AIs from literally stealing protected information will halt the rise of smarter AI. There is plenty of free use information out there that is LEGAL TO USE. Just curate your dataset to prevent the ILLEGAL use of other material.
Depending on the outcome of the NYT case, I.P. may in fact be the death of (profitable) AI.
One problem with AI (vs e.g. Google search) is that there is no mechanism for attribution. So it is impossible for rights holders to understand whether their works are being monetised (unlike, say, Spotify).
Should NYT win, then either AI services will have to pay into a (very large) common pot for rights holders, which likely will make AI uneconomic (even if it does turn out to have some business benefit), or (paid) AI services will simply be banned.
>> One problem with AI (vs e.g. Google search) is that there is no mechanism for attribution.
Sorry if this is a silly question, but is that really the case? Can we not train an LLM on successively larger training sets, each will uniquely ID's model and associated better performance?
There’s no way AI services are going anywhere. If the US kneecaps itself for parasitic rent seekers then China will win and benefit from the productivity these tools produce.
It’s not impossible. AI developers can 100% trace node that contribute to a result during the processing of a prompt, therefore they can be attributed.
They’ll never do this, of course, cause they’re riding high on wins of stupid people pushing “you can’t work backward from a result to find the nodes” (which is probably somewhat grounded in reality).
Attribution is absolutely not impossible, they just want to think it’s not possible, because enabling such a thing would show just how egregious the copying actually is.
As far as I'm aware, what you propose really isn't a solved issue. Most naive approaches you'd think might work are either infeasible or give chaotic nonsense results.
Partial matching between the output text and the training data is possible as a step after generation (though doesn't determine whether it's coincidence).
I'm not sure I like the direction AI is going anyway.
Right now we're on a path towards just a handful of megacorps with AI agents that will make large numbers of white collar workers redundant. Imagine something like closed-source software only much worse.
In the last six months model size has come way down and quality has gone way up through different training styles/methods. There is an incentive to release separate models capable of running on 8/16/32gb ram. On my laptop I have something like six free/open source models installed. LLMs are getting so easy to make that people are doing this in their spare time as a hobby. This is starting to feel like a small stepping stone towards AGI but only time will tell.
So essentially ai is profitable only if it resells stolen property. Almost as if it’s not intelligent and it only procedurally generates output. The more the better.
Just because it's publicly viewable does not mean it's free for reuse. Imagine an author who uploads chapters of their book to their own website. You cant take that and claim it's yours.
Similarly with all the publically viewable source codes on GitHub/Lab. Just because you can view the code doesn't mean there aren't licenses in play for how you are able to use it.
Further, what happens if stolen information is posted publicly? (E.g. wholesale copies of full books on html pages?
Just because it's public doesn't mean the viewer has freedom to use it as they see fit.
That 'usage' is taking statistical notes about the work (creating factual statements about the work) and imputing those notes into a database, averaged with a few billion other notes about other works. That is a usage that copyright under current law simply doesn't cover or protect for. It doesn't even need the analysis if 'fair use' because, there's no copying or public performance happing in the creation/training of the model.
Where infringement arguably can happen is when that model is used in the generation of content - and if the user is prompting regenerate a protected work, then that is where the infringement happens. But not before. Maybe the various ai services can adequately guard against that illicit usage. Maybe not. And if not, its those live services that would need to be shut down.
But the creation and training of a model, and even distributing that model for people to use with their own computers in private does not engage in copyright infringement.
All digital content is statements about the work. Pictures are pixels with facts about colours. Books are characters describing the words.
AI models are compressed versions of those.
Downloading, ingesting into models, and redistributing them is copyright infringement.
Even if they weren’t compressed; downloading them without permission is not permitted. Every now and then you can read about ai criminals discussing techniques for illegally acquiring books, images and other protected content even here on HN. It wouldnt be an issue if it were licensed content, or a dudette training her local ai. These are companies that want to monetise people’s work. Somehow, just like thieves, they feel they are entitled to snatching someone else’s property.
If I create a sculpture and put it on my front lawn, I think it's fair for me to complain if you scan it and start 3d printing and selling copies, even if I intended it to be in public view. Just because I want the public to be able to enjoy my version of something, doesn't mean I want people to feel free to ingest it and monetize it for personal profit.
Right, but if I look at your sculpture and get ideas on how to make a better sculpture, then make an entirely different sculpture using yours as inspiration, and you get upset that I'm 'copying your style' we generally don't give you much sympathy.
What if I hand-painted an image (without AI) and God himself came down from heaven and explained to the world that my painting was derived 10% from my observations of the collected works of Lisa Frank, 3% from looking at Davinci paintings, and 87% a long tail of various sources. And then I sell my painting for $100.
No, because copyright law says that if you hand-paint an image (without AI), then as long as the images isn’t deemed (by a court, if necessary) to be a derivative work, then you, and nobody else, own the copyright. The word of God is not necessarily considered by the law. And even if the court takes the word of God as gospel, a 10% derivative work might not exceed the necessary threshold of infringement.
Alternatively, if I created a large body of works that were 90% Lisa Frank and brought it to a large market distribution, making significant revenue, then she could sue me for damaging her market and her brand. Which she has done to others in the past and rightly won. Go Lisa Frank!
But, if I used an AI tool in the process of making an image that came out 10% Lisa Frank, she wouldn't have a strong case against me. Nor does she have a case against the million random teenagers hand-copying her style but not making significant revenue from their copies.
Either way, AI doesn't really factor in here. The results are the results and the markets are the markets regardless of AI or paint brushes.
Do you think it's reasonable for artists to want a way to forbid major AI corporations from training on their work? Because some people will stop producing/stop making available their work if they don't get access to a different way of doing so.
Much like how some artists are uncomfortable with making high-quality scans of their work available online, because it becomes prohibitively difficult to prevent people from selling prints. Forget the current legal framework, it seems fair to me to say "I don't want that happening to my art" when "that" is a corporation finding a way to monetize it. Even if the AI is using less than a pixel of your work, I think it's fair to ask that it not.
I get that the concept is upsetting. And, I'm not going to tell people to not be upset or not state their will. Ex: DeviantArt has instated an opt-in checkbox for "It's OK to train on my art" and I think that's great.
But, everything I've heard and learned about how AI works leads me to the conclusion that such a move is almost entirely symbolic. Counter to what our egos would like to believe, the vast majority of the training images and the vast majority of the learning value for AI comes from random, crappy photographs of various things. Without a million photos of dogs and trees, the AI can't make a stylized dog distinguishable from a stylized tree. And, it's turning out that, from that base of crappy photographs, it doesn't take many high-aesthetic images to make a high-aesthetic AI.
Quality aside, art still has copyright for these exact reasons (and photographs are no different, long settled case law). So you can't just say "well no one cares about this blurry dog photo". Courts need more precise language than "care".
If it needs millions of photos of dogs and trees you
1. Take them you self
2. Pay someone with a db of dog and tree photos soem agreed upon amount (Aka someone did the work for you already and you value their work)
3. You scrape only on photos you are sure are CC0 or any other kind of permissive license. I'm sure you can find a few million dog photos this way.
But clearly they value #2 without payment so lawsuits were inevitable. Maybe because #3 took too much time and they decide #2 is worth the legal fees to be first to the race (something else I'd want fixed, but that's another rabbit hole).
In the mean time, we have and even worse situation: "AI" models like LLMs are effectively allowed to launder IP. As long as companies like OpenAI are allowed to be ignorant of copyright, they get to monopolize that ignorance.
The result is that the rest of us are in an even worse position, because we have to respect DMCA, but can't benefit from it by monopolizing our own IP.
Copyright is not a real thing. It's a fiction. Subverting copyright law is not like subverting the laws of physics. What happens when you subvert copyright law is some guys with guns, in either an overt way or a subtle one, force a trade of your imaginary resource (copyright) for the real resource (money), through licensing or fines.
Debate about what AI means for copyright is really a debate about what we choose to do, what we choose to allow. It's a debate about whether the guys with guns will stop the development of AI, or how much.
It's a bit of clever misdirection to declare the output of AI's "copyrightable" or "not copyrightable." Copyrightable is made up! It's fiction! It is whatever we want it to be!
Now, I don't have a college degree, but with my 60+ years life experience I can say, yes, copyright is a real thing, it protects people who create, from assholes and AI.
Am I the only one that loves a blunt and honest comment?
Funny how advocates against property rights have no issue with corporations protecting their own. That’s fine. What they want is to legally steal from the average joe.
Are you stealing from your lord, serf? Perhaps you should be churning the butter, not thinking for yourself. (I kid, I kid!)
It's going to be just as easy to make a movie or a song as it was to write this comment. I'm working and researching at the edge, and I promise you that the entire world will be in awe of these next 12 months. Go play with Suno for a taste.
I'm a senior engineer, and now I'm tab completing Rust code 70% of the time. It's hard to believe we've come this far, and it's only going to keep climbing in capability.
You're watching breakneck progress, and the genie isn't going back.
More art will be created per month than the entire human history to this point. Gen alpha is growing up on this stuff and using it to great effect to communicate amongst their peers. The things they build will be incredible.
I don't understand why the autocomplete meme is so compelling to people. How is drawing a picture, which modern AI systems are already capable of, "autocomplete"?
(1) I was responding to the actual thing OP said (70% code completion in Rust).
(2) It literally is autocompleting - it draws each pixel because it statistically determines that <some color value> is the best fit given the prompt and the prior pixels it drew. It's a more advanced robot, but it's still a robot.
AI systems are not “drawing a picture”. They procedurally generate output given vast amounts of input, much of which has been stolen. Without it there would be no credible output. I dont understand why are ai cultists hell bent on theft.
Llama, ollama, etc are not the issue here. Nor is ai. The issue is theft for training.
If it's theft, then your brain belongs to Disney and you need to start forking over your salary. Your connectome is littered with their platitudes.
This is the same argument the horse people had about the cars taking over the streets. In the end, progress will win.
Do we really want people spending hours and hours practicing and performing legacy acts of creation, with some people never able to summon the opportunity cost to even begin? For people to endlessly labor over creating a small amount of unrepeatable output?
Or do we want a world where everybody on the planet is able to turn their thoughts into something visible and tangible quickly? For society as a whole to have orders of magnitude bigger and better works?
This is a horse to car moment, but for thought itself.
> If it's theft, then your brain belongs to Disney and you need to start forking over your salary. Your connectome is littered with their platitudes.
I think it's quite telling that sooner or later, every discussion about "AI" and copyright ends up equating people to statistical models controlled by predatory megacorporations.
My connectome, in the eyes of the law, has a special legal status which is not shared by inanimate and non-sentient objects. In fact, some would argue that the entire point of the law should be to protect the special legal status of my connectome from those that would draw an equivalence between people and property.
"Open-source AI" is interesting in multiple ways, but not relevant to the legality of the models.
You know the "If ML is IP infringement then your brain is owned by Disney" take is just as sensible (and specious) for pre-ML stuff too, right? You're allowed to memorize the full script of any movie you want, without anyone trying even a little to claim that doing so should reduce your legal status to chattel, but obviously you're not allowed to just rip the subtitles from a DVD and sell it as a book.
The parent makes the mistake, to assume an adversarial relationship between an mp3 download from a torrent and the musician. Or an adversarial relationship between training an A.I. statistical engine to a painter and reproducing the style.
That's not correct. We are now increasing the capabilities of everyone creative, to achieve much more, almost free and instantly. The painter now, will have a 100 million film studio on his fingertips to create movies. The musician will be able make a high quality album, just from his snoring.
One usual misconception is that talent is not important anymore. That's certainly not true. Talent is not going anywhere. People not so talented can get some results which are ok or good enough, but talented people can create magnificent art, without even trying.
Also "tab completing Rust code 70% of the time" is breadcrumbs. I am working in making, Rust specifically, 100% code generated.
> I am working in making, Rust specifically, 100% code generated.
Generating code is trivial. Generating code that works is not. I suspect you are at around 10-15% of reaching that. All models out there are probably at 2-3% as they barely equal a google prompt in reliability.
Generating code that works, 100% percent computer-generated, i have already done that. I created a notification service for HN [1]. In the repo i have the ChatGPT conversation.
I don't put it to work for the moment, cause i am trying to create a TUI for HN, or a TUI forum more generally, like HN, maybe even backed by version control like Pijul instead of a database. 100% computer generated obviously, that goes without saying.
My biggest problem so far, is that GPT doesn't write idiomatic Rust. The code it generates is very C or Pythonic, and i am trying to figure out a way to guide it, to write beautiful code, instead of just correct code. When the machine generates just correct code, it doesn't cut it for me, is what i am saying. I am reading "Command Line Rust" to get a sense of idiomatic Rust, and then i have to study github projects as well to really figure out how beautiful idiomatic Rust code really is structured and written. Like, i know what idiomatic Rust is, just not that well at the moment.
The tool i want to create, to use GPT instead of Copilot, is something like Aider [1] but not in Python and maybe a TUI.
> Funny how advocates against property rights have no issue with corporations protecting their own
Are you just not aware that the most outspoken advocates against private property (and rights to it, including IP/copyright), also strongly oppose capitalism and the idea of corporations?
The movement against ip has been hijacked by those who wish to exploit creators. People advocating for the abolition of property are useful idiots. Corporations know that the biggest threat of ip is to them not to us. By being able to protect what’s ours means we can leverage the vast knowledge we’ve created by building open source models that make services provided by many knowledge corporations redundant. Ironically the only way to abolish ip according to our terms is by not abolishing ip. We can leverage it to explicitly prevent corporations from using open licensed code unless they too open their models. That’s the idea.
