Nobody is going to "generate training data" if they get precisely nothing out of it.
The complete disregard for intellectual property rights being exhibited by the "AI" community will inevitably come back to bite them in the arse one day.
The complete disregard for intellectual property rights being exhibited by the "AI" community will inevitably come back to bite them in the arse one day.
For example, in the early days Amazon famously "leveraged" the purchase of out of print books from their wholesalers to obtain orders that would otherwise be too small to ship.
That's just how capitalism works. You have to be either too small to bother with or too big to fail -- the problem area is in the middle.
For example, there's this widely told story about how Airbnb in the beginning exploited Craig's List to list their properties. This is not illegal, it's not that they hacked into Craig's List, but it's for sure not ok. It's ok to break the rules as long you are not caught. Imagine if everybody would start doing that. Therefore, I'm also not surprised that they don't give two shits about city regulations and so on.
Right, that's what "hack" doesn't imply to me. I mean, this is "Hacker News", but it is not about nefarious activities. That said, I do understand that's what it implies to the younger generations.
> It's ok to break the rules as long you are not caught.
This is an attitude that seems to be common in SV, and I'm disappointed if YC doesn't reject applicants who demonstrate it.
In socialism, as far as I can tell, anyone would totally be able to ingest all these books, as there would be no concept of intellectual property rights. Thus, what you complain about with respect to capitalism, would be a totally normal and acceptable thing to do in socialism.
Do you have some other, alternative system in mind, which would not allow a wholesale exploitation of private intellectual property?
"Der Tod eines Menschen: das ist eine Katastrophe. Hunderttausend Tote: das ist eine Statistik!"
If you owe the bank $10,000, it's your problem. If you owe the bank $100m dollars, it's the bank's problem.
But do you think that China and Russia are going to do the same?
The western world keeps getting crippled by ransomware attacks and other malware because nobody can practice the dark arts domestically without risking serious jail time. The only people who can advance these skills enough to know how to defend against it (or return fire!) are either CIA, Israeli contractors, criminals, or foreigners abroad. An American will go to jail for hacking a Russian bank, but no Russian goes to jail for hacking an American bank.
Train AI on foreign content only. Understand their culture better than they do. Then learn how to subvert it. It's what they've been doing to us.
https://www.newyorker.com/business/currency/what-ever-happen...
Now I think we'd all agree to something more realistic to actually protect authors, like 10-30 years protection from publication. But every time someone brings "copyright! IP!" to ML I just laugh since it's one of the more corrupt laws that we have currently.
also how many people are responsible for stuff like ChatGPT... 300? getting the bulky of their trainning data from, again, who?
it feels the monopoly history is re-inventing itself, now with fancy computers replicating human behaviour
I guess the only solution is to get rid of copyright law so people don't exploit it so much.
Problem solved. Let's all stop protecting this, and allow anyone to innovate by making cool and new awesome things.
How about Make the model so big you can only run it in a data center?
My older brother decided to go to art school.
It wasn’t until he was 40 that his artwork started to be noticed and worth something.
His images have been bootlegged endlessly by Asian companies placing them on t-shirts and selling them for $5 on Etsy and every other marketplace.
Fortunately for him, he’s designed a very popular set of kids books that have allowed him to move out of my parents, get a place of his own, pay down debt and buy a house.
Intellectual property is how he takes care of his family. He should be able to at least provide for his kids college after his death if his IP still has value in the marketplace.
…And so should other creators.
Not everyone is Disney.
About the inheritance I disagree, when you are a farmer and sell a potato you sell it once and then it's gone, if you are a carpenter you make a furniture piece and sell it and its gone. Why should art be a "legacy" that can be kept selling forever?
As an artist starting out in the beginning of my career, I made a rationalized choice to never post my work online - in retrospect this seems to have been the right choice. My work is no AI’s whetstone.
It is a luxury career supported exclusively by surplus. There will always be demand for it, but it is highly elastic and heavily influenced by trends and skewed by the top end.
Sorry if I framed it confusingly, this isn't only for artists, it's for everyone. They don't care what you do, only how much money you make.
> how it's not the same as a basic unemployment
It pretty much is, though unemployment has more strings attached. Anything you get from unemployment is subtracted from social security, and vice versa. For unemployment the online form is simpler and money comes much quicker, but you have to attend unemployment office events and stuff.
> whether there is a maximum amount of people who can be on this programme
The true maximum is of course however much the system as a whole can support, through taxing people's income. My back-of-the-napkin math puts the cost of the whole program at about $1/mo per taxpayer. So all artists and bums combined. Might not include basic retirement, which is slightly less.
> how qualification works
It's based on income/assets. If you make less than $10k/y plus $150/mo, the program covers the difference for living expenses.
In practice, there are limits to how much they're willing to pay for things like rent and utilities. In my city the max rent is $560/mo. On this winter's coldest month they complained that my electricity bill was too high (electric heating), but still paid it in full when I complained back.
To qualify, you have to attach to your application all of your bank accounts' statements from the past 3 months, so they can see how much money you've got. When you get a bill, you send the bill and they cover it.
The social security program covers: rent, water, electricity, home insurance, Internet, healthcare, moving van, security deposit of new apartment. On top of that you get about $15/day for food and other stuff like clothes, cleaning products, what have you. In my experience food costs some $10/day to live comfortably so some $150/mo remains for everything else. Second hand clothing stores sell decent clothes for a few dollars a piece, sometimes less.
You can request additional funds for infrequent, dire needs, like a new bed, vaccuum cleaner, etc.
Feel free to ask more questions, I'm happy to elaborate :)
Ah, OK, that was what I was wondering. Some social security schemes for artists that I am aware of require proof that you are in fact an artist, which can be problematic.
I think it doesn't go quite far enough though. I still believe a universal basic income would do better and be easier to administer.
What is your career, precisely?
My point was that it's a career whose demand is dictated by fluctuations in the economy and trends. And can be almost completely shut down (in theory) if the situation dictates.
Wherever you stand on copyrights, it would be a mistake to underestimate the central importance of this issue going forward.