However, I am strongly against using artist or writer work against their will. That also benefits corporations. Also ironic that we can use against them the play they used against us. Imagine open models that the artist or legal or writing community can use but not corporations unless they open their models.
In just about a year, kids will be making entire Pixar movies from home. (The content will probably be Skibidi toilet related, but of Pixar scale and scope.)
How that reconciles with copyright, you've got me. It doesn't. It's a domain mismatch. It's completely outmoded by what's coming.
And the music industry is equally toast. I can already make an excellent banger song about superhero rodents in under 30 seconds.
The YouTube watch history of any 13-year-old provides a handy counterexample to your assertion.
Slightly less flippantly, the days when the nation gathered around their TVs to all watch "Leave It To Beaver" at 8pm are long gone, the media landscape has been fragmenting for decades and this is just the next step. My kids don't watch TV shows, they follow YouTubers.
A six-year-old filmmaker has as much claim to copyright protection as Spielberg and Tarantino. Just because one uploads to YouTube and one is paid millions of dollars by a major movie studio, it doesn't mean that they're different in the eyes of the law.
From what I understand, once a work is created, copyright is assigned to that work's creator. That creator may then license that work however they want. Quality doesn't factor in.
Now, I don't have a college degree, but with my 60+ years life experience I can say, yes, copyright is a real thing, it protects people who create, from assholes and AI.
There, I've stolen your post. Did copyright protect you? No, it did not. Copyright is an idea we came up with that sounded good at the time, and it's come to pass at this point that the original idea was deeply flawed in ways we could never have predicted, but now can see clearly in hindsight.
The truth is, intellectual property should be protected by those creating it. Coke does a great job of this, as do many, many other companies. We call these "trade secrets", but ultimately the concept is the same. You're protecting the work you deem worth protecting.
I don't buy the notion that copyright ever protected the creator. What it really protected was the interests of the entities who effectively enslave the creators via contract, and is not of any tangible benefit to the creators themselves. If one truly cares about the artisans among us, one cannot justify the existence of our ideas surrounding copyright.
Yes, removing those laws from our doctrine would cause upheaval, as the market must then rebalance itself in the absence of the artificial pressure we've put on it, but in time all things find equilibrium again, and placing the value and responsibility back on the individual is, in my mind, simple human decency.
> The truth is, intellectual property should be protected by those creating it.
The whole point of government and laws and societal norms is that not everyone has to be deeply involved and specialized in protecting their rights. We default to people doing the right thing and seek out specialists (e.g. lawyers) when we're wronged.
Some countries have roundabouts. Some countries let you make a right turn on a red light. Traffic law isn't any more universal than copyright law. All law is made up but that doesn't make it less real.
I can drive my car over the land border between two countries and have to switch sides of the road. Or the speed limit can go up/down. Or they can mandate different safety equipment.
I always found it fascinating how universal copyright law is. Countries that can't agree on almost anything else have all deigned to give (okay, now that it's 2024, we can't use the rodent anymore, what's the new symbol of eternal copyright?) holy sanction for decades at a time.
Much of the heavy lifting is done by treaty-- not so much that anyone is asking legislatures directly to establish and extend copyright, but that it comes part-and-parcel with other international obligations.
Without the weight of such treaties, I wonder how many countries would do without an IP regime at all, especially if they don't see any IP-related export opportunities, so copyright and patent merely limits their ability to make cheaper versions of foreign goods.
U.S. copyright also has a constitutionally defined purpose:
`[The Congress shall have Power . . . ] To promote the Progress of Science and useful Arts, by securing for limited Times to Authors and Inventors the exclusive Right to their respective Writings and Discoveries.`
Does the current scheme really promote the progress of science and useful arts? Is the current length of copyright really "limited Times"?
>What happens when you subvert copyright law is some guys with guns, in either an overt way or a subtle one
That just seems like a sociopathic argument to circumvent perfectly clear laws intended to protect people who expend energy into offering new ideas and also trying to diminish what is already delineated as right and wrong. In the blatant sense.
I'm not sure how your gun analogy funnels along to fair use.
As with anyone who doesn't seem to respect copyright laws, I'd say to them, give me a copy of everything useful you've ever done in your life. Going by the same rule.
Most copyrights are not held by the people who expended energy into the labor of producing new work, they're held by the employers of that labor. Most people have no practical choice in finding the means for their survival but through wage labor and selling the ownership of what they produce.
The "gun analogy" is the parent saying the the government (who is supposed to be the only entity able to use force) is supposed to protect IP. You're saying the same thing.
You shouldn't say anything to anyone who doesn't respect copyright laws - or, rather, you shouldn't have to. Your tax dollars, and the entity that you authorize to be the one to use force where you live, should be able to say whatever you, as the governed citizen, want to say to the person who doesn't respect copyright.
And if you look at open source it does exactly that. Most AI models are open source/weights. People are giving away some of their best work because they want to cooperate with others to build things.
Scientific publications also share knowledge openly. Only private journals are locking up papers, but they are losing ground. At least AI papers are free to read and implement. They share even code, datasets and trained models.
I think the ethos of progress is with open culture not closed copyrights.
It may be a good ethos but you can't simply undermine people's life work for your ethos especially when there's laws protecting their life work.
Put yourself in their shoes. Maybe you use some AI derivative work for your own life's work and then when you've finally accomplished what you want to do, someone else duplicates it, or uses the majority of your work?
What do you expect out of it. If you don't expect to be paid, great - I guess you'll be spending X other hours supporting yourself and possible family.
And if you do, you don't expect the hard work of others to be compensated? Rhetorical question.
Use AI to build things for yourself. You get the benefit in those things, it doesn't matter if anyone else copies, as long as you're not selling content.
But there is differences between technological evolution and wholesale taking of other people's intellectual property.
You can go to Watt's steam engine or some such. I doubt he'd argue that his engine had intellectual property on the manual labour people did since time immemorial. But yet it did make their jobs obsolete.
Maybe some of them died while his patents were still in play. All the same, his invention, his idea, plays by the rules.
You can find all the code I've ever written on Github, licensed MIT. I only wish I could do the same for the code I've written for pay. I cannot release that, because people more powerful than me have decided that I don't own it.
Do you not believe that societal constructs are real? What makes (fiat) money a real resource and copyright an imaginary one? What about this logic:
Debt is not a real thing. It's a fiction. Subverting contract law is not like subverting the laws of physics. Simply don't pay your bills! What happens when you subvert contract law is some guys with guns, in either an overt way or a subtle one, force a trade of your imaginary resource (money) for the real resource (labor), through a contract.
It's a bit of clever misdirection to declare yourself "in debt" or "debt free." Money is made up! It's fiction! It is whatever we want it to be!
The parent comment basically says "copyright is not a physical good or law, therefore we may disregard it".
As if the same couldn't be said for any verbal agreement, sexual consent, property rights, whatever.
They ought to be arguing why this _particular_ verbal agreement/social contract is not worth enforcing or practicing, rather than dismissing the entire category.
It can and should be said about every fiction, as a healthy reminder to yourself. Fiat money, government and property rights are complete abstractions, completely abused and tarnished beyond all limits. A verbal agreement is on a much closer level to reality. Even people who have never heard of money or owned property will take a man up on his word. As for sexual consent it is not something abstract at all. It's not words. It is the will of a person and something that can never be misunderstood.
A bit late, but I don't find that analogy apt here. Your territory refers directly to the map.
And if you don't believe social constructs to be real, then I am confused as to why in your original comment "money" is given as something which is real. I think one could reasonably say that all social constructs are "imaginary" (whether that is useful is another question, but I would say it is not), but I find it more difficult once you engage in line-drawing within social constructs, especially so when you don't specifically articulate a boundary.
> I do not believe any of these social constructs are real.
You are redefining the word "real". You know this. Some people get a delight from redefining words and seeing the reaction. It is rather tiresome; it doesn't help constructive debate. What have we accomplished here? Have you convinced anyone?
Feel free to disagree with a particular social norm or law and present your thinking.
If you want to get into epistemology, please do. But we have to find a shared language that makes sense.
Thank you, kopecs! It is nice to see a voice of reason here. Sometimes it helps me to remember that people are not that far evolved from other primates. Our abilities to form useful conceptualizations and error-correct are far from uniform. Forming and reforming mental models and continually seeking truth is not something that comes easy -- or possible -- for many.
This is funny, the moment the absence of IP can benefit large companies it's just a fiction, a made up concept, not too serious at all.
Of course you go onto the OpenAI corporate pitch page and it will proudly say "respects your IP by not training from it!", that's not the part we are looking to get rid of apparently? That's very real?
This is a clever trick, making out corporations to be the beneficiary - who is the real beneficiary? The one who prompts, because they can use the output of the model to solve their problems. In the meantime AI developers make cents on 100K tokens and are mostly in the red.
Why not put the issue where it is - it's a debate about public empowerment with AI vs incumbents protecting their copyrights. Especially for open source models - this situation is not about corporations. Yes, corporations are the ones with deep pockets and easier to sue, but that's just how lawsuits go, you don't waste money suing someone who can't pay up.
What is more important in the next decades - public empowerment or extending copyright to block training AI models? Can we agree to limit copyright to exact reproductions or should it cover all content generated from a model, even if when it looks different from all training examples?
I think going the copyright maximalist way will indirectly hurt creatives because all their works will be checked with the same tools we use to check AI. Anyone could be secretly using generative models. The AI attribution methods will reveal all sorts of things we don't like to see.
The outrage as packaged by the media really does mirror the anger against Napster from 25 years ago. I seem to recall the small-time musicians were more excited about Napster, though.
I remember when BitTorrent users were getting sued left and right, I was thinking about how can a P2P network allow downloading without revealing what was downloaded. Now we got the answer - the LLM - you can download the same LLM like anyone else and nobody knows what you're getting from it.
In the sense of Sapiens†, yes, copyright is a fiction, but so is money, or any law or contract or agreement. In as much as it affects people's lives and societies and economies it is "real". The question of the article isn't about realness, it's whether it can evolve or survive this latest shift.
Can you please clarify this cryptic comment? The OP just used a common variation of asterisk, and I am really confused what your beef with it is.
"in the manner it is being put after the names of dead people " is a total non sequitur in this context, how is that in any way related to the discussion or anything the OP said?
To derail us even further: You know, I just realized, I think the dagger symbol marking dead people is usually shown at full size, while the dagger symbol marking a footnote is shown as a superscript. Maybe the commenter got confused by that.
I agree but I would word it a bit differently: Copyright isn't some fixed, timeless thing. It has been continuously modified to suit the times and to adapt to technological change.
Rather than try to reason about how existing copyright law applies to AI models, we should be focused on changing copyright law to work well in a world with AI.
We can balance the incentives of content creators with the incentives of the users and creators of AI models. I'm confident we can do it because we've done it every time in the past when technology has changed, and AI models will be no different.
> It has been continuously modified to suit the times and to adapt to technological change.
And to suit the whims of sufficiently large and influential media conglomerates.
Yes, you, as an indie creator, have the exact same intellectual property rights as Disney. But if Disney steals something you made, who's going to win that fight?
Leaving aside the semantics of what is "real" - yes, that is the point, that is why the article exists. "Is A.I. The Death of Gravity" is a nonsense question, "Is A.I. the Death of I.P." is not, precisely because humans get to decide the answer. We have developed certain norms and institutions over hundreds of years and now we must decide whether to scrap them in the name of technological progress. How we decide the question will have huge implications for how wealth and power are distributed in our society.
Contracts are not a real thing. It's a fiction. Subverting contract law is not like subverting the laws of physics. What happens when you subvert contract law is some guys with guns, in either an overt way or a subtle one, force a trade of your imaginary resource (a contract) for the real resource (money), through licensing or fines.
Well that’s a stupid argument. A good argument is that AI is in accordance with copyright law already since there are no reproductions. In the rare cases where there are reproductions, they can be DMCA taken down like we already do with post LLM output filters.
I think you're sowing the seeds to undermine your own arguments, because, literally, everything of social utility is a fiction: money, law, "private property", contracts, and so forth.
All of them are needed for society to function, but at the same time all of them are convenient fictions for us to build upon.
My point is that in these discussions, don't make the mistake of considering the artificial thing ("it's copyrightable/it's not copyrightable"). Rather, consider the real thing ("You will pay a licensing fee to use your GPU").
I'm still confused. Sure, a GPU is a real, physical thing. But licenses and fees are no more or less fiction than copyright. A license says "I have permission from the company makers to use this product", similar to how copyright is "you must get permission from me to use my product".
But for the sake of this discussion: I don't think anyone can assume those permissions when using a tool to generate the base image. Ironically enough, it may make more sense to have artists for concepting, copyrighting the character or world so you obtain those permissions, and then using AI to ramp up production. That's an inversion of what's happening for early AI usage.
The main question is how much concepting is needed to copyright a character? Making a few sketches in private? Releasing a full work first with no AI on what you want to copyright?
I know that everyone is chomping at the bit to logically own me in this thread, and that's fine -- have your fun. There are a dozen comments informing me that "IP law around AI" and "fiat currency" are both made-up, and that's obviously true.
Right now "fiat currency" or "contracts" is settled. We are not radically redefining what those are (despite the web3 crowd's insistence). The story around IP with regards to AI is quite different. We are at a fork in the road. It is actually up to us to decide what AI fundamentally is, as it relates to IP. It's genuinely up in the air, and gives us an opportunity to re-examine some aspects of how we treat ownership of information. In my opinion, this is a good thing and an exercise worth doing. A chance to re-do some of what the internet failed at. Remember Aaron Swartz?
Personally, I'm quite cynical: I think the existing social power structure will win and absorb AI into its IP regime. AI will be locked down even faster than the internet was, because value can so obviously captured. The winners will be Disney, Sony, Warner, Elsevier... we'll all be poorer for it. But yeah! Good thing people can't make images of Spider Man?
I apologize for unknowingly dogpiling on you (there were maybe 3 replies when I made mine). I genuinely wanted to understand your POV on what man made concepts were "real" vs. "fake". I think dividing it into "settled" vs "unsettled" makes a lot more sense. We can argue which concepts are really "settled", but given IP concepts as recent as DMCA it's safe to call software IP unsettled.
>Personally, I'm quite cynical: I think the existing social power structure will win and absorb AI into its IP regime
Yes, that's my fear as well. Which is why I hope these LLVMs get reigned in before it gets out of hand, as it did with DMCA (very rarely used to make sure proper parties are compensated and used just as often to make digital creators kowtow to the whims of the elite).
In your words, I don't see a future where AI wins these cases and the little guy gets to make images of Spiderman without Sony/Marvel/whoever squashing them down. All while probably using that artist's own prompts and existing art to make Spiderman 57.
It's only real if you believe it has value. If you choose to say copyright is made up and doesn't have real value then it isn't real to you, but recognize that it's arbitrary as others will believe in the value of copyright so still has value in their eyes. e.g. Taking a copyright from someone (if that were possible) would cost them like taking a fee costs them--losing something they value.
Sure, you can make a slippery slope argument, but there is something substantially more "fictional" / artificial / unnatural about copyright/IP than currencies and other social constructs you are equivocating with.
Storytelling is as old as language itself and might even be older than our particular species, and is basically copyright infringement. The same goes for tool-making, which is thought to be one of the sparks that gave rise to our species' vast intelligence. That's how imaginary copyright and IP in general is, let's take the thing we've been doing since literally the dawn of time, copying and sharing information for free and building upon and improving it, and portray this practice as "unnatural" and "unlicensed" and in fact let's set up a restrictive framework under which most of the technological and cultural achievements in human history and before would have been impossible or severely hampered. It reeks of artificiality, way more than money and other things do.
It was created by corporations to make it easier for them to form monopolies around unchallenged control over a particular intellectual property or idea, and now that it is becoming inconvenient for them with the advent of AI, they will probably get rid of it or re-invent it in some way that even further benefits them.
It also matters very little what we do because in the end we will be out-competed by other countries like China that don't draw these artificial intellectual lines that hamper progress.
I think your point is, that copyright is whatever we decide it to be, as with other things we use in society, but we should decide by what is most useful for the society as a whole, in this case it might be that, we have to not care about it, because it might be societal danger to block AI evolvement as competitors who don't care will get ahead. This might override the usual benefit of protecting people's output so they would be more incentivised to share and create their output in the first place.
> Sure, you can make a slippery slope argument, but there is something substantially more "fictional" / artificial / unnatural about copyright/IP than currencies and other social constructs you are equivocating with.
I think there's probably some modern bias in this perspective, about how "natural"/"real" money is. Early human populations certainly didn't have the same concepts of currency, land ownership, etc. that we do today. Communally sharing resources to help each other/ourselves is older than our species, in fact common to multiple species, and the other social constructs like "money" get in the way of it too.
> That's how imaginary copyright and IP in general is, let's take the thing we've been doing since literally the dawn of time, copying and sharing information for free and building upon and improving it, and portray this practice as "unnatural" and "unlicensed" and […] It was created by corporations to make it easier for them to form monopolies around unchallenged control over a particular intellectual property or idea, and now that it is becoming inconvenient for them with the advent of AI, they will probably get rid of it or re-invent it in some way that even further benefits them.
Wikipedia says that King Diarmait Mac Cerbhaill adjudicated what was basically a copyright case over the Cathach as early as the sixth century, and the legal basis of copyright is founded in moral, economic, and property rights that existed in ancient Jewish, Greek, and Roman societies. Apparently, it took the printing press for unauthorized copying to become a big enough issue to be codified in law, but the poet Martial was complaining about missed profits from unauthorized copying and "book sellers would sometimes pay a well-regarded author for first access to a text for copying" (I.E. there was something resembling IP licensing, enforced by physical access rather than law) back during Roman times.
The discussion so far is rather dull and dogmatic. Wouldn't it be better to read something produced by scholars? Has anyone in this thread dedicated thousands of hours to understanding these topics deeply? (I would assign a probability estimate of less than 1%.) So, instead of swimming in the kiddie pool here, go read something substantive from people who "went pro" in these areas. I suggest:
## "Rhetoric and Reality in Copyright Law" by Stewart E. Sterk
> Why give authors an exclusive right to their writings? Copyright rhetoric generally offers two answers. The first is instrumental: copyright provides an incentive for authors to create and disseminate works of social value. By giving authors a monopoly over their works, copyright corrects for the underincentive to create that might result if free riders were permitted to share in the value created by an author's efforts. The second answer is desert: copyright rewards authors, who simply deserve recompense for their contributions whether or not recompense would induce them to engage in creative activity.
> The rhetoric evokes sympathetic images of the author at work. The instrumental justification for copyright paints a picture of an author struggling to avoid abandoning his calling in order to feed his family. By contrast, the desert justification conjures up a genius irrevocably committed to his work, resigned - or oblivious - to living conditions not commensurate with his social contributions. The two images have a common thread: extending the scope of copyright protection relieves the author's plight.
> Indeed, the same rhetoric· - emphasizing both incentives and desert - consistently has been invoked to justify two centuries of copyright expansion. Unfortunately, however, the rhetoric captures only a small slice of contemporary copyright reality. Although some copyright protection indeed may be necessary to induce creative activity, copyright doctrine now extends well beyond the contours of the instrumental justification. ...
## "Copyright Nonconsequentialism" by David McGowan
> This Article explores the foundations of copyright law. It tries to explain why those who debate copyright often seem to talk past each other. I contend the problem is that copyright scholars pay too much attention to instrumental arguments, which are often indeterminate, and too little to the first principles that affect how one approaches copyright law.
> Most arguments about copyright law use instrumental language to make consequentialist arguments. It is common for scholars to contend one or another rule will advance or impede innovation, the efficient allocation and production of expression, personal autonomy, consumer welfare, the "robustness" of public debate, and so on.' Most of these instrumental arguments, though not quite all of them, reduce to propositions that cannot be tested or rejected empirically. Such propositions therefore cannot explain existing doctrine or the positions taken in debate.
> These positions vary widely. Consumer advocates favor broad fair use rights and narrow liability standards for contributory infiringement; producer advocates favor the reverse.' Most of the arguments for both consumers and producers prove too much. It is easy to say that the right to exclude is needed to provide incentives for authors. It is hard to show that any particular rules provide optimal incentives. It is easy to point to deviations from the model of perfect competition. It is hard to show why these deviations imply particular rules.
"Society" is the most abstract illusion of them all. It doesn't exist and has never existed anywhere, it has always been a complete fiction and pretend. So, maintaining a bunch of fictions in order to maintain another fiction becomes circular reasoning.
That's… Interesting. So, what's your axiom for deriving realness from? Because I tend to think along two axes:
1. You can derive realness from materialistic, physical phenomena. So in this case, quark-gluon-electron interactions are real, and basically everything else is like 99% imaginary. "Society" is no more and no less real than anything else.
And/or:
2. You can derive realness from minds' conscious perception, experience, and interaction. In this case, "society", being an aggregate of many minds' mutual and conflicting experiences and interactions, is probably one of the most real things there are.
And of course you can unify the two, by framing the latter within the former based on the information-theoretic complexity of the physical substrate of the minds, and by framing the former within the latter as the common, shared basis of perceived and lived experiences.
---
But I'm not sure what concept of realness would let you say ""Society" is the most abstract illusion of them all".
Or do you mean that the rigidity of currently existing societies and their specific laws and customs, as universal and immutable constants, is what "has always been a complete fiction and pretend"? In that case I would fully agree; You can have multiple possible societies, in any number of shapes and configurations. But that does not make extant societies, as they are right now, any less real for those that live in or otherwise interact with them.
> So, maintaining a bunch of fictions in order to maintain another fiction becomes circular reasoning.
Uh, no. Not circular reasoning. Fear of starvation, more like. Fear of marauders killing your loved ones to take their resources. Fear of entropy.
Agriculture, presently existing as a function of society, is certainly very real– assuming your body obeys the laws of thermodynamics.
All materialistic, physical objects and phenomena are real. The physical world is one category of what exists. Think of what would exist if humans were not conscious beings, starting tomorrow. Without any complex thinking or reasoning. Sure, money would exist as the paper it's printed on, and your government would exist as some concrete buildings and some papers in a filing cabinet. That is only what these things are. They are in physical reality nothing more.
The second category of what exists in reality is human souls / consciousness and animal souls / consciousness. Beyond those two categories, there is nothing that exists, unless you enter supernatural territory. Which is not to be discarded, but here we're talking about man-made stuff. We can talk about money, society, ownership of land, and such. But they are always only imagined by people, they themselves can never transcend from the realm of idea and imagination into the realm of the real. Even though we have to relate to these ideas in everyday life because how they control the behaviours of our peers and ourselves, we can never forget that they are not real. But most forget.
In the past and in some regions even today, people held deep beliefs about curses and individual taboos. That a jealous neighbour or romantic rival would curse them, or more likely that a Shaman at their birth informed their family that this person can not eat this or do that, or they would die. And from accounts I've read, people believed this so deeply that they would have convulsions and even die if they broke their taboo by mistake. Just by the power of a false belief. Or the countless who waste their lives believing in some lie about "society". Or people who live under the curses of shame and guilt. And the curse of pride for that matter.
It is much more beneficial and easy on the soul to disregard all and any idea of society and instead deal with people as individuals. Because you're almost always dealing with individuals. And stay away from those who are too deep in the illusion and false belief of "society" or any such thing, because they only know how to bow to those they perceive as higher and kick down on those they perceive as lower.
As humans and individuals, the best thing we can do for ourselves is cleanse our mind from false beliefs. That doesn't mean that we will forget how these ideas work on people, it means we can see them from a higher perspective and that they do not exist. They can be used as pleased, but should never control a superior being.
Agriculture is not a function of society, it's a function of tilling the earth with a plow and sowing seeds. Nothing about it is abstract. I frankly didn't understand that example.
The Bible goes on and on about not having other gods besides God, but it also says that you should not _make_ idols, which I think is a very important and overlooked part. Anything that is undeniably created by humans is a false idol and worshipping it is a mistake. Most people alive right now worship their government, or more accurately their idea of what their government is. Because most do not know how their own government works on paper. Most people would die and kill for their government, while at the same time being nonplussed by other things that should make their blood boil. The slightly smarter ones worship money. One day when they're old, they'll be rich and have their own yacht. If they forgot about their fake money god for a moment, they could go sail the oceans as crew on a yacht while they are young and can enjoy the experience. Can they buy back their youth with some fake money printed in limitless quantities? No, because reality does not bow to the non-real, even if every single person on the planet believed in a falsehood. But a lot of people will have a whole lot of suffering from false beliefs. That's how literal hell on earth is created.
The point AIUI is that it can be whatever we want to define it to be. There is no objective answer to AI vs copyright. Hell, we could invent copyright v2 which applies to AI on days divisible by 4 and v3 on all other days, but only if someone named Bob in Texas wears a beret.
Therefore, people talking about morals or ethics or creativity or human inspiration or whatever are wasting their time on a false premise.
As many other comments mention, copyright is as real as contract law or money.
A more correct statement with the same thrust would be "copyright is not _natural_ -- it's an artificial, bolt-on construct."
And yet lots of artificial constructs are useful for incentivizing / de-incentivizing massed behavior in a way that is maximally beneficial to all. For sure, getting it right is difficult but let's not write off all artificial constructs as worthless or impossible to get right (especially when you just have to get it right _enough_).
Maybe, but you can also say that about money or about being married. Social fictions work as collective agreements on what is seen as useful for society. One can obviously critique their utility, but pointing out the ficitonal component isn't in itself a criticism.
The thing that separates copyright from the other fictions mentioned is that there are pretty good reasons to think this particular fiction is not useful.
Property and debt have much longer histories and seem more clearly defensible.
Of course there is one really big fiction out there that a lot of people no longer really believe, but that doesn't mean longevity isn't a thing to consider.
> Copyright is not a real thing. It's a fiction. Subverting copyright law is not like subverting the laws of physics. What happens when you subvert copyright law is some guys with guns...
That reasoning applies to murder, too. After all, a bullet through your mom's brain is just a physically allowable reconfiguration of the position of some atoms.
So you're making a true point, but a useless one, since you're operating on the wrong level of abstraction.
No, but media companies like Disney, Warner Bros, and Paramount might find it a useful lever to squeeze independent creators even further than they already are.
Here’s a story from my days at Motorola Labs around 2009 that you might find relevant: we were looking at ways for streaming content to homes (Moto owned a set-top box business at that time) and there were big debates about the future of streaming. The position that prevailed was that consumers would not be able to stream content to their heart’s content en masse because content providers would never let it. I distinctly remember a presentation poster that said you couldn’t stream because it’s against the law, hence use our solution, etc.
Moral of the story for me is that if there’s adequate money to be made laws, protocols, etc. can totally be changed.