I don't think my comment implies that at all. I'm not convinced it'll ever grow that large though. See: Content is not king [1] for a good explanation of why.
I think it's natural that it grows now to levels never seen before.
What am I underestimating? I agree it is an important issue; my comment is orthogonal to it though.
The US has cultural poverty because it has decided to support litigation machines.
I mean a good thought experiment here would be to replace "artists" with "entrepreneurs" or "founders" ("historically, [...] were independently wealthy" would expose some of the myths we attach to the idea of "self-made" which we rarely attach to artists and authors) here and rethink the history of Western commerce from that perspective.
Can you explain this position? My understanding is most artists don't thrive economically because there's not much demand for the art they make. I'm not sure that's correct, but "lack of IP protections" seems even less likely for most artists. What protections do you think would help?
It seems to me that the current system primarily benefits corporations who acquire a vast library of IP and can afford to legally defend it all as necessary.
I agree that a different set of policies would result in more art being created and more artists who are able to support themselves doing art, but my immediate assumption of what that would look like is more like funding art educations and exhibitions (of various kinds).
https://qph.cf2.quoracdn.net/main-qimg-53ec1e395a4b59adb8391...
Yeah, the law may be more on their side, but overall it seems unless you're somebody on the level of Akira Toriyama you likely barely managed to retain your sanity while cranking out your world famous works, and even the famous ones suffer greatly during the process.
You think we should have 1000x more artists than we do? I think there's another economic problem with that idea...
> The US has chosen relative cultural poverty compared to other cultures that find non-market mechanisms to support artists.
The USA has the largest market in the world for creative products and the most rich artists.
> My work is no AI’s whetstone.
Are you the same way with juniors? "I paid dearly to learn this technique - you should too!"
Aside from overtraining issues, the AI can't store your work anymore than you can store representations of everything you've trained on, it's vastly smaller than the sum of its training data. It distills out features and their combinations.
Some bigname artist is upset because he thinks he's the first one to put certain bat and lizard features on a dragon and that he now owns that entire sort of creature. Turns out though, that given an old picture of a dragon and that single sentence of mine that he could be copied by almost anyone. The only way to keep the AI from "copying" his work is to make sure that, even if not trained on his work, nobody asks it for those features. To satisfy these people it'll have to have a big red sign that says "Dragons are off limits, Bob owns them because you might put claws on the wings!".
What is the market price for a potato?
What is the market price for a chair?
What is the price of developing an art career over 40 years? Unless you believe that art, music, poetry, literature, and the rest have no value to society then IP protections are what incentives a portion of humanity to do something crazy like go to art school…
…instead of you know doing the “responsible” thing and getting a CS degree so you can one day get an AI gig stealing “training data” in the name of profit
If you want to encourage more creative work, maybe look for other mechanisms, because going into creative work in the hope of making it is pretty much gambling with awful odds.
Sure, we can say that still isn't "good enough", but that kinda proves the smaller point that the actual creative output of most creatives is not valued, and at the same time ignores that the main point I made above is that it is irrational for people to gamble on being valued, because the vast majority of "successful" creators as measured by relative rank within their peer group has no realistic prospect of that happening today.
* Rejecting intellectual property necessitates rejecting that art provides any value to society.
* The only reason artists make art is to make money.
* (Implicitly) Without IP artists would have no incentive to make art, monetary or otherwise.
* Making art requires going to art school.
* Getting a CS degree and making art are mutually exclusive.
Of course, with weaker IP protections, there would be change. Some very specific business models that exist today would go extinct. And other business models (e.g. live performances) might see a renaissance in popularity.
> IP protections are what incentives a portion of humanity to do something crazy like go to art school
Hard disagree, most of the art fields are first a passion and then a profession. People who go to art school is normally because they love it so much that they cannot imagine themselves doing something else, even though they already know it pays little. I've never ever heard of someone going to art school "because of IP protections".
One of the reasons it's so hard to make money is because there's many people doing it as a hobby that are already great at it, and so would jump at the chance of getting some money for it, leaving people who want to make livable wages outcompeted. Which is fine, this way society as a whole benefits greatly, sure it's unfortunate there are no more people living off it, but in exchange there are many, many amateurs experimenting and doing art, and from time to time one finds a formula that allows them to live off it (or "shills" to corporate).
I really didn't get any vibe of bias from them.
Either way, it seems you miss their point entirely, which is that you cannot equate a creative idea as the same as a product.
The furniture example all boils down to if the work is allowed by the author to be mass produced (if it even can be done) but the creative idea is still theirs.
>I've never ever heard of someone going to art school "because of IP protections".
Similar with my first point you misunderstood their point. Their point is that thanks to ip protection they can invest the time needed to become a successful artist within their lifetime.
> "He should be able to at least provide for his kids college after his death if his IP still has value in the marketplace."
Why should he, besides because that's a good desire for a family member? If it was a 3rd party wanting this, usually we'd call that rent-seeking behavior and it's usually seen negatively.
> But the creative idea is still theirs
Why is this (besides copyright law ofc)? The carpenter doesn't forbid me from creating a similar chair of him, and if we are talking about a painting and e.g. I was learning I could definitely copy it for learning purposes and everyone would be fine with that, so at which point should it "not be okay" to copy it? To lend to a friend in private? Can I share (in person, no copies) to a group of friends? I bought a painting, can I print it on a t-shirt for myself? What I can do after I purchase a copy of the idea in private should not be up to the author IMHO, and that line is often very blurry.
Technically as long as you don't sell a product that is a copy, you are "fine" to have said product.
However if you do sell it, it would incentivize a parasitic market.
Something you see sometimes in open source software, esp with IaaS.
You sure about that, legally?
>…instead of you know doing the “responsible” thing and getting a CS degree so you can one day get an AI gig stealing “training data” in the name of profit
You're right about this, though. But this strikes me as a discursive issue that has more to do with the generalized butthurt over schooling that you see here in HN and tech circles - I think it's funny that it's Local Boob Paul Graham talking about writing (which to be fair he cannot shutup about) that set this whole thing off.