> Bellos, a comparative-literature professor at Princeton, and Montagu, an intellectual-property lawyer, find this kind of rent-seeking objectionable. They complain that corporate copyright owners “strut the world stage as the new barons of the twenty-first century,” and they call copyright “the biggest money machine the world has seen.” They point out that, at a time when corporate ownership of copyrights has boomed, the income of authors, apart from a few superstars, has been falling. They think that I.P. law is not a set of rules protecting individual rights so much as a regulatory instrument for business.
Remove copywriter protections and small time authors have the potential to earn EVEN LESS.
Why write books or any creative text based content when the second it's available it will be hoovered into the next AI training cycle and can be reproduced in part or in whole by users who almost definitely do not know who you are in the first place.
Is that even the case? Who would read a small time writers fiction book through chatgpt? Most of the use cases I've seen for chatgpt, e.g. Translation, conversation, brainstorming, reformulation, summarization are wholly different from the dataset of books. I would say that the art generators have a stronger claim of competing with their work than llms
OpenAI/other is earning money (by selling subscriptions, ...) with a product that has been trained on and utilizes the protected work to produce their results.
Personally I think the only way forward is for these AI companies to curate their dataset to only material that is legal for use. E.g. only GitHub repos with licenses stating such. Just because there is not a license doesn't mean it's free to hoover.
Or there could be a very cheap license for AI training that doesn't allow other uses, much like radio licenses for music.
They could buy a bunch of content created (or specifically licensed) for purpose, the artists get paid, and the artists that don't want their work to go into the pipeline get what they want too.
But too many people consider the development of AI to be such a moral imperative that no objections can possibly be relevant.
I used the new Instagram "AI" yesterday to "Add a famous duck and his famous mouse friend" to my image which gave me a perfect Donald Duck in the background (I guess it forgot the famous mouse). Surely lawyers just need to work on the replication prompts to demonstrate the model was trained from a copyrighted image and replicates it so accurately to make their case?
I find it revealing that some decades ago when teenagers were copying mp3s in their rooms for their own enjoyment it was piracy, crime, reprehensible, police, prisons, etc. When corporations are doing mass copyright infringement, we are talking about death of IP or changing the copyright laws to accommodate them.
This is the right take — whatever your position on the future of this technology or its costs and benefits for humanity, the hypocrisy mentioned here shouldn't go unacknowledged.
There are hundreds of millions or billions of us, though, depending on how broad a brush you use. Any given issue or value is going to have someone, in fact quite a lot of someones, on every conceivable side of it. Even if you limit your scope to the people who have a substantial voice in the culture.
Yeah, but it's not a good look when the influence-weighted average of all those people applies double standards for different groups in similar situations. I mean, even when you take just a single person, they're probably going to have a couple conflicting feelings towards any one topic. And yet, if the emergent result of that internal conflict doesn't produce some level of self-consistency in resulting opinions and actions, then something's not quite right.
Like… the entire idea of words like "society", "culture", "zeitgeist", etc. is that they imply a greater level of organization, shared values and ability to act as though one autonomous entity or whatever, than if you merely took "hundreds of millions or billions of" people and dumped them all in one place.
A society or zeitgeist can absolutely have 'hypocrisy'. If it can't, then it's not a culture, just a mass of hominids.
Plus, though it's hard to track, there's probably some degree of hypocrisy aggregated from an individual level too, where people who took one side before are taking the other side now.
Depends on your lens. If you view it as "crime", sure. If you view it as "obtaining quality content", you can see why and where these users' priorities are. The web is already this riddled SEO optimized spam-Festa that makes it hard to find what you want or desire, now AI comes in and increases the spam tenfold.
AI for those people just make their jobs of seeking what's worth pirating that much harder. Especially for those that can now fire half their staff and put most of that remaining money into said SEO optimization for their spam.
In a vacuum this shouldn't be an issue at all. We make stuff, we honor and benefit the creators, we share as needed and progress humanity as a whole. So I can understand the more optimistic approaches to AI as a tech revolution.
But in reality, everytime small entrepreneurs always thnk this will be the tech that let them usurp the big boys, and they end up selling out (not necessarily blaming them. I'd do it too). They have more resources, can scale up faster, produce more, and hire more people. It's a hopelessly risky gambit to think you can topple such advantage. And if you get an edge in, they just take it in for themselves and you're back at square one.
But you may not even get that far. They try to make use of this new commons and the Big Boys will litigate the gates back closed for them specifically. They got their value, made the next iteration, and then decided not to give back to the commons (a tragedy, if you will). They may even seem benevolent at first, but time goes on, priorities or leaders change, and the focus on market capture eclipses any public goodwill.
The same players as yester-decade are at the helm, so I don't know why anyone would want to trust them after giving decades of scrutiny as they gradually unmasked themselves. I don't necessarily fear the tech per se, I fear the same "Embrace, Extend, Extinguish" playbook happening all over again, maybe in such a way to make the new gap insurmountable.
> when teenagers were copying mp3s in their rooms for their own enjoyment it was piracy, crime, reprehensible, police, prisons, etc. When corporations are doing mass copyright infringement, we are talking about death of IP or changing the copyright laws to accommodate them
It's been over a decade since an individual was prosecuted for digital piracy [1]. Longer since anyone was threatened with jail time.
When it was new, both individuals and companies were prosecuted. The tides shifted and law enforcement responded. This isn't a story of different standards, but one of evolving ones.
> GP's point was it's different rules for big co vs individuals breaching the same idea
My point is it isn't. Individual copyright infringement is virtually unenforced today. To the extent it was in the last decade, the penalty was a fine. Meanwhile, OpenAI is being sued by the New York Times [1] and various writers [2].
I don't think you have a point, the original poster was talking about individual indiscretions (20 years ago also) vs wholesale scraping of billions of people's copyrighted work for corporate gain, today.
> individual indiscretions vs wholesale scraping of billions of people's copyrighted work...some lawsuits don't change that fact
Over the last decade, individual indiscretions have not been punished. Wholesale scraping is being punished. Some lawsuits do challenge the hypothesis that corporates' copyright violations are being treated more leniently than individuals'.
Why are you introducing this 'last decade' thing. Was the decade before not relevant? The original point was not about lawsuits anyway. Are they American lawsuits you're talking about?
You seem to be conflating a whole bunch of things, seems like misdirection.
> Why are you introducing this 'last decade' thing. Was the decade before not relevant?
I'm arguing that policy preferences around copyright infringement have changed in general. In general, in 1990s, copyright infringement meant "crime, reprehensible, police, prisons, etc." for both individuals and coporations. In general, in the past decade, it's meant none of those things for either individuals or corporations. Yet it's meant fines and lawsuits for corporations with virtually none I can find, in America, aimed at individuals.
Also, LLMs were basically invented less than a decade ago [1].
> seems like misdirection
"Please don't post insinuations about astroturfing, shilling, brigading, foreign agents, and the like" [2].
You seem to have a very narrow view of what is a relevant or a valid comment. Just because a counterargument doesn't completely refute the original comment, or "introduces" new concepts, doesn't make it irrelevant or "misdirection".
Someone compared treatment of X 20 years ago to treatment of Y today -- seems pretty natural to bring up treatment of X more recently. You can't just say "the original comment didn't mention it so you can't mention it either".
I don't see how your accusations of bad faith are warranted.
If your doubts are true, why did you have to introduce an analogy?
I feel I have a pretty open view.
And, there's no definition of a 'valid' comment. I'll address points, raise points or whatever. Sorry, I'm not in the realm of 'valid' comments, never was.
And plenty of people are threatened with lawsuits for seeding on an ongoing basis, especially in some markets like Germany where an entire industry stalks public torrents looking for German IPs.
> plenty of people are threatened with lawsuits for seeding on an ongoing basis, especially in some markets like Germany
Fair enough, I'm talking about America. To my knowledge, individuals downloading pirated content have not been threatened with lawsuits. And to the degree seeders have been threatened, it's only that--threats. When was the last distributor actually sued?
I know porn owners go after seeders in the US, I've seen news articles from torrentfreak.com to that extent.
But the fact that they're not being sued is a weird delineation to make. They're not being sued because they can't afford to defend so are settling. It's not a case that IP owners aren't enforcing copyright, they're just not escalating to actual lawsuits because they don't need to.
> not being sued because they can't afford to defend so are settling
I’m asking for evidence of these settlements.
The claim is corporations are being given a free pass relative to how individuals are being treated. I’m not seeing that signal. What I am seeing is Americans becoming more sceptical about intellectual property as a result of the MPAA and RIAA overplaying their hands in the 90s.
Try downloading pirated movies and see how long it takes for you to receive a letter from your ISP telling you to stop or else... It's automatic, that's why you don't hear about lawsuits anymore: not necessary, the law enforcement is semi automatic now.
We're not exactly tearing everything apart to combat piracy nowadays even though piracy still exists, so I'm not sure exactly what it reveals other than attitudes change?
If you ask it to complete text and give it the first line, it will sometimes continue the lyrics. Seems to work best for particularly famous stretches of lyrics, like Lose Yourself by Eminem, Bohemian Rhapsody or Hurt.
When I told it to complete the text "Obie Trice, real name no gimmicks", it said I'm sorry I can't reproduce the lyrics to Without Me by Eminem. And offered to tell me more about Obie Trice.
When I just asked it to reproduce the lyrics to Bohemian Rhapsody, it once again refused and offered to analyse the lyrics.
Seems like there's clearly song lyrics in the training data, and that they've at least made attempts to prevent it from regurgitating them.
Remember those very dramatic warnings before movie trailers equating piracy to real-world theft (You wouldn't steal a car; you wouldn't rob a bank; piracy is a crime...)? Seems laughable now.
Rules for thee and not for me? Perhaps V, perhaps.
I want to gently reframe the debate, for while I agree with the hypocrisy it rather misses a very key point.
Intellectual law, as bedrock principle, explicitly doesn’t recognize nonhuman creators. So at the heart of the issue isn’t changing copyright laws or the death of IP, but what to do with non-human creativity? This is an interesting issue now, but most of the debates and options being debated thus far won’t survive an AGI let alone a world full of advanced AGIs.
But it’s a critical distinction between “so now it doesn’t matter when it was theft before,” and “what do we do with non-human intelligence when our entire system of creativity protection is build around humans and non-humans now exist?”
It’s a paradigm shift that gets somewhat denied by the hypocrisy argument. Most people look at AI as “technology” we have developed, but if any science fiction writers are right it’s actually a bona fide digital intelligence that’s getting developed here — essentially the possible digital twin of human intelligence — and that’s a whole different set of considerations.
As an NLP researcher, I've come across no convincing research indicating LLMs are "the possible digital twin of human intelligence."
Claims certainly get made, but they are subsequently shot down by future research. This has happened for discussions of emergence [1], reasoning [2], and other human attributes people have ascribed to LLMs.
I think you’re discussing the logical extension of our current AI efforts, and I’m discussing the fully realized science fiction reality that could take decades to perfect or that would be driven by breakthroughs beyond the forecast of existing systems.
That’s the difference between seeing it as a technology vs a larger directional shift in human development. Put more directly we have been experiencing the age of miniaturization, and now we are entering the age of digital intelligences.
Very often early technology looks toy-like and lacks substantive real world application (despite the hype). It’s only later that deliver substantive applications and real-world change. This generation of AI is following the Gartner hype cycle perfectly thus far:
That said Gartner tends to be part of the hype cycle these days, but the pattern of new technology holds true.
So in short I would expect anything like a digital twin to emerge from post trough applications not the current generation, and I don't mean to imply that the current technology will create such a system but just that science fiction and futurists generally forecast or even predict this technology will generate a digital twin (in the form of an AGI) from this civilization-level line of technology exploration.
All that said, I don’t think many of us technologists were thinking we would be anywhere near this level of AI technology in 2018. In five or six years we have gone what people thought would take 10-20. So, as always, linear extension predictive and mental models of technology advancement once again fails to predict the true course of technology progress, which seems to be much more dynamic than people generally consider when talking about the future.
"The law today treats companies as "authors," and classifies things like the source code of software as "literary works," giving software a much longer period of protection than it would have if it were classified only as an invention and eligible for a patent (now good for twenty years, with some exceptions)."
Microsoft has been arguably the largest beneficiary of software copyright. Gates argued for it in the early days of MicroSoft. Microsoft's anti-piracy campaigns are unparalleled. Anyone remember the "Business Software Alliance".
Now the company wants to be the largest beneficiary of "AI".
If anyone wants to train their AI thing on my poetic output ... they're more than welcome to![1] I've been working for decades to get people's eyeballs bleeding from reading (too much of) my poetry; the thought that someone would even want to train a machine to churn out eyeball-bleeding poetry influenced by my work - it makes me happy!
The most powerful force in the world is when a big group of humans unite behind a story, fictional or otherwise, and march forth under a common banner. The question is which group of humans will win this one. My vote is for the ones arguing correctly that knowledge in all of its forms routes around censorship with or without our approval, for better or worse. There was never a lock. We just pretend like there is.
308 comments
[ 51.6 ms ] story [ 647 ms ] threadI suspect the actual answer is “those with wealth will continue to amass it and those without will struggle more”
But your statement brings up (or at least makes me think of) what I think is an interesting point: there's the notion of mass media, which started when broadcasting became a thing. Play 1 recording via a method where potentially millions can hear or see it. It was after the development of broadcast technology that copyright maximalism took hold. IMHO the current IP regime is for that purpose, and cracks are showing in it because media is becoming more fragmented and more personalized. Would mass media ever become a thing of the past?
Of course, they have relatively little to lose and can just reverse course in a few years, likely undercutting the market despite talent being more efficient and self sufficient and able to generate more funds for the company (if I take the charitable end road for this AI stuff).