They just show it to the AI, it doesn't store it - just data about it. The same as you when you look at something, be it a painting or a chair.
The only reason AI companies store the data is they want repeatability. If you want to train two copies on the same data you can't rely on the same sites being up. I commit the same crime by saving websites I've read because I don't want my sources to dry up on me.
> What is the price of developing an art career over 40 years?
Market price. How does it benefit the customer here and now?
It's no different than being a furniture maker with years of carpentry skills. The value to the customer is still just the marginal benefit over an IKEA stool.
In general for machine learning, yes, but deep learning models with large numbers of parameters such as transformers and stable diffusion do actually memorise part of their training data verbatim. The problem is mainly duplicate data points. E.g.
https://arstechnica.com/information-technology/2023/02/resea...
https://arxiv.org/pdf/2202.07646.pdf Quantifying Memorization Across Neural Language Models
I was translating code with one of the code- models at OpenAI and it spit out about thirty words of someone's code of conduct from Github, almost verbatim. I found the repo by searching what it had produced. That's when I changed my prompts to remove comments, but it stuck with me.
In general though, I think the pigeon-hole principle is relevant. The model is fairly small compared to the training material and couldn't contain much actual full content or it wouldn't be able to contain anything about the rest.
I've heard people talking about using the model to judge similarity of new training data to what it's already seen, mainly to train more efficiently, but also to avoid this.
The point of copyright is to encourage more work. At this point he is rent seeking.
Every entrepreneur should also be forced to sell their business instead of passing that equity to their children. No more rent-seeking, multi-generational family businesses.
Matter of fact, let’s force the same on anyone with a 401k, IRA, or Roth as well because at the end of the day it’s all equity, right?
All equity stakes much be converted to cash upon death and either returned to the state OR given to charity.
That’s the way to make “Single serving wealth” happen and put an end to rent-seeking.
Meanwhile there's no effort involved in collecting royalties from your father's work, and no social benefits are provided by it.
That said, your comparison to forced confiscation of retirement plans by the state or charity is very strange, because it's not as if on your brother's death the state would come in and takes every cent he's ever made as a producer of IP.
In fact, the state literally does the opposite, and gives your brother's children and grandchildren the entire backing of both the civil and criminal justice systems. This is why we spent the 90's and 2000's sitting through VHS tapes telling us that unauthorized copying will be met with an FBI investigation, $250k in fines, and 5 years in prison.
There's nothing stopping an extremely successful and wealthy artist from handing down the money he made to his children. If they want to use that money to continue doing artwork in his style, using everything they have learned from him, there's nothing stopping them.
And why not, really? How does that affect your brother? Why should your brother have a say on what people can or cannot put on a T-Shirt? Creating pieces of art professionally should be paid for. But having some weird right of ownership on what is essentially just a huge number is, frankly, a clutch at best and outright criminal at worst. Digital artwork can be copied at zero cost and artists have to deal with that, not the world has to bend to one particular model of generating revenues for artists.
We need to stop with these schemes at one point and shift to a better market or we will effectively have to ceed all control of data and computing to copyright-control companies.
What's the happy middle ground?
Of course they will. But so can others. Without copyright Disney won't have a legally enforced monopoly.
It too can be copied for zero cost.
Why should MS have a say on what people can or cannot put on the harddrive (Well ssd)
If we were to get rid of it entirely then your success in the market largely depends on what resources you already have. Large companies have the upper hand as they already have the user base, along with the marketing budget to sink you (not that the current situation differs all that much).
If your argument is "thats the way it should be, let the market decide who succeeds" then that's a whole different conversation.
Either way I don't see an obvious solution that doesn't benefit the very companies you say we need to stop, though I agree on the sentiment.
and then, imagine if a company arrives that offers a product, let's just say, imagery. artwork. at a low low price of 'free'. not just "free copying of existing artwork", but complete "professional creation". for free. it's real hard to compete with 'free'. you may have been selling your artwork for some price, but when a 'free' alternative comes around, what do you do? how do you solve for that "challenge" on the market? tough shit. and also, since it's 'free', it's also 'whatever price they want to set', so they can undercut you at any price point, because they can just make up any price point. capturing customers, capturing profit, while you just get displaced in the market. good luck trying to explain to people that "creating pieces of art should be paid for" and that "you" "should be paid. people can just get art for free! or, at a low low price that undercuts you.
see a problem that arises here? this is what those annoying art people have been trying to argue about.
Does the system really work if he had to invest in himself for so long to even start having hopes if sustaining himself with his creative output?
If we didn't assign value to abstract concepts, ownership contracts for tangible assets would only be worth the paper they're written on. Abstract ideas are inherently valuable.
This is a very young/naive way of seeing it, as the Spanish saying dutifully notes, "physically having something is 9 parts out of 10 of the law for ownership" ("La posesión es nueve décimas partes de la ley"). Meaning ownership is not as much what is on paper but it is having something, claiming it's yours, and no one else claiming it's theirs basically (not always, but that's why the saying says 9 out of 10).
https://primaveraeuropea.eu/la-posesion-es-nueve-decimas-par...
This is already an English saying. "Possession is nine tenths of the law."
> Intellectual property is how he takes care of his family.
Art is how he makes money. Money is how he takes care of his family. People have been making these things forever -- they are older than human language itself! "Intellectual property" otoh is a relatively new idea, and so can't have everything to do with it.
To think "Intellectual Property" laws are corrupt and benefit the already-rich should not be incompatible with saying some copyright term might be good: If copyright terms were half what they are, your brothers' story would be exactly the same, but Disney's would be a lot different!
I'm not saying it's a good argument (it's sentimental, for starters, not that OPs is much different - "proprietors" didn't invent IP law, and Disney is apparently giving its pound of flesh back to the powers that helped it game the system), but I'm pretty sure the same one was used in defenses of generative work and, notably, NFTs - "this will help the poor artist, who would otherwise have no market." Rogers v Koons put paid to that argument (indirectly), and I'm hoping the current class actions succeed, your brother notwithstanding.