I guess history really does rhyme.
Personally, I don't think this is sustainable. Once you make the injustice obvious, people will give-up on following the rule.
1: https://en.wikipedia.org/wiki/Betteridge's_law_of_headlines
Hell yeah!
I do wish I.P. dead. Or at least, I wish for a much shorter protection period, and for original authors / artists only. For all the wrong reasons, but primarily, because nothing's sacred. Amen.
I get corporate IP is used for a lot of bad. But intellectuals and artists create a lot of value for society in ways that are very hard to monitize as a individual. If we completely kill IP what becomes of our authors, philopshipers, historians, artists etc.?
They are already having it hard as it is with existing ip protections vs. big tech, big pharma, big record labels, etc.
<https://en.wikipedia.org/wiki/Criticism_of_copyright>
<http://www.dklevine.com/general/intellectual/againstfinal.ht...>
People do abuse the protections, which are too lengthy, makes possible to sell rights, etc.
In the end, corporations (ab)uses (buy and sell rights) protections for amassing fortunes much bigger than the original creator ever got. That's not right, not fair, at all. System's broken, need fix, throw it all out and build something new, centered around creators.
Give back to creators only, for less time.
At the moment, most creators are starving, while everyone, AI included, are feeding on them - big time. Successful then powerful creators only may dream about making AI pay - to them only.
Maybe it's time for some minimum revenue scheme for all creators, on top of heavy limitations on IP. Maybe AI corps can pay for that?
I'm very very tentatively hopeful that we might see some reduction of IP rights in my lifetime. It's going to be a long hard fight but maybe, just maybe we will see people wake the fuck up and realize how important the building up the fucking commons is to the progress of all humanity.
Rent-seeking by controlling AI models is anti-commons, the models themselves are not, neither is training them. Don't delude yourself into thinking that Miyazaki didn't study Moebius didn't study Mucha. Even if you got your way and all training on copyrighted material was stopped today and all existing models were deleted it would only be a matter of time before we could synthesize arbitrary styles (at least as well as we can today) by identifying and controlling the correct axes in tensor space using a model purely informed by public domain data.
Your art is mostly not yours, your art is mostly an amalgamation of other people's art which itself is also the same, all the way down. You are a sprinkling of flavor on top. Your art is stolen styles, techniques and vibes mashed together. And here you are, trapped arguing about where we draw the line on what's okay. 75 years? now 95? Look at what you're saying! Introspect and realize that your understanding of the context is defined by Disney and Bono wanting to collect rent on ideas. You're just focused on getting your crumbs of the pie in a game that's rigged to centralize power and keep it centralized.
original or "original"? maybe AI companies need to hire people to make "original" content for AI.
You might think I'm being rude or overly hostile to you, but frankly I think you deserve it. Your views are atrocious.
The "rent-seekers" is the problem. We collectively inherit and own our shared culture, but large corporations have always wanted to sell it back to us. AI companies are arguing they should have no limitations on their usage of the culture, but that the same shouldn't apply to them. Selling tickets to the commons is anti-commons.
Perhaps if these companies were themselves arguing for the end of copyright and IP for everyone, the conversation would be different.
In order to make attribution models we need text+author+timestamp. We can get that from books, newspaper articles, scientific papers and social network posts. Then we extend to the rest of the training set.
But then we can also make AI models that cleverly avoid infringement while the same strict checking is going to be applied to human made content. Humans are not that good at avoiding pitfalls.
Do it then.
> Your art is mostly not yours, your art is mostly an amalgamation of other people's art which itself is also the same, all the way down. You are a sprinkling of flavor on top. Your art is stolen styles, techniques and vibes mashed together.
Eat shit. People deserve to have some ownership over the product of their labor. Imagine doing all that work to learn all these skills and some rat fuck nerd tells you they have the right to alienate you from that work, making some greasy fucking argument that the multi-billion dollar program they made is doing the same thing you do.
Corporate abuse of copyright isn't an excuse for corporate abuse of people who can actually make things.
There's a limited amount of "attention" on the internet that's available in online art spaces. I theorize that artists need to maintain a public portfolio to find work, either creative work for advertising or entertainment companies or commissions or patreon subscription. You need to be out there for people to hire you if you want to make a living doing art.
AI art is disruptive and harmful to that ecosystem in a few ways. It competes with artists not only for dollars and jobs for commercial art (ie ad agency work), but more fundamentally it competes with artists for attention on the internet, which at a macro scale makes it so there's less money in the system for real artists to practice and master their skills. It's also not necessarily that the AI art is better, it's just good enough to pollute the signal to noise ratio. And as you said, it's fast. That can quickly overwhelm sites like DeviantArt or Art Station, which moves those communities and resources away from their original purpose.
Additionally, AI art models need data from the commons to function. I'm skeptical that synthetic data is good enough to use for improving these models, so theoretically, one of the main ways to improve these models is getting more and better data from humans. If the economy of these models is structured in a way that discourages artists from posting their work publicly, or working at all, then the pace of improvement decreases. I think a lot of artists are pretty pissed off that their work is being used to produce commercial software that creates substitutes for their work without their permission or knowledge.
So that's what I mean by the tragedy of the commons. The online art ecosystem is more fragile than many people think it is and ai companies are over exploiting it. The content remains but the thing you're actually trying to sample, human knowledge and skill, withers on the vine.
Many in the software world are in awe how much theft ai people do.
I really don't understand open source people combating generative AI. I've never seen this ethos since the javascript framework wars. Take a look at llama.cpp, ollama and vllm repos and their friends. They got such a sustained rate of development and participation.
Do we really want to deny people something that costs nothing in order to preserve the status quo? Perhaps it's the system of value assignment that's broken.
I'm not necessarily opposed to moving to a post scarcity model without IP, but we need to actually be post scarcity first.
I'm the only one who's identified a problem. You haven't. Production costs aren't the only type of cost.
Artificial scarcity is bad. I am not comfortable denying people things that cost nothing to reproduce. This is a problem.
If writers aren't able to find profit in writing then there will be less people spending the time and effort to produce good writing, which in turn means less quality material for readers to read (and, as an aside, less quality material for an AI to train on in the future).
We could arguably reach some type of feedback loop where less quality writing leads to less interest in reading, which leads to smaller reading audiences, which leads to less quality writing being produced, etc.
People do love free stuff, but free stuff often isn't sustainable under our current economic system.
I disagree - some people live their whole lives on benefits or similar entitlements. That's an incredible economic achievement, as well as being a brilliant show of hard work by the net contributors.
Except, of course, UBI.
That's too much money for anybody in charge to let die. They're much more likely to nuke AI.
What would you propose the US workforce did with some 47 million knowledge workers with suddenly useless skills if you removed the only widespread economic mechanism our society has to support them?
Realistically, it's a part of the reason IP is "too big to fail". The fallout would effectively destroy the US.
I’ve encountered many people over the years that hardline shit like that—they’re almost exclusively middle to upper middle class suburbanites that are edge lording that stuff to subconsciously quiet their deep-seated insecurities about how incredibly soft their comparatively unchallenged existence has made them.
The real reason is that anybody can do A.I. and it can't be very patent-encumbered, being a number of abstract mathematical techniques. If it becomes an absurdly productive technology, you won't easily be able to keep people from using it in private. Maybe the real Butlerian Jihad will be the government Office of A.I. Copyright Royalties sending agents and their silicon-sniffing dogs to kick in people's doors following rumors of Aggravated Infringement.
Maybe it will be what finally gets general purpose computers banned?
The good news is that this probably still won't much affect your personal projects or anything. Anything you can already self-host and do today isn't going anywhere and you've probably got at least another two or three generations in this fast-moving space before the law clamps down. It is likely that many of the referenced AI aspects will be essentially "solved" before then, such as voice cloning. They aren't going to go after every last little AI user for every last thing because that's squeezing blood from a stone and not cost-effective. But don't plan on building a multi-million-subscriber YouTube channel out of it.
And if you are interested in the freedom to do things the copyright owners don't want you to do, be sure to self-host and archive as much as possible. The ability to do this on platforms others host is going to disappear rapidly, if it isn't already.
On the other hand, this is problematic because it'll slow down the development of the self-hostable models outside of the community moving to a darkweb like model where they stay beyond the reach of the legal system.
Keeping AI performance workable on personal-level hardware is going to be a big deal. Open source work on keeping performance at that level may be long-term more important that this or that feature.
Watch out for calls for AI ethics to turn into calls for controls on hardware capable of running AI. I'm half surprised we aren't already hearing calls for that. I wouldn't be surprised to see it by the end of the year, though my expectation would be at least next year if not the one after. The real goal behind hardware control will be this sort of thing (though copyright control may actually be a secondary concern versus just keeping the really good tools out of the hands of the plebs), not AI ethics. (The elites know AI ethics limitations have no meanings as they have no intention of limiting themselves.)
Both will win. As corporations will make more and more advanced models, training data generated by their models will uplift open models. They only lag a few months. We already can shrink most of chatGPT's abilities in a 7B model. All open models trained after 2022 are benefiting from their big brothers. It's too easy to exfiltrate skills from proprietary models.
TBF Napster and torrents are another and earlier angle.
Purely IMO, big tech normalised the taking of other people's content, and it's getting to the point where content creators (as in, ones that are capable of producing unique content, not wealthy YouTubers per se) have simply had enough. The idea of authorities and primary sources is rapidly diminished with the hodge podge we have to navigate through in information discovery.
This was normalized by large media companies long before big tech.
What had sprung to mind for me was the wholesale rip off of European IP by the US in its commercial development in the 19th century and China now with the US.
I guess what I'm thinking of is at the individual level, no one cares much for re-appropriating someone elses work. It has been normalised.
That's what they say
Everything I ever done
Gonna give it away
Someone hit the big score
They figured it out
That we're gonna do it anyway
Even if it doesn't pay"
- 'Everything is Free', Gillian Welch, 2001
"No one's gonna notice if you're never right or wrong
And if you and your next neighbour, yeah, ya don't quite get along
No one's gonna notice if you're singing anyway
Those not coming in for free will learn they gotta pay"
Ian Brown, 2004.
Let's be honest here. In order to get that way, they needed to stomp out and eradicate every trace of "evidence" that natural medicine was efficacious, because herbs, botanicals, and other natural substances can't be patented, can't be synthesized in factories, and can't be exploited for profit at megascale.
If word ever leaked out that, for example, selenium, valerian root or St. John's wort were equivalent/better than synthetic, patented medications, we'd all be in trouble.
And I'd love to see a natural replacement for insulin that doesn't kill people, but I'm pretty sure it doesn't exist. So maybe pharmaceutical technology does a few things that natural remedies can't?
No sugar, no alcohol, low carbohydrate diets are the answer to the diabetes epidemic. If you want to regulate something, start with the corporations marketing processed, poisoned garbage to people as "food" and "beverages".
Napster didn't kill the music industry, torrents didn't kill the movie industry because people still want to see movies on a big screen experience. And so, actors, directors, VFX artists, cutters, riggers, lighting staff, costume/mask people, they all continue to have jobs.
But now, with AI being already able to do very realistic photos, and in a few years likely to create short movie sequences all based upon a prompt, not only will all these people be replaced by one prompt artist feeding a massive AI engine... but the diversity the AI can generate will always be limited by the diversity of its training material. It is by definition incapable of trying something entirely new, and since there will be no economic incentive to try anything entirely new there will be no expansion. (Oh, and guess what, movie prices aren't going down but since the expenses for all the humans vanish the profit concentration will accelerate!)
Widespread AI will freeze our culture in an era of about 2020-2030 forever simply because capitalism will not offer any incentive to feed the AI with new, creative things.
Given you believe the reason torrents didn't kill the movie industry is because people weren't satisfied with smaller screens: would you be satisfied with content that is stuck in "2020-2030 forever" and can never produce anything "entirely new"?
Personally I think novelty in art thrives when the entry barrier is low and experimentation is possible without a huge budget - which AI can help achieve.
I fear the enshittification of movies - similar to, say, the situation in video games or residential ISPs: when everyone has silently agreed to a common level of base bullshit, customers won't move (because why move from a pile of cow dung to a pile of dog dung? it's all dung in the end), and everyone can keep fleecing the customers.
> Personally I think novelty in art thrives when the entry barrier is low and experimentation is possible without a huge budget - which AI can help achieve.
Oh even now even utter novices and two-three person teams can produce hours of high quality content, youtube is enough proof of that. Experimentation is not the problem, the problem is all the marketing to get people to sit in the cinema. For a lot of modern movies, music and video games the cost of production is at least on the same order of magnitude as the cost of marketing - even a blockbuster movie like Avengers Endgame [1] reportedly had 356 M$ of production cost vs 200 M$ of marketing cost, and of the production cost at least 100 M$ were compensation for the star actors.
[1] https://collider.com/avengers-endgame-box-office-budget/
So, if I'm understanding, you believe you wouldn't be satisfied with the movies, but wouldn't have a choice because of a silent agreement (as with residential ISPs) leaving customers with no good options. Is lowering barriers to competition not one of the ways that such oligopolies are fought?
To me this seems far more of a risk in our current environment where the vast majority of high-production-value movies come from a small handful of companies that have the budget to make them (and give us plenty of safe low-risk sequels).
> Oh even now even utter novices and two-three person teams can produce hours of high quality content, youtube is enough proof of that
There are reasons why working-class people are vastly under-represented in arts - the time and resources required to create "hours of high quality video content" is a hefty investment even for Youtube videos, not to mention a feature film that people will want to watch.
Disney is already following that thesis without AI, yet it's never looked so vulnerable to the whims of the free market.
It depends on the director and style of the era, but average "shot length" runs between 5 and 14 seconds. Six months ago companies were already demonstrating AI generated, 20 second shots (albeit with low motion, like slow pans) with the background staying mostly the same. I would probably upgrade this statement from "likely to create" to "almost near certainty".
Only if the AI doesn't see any usage. But when you start using it, you get AI+human generating things that can be outside its training scope. Humans can input novel ideas into the prompt or include new information. Then AI gets to retrain on the data generated with human in the loop and improve. There is a chance for humans and AI to explore new directions.
If you got a simple way to test, like the DeepMind geometry problem solving model, then AI models can improve on their own by doing massive search and validation. Kind of similar to the scientific method - formulate hypothesis, validate, observe outcomes, and repeat. Works for code too, or any AI output that can be "executed" and generate a validation signal.
If tomorrow we discovered a source of infinite free energy, oil and energy companies could try anything they wanted, their current business model would end up obsolete sooner or later.
Maybe even if the big tech could pass on some (most) of the proceeds because of the (content share), it'd be far less a thing. Otherwise, 90% of people being made aware of an idea see it somewhere beyond the original source and those secondary sources capitalise because... simply because it's the norm.
A better example is solar power, which is slowly but surely whittling away at fossil fuels.
If you have smart robots in the home, much less AGI, what's the plan on restricting them? Sorry robot, you may only learn this limited set of skills, you're not allowed to create, because you've seen too much and might be too good at learning and producing.
They'll be incredibly metered if they're not allowed to learn and produce at a high capability.
And if we're talking about AGI, you can never realistically reach it if you put it in the straightjacket required to protect IP as things are today. Sorry all-knowing AGI god, you may not make images that look too much like XYZ. In fact, you're not allowed to even see any of it because of your capabilities. In fact, Disney has decided through the copyright act of 2042 that you're not even allowed access to the Internet (or magazines, or radio, or music, or TV).
It can't be AGI if it can't create. Forget about trained on ('just do not train it on material with copyright'), it'll self-train, and then what? Well then you can't even let it self-train, ie no AGI is possible.
It's going to be either or. Either the IP laws get changed, or you get no AGI. The only thing possible would be extremely boxed off, narrow, high capability AI.
Further, I doubt that preventing AIs from literally stealing protected information will halt the rise of smarter AI. There is plenty of free use information out there that is LEGAL TO USE. Just curate your dataset to prevent the ILLEGAL use of other material.
One problem with AI (vs e.g. Google search) is that there is no mechanism for attribution. So it is impossible for rights holders to understand whether their works are being monetised (unlike, say, Spotify).
Should NYT win, then either AI services will have to pay into a (very large) common pot for rights holders, which likely will make AI uneconomic (even if it does turn out to have some business benefit), or (paid) AI services will simply be banned.
Sorry if this is a silly question, but is that really the case? Can we not train an LLM on successively larger training sets, each will uniquely ID's model and associated better performance?
They’ll never do this, of course, cause they’re riding high on wins of stupid people pushing “you can’t work backward from a result to find the nodes” (which is probably somewhat grounded in reality).
Attribution is absolutely not impossible, they just want to think it’s not possible, because enabling such a thing would show just how egregious the copying actually is.
As far as I'm aware, what you propose really isn't a solved issue. Most naive approaches you'd think might work are either infeasible or give chaotic nonsense results.
Partial matching between the output text and the training data is possible as a step after generation (though doesn't determine whether it's coincidence).
Right now we're on a path towards just a handful of megacorps with AI agents that will make large numbers of white collar workers redundant. Imagine something like closed-source software only much worse.
Similarly with all the publically viewable source codes on GitHub/Lab. Just because you can view the code doesn't mean there aren't licenses in play for how you are able to use it.
Further, what happens if stolen information is posted publicly? (E.g. wholesale copies of full books on html pages?
Just because it's public doesn't mean the viewer has freedom to use it as they see fit.
Where infringement arguably can happen is when that model is used in the generation of content - and if the user is prompting regenerate a protected work, then that is where the infringement happens. But not before. Maybe the various ai services can adequately guard against that illicit usage. Maybe not. And if not, its those live services that would need to be shut down.
But the creation and training of a model, and even distributing that model for people to use with their own computers in private does not engage in copyright infringement.
All digital content is statements about the work. Pictures are pixels with facts about colours. Books are characters describing the words.
AI models are compressed versions of those.
Downloading, ingesting into models, and redistributing them is copyright infringement.
Even if they weren’t compressed; downloading them without permission is not permitted. Every now and then you can read about ai criminals discussing techniques for illegally acquiring books, images and other protected content even here on HN. It wouldnt be an issue if it were licensed content, or a dudette training her local ai. These are companies that want to monetise people’s work. Somehow, just like thieves, they feel they are entitled to snatching someone else’s property.
What if there was?
What if I hand-painted an image (without AI) and God himself came down from heaven and explained to the world that my painting was derived 10% from my observations of the collected works of Lisa Frank, 3% from looking at Davinci paintings, and 87% a long tail of various sources. And then I sell my painting for $100.
OK. Now what?
No, because copyright law says that if you hand-paint an image (without AI), then as long as the images isn’t deemed (by a court, if necessary) to be a derivative work, then you, and nobody else, own the copyright. The word of God is not necessarily considered by the law. And even if the court takes the word of God as gospel, a 10% derivative work might not exceed the necessary threshold of infringement.
Alternatively, if I created a large body of works that were 90% Lisa Frank and brought it to a large market distribution, making significant revenue, then she could sue me for damaging her market and her brand. Which she has done to others in the past and rightly won. Go Lisa Frank!
But, if I used an AI tool in the process of making an image that came out 10% Lisa Frank, she wouldn't have a strong case against me. Nor does she have a case against the million random teenagers hand-copying her style but not making significant revenue from their copies.
Either way, AI doesn't really factor in here. The results are the results and the markets are the markets regardless of AI or paint brushes.
Much like how some artists are uncomfortable with making high-quality scans of their work available online, because it becomes prohibitively difficult to prevent people from selling prints. Forget the current legal framework, it seems fair to me to say "I don't want that happening to my art" when "that" is a corporation finding a way to monetize it. Even if the AI is using less than a pixel of your work, I think it's fair to ask that it not.
But, everything I've heard and learned about how AI works leads me to the conclusion that such a move is almost entirely symbolic. Counter to what our egos would like to believe, the vast majority of the training images and the vast majority of the learning value for AI comes from random, crappy photographs of various things. Without a million photos of dogs and trees, the AI can't make a stylized dog distinguishable from a stylized tree. And, it's turning out that, from that base of crappy photographs, it doesn't take many high-aesthetic images to make a high-aesthetic AI.
If it needs millions of photos of dogs and trees you
1. Take them you self
2. Pay someone with a db of dog and tree photos soem agreed upon amount (Aka someone did the work for you already and you value their work)
3. You scrape only on photos you are sure are CC0 or any other kind of permissive license. I'm sure you can find a few million dog photos this way.
But clearly they value #2 without payment so lawsuits were inevitable. Maybe because #3 took too much time and they decide #2 is worth the legal fees to be first to the race (something else I'd want fixed, but that's another rabbit hole).
(Not just delayed briefly by starving investment or something.)
Furthermore, rights holders did it for file sharing, why not AI too?
In the mean time, we have and even worse situation: "AI" models like LLMs are effectively allowed to launder IP. As long as companies like OpenAI are allowed to be ignorant of copyright, they get to monopolize that ignorance.
The result is that the rest of us are in an even worse position, because we have to respect DMCA, but can't benefit from it by monopolizing our own IP.
Take xx billion pages of unique human content, do something, pretend that there's no monetary value to that content and monetise it. Farcical.
Debate about what AI means for copyright is really a debate about what we choose to do, what we choose to allow. It's a debate about whether the guys with guns will stop the development of AI, or how much.
It's a bit of clever misdirection to declare the output of AI's "copyrightable" or "not copyrightable." Copyrightable is made up! It's fiction! It is whatever we want it to be!
Now, I don't have a college degree, but with my 60+ years life experience I can say, yes, copyright is a real thing, it protects people who create, from assholes and AI.
Funny how advocates against property rights have no issue with corporations protecting their own. That’s fine. What they want is to legally steal from the average joe.
It's going to be just as easy to make a movie or a song as it was to write this comment. I'm working and researching at the edge, and I promise you that the entire world will be in awe of these next 12 months. Go play with Suno for a taste.
I'm a senior engineer, and now I'm tab completing Rust code 70% of the time. It's hard to believe we've come this far, and it's only going to keep climbing in capability.
You're watching breakneck progress, and the genie isn't going back.
More art will be created per month than the entire human history to this point. Gen alpha is growing up on this stuff and using it to great effect to communicate amongst their peers. The things they build will be incredible.
Or we hit the top of the sigmoid curve and just have fancy autocomplete.
(2) It literally is autocompleting - it draws each pixel because it statistically determines that <some color value> is the best fit given the prompt and the prior pixels it drew. It's a more advanced robot, but it's still a robot.
Llama, ollama, etc are not the issue here. Nor is ai. The issue is theft for training.
This is the same argument the horse people had about the cars taking over the streets. In the end, progress will win.
Do we really want people spending hours and hours practicing and performing legacy acts of creation, with some people never able to summon the opportunity cost to even begin? For people to endlessly labor over creating a small amount of unrepeatable output?
Or do we want a world where everybody on the planet is able to turn their thoughts into something visible and tangible quickly? For society as a whole to have orders of magnitude bigger and better works?
This is a horse to car moment, but for thought itself.
I think it's quite telling that sooner or later, every discussion about "AI" and copyright ends up equating people to statistical models controlled by predatory megacorporations.
My connectome, in the eyes of the law, has a special legal status which is not shared by inanimate and non-sentient objects. In fact, some would argue that the entire point of the law should be to protect the special legal status of my connectome from those that would draw an equivalence between people and property.
"Open-source AI" is interesting in multiple ways, but not relevant to the legality of the models.
You know the "If ML is IP infringement then your brain is owned by Disney" take is just as sensible (and specious) for pre-ML stuff too, right? You're allowed to memorize the full script of any movie you want, without anyone trying even a little to claim that doing so should reduce your legal status to chattel, but obviously you're not allowed to just rip the subtitles from a DVD and sell it as a book.
That's not correct. We are now increasing the capabilities of everyone creative, to achieve much more, almost free and instantly. The painter now, will have a 100 million film studio on his fingertips to create movies. The musician will be able make a high quality album, just from his snoring.
One usual misconception is that talent is not important anymore. That's certainly not true. Talent is not going anywhere. People not so talented can get some results which are ok or good enough, but talented people can create magnificent art, without even trying.
Also "tab completing Rust code 70% of the time" is breadcrumbs. I am working in making, Rust specifically, 100% code generated.
Generating code is trivial. Generating code that works is not. I suspect you are at around 10-15% of reaching that. All models out there are probably at 2-3% as they barely equal a google prompt in reliability.
I don't put it to work for the moment, cause i am trying to create a TUI for HN, or a TUI forum more generally, like HN, maybe even backed by version control like Pijul instead of a database. 100% computer generated obviously, that goes without saying.
My biggest problem so far, is that GPT doesn't write idiomatic Rust. The code it generates is very C or Pythonic, and i am trying to figure out a way to guide it, to write beautiful code, instead of just correct code. When the machine generates just correct code, it doesn't cut it for me, is what i am saying. I am reading "Command Line Rust" to get a sense of idiomatic Rust, and then i have to study github projects as well to really figure out how beautiful idiomatic Rust code really is structured and written. Like, i know what idiomatic Rust is, just not that well at the moment.
The tool i want to create, to use GPT instead of Copilot, is something like Aider [1] but not in Python and maybe a TUI.
https://github.com/paul-gauthier/aider [1] https://github.com/pramatias/hnotify
People were saying the same thing 12 months ago.
Are you just not aware that the most outspoken advocates against private property (and rights to it, including IP/copyright), also strongly oppose capitalism and the idea of corporations?
However, I am strongly against using artist or writer work against their will. That also benefits corporations. Also ironic that we can use against them the play they used against us. Imagine open models that the artist or legal or writing community can use but not corporations unless they open their models.
Dowry payments used to be a real thing.
Butter churning used to be a real thing.
Copyright law is, and will be, mutable.
In just about a year, kids will be making entire Pixar movies from home. (The content will probably be Skibidi toilet related, but of Pixar scale and scope.)
How that reconciles with copyright, you've got me. It doesn't. It's a domain mismatch. It's completely outmoded by what's coming.
And the music industry is equally toast. I can already make an excellent banger song about superhero rodents in under 30 seconds.
Slightly less flippantly, the days when the nation gathered around their TVs to all watch "Leave It To Beaver" at 8pm are long gone, the media landscape has been fragmenting for decades and this is just the next step. My kids don't watch TV shows, they follow YouTubers.
A six-year-old filmmaker has as much claim to copyright protection as Spielberg and Tarantino. Just because one uploads to YouTube and one is paid millions of dollars by a major movie studio, it doesn't mean that they're different in the eyes of the law.
From what I understand, once a work is created, copyright is assigned to that work's creator. That creator may then license that work however they want. Quality doesn't factor in.
There, I've stolen your post. Did copyright protect you? No, it did not. Copyright is an idea we came up with that sounded good at the time, and it's come to pass at this point that the original idea was deeply flawed in ways we could never have predicted, but now can see clearly in hindsight.