For every case like your brother's, how many people out there are throwing their lives away chasing the illusion that they will ever "make it"?
For the general expression of ideas in a corporate setting, copyright is an asset that is backed by a legal team. You're betting that a property gives you a competitive advantage.
I don't.
I don't think my open source stuff is every going to be worth a penny. I still want it to be protected so that its license is respected. Mine, and everyone else's.
Yes it does.
I went to a meeting with one of our clients that was just signing with us. The main reason that he was contracting with us was, I quote, "the last time we went to China we found a Huawei rep making a demo of an application that worked exactly like ours - which was meant to run on Cisco gear. Even the terminal escape sequences to get to debugging mode was the same. The Chinese were stealing from us and there was absolutely nothing we could do about it."
Copyright law might be enough of a deterrent to make Joe Nobody pay $9.99 to Netflix instead of torrenting movies, and (on occasion) it may be enough to force some high-profile artist to give credit to some other lesser known musician who "served as inspiration". Outside of that, copyright law is as effective to "stop bad actors" as the "War on Drugs" is effective to stop organized crime and drug trafficking.
Let me spell it out: I think that the cost of trying to police "copyright violations" far outweighs any of the potential losses and damages that might be incurred for losing such "protections". This is not the case for murder. The cost of life is incalculable, so any measures to prevent loss of life is justifiable.
Laws that try to protect citizens from physical harm and are meant to preserve human life are of a completely different nature than "copyright violation". It's perfectly reasonable to argue for decriminalization of copyright infringement without having to accept that anything should be decriminalized.
To go further: I see no moral justification to have copyright infringement as a violation worthy of criminal persecution. I believe that "intellectual property" is a misnomer and I think that any type of current laws are meant only to protect large corporations and are not made with the best interests of society in general.
Is that enough for you, or are you going to continue pushing false slippery slopes?
You misunderstood. I was not trying to portray your argument as a slippery slope. I was trying to portray your argument as just generally weak, which it obviously was ("but china" is not even an argument). Given that you completely misunderstood my point, the rest of your comment doesn't even apply.
That you had to resort to subtly calling me an idiot says a lot.
> That you had to resort to subtly calling me an idiot says a lot.
That you didn't mind me subtly calling you dishonest says even more, and shows that I'm done here.
Which is why all AI code and datasets are freely available and legally usable by anyone for any purpose?
If you had created anything other than internet forum opinions in your life you'd have a different perspective.
Yes.
> Do you really believe they did it without knowing what they were doing?
Of course not, but I also believe they thought it would be more effective at making their point than it was.
It is tough to respond to something that is crass and meant to be offensive without resorting to the same, so by using the 'annoyed teacher' style I can get my point across without crossing the line to snark or contemptuousness.
I see. I suppose this is where we differ, I'm sure they don't care about making a point at all.
That it’s been abused by corporate entities is another matter. The fact that some famous musicians sell their whole catalog to companies is also a sign of times changing again.
AI growth as it is here based on preexisting work is directly in confrontation with copyright. And something will have to give in some way. But law will likely not catch up before at least another 10 years, the time needed for most people to really understand/experience the impact of the changes.
It is a necessary protection against moral hazard, and a way for artists who may well be poor during their lifetime to be able to leave something behind for their children.
In your world, if one wants to print Calvin-peeing-on-things t-shirts without having to deal with legal hurdles, all they'd have to do is shove Bill Watterson into traffic to force his creations into the public domain.
There is no point to creating anything if someone else can throw you under an actual bus to immediately claim rights to (and start profiting from) the years of time and effort you spent building the "brand." Forcing a delay of 50 years goes a long way toward discouraging this sort of literal piracy.
This is the premise of the current AI debate, only it's forgery at scale instead of manslaughter.
IP laws are legal constructs and subject to changes like everything else depending on our needs. When and if enough of us see benefits from AI hugely outweighing benefits from maintaining IP framework, I'm pretty sure laws will be passed to make needed adjustments.
Authentic human generated information has irreducible newness.
I believe that point is open to debate.
"Human brain is magic." <> "Human brain is not simulatable by Turing machines."
There are other possibilities. I'm sure you can think of a few.
Sure, a style like cubism is unlikely to appear from current AI art generation, but ultimately there's a human still in the loop. The human picks the direction the generation goes in and you can definitely see "styles" or even "characters" appear in certain types of prompts.
If you hang around people that do this a lot you'll see that some of them end up generating hundreds, if not thousands, of images of what seems to be the same character in varying situations. That indicates to me that you could potentially find a type of prompt that would lead to an art movement.
Hell, AI art itself might constitute an art movement.
Think language change every 1000 years. Would AI adapt to that? What new training data would it require? Surely obviously it would require some new inputs. It surely requires internet connected speakers of the developing language.
I think the “surely it would require new inputs” is enough to favour the irreducible newness point.
“Irreducible newness” may be just overhyped “changing relevancy” though, that’s what I’m stuck on.
You see, you're part of the system, not outside it. If it were not for my deep, deep hatred of the farce we call IP law, I must admit, I'd seriously consider walking over to the other side; because I look at these AI folks scouring training data without so much as asking the original creator for consent, or even sparing a couple cycles to think about whether they should, and it really rankles me.
In fact, the entire damn computing industry is filled with people with some seriously flawed ideas around the concept and mechanics of the act of obtaining consent.
Consent for… what exactly? Why is anyone entitled to some notion of ownership & power over who may and may not make use of information?
I understand to some degree practical arguments around recouping cost of development and distribution of information, but I don’t think anyone should really concern themselves ethically with what some random person wants them to do with the information. This isn’t like ensuring a sexual partner is an enthusiastic participant, this feels to me more like I’m not going to ask mommy for permission every time I want to play outside (let alone care if I’m told “no”). What practical defense is that anyways?
Maybe my understanding of your framing is off?
You mean like copyright holders? Patent holders? Somebody with a trade secret?