The truth is, intellectual property should be protected by those creating it. Coke does a great job of this, as do many, many other companies. We call these "trade secrets", but ultimately the concept is the same. You're protecting the work you deem worth protecting.
I don't buy the notion that copyright ever protected the creator. What it really protected was the interests of the entities who effectively enslave the creators via contract, and is not of any tangible benefit to the creators themselves. If one truly cares about the artisans among us, one cannot justify the existence of our ideas surrounding copyright.
Yes, removing those laws from our doctrine would cause upheaval, as the market must then rebalance itself in the absence of the artificial pressure we've put on it, but in time all things find equilibrium again, and placing the value and responsibility back on the individual is, in my mind, simple human decency.
The whole point of government and laws and societal norms is that not everyone has to be deeply involved and specialized in protecting their rights. We default to people doing the right thing and seek out specialists (e.g. lawyers) when we're wronged.
Copyright law is already different from country to country.
I just wanted to say that laws are subject to change.
Much of the heavy lifting is done by treaty-- not so much that anyone is asking legislatures directly to establish and extend copyright, but that it comes part-and-parcel with other international obligations.
Without the weight of such treaties, I wonder how many countries would do without an IP regime at all, especially if they don't see any IP-related export opportunities, so copyright and patent merely limits their ability to make cheaper versions of foreign goods.
`[The Congress shall have Power . . . ] To promote the Progress of Science and useful Arts, by securing for limited Times to Authors and Inventors the exclusive Right to their respective Writings and Discoveries.`
Does the current scheme really promote the progress of science and useful arts? Is the current length of copyright really "limited Times"?
This was argued before, and the Supreme Court decided "Yes, it is, somehow."
That just seems like a sociopathic argument to circumvent perfectly clear laws intended to protect people who expend energy into offering new ideas and also trying to diminish what is already delineated as right and wrong. In the blatant sense.
I'm not sure how your gun analogy funnels along to fair use.
As with anyone who doesn't seem to respect copyright laws, I'd say to them, give me a copy of everything useful you've ever done in your life. Going by the same rule.
You shouldn't say anything to anyone who doesn't respect copyright laws - or, rather, you shouldn't have to. Your tax dollars, and the entity that you authorize to be the one to use force where you live, should be able to say whatever you, as the governed citizen, want to say to the person who doesn't respect copyright.
Scientific publications also share knowledge openly. Only private journals are locking up papers, but they are losing ground. At least AI papers are free to read and implement. They share even code, datasets and trained models.
I think the ethos of progress is with open culture not closed copyrights.
Put yourself in their shoes. Maybe you use some AI derivative work for your own life's work and then when you've finally accomplished what you want to do, someone else duplicates it, or uses the majority of your work?
What do you expect out of it. If you don't expect to be paid, great - I guess you'll be spending X other hours supporting yourself and possible family.
And if you do, you don't expect the hard work of others to be compensated? Rhetorical question.
You can go to Watt's steam engine or some such. I doubt he'd argue that his engine had intellectual property on the manual labour people did since time immemorial. But yet it did make their jobs obsolete.
Maybe some of them died while his patents were still in play. All the same, his invention, his idea, plays by the rules.
Debt is not a real thing. It's a fiction. Subverting contract law is not like subverting the laws of physics. Simply don't pay your bills! What happens when you subvert contract law is some guys with guns, in either an overt way or a subtle one, force a trade of your imaginary resource (money) for the real resource (labor), through a contract.
It's a bit of clever misdirection to declare yourself "in debt" or "debt free." Money is made up! It's fiction! It is whatever we want it to be!
The parent comment basically says "copyright is not a physical good or law, therefore we may disregard it".
As if the same couldn't be said for any verbal agreement, sexual consent, property rights, whatever.
They ought to be arguing why this _particular_ verbal agreement/social contract is not worth enforcing or practicing, rather than dismissing the entire category.
The map: debt.
The territory: guys who come to your house and kick you out for not paying your debt.
And if you don't believe social constructs to be real, then I am confused as to why in your original comment "money" is given as something which is real. I think one could reasonably say that all social constructs are "imaginary" (whether that is useful is another question, but I would say it is not), but I find it more difficult once you engage in line-drawing within social constructs, especially so when you don't specifically articulate a boundary.
You are redefining the word "real". You know this. Some people get a delight from redefining words and seeing the reaction. It is rather tiresome; it doesn't help constructive debate. What have we accomplished here? Have you convinced anyone?
Feel free to disagree with a particular social norm or law and present your thinking.
If you want to get into epistemology, please do. But we have to find a shared language that makes sense.
Of course you go onto the OpenAI corporate pitch page and it will proudly say "respects your IP by not training from it!", that's not the part we are looking to get rid of apparently? That's very real?
Why not put the issue where it is - it's a debate about public empowerment with AI vs incumbents protecting their copyrights. Especially for open source models - this situation is not about corporations. Yes, corporations are the ones with deep pockets and easier to sue, but that's just how lawsuits go, you don't waste money suing someone who can't pay up.
What is more important in the next decades - public empowerment or extending copyright to block training AI models? Can we agree to limit copyright to exact reproductions or should it cover all content generated from a model, even if when it looks different from all training examples?
I think going the copyright maximalist way will indirectly hurt creatives because all their works will be checked with the same tools we use to check AI. Anyone could be secretly using generative models. The AI attribution methods will reveal all sorts of things we don't like to see.
† https://en.wikipedia.org/wiki/Sapiens:_A_Brief_History_of_Hu...
"in the manner it is being put after the names of dead people " is a total non sequitur in this context, how is that in any way related to the discussion or anything the OP said?
Rather than try to reason about how existing copyright law applies to AI models, we should be focused on changing copyright law to work well in a world with AI.
We can balance the incentives of content creators with the incentives of the users and creators of AI models. I'm confident we can do it because we've done it every time in the past when technology has changed, and AI models will be no different.
And to suit the whims of sufficiently large and influential media conglomerates.
Yes, you, as an indie creator, have the exact same intellectual property rights as Disney. But if Disney steals something you made, who's going to win that fight?
All of them are needed for society to function, but at the same time all of them are convenient fictions for us to build upon.
But for the sake of this discussion: I don't think anyone can assume those permissions when using a tool to generate the base image. Ironically enough, it may make more sense to have artists for concepting, copyrighting the character or world so you obtain those permissions, and then using AI to ramp up production. That's an inversion of what's happening for early AI usage.
The main question is how much concepting is needed to copyright a character? Making a few sketches in private? Releasing a full work first with no AI on what you want to copyright?
> - I think you're sowing the seeds to undermine your own arguments
> licensing fee...
It took them exactly one reply. New record
I know that everyone is chomping at the bit to logically own me in this thread, and that's fine -- have your fun. There are a dozen comments informing me that "IP law around AI" and "fiat currency" are both made-up, and that's obviously true.
Right now "fiat currency" or "contracts" is settled. We are not radically redefining what those are (despite the web3 crowd's insistence). The story around IP with regards to AI is quite different. We are at a fork in the road. It is actually up to us to decide what AI fundamentally is, as it relates to IP. It's genuinely up in the air, and gives us an opportunity to re-examine some aspects of how we treat ownership of information. In my opinion, this is a good thing and an exercise worth doing. A chance to re-do some of what the internet failed at. Remember Aaron Swartz?
Personally, I'm quite cynical: I think the existing social power structure will win and absorb AI into its IP regime. AI will be locked down even faster than the internet was, because value can so obviously captured. The winners will be Disney, Sony, Warner, Elsevier... we'll all be poorer for it. But yeah! Good thing people can't make images of Spider Man?
>Personally, I'm quite cynical: I think the existing social power structure will win and absorb AI into its IP regime
Yes, that's my fear as well. Which is why I hope these LLVMs get reigned in before it gets out of hand, as it did with DMCA (very rarely used to make sure proper parties are compensated and used just as often to make digital creators kowtow to the whims of the elite).
In your words, I don't see a future where AI wins these cases and the little guy gets to make images of Spiderman without Sony/Marvel/whoever squashing them down. All while probably using that artist's own prompts and existing art to make Spiderman 57.
Storytelling is as old as language itself and might even be older than our particular species, and is basically copyright infringement. The same goes for tool-making, which is thought to be one of the sparks that gave rise to our species' vast intelligence. That's how imaginary copyright and IP in general is, let's take the thing we've been doing since literally the dawn of time, copying and sharing information for free and building upon and improving it, and portray this practice as "unnatural" and "unlicensed" and in fact let's set up a restrictive framework under which most of the technological and cultural achievements in human history and before would have been impossible or severely hampered. It reeks of artificiality, way more than money and other things do.
It was created by corporations to make it easier for them to form monopolies around unchallenged control over a particular intellectual property or idea, and now that it is becoming inconvenient for them with the advent of AI, they will probably get rid of it or re-invent it in some way that even further benefits them.
It also matters very little what we do because in the end we will be out-competed by other countries like China that don't draw these artificial intellectual lines that hamper progress.
I think there's probably some modern bias in this perspective, about how "natural"/"real" money is. Early human populations certainly didn't have the same concepts of currency, land ownership, etc. that we do today. Communally sharing resources to help each other/ourselves is older than our species, in fact common to multiple species, and the other social constructs like "money" get in the way of it too.
> That's how imaginary copyright and IP in general is, let's take the thing we've been doing since literally the dawn of time, copying and sharing information for free and building upon and improving it, and portray this practice as "unnatural" and "unlicensed" and […] It was created by corporations to make it easier for them to form monopolies around unchallenged control over a particular intellectual property or idea, and now that it is becoming inconvenient for them with the advent of AI, they will probably get rid of it or re-invent it in some way that even further benefits them.
Wikipedia says that King Diarmait Mac Cerbhaill adjudicated what was basically a copyright case over the Cathach as early as the sixth century, and the legal basis of copyright is founded in moral, economic, and property rights that existed in ancient Jewish, Greek, and Roman societies. Apparently, it took the printing press for unauthorized copying to become a big enough issue to be codified in law, but the poet Martial was complaining about missed profits from unauthorized copying and "book sellers would sometimes pay a well-regarded author for first access to a text for copying" (I.E. there was something resembling IP licensing, enforced by physical access rather than law) back during Roman times.
https://en.wikipedia.org/wiki/History_of_copyright#Early_dev...
## "Rhetoric and Reality in Copyright Law" by Stewart E. Sterk
Benjamin N. Cardozo School of Law. https://repository.law.umich.edu/cgi/viewcontent.cgi?article...
> Why give authors an exclusive right to their writings? Copyright rhetoric generally offers two answers. The first is instrumental: copyright provides an incentive for authors to create and disseminate works of social value. By giving authors a monopoly over their works, copyright corrects for the underincentive to create that might result if free riders were permitted to share in the value created by an author's efforts. The second answer is desert: copyright rewards authors, who simply deserve recompense for their contributions whether or not recompense would induce them to engage in creative activity.
> The rhetoric evokes sympathetic images of the author at work. The instrumental justification for copyright paints a picture of an author struggling to avoid abandoning his calling in order to feed his family. By contrast, the desert justification conjures up a genius irrevocably committed to his work, resigned - or oblivious - to living conditions not commensurate with his social contributions. The two images have a common thread: extending the scope of copyright protection relieves the author's plight.
> Indeed, the same rhetoric· - emphasizing both incentives and desert - consistently has been invoked to justify two centuries of copyright expansion. Unfortunately, however, the rhetoric captures only a small slice of contemporary copyright reality. Although some copyright protection indeed may be necessary to induce creative activity, copyright doctrine now extends well beyond the contours of the instrumental justification. ...
## "Copyright Nonconsequentialism" by David McGowan
Missouri Law Review. https://scholarship.law.missouri.edu/cgi/viewcontent.cgi?art...
> This Article explores the foundations of copyright law. It tries to explain why those who debate copyright often seem to talk past each other. I contend the problem is that copyright scholars pay too much attention to instrumental arguments, which are often indeterminate, and too little to the first principles that affect how one approaches copyright law.
> Most arguments about copyright law use instrumental language to make consequentialist arguments. It is common for scholars to contend one or another rule will advance or impede innovation, the efficient allocation and production of expression, personal autonomy, consumer welfare, the "robustness" of public debate, and so on.' Most of these instrumental arguments, though not quite all of them, reduce to propositions that cannot be tested or rejected empirically. Such propositions therefore cannot explain existing doctrine or the positions taken in debate.
> These positions vary widely. Consumer advocates favor broad fair use rights and narrow liability standards for contributory infiringement; producer advocates favor the reverse.' Most of the arguments for both consumers and producers prove too much. It is easy to say that the right to exclude is needed to provide incentives for authors. It is hard to show that any particular rules provide optimal incentives. It is easy to point to deviations from the model of perfect competition. It is hard to show why these deviations imply particular rules.
1. You can derive realness from materialistic, physical phenomena. So in this case, quark-gluon-electron interactions are real, and basically everything else is like 99% imaginary. "Society" is no more and no less real than anything else.
And/or:
2. You can derive realness from minds' conscious perception, experience, and interaction. In this case, "society", being an aggregate of many minds' mutual and conflicting experiences and interactions, is probably one of the most real things there are.
And of course you can unify the two, by framing the latter within the former based on the information-theoretic complexity of the physical substrate of the minds, and by framing the former within the latter as the common, shared basis of perceived and lived experiences.
---
But I'm not sure what concept of realness would let you say ""Society" is the most abstract illusion of them all".