>but I don’t think anyone should really concern themselves ethically with what some random person wants them to do with the information. This isn’t like ensuring a sexual partner is an enthusiastic participant, this feels to me more like I’m not going to ask mommy for permission every time I want to play outside (let alone care if I’m told “no”). What practical defense is that anyways?
But yes, anyone should. You're just saying "I don't think people should disagree with me" not "information distribution and consent aren't ethical concerns" - of course they are! Why wouldn't you propose as counterexample a model like medicine's "informed consent" instead of sexual consent or "asking mommy"?
These are legal fictions that exist because the state’s monopoly on violence. Like I said, I understand the motivation to enable e.g. book writers to seek profit for their work. That’s about where my sympathy ends, especially when those mechanisms are abused for petty control by entities making billions. I hardly view them as ethical constructs, in fact I think the vast majority of restrictions based on these are unethical and immoral.
> Why wouldn't you propose as counterexample a model like medicine's "informed consent"
I think there’s a disconnect here. “Consent” in a sexual or medical context is ethically necessary due to the fact that the asker wants to perform acts directly upon/with the concerned party. Consent to play outside or make use of information does not involve the intimate invasion of one’s person, and really does not need to involve this other arbitrary party at all. This is what I was trying to express in my previous reply.
I don’t see why anyone is ethically entitled under these conditions to insert themselves as a governing authority of how people use information. It’s not even a realistic expectation to hold, information wants to be free and people will do what they want with it regardless.
> You're just saying "I don't think people should disagree with me"
I mean, we are talking about ethics here after all. The best I can do is disagree with your conception of consent and make my own arbitrary value judgement.
Well, ok I get you there. But they're not really "fictions" when backed by the force of that violence. This critique makes sense, I just don't know if it really can ignore the fact that the legal regime applies in most cases.
Did you pay for the product? See, that's what you get under Rights of First Sale. If you pay for it, it's yours. Free and clear, no entanglements whatsoever. You still generally have to ask and negotiate a sale.
>This isn’t like ensuring a sexual partner is an enthusiastic participant.
I wasn't even going to bring it up, but since you opened the door... Someone has to bear significant cost to bring into being something (this is the part you claim to understand, do I'll hold you to your given); you come along and enter a business relationship with them (they are a supplier), set their compensation at non-negotiable 0, use their output as your input, to enable you to do things that they might not be at all comfortable contributing to, even indirectly, which again, you don't even know either way, because you didn't ask. Just saying, you've got more similarity here than you may be comfortable with on further deconstruction. Can't blame you for not wanting to do it, but as someone who has, maybe you should reevaluate?
Besides which, why does everyone think "consent" only applies to sex anyway? Consent is a fundamental underpinning of a liberal government of, by, and for the people.
>This feels to me more like I’m not going to ask mommy for permission every time I want to play outside (let alone care if I’m told “no”).
Live somewhere in the midst of a warzone, political upheaval, or a high crime area, and this could get you shot. In the case you were a child, where your mother has custodianship of you, she bears responsibility for your actions. That you, in that situation, would not care if she said no, just sort of proves my point.
Stakes may be low, but there is a flow to the interaction. If it were not an issue in our industry "it is easier to ask forgiveness than permission" wouldn't be such a sticking point of our industry.
It's reflected everywhere. Even the innocuous places. Our biggest industry actors have made their fortunes on tracking people whether they wanted to be tracked or not. We have browser fingerprinting, which is the equivalent of "hold on a minute, need to figure out what your living room looks like so I can forward the description to all my buddies so we'll know who you are when we next see you." Except that isn't ever outright said or explained to the end user, because if it was, no one would ever consent. Which is why nobody asks and just does, which proves our industry has a problem with consent.
These are exactly the types of thing to get tied in ethical knots over, because damnit, we won't be here forever, and we're setting the example for down the line. While it'd be nice for our practices around IP to be less rigid, there is still a line to be drawn such that it is not okay to just run off with digital info because digital.
Every vice we maintain for ourselves individually, will be abused collectively; likely to a far more destructive degree. Asking first is not a huge ordeal. Lord knows. We might get so tired of being asked we finally figure out some default terms so that people can stop asking. That never happens if no one bothers to get the ball rolling though. You have to ask to get the iteration started.
https://ustr.gov/sites/default/files/2019_Special_301_Report...
Those "rights" should not even exist to begin with. Nobody is interested in bending over backwards just to appease a bunch of aging monopolists. We want to explore the full potential of technology, not see it crippled by artificial limitations due to "concerns" over imaginary property.
If Disney trains an AI good enough to just generate a good story then I think that everyone who fed training data into that model should get a piece of the revenue.
Except right now models are being released openly, with anyone being able to download and fine-tune the weights. Yes, they have licenses, but it's not like the licenses protect some core part of their business models. Whatever incentives they had to release those weights, they would mostly have the same incentives if copyright law disappeared overnight.
So, if Disney decides to incorporate a sailing boat in ther stories, does that mean that anyone who has been making sailing what it is in our culture (i.e. anyone ever been on a boat, built a boat, talked about boats etc) should get a cut? If not, why not?
Disney will just commission this art or figure out a neat trick to get artists to give them this permission. Then afterwards they'll turn around and license the model back to all those artists. And those artists are going to pay to remain competitive.
Congratulations, you've now created a monopoly or oligopoly. And the people who lose the most from it are the same artists that demanded this system, because free models will be illegal.
We're in the 21st century, the age of information and networked computers, and these people are trying to sell bits. It makes no sense. If your work is making bits, you need to figure out a way to get paid for the labor of discovering those bits, not the bits themselves.
Because god knows how easy it is to copy those bits after they've been found.
Writers make basically all of their money from copyright-related sources, but the median income from writing for full-time authors in the US is 20k [2].
My point is, copyright is a way to compensate creators for their work, but it's not the only way, and in practice for most people it doesn't do a stellar job.
[1]: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2199058 [2]: https://authorsguild.org/news/authors-guild-survey-shows-dra...
Property rights are the basis for any society in which people have independence and autonomy. I think we should put software developers on sponsorship and donations for a year so that they understand the consequence of an economy like this.
Most of what we see today as the greatest art of all time was produced under this arrangement, so if we look purely at the outcome for society, the arrangement can't be that bad.
As if it's any different now. Wanna create something and need money to make it happen? Be prepared to enter into arrangements with rich investors and corporations and producers. The people with the money will literally get credited in your works.
> It's pretty funny how artificial is a pejorative in your world when it precedes 'scarcity' but not when it precedes 'intelligence'.
The "artificial" in intelligence is a technological limitation: it's not a real intelligence because we haven't been able to make one. The "artificial" in artificial scarcity is a condition humans impose on abundant resources in order to create an economy where there is none: we have the technology to copy and distribute these abundant resources but the monopolists won't have it.
So you bet it's pejorative.
> Property rights are the basis for any society in which people have independence and autonomy.
Sure. Real property like land and physical possessions. Not this imaginary property that only exists to let people turn abundance into scarcity.
> I think we should put software developers on sponsorship and donations for a year so that they understand the consequence of an economy like this.
I don't even disagree. Software developers are also creators that exploit their monopolies for gain. Personally I don't even believe in copyleft licenses, they only exist to fix copyright and turn it to our advantage. There should be no copyrights to fix in the first place.
There is no difference between the two. All property is 'imaginary'. Turning abundance into scarcity is literally the only reason to have property rights all the time, including in the physical world. Land isn't actually scarce in most places.
The reason we introduce scarcity and ownership for land as well as digital goods is exactly the same. because it gives the owner an incentive to tend to it. It lets you capture the value of the improvements you add to it. The enclosure of the commons didn't happen because the commons were scarce, it happened to create conditions for their development rather than their abuse.
Software developers who don't own their code earn 800 dollars on patreon will Microsoft makes millions of their labor. This is literally the economics of free software. It's the largest upwards distribution of wealth ever invented. Imaginary property, that is to say intellectual property, is the only thing that puts you on even ground with Microsoft. You want a world without it where hardware is the only thing that matters? Good luck competing.
No way. Physical things are scarce. This is basic physics. You cannot duplicate energy or matter. Two physical bodies cannot occupy the same space. Physical property is not even in the same conceptual space as intellectual property. Land isn't scarce in most places? Really.
Exactly none of the properties I just mentioned apply to any form of intellectual property. You can duplicate bits infinitely at negligible costs.
> Software developers who don't own their code earn 800 dollars on patreon will Microsoft makes millions of their labor.
Microsoft makes a zillion dollars off of free software labor because those developers were tricked into giving up their rights while simultaneously allowing Microsoft to retain theirs. It makes no sense to give away code unless everyone does. I have three answers to this:
Abolish copyright straight up. Now everyone's on equal footing with Microsoft. No longer a crime to use their proprietary code when it inevitably leaks and even make money off of it.
License everything under AGPLv3. They can't exploit your free labor if they refuse to touch your free software code. This is the approach I ended up choosing because I don't have the power to enact the previous solution. I don't think I'll ever license anything under a "permissive" license ever again.
Simply don't publish the code. Nothing wrong with keeping it private, for one's own use only. I do this too.
Are mathematical theorems property? No, because someone who proved the theorem "owns" it just as much as someone who has studied the proof and learned it by heart. And what about independently discovered theorems? Who owns the idea in that case? Curiously when we talk about science where there is no money, the knowledge is ownerless, but when we talk about engineering where there is money to be made, suddenly inventions are property.
The moment "your" idea travels from your brain to my brain it becomes "my" idea too. Just like a digital file that gets copied from one computer to another is the exact same file, bit for bit. It's senseless to talk about "original" in that case.
>Also the presentation of something is an idea just like the something itself, so
This is exactly what copyright is. Copyright is based on the argument that the presentation of an idea is a unique correspondence to the idea, and constitutes a tangible item to which property value is attached. It's not intangible and ephemeral (an idea) or concrete and unremarkable (a piece of paper that the idea will be written on).
>Curiously when we talk about science where there is no money, the knowledge is ownerless
USDA grant disbursements for ag science alone would make your eyes bleed.
We need more countries to be strong and brave and say no to westerners forcing them to be unable to copy technologies... like China has done to some extent... but that might take too long.
The path where AI lawyers can sort IP disputes instantly and hence competitors can easily find out how to copy technology without violating laws would be another nail in the coffin of this corrupt practice.
> Nobody is going to "generate training data" if they get precisely nothing out of it.
The people who we don't want generating training data will not generate it... and that is a good thing. You have been generating too much crap in the name of innovation anyway. Real tangible useful innovation gets copied naturally in the marketplace in absence of corruption.
The idea that complete original "art" is needed to run image generators is a very temporary state of affairs that bootstraps the proof of concept, but is in no way necessary to the process. The entire process can be adversarially trained solely against human aesthetic values - we know this, because image generators are already a derivative technology of building image classifiers - i.e. they functionally are adversarial against a given textual description in a classifier.
The real limitation right now is speed: once we can diffuse at > 1 frame-per-second, data collection can be run solely from the consumer side.
How much did you get paid to write this comment?
In my case, the content I write involves hours of research, and correspondence with a network of experts I have built up over 5 years. I constantly update and revise that content.
The website on which it's hosted was also developed with great attention over the same period.
I don't think that I'd keep doing if I was deprived of revenue, and from the kind feedback of readers. I certainly couldn't do it full time.
We don't employ roomfuls of people to solve math equations by hand to calculate trajectories anymore, because the computer can do that faster.
If GPT can't do research and produce content with the same level of accuracy as you (which, so far, it can't) than surely people would be happy to continue compensating you to do so.
I tried ChatGPT. I asked it questions about the work I do, and it was about as correct as the cheapest copywriter. The writing was very good (if verbose), but the information was shallow but credible bullshit.
There is no efficient market that allocates money to the best products. In reality, the winner takes it all.
Therein lies the problem. If people find credible bullshit before valid information, they might stop there. It deprives them of valid information, and me of revenue. There is plenty of credible bullshit already. GPT just makes it cheaper to produce.