Or do you mean that the rigidity of currently existing societies and their specific laws and customs, as universal and immutable constants, is what "has always been a complete fiction and pretend"? In that case I would fully agree; You can have multiple possible societies, in any number of shapes and configurations. But that does not make extant societies, as they are right now, any less real for those that live in or otherwise interact with them.
> So, maintaining a bunch of fictions in order to maintain another fiction becomes circular reasoning.
Uh, no. Not circular reasoning. Fear of starvation, more like. Fear of marauders killing your loved ones to take their resources. Fear of entropy.
Agriculture, presently existing as a function of society, is certainly very real– assuming your body obeys the laws of thermodynamics.
The second category of what exists in reality is human souls / consciousness and animal souls / consciousness. Beyond those two categories, there is nothing that exists, unless you enter supernatural territory. Which is not to be discarded, but here we're talking about man-made stuff. We can talk about money, society, ownership of land, and such. But they are always only imagined by people, they themselves can never transcend from the realm of idea and imagination into the realm of the real. Even though we have to relate to these ideas in everyday life because how they control the behaviours of our peers and ourselves, we can never forget that they are not real. But most forget.
In the past and in some regions even today, people held deep beliefs about curses and individual taboos. That a jealous neighbour or romantic rival would curse them, or more likely that a Shaman at their birth informed their family that this person can not eat this or do that, or they would die. And from accounts I've read, people believed this so deeply that they would have convulsions and even die if they broke their taboo by mistake. Just by the power of a false belief. Or the countless who waste their lives believing in some lie about "society". Or people who live under the curses of shame and guilt. And the curse of pride for that matter.
It is much more beneficial and easy on the soul to disregard all and any idea of society and instead deal with people as individuals. Because you're almost always dealing with individuals. And stay away from those who are too deep in the illusion and false belief of "society" or any such thing, because they only know how to bow to those they perceive as higher and kick down on those they perceive as lower.
As humans and individuals, the best thing we can do for ourselves is cleanse our mind from false beliefs. That doesn't mean that we will forget how these ideas work on people, it means we can see them from a higher perspective and that they do not exist. They can be used as pleased, but should never control a superior being.
Agriculture is not a function of society, it's a function of tilling the earth with a plow and sowing seeds. Nothing about it is abstract. I frankly didn't understand that example.
The Bible goes on and on about not having other gods besides God, but it also says that you should not _make_ idols, which I think is a very important and overlooked part. Anything that is undeniably created by humans is a false idol and worshipping it is a mistake. Most people alive right now worship their government, or more accurately their idea of what their government is. Because most do not know how their own government works on paper. Most people would die and kill for their government, while at the same time being nonplussed by other things that should make their blood boil. The slightly smarter ones worship money. One day when they're old, they'll be rich and have their own yacht. If they forgot about their fake money god for a moment, they could go sail the oceans as crew on a yacht while they are young and can enjoy the experience. Can they buy back their youth with some fake money printed in limitless quantities? No, because reality does not bow to the non-real, even if every single person on the planet believed in a falsehood. But a lot of people will have a whole lot of suffering from false beliefs. That's how literal hell on earth is created.
Therefore, people talking about morals or ethics or creativity or human inspiration or whatever are wasting their time on a false premise.
A more correct statement with the same thrust would be "copyright is not _natural_ -- it's an artificial, bolt-on construct."
And yet lots of artificial constructs are useful for incentivizing / de-incentivizing massed behavior in a way that is maximally beneficial to all. For sure, getting it right is difficult but let's not write off all artificial constructs as worthless or impossible to get right (especially when you just have to get it right _enough_).
Maybe, but you can also say that about money or about being married. Social fictions work as collective agreements on what is seen as useful for society. One can obviously critique their utility, but pointing out the ficitonal component isn't in itself a criticism.
Property and debt have much longer histories and seem more clearly defensible.
Of course there is one really big fiction out there that a lot of people no longer really believe, but that doesn't mean longevity isn't a thing to consider.
That reasoning applies to murder, too. After all, a bullet through your mom's brain is just a physically allowable reconfiguration of the position of some atoms.
So you're making a true point, but a useless one, since you're operating on the wrong level of abstraction.
Moral of the story for me is that if there’s adequate money to be made laws, protocols, etc. can totally be changed.
Couldn't agree more
Why write books or any creative text based content when the second it's available it will be hoovered into the next AI training cycle and can be reproduced in part or in whole by users who almost definitely do not know who you are in the first place.
Personally I think the only way forward is for these AI companies to curate their dataset to only material that is legal for use. E.g. only GitHub repos with licenses stating such. Just because there is not a license doesn't mean it's free to hoover.
They could buy a bunch of content created (or specifically licensed) for purpose, the artists get paid, and the artists that don't want their work to go into the pipeline get what they want too.
But too many people consider the development of AI to be such a moral imperative that no objections can possibly be relevant.
One can only hope.
Like… the entire idea of words like "society", "culture", "zeitgeist", etc. is that they imply a greater level of organization, shared values and ability to act as though one autonomous entity or whatever, than if you merely took "hundreds of millions or billions of" people and dumped them all in one place.
A society or zeitgeist can absolutely have 'hypocrisy'. If it can't, then it's not a culture, just a mass of hominids.
Plus, though it's hard to track, there's probably some degree of hypocrisy aggregated from an individual level too, where people who took one side before are taking the other side now.
Former Microsoft CEO Steve Ballmer: "Piracy reduction can be a source of Windows revenue growth." [1]
Current Microsoft CEO Satya Nadella: "Well, we’ve always had freemium. Sometimes our freemium was called piracy." [2]
[1]: https://www.zdnet.com/article/only-consumers-pressure-can-cu...
[2]: https://www.cnbc.com/2014/10/20/cnbc-exclusive-cnbc-transcri...
AI for those people just make their jobs of seeking what's worth pirating that much harder. Especially for those that can now fire half their staff and put most of that remaining money into said SEO optimization for their spam.
But in reality, everytime small entrepreneurs always thnk this will be the tech that let them usurp the big boys, and they end up selling out (not necessarily blaming them. I'd do it too). They have more resources, can scale up faster, produce more, and hire more people. It's a hopelessly risky gambit to think you can topple such advantage. And if you get an edge in, they just take it in for themselves and you're back at square one.
But you may not even get that far. They try to make use of this new commons and the Big Boys will litigate the gates back closed for them specifically. They got their value, made the next iteration, and then decided not to give back to the commons (a tragedy, if you will). They may even seem benevolent at first, but time goes on, priorities or leaders change, and the focus on market capture eclipses any public goodwill.
The same players as yester-decade are at the helm, so I don't know why anyone would want to trust them after giving decades of scrutiny as they gradually unmasked themselves. I don't necessarily fear the tech per se, I fear the same "Embrace, Extend, Extinguish" playbook happening all over again, maybe in such a way to make the new gap insurmountable.
It's been over a decade since an individual was prosecuted for digital piracy [1]. Longer since anyone was threatened with jail time.
When it was new, both individuals and companies were prosecuted. The tides shifted and law enforcement responded. This isn't a story of different standards, but one of evolving ones.
[1] https://en.wikipedia.org/wiki/Sony_BMG_v._Tenenbaum
My point is it isn't. Individual copyright infringement is virtually unenforced today. To the extent it was in the last decade, the penalty was a fine. Meanwhile, OpenAI is being sued by the New York Times [1] and various writers [2].
[1] https://www.nytimes.com/2023/12/27/business/media/new-york-t...
[2] https://www.reuters.com/technology/more-writers-sue-openai-c...
Some lawsuits don't change that fact.
Over the last decade, individual indiscretions have not been punished. Wholesale scraping is being punished. Some lawsuits do challenge the hypothesis that corporates' copyright violations are being treated more leniently than individuals'.
You seem to be conflating a whole bunch of things, seems like misdirection.
I'm arguing that policy preferences around copyright infringement have changed in general. In general, in 1990s, copyright infringement meant "crime, reprehensible, police, prisons, etc." for both individuals and coporations. In general, in the past decade, it's meant none of those things for either individuals or corporations. Yet it's meant fines and lawsuits for corporations with virtually none I can find, in America, aimed at individuals.
Also, LLMs were basically invented less than a decade ago [1].
> seems like misdirection
"Please don't post insinuations about astroturfing, shilling, brigading, foreign agents, and the like" [2].
[1] https://arxiv.org/abs/1706.03762
[2] https://news.ycombinator.com/newsguidelines.html
>I'm arguing that policy preferences around copyright infringement have changed in general.
Perhaps they have, and I take on your take on that. In the end you were replying to me and the original poster so respect the spirit of those posts.
Someone compared treatment of X 20 years ago to treatment of Y today -- seems pretty natural to bring up treatment of X more recently. You can't just say "the original comment didn't mention it so you can't mention it either".
I don't see how your accusations of bad faith are warranted.
I feel I have a pretty open view.
And, there's no definition of a 'valid' comment. I'll address points, raise points or whatever. Sorry, I'm not in the realm of 'valid' comments, never was.
GP made a comparison! How do you refute a comparison without criticising the comparison?
> I'm not in the realm of 'valid' comments, never was
Claiming misdirection is an argument about validity.
Fair enough, I'm talking about America. To my knowledge, individuals downloading pirated content have not been threatened with lawsuits. And to the degree seeders have been threatened, it's only that--threats. When was the last distributor actually sued?
But the fact that they're not being sued is a weird delineation to make. They're not being sued because they can't afford to defend so are settling. It's not a case that IP owners aren't enforcing copyright, they're just not escalating to actual lawsuits because they don't need to.
I’m asking for evidence of these settlements.
The claim is corporations are being given a free pass relative to how individuals are being treated. I’m not seeing that signal. What I am seeing is Americans becoming more sceptical about intellectual property as a result of the MPAA and RIAA overplaying their hands in the 90s.
Objectively, yes. Paying a settlement in private is better than being publicly prosecuted and then put in jail.
That said, it's still no cakewalk. Do you have a source for individuals settling copyright claims? I'm not finding any recent stories nor surveys.
If you ask it to complete text and give it the first line, it will sometimes continue the lyrics. Seems to work best for particularly famous stretches of lyrics, like Lose Yourself by Eminem, Bohemian Rhapsody or Hurt.
When I told it to complete the text "Obie Trice, real name no gimmicks", it said I'm sorry I can't reproduce the lyrics to Without Me by Eminem. And offered to tell me more about Obie Trice.
When I just asked it to reproduce the lyrics to Bohemian Rhapsody, it once again refused and offered to analyse the lyrics.
Seems like there's clearly song lyrics in the training data, and that they've at least made attempts to prevent it from regurgitating them.
But look at this...
https://poe.com/s/Y73se7ueRrQy9xHRPuoh
I swear the chaps at OpenAI have it all as exact copies. Articles and whatnot. Why wouldn't they if they have the lyrics precisely cloned.
Opinion articles.
I want to gently reframe the debate, for while I agree with the hypocrisy it rather misses a very key point.
Intellectual law, as bedrock principle, explicitly doesn’t recognize nonhuman creators. So at the heart of the issue isn’t changing copyright laws or the death of IP, but what to do with non-human creativity? This is an interesting issue now, but most of the debates and options being debated thus far won’t survive an AGI let alone a world full of advanced AGIs.
But it’s a critical distinction between “so now it doesn’t matter when it was theft before,” and “what do we do with non-human intelligence when our entire system of creativity protection is build around humans and non-humans now exist?”
It’s a paradigm shift that gets somewhat denied by the hypocrisy argument. Most people look at AI as “technology” we have developed, but if any science fiction writers are right it’s actually a bona fide digital intelligence that’s getting developed here — essentially the possible digital twin of human intelligence — and that’s a whole different set of considerations.
Claims certainly get made, but they are subsequently shot down by future research. This has happened for discussions of emergence [1], reasoning [2], and other human attributes people have ascribed to LLMs.
[1]: https://neurips.cc/virtual/2023/poster/72117
[2]: https://m-cacm.acm.org/blogs/blog-cacm/276268-can-llms-reall...
That’s the difference between seeing it as a technology vs a larger directional shift in human development. Put more directly we have been experiencing the age of miniaturization, and now we are entering the age of digital intelligences.
Very often early technology looks toy-like and lacks substantive real world application (despite the hype). It’s only later that deliver substantive applications and real-world change. This generation of AI is following the Gartner hype cycle perfectly thus far:
https://www.gartner.com/en/articles/what-s-new-in-artificial...
That said Gartner tends to be part of the hype cycle these days, but the pattern of new technology holds true.
So in short I would expect anything like a digital twin to emerge from post trough applications not the current generation, and I don't mean to imply that the current technology will create such a system but just that science fiction and futurists generally forecast or even predict this technology will generate a digital twin (in the form of an AGI) from this civilization-level line of technology exploration.
All that said, I don’t think many of us technologists were thinking we would be anywhere near this level of AI technology in 2018. In five or six years we have gone what people thought would take 10-20. So, as always, linear extension predictive and mental models of technology advancement once again fails to predict the true course of technology progress, which seems to be much more dynamic than people generally consider when talking about the future.
What happens when AI systems can work around IP?
This gets done all the time now. It is going to be way cheaper.
Microsoft has been arguably the largest beneficiary of software copyright. Gates argued for it in the early days of MicroSoft. Microsoft's anti-piracy campaigns are unparalleled. Anyone remember the "Business Software Alliance".
Now the company wants to be the largest beneficiary of "AI".
It is nature after all, to spend the least amount of energy to attain a goal.
It would be convenient to accept it as par for the course.
The problem seems to be is that copyright laws are basically ignored nowadays.
[1] - https://rikverse2020.rikweb.org.uk/blog/copyrights
(sarcasm only half-intended)