If people consume "credible bullshit" and can't tell the difference, without consequences, then your thoughtful work isn't as valuable as you thought.
If on the other hand people consume BS and it comes to bite them in the ass (through reduced socioeconomic status and, ultimately, reduced reproductive success), then competing approaches become viable.
We may not be around to experience it – a variation of "the market can stay irrational longer than you can stay solvent" – but you may find comfort in the fact that Nature has a way of optimizing away cruft. Most often through repurposing the obsolete-but-already-existing system for another function, and failing that, through atrophy and death.
Why is this type of dismissal so prevalent on HN? Do you really believe that?
In the real world, core infrastructure depends on code that does not raise enough money to buy groceries. Do you remember OpenSSL and Heartbleed? The entire internet paraphrases the same few nuggets of original research, with affiliate links slapped on top.
The market has been irrational for as long as I have been alive, and it's getting more irrational every year. When will the invisible hand of the free market correct things? Why do you insist that we live in a libertarian utopia where this statement holds any semblance of truth?
And just to be clear, people do write long, well-researched content for free all the time.
I'm all for holding Microsoft et al. to the same copyright laws they expect others to follow when it comes to their own works, but ultimately it would be in societies interest to not limit creators. Sure, maybe there are areas where we need to find ways to encourage creators but let's not do that by limiting everyone else.
Yes, I do.
In fact, I see that statement as nearly tautological.
I also believe that throwing people in jail if they don't value the same things as you is wrong.
Both of which are largely orthogonal to my original point though – which is that determination of value is a self-correcting process over evolutionary time spans. But I understand you wanted to shake your fist at the sky, in these difficult times.
Same with your OSS reference – I happen to publish several open source packages, used by thousands of companies around the world, which in total bring me about grocery-level income (~€500/month, pre-tax). So what?
If the "AI" community chose not to buy nicbou's work and chose to use materials in the Public Domain instead, that's fine. If the "AI" community chose not to buy nicbou's work and chose to use Radim's work instead under terms agreeable to Radim, that's fine.
But that's not what happens. What happens is the "AI" community chose not to buy nicbou's work and chose to use nicbou's work anyway. That's where the problem is.
Just like us, really.
The reward was mere internet points and the hardwired gratification that comes from social interaction.
People sometimes go to great lengths for non-monetary rewards.
Everyone that has a job and gets paid for it generates (with the appropriate monitoring) some kind of data that could be training data in the course of doing their job. (And doesn’t raise the issue of “if they get precisely nothing out of it”.)
Once no humans have jobs because AIs are better at them (should that occur), then the AIs themselves will be the best source of training data for the next generation.
There are many problems to address with AI; “Where is training data going to come from,” however, is nowhere on the list. It is a distraction from real issues.
It's using common crawl and wikipedia among other sources. The original purpose is something else.
Another anecdote that I find amusing is that Google Translate was trained on EU legal texts (among other sources), which are (generally) translated into 26/27 languages of the EU members. A huge, multinational bureaucracy, generated the texts by hand, but Google gets all the credit. I'm guessing that's part of the reason that Chinese isn't so accurate compared with other translations.
Lots of subtleties: fair use, what’s not fair use, etc?
It does seem like we’re in new territory, I can see merits on both sides and I guess we’ll figure it out via the courts / new laws over the next couple of years.
You can't just go "computer did it" and handwave away existing IP law.
Ok, now image that it isn't giving out verbatim results. Which is a lot of content.
Totally fine then right?
Unless the original was a bestseller and you are a bestseller, you almost certainly will not. This obviously happens all the time as most writers suck and do actually deliberately copy things and then self publish. AI is actually helping to now not do that.
Likewise, someone could go on Hacker News to start an argument about intellectual property and half of the responses would be nearly identical to comments on Slashdot twenty years ago.
No.
You can’t equate something you understand with something you don’t understand. Something you can create with something you can’t.
I'm sorry for you if you can't do better than an LLM.
On the otherhand, if the AI can be coxed into producing nearly complete getty image with only minor modifications, then maybe there would be grounds for a copyright lawsuit.
I think AI detractors should focus on passing new laws. Copyright is not the right tool for this.
"Before accessing your account, please solve this puzzle to verify you're not a robot."
And people post stuff on the public internet and get "nothing" out of it.
It's not obvious to me that everyone who has shared something permissively is unhappy if someone else uses it for training data. In many cases, it's just loud people who want a new revenue stream, which is the problem with intellectual property generally, the concept breeds entitlement. Why should I retroactively profit from what I willingly shared because somebody else found a use for it.
Now, let's say that on my morning walk, I see someone selling a bike I gave them a week ago.
Is it fair to be offended by this?
Re the actual bike thing (which has nothing to do with the discussion) I wouldn't be offended if the person sold it. In the hypothetical example, I got the enjoyment of fixing the bike, apparently gave it to someone no strings attached, and he's found a way that maximizes the utility he gets out of it. What's the problem?
The hypocrisy completely lost on them.
0.1% chance the Feds get involved and sue AI companies for criminal copyright infringement.
It has no problem reproducing song lyrics for me, which I’m pretty sure are absolutely copyright. Interestingly, it’ll start to reproduce a famous book for me, and then cut off in such a way as to suggest there’s a hard filter to stop it proceeding
They are slighting the law just like airbnb and uber did. They offer a better form of consumption. This is disruption at it's finest, and it will be in the courts forever.
This is trivially false. Eg, fanfiction exists.
Maybe monetizing art will become hard as cheap art becomes ubiquitous, but that doesn't mean people will stop creating stuff.
What? What do you call literally the entire training dataset of stable diffusion, dalle, gpt3, and gpt4? You think they compensated anyone that went into the dataset? Did you get paid for your comment, because guess what? Your comment will end up in the gpt4.1 or gpt5 model.
Training data will likely continue to include everything public on the internet. Not everyone will stop posting things online for free, therefore humans will still generate training data for models.
Sure, random comments of insignificant value like these will probably continue being created. Valuable content, though? That's a different story.
"This is why we can't have nice things." is a saying that accurately reflects what could happen if the "AI" community doesn't stop behaving like uncivilized barbarians. Personally, I'd prefer to have nice things.
My priors tell me that great art will continue to be made, both with and without these new tools; and that people will simply re-adjust to a new normal of sorts.
What are you suggesting exactly? That all the creators are so weak-willed that they simply stop creating? Or are you implying that they are going to all go live on some sort of "no-internet-allowed" island together?
Because it's not intelligent. It's definitely artificial, but it's not intelligent in any meaningful sense of the term. "Automatic Interpolation" or something similar would be a far more suitable use for the acronym.
>What are you suggesting exactly?
Noone is going to spend time making something if the next guy immediately goes "You made this? ... I made this." like clockwork. Even moreso if someone was making something professionally.
If we do see creators and inventors go away, it would be because of their own volition and desire to find greener pastures worthy of their time rather than "AI" usurping their jobs.
Yes they will.
Not everyone is like you who will collapse into anger and outrage at the thought of other people creating new things based on your work.
I do not care if anyone uses any of my intellectual property that I have ever created.
In fact, I would be flattered if people use the stuff that I have built to them make newer cooler things.
I support anyone and everyone "stealing" any intellectual property that I have every created.
Instead I am refuting your completely false statement of"Noone is going to spend time making something" that I was responding to.
So great, you agree with me that yes there are people like me who will continue to produce intellectual property and release it to the public.
> Particularly those who create things for a living.
I create things for a living to the tune of hundreds of thousands of dollars a year.
So yes, I speak, at least partily, for those people and am totally fine with people stealing my work.
If it's artists, I think you are underestimating the impact a "real" author has on a given work. We're all human, and art tends to be about capturing a shared experience. This means the viewer of the art can make assumptions about the author's humanity and (sometimes) connect with them emotionally in this regard. It's about communication _with people_ - not with an unfeeling language model that is very clearly trying to win a game of sorts.
So the artists incorporate these tools, some of them don't, new genres form, and new ideas are captured. All the artist has to do is assume they still have an audience (they will, because humans want to see art from other humans).
I think the biggest problem is that indeed, _other people_ will use these tools to trivialize the works of others and conceal the origin. But again, the artist can just ignore this world - as before. It's not like copyright law was actually ever protecting them anyways. Same argument applies to licensing on Github code; while you can get on a soapbox and complain about how your code is GPL licensed - that won't prevent hundreds of developers from copy and pasting it when they need a working solution.
The most pragmatic solution before, and now, was always "don't post things on the public internet for free if you don't want theft to occur". This is a lesson people keep having to learn since Napster.
In any case, if the end result is that artists do indeed start charging for their creations - then great! I don't really see that magically solving other issues (surplus of artists, shortage for the labor they would be good at), however.
They are; you can't use or incorporate an artist's products in public without their permission (aka a license).
>In any case, if the end result is that artists do indeed start charging for their creations - then great!
That's the thing: Artists are charging and the "AI" community doesn't give a shit.
Sooner or later, that house of "AI" cards is going to come crashing down with noone to blame but the "AI" community themselves.
>while you can get on a soapbox and complain about how your code is GPL licensed - that won't prevent hundreds of developers from copy and pasting it when they need a working solution.
Exactly. Sooner or later, someone who's been maintaining that one single block holding up all of modern humanity is going to say "fuck this and fuck you all" and leave everything to come crashing down; no, we can't blame him either.
We've already seen creators disgruntled with insufficient compensation (putting aside whether their demands were just) taking their wares and leaving for greener pastures to the dismay of everyone else. Didn't need "AI" for that, either.
I think we are mostly in agreement actually. Just that you are indicating this will amount to the majority of creators and I and the sibling commenter are seeing the world as it is, not as it should be. Realistically? I mean come on - think about it. Hustlers will hustle, sure. But artistic expression and creation doesn't _only_ happen because of money - it happens because some people would rather die than do anything else.
This is what I'm talking about, and what the sibling comment is talking about. I put all of my code on GitHub. These are my creations. If I wanted to be compensated for them, I wouldn't have posted it online. I use permissive licenses and don't need any credit for my work. This is humility - there is nothing special about my work per se. I mean, sure, maybe there is - but it isn't anything that I am the only in the world to pull off. Further, by putting the code online and giving it away - you are democratizing a creation that shouldn't really by _owned_ by anyone, in my opinion.
I gather you don't share that view - but you should know there are a lot of us out there. It's called having passion; and I don't know of any other way to accomplish things at work or for free.
Releasing something free-as-in-libre as a specific concept is irrelevant; I would publish certain things free-as-in-libre, others free-as-in-beer but with certain rights reserved, others commercially with certain rights reserved, others with all rights reserved. It depends on what it is I'm doing.
Insisting on releasing everything free-as-in-libre just because it's libre is, as far as I'm concerned, nonsensical. Licensing is a tool to be used appropriately, like any other tool; it's a means to an end and not the end itself.
>It's called having passion
I argue there is nothing as demotivating as having your passion denied by others shamelessly taking what you made and doing whatever with it without a single care for you, the creator.
Abiding by licenses is a matter of respecting the fact that someone, somewhere took their time to make something. The least you could do is respect their wishes, be they libre or closed, free or commercial, or anywhere in-between.
I'm willing to bet that a majority of the artists lambasting "AI" stealing and abusing their creations for training materials would be happy to provide training materials if they were respected and compensated for their time instead of getting treated like free real estate.
How is that "having your passion denied"? You can just ignore it, and then you're in the same position you would be in if they hadn't done it, which seems to be the alternative you have in mind anyway.
> I mean, sure, maybe there is - but it isn't anything that I am the only in the world to pull off. Further, by putting the code online and giving it away - you are democratizing a creation that shouldn't really by _owned_ by anyone, in my opinion.
That just sounds like slavery with extra steps.
Also, Art doesn't just come out of a LLM -- it depends on the times and conditions of the artist. "AI" models suffocate all of it.