[HN Gopher] Bohemians at the Gate?
___________________________________________________________________
Bohemians at the Gate?
Author : surprisetalk
Score : 39 points
Date : 2025-05-31 13:41 UTC (2 days ago)
(HTM) web link (inferencemagazine.substack.com)
(TXT) w3m dump (inferencemagazine.substack.com)
| philipwhiuk wrote:
| The analogy is reduced to the point of irrelevancy.
|
| OpenAI isn't trying to produce art. They are trying to replace
| human creation with artificial creation, but an artificial
| creation whose entire input was the human creation.
|
| Kawano was not trying to replace Mondrian. And Mondrian was dead.
|
| Personally I'm happy to restrict copyright expiry and AI input to
| the work of dead people. Right now copyright is too long and AI
| is happy to ignore entirely. Let's allow AI and humans to both
| reproduce the work of dead people but leave the living to their
| benefits of their creation so long as they are alive.
| bloak wrote:
| For a number of reasons, I don't like the duration of copyright
| depending on how long the author lives.
|
| * It's age discrimination against older people.
|
| * If the copyright ends up being owned by some corporation (or
| effectively owned: some corporation has exclusive rights),
| which is what usually happens for any valuable copyright, it
| seems weird that the duration should be based on the lifetime
| of the author.
|
| * It creates problems later when it's not obvious who the
| author was or when the author died.
|
| * It could easily be abused by authors adding their
| grandchildren as co-authors. (I'm a bit wary of mentioning this
| in case I give anyone ideas. It has happened already, but so
| far not very often, as far as I can tell.)
| TheOtherHobbes wrote:
| If you're going to make these arguments about IP, you can
| also make them about property rights in general.
|
| What is it about creative IP that makes it different to other
| kinds of family inheritance, or other kinds of rentier
| ownership?
|
| Why does the argument that "Ownership rights should expire as
| soon as possible for the common good" not apply to huge land
| and property holdings, some of which have been passed down
| for centuries, with far more obvious economic and political
| effects than - say - Tolkien's descendants not needing a day
| job?
| zajio1am wrote:
| > If you're going to make these arguments about IP, you can
| also make them about property rights in general.
|
| No, physical property are both excludable and rivalrous
| goods [1], for them ownership is a natural right.
|
| OTOH, IP are naturally non-rivalrous, and only semi-
| excludable through establishment of copyright law. IP
| ownership is just societal construct trying to fit market
| mechanisms to them.
|
| [1] https://en.wikipedia.org/wiki/Goods
| pixl97 wrote:
| Also it's important to point out there is another class
| called anti-rivalrous goods which may be important for
| the discussion.
|
| https://en.wikipedia.org/wiki/Anti-rival_good
|
| >An anti-rival good is one where the more people share
| it, the more utility each person receives. It is the
| opposite of a rival good. Examples include software and
| other information goods created through the process of
| commons-based peer production. The term was coined by
| economist Steven Weber.
| aesch wrote:
| It is unfortunate and confusing that copyright and patents
| are described as intellectual "property" because
| intellectual property is fundamentally different than other
| property rights. Property laws as originally formulated
| applied to rivalrous goods where consumption by one
| consumer prevents consumption by others. Copyright and
| patents apply to non-rivalrous goods.
|
| Rivalrous goods are relatively easier to protect and
| monetize without the assistance of the state. Non-rivalrous
| goods are pretty much impossible to protect/monetize
| without the assistance of a state.
| tough wrote:
| Thanks I never had learned non-rivalrous goods and its'a.
| great way to explain the IP != property
|
| you wouldn't steal a car, err i mean download, right!
| leereeves wrote:
| > If you're going to make these arguments about IP, you can
| also make them about property rights in general.
|
| In general, property rights don't end when someone dies,
| they're just transferred to someone else. People can even
| take advantage of the fact that property rights endure by
| selling the right to take possession after their death.
| roughly wrote:
| > OpenAI isn't trying to produce art. They are trying to
| replace human creation with artificial creation, but an
| artificial creation whose entire input was the human creation.
|
| And they're trying to build a business off it as well. There
| are a whole lot of things we generally excuse when they're
| noncommercial (morally, even if not strictly legally), but
| OpenAI is a business, not a hobby. Similarly, we cast a gentler
| eye towards someone working towards self-improvement or
| personal discovery - artists are known to duplicate the work of
| other artists to hone their techniques, and Kawano was copying
| Mondrian's style to understand the use of new technology for
| making art. OpenAI is doing none of that - the point of their
| system is to skip all the hard work of learning and self
| improvement.
| parpfish wrote:
| I think you could argue that openAI isn't in the business of
| making art -- it's their users that make AI art that are in
| the business of making art.
|
| There are plenty of reasons for a model to ingest the art
| that are not directly related to creating new derivatives of
| that art. If you want the model to be able to recognize
| famous paintings or discuss styles/genres, it needs to get
| those examples.
| BobaFloutist wrote:
| But it sure doesn't need to be able to generate them for
| anything other than generating them.
| zzzeek wrote:
| Sudden uptick in mysterious artist deaths still unexplained
| wslh wrote:
| > but leave the living to their benefits of their creation so
| long as they are alive.
|
| Or pay somehow in the future for the use of them?
| fellowniusmonk wrote:
| Should the people who created the data set that was required
| for training reap the rewards or be laid off permanently? That
| is the real world scenario that is already playing out and will
| continue to ramp. That's the world we are currently living in.
|
| If commercial generative Ai groups that design the algo that
| relies on that data are heavily reliant on open publishing and
| open source and depended on and iterated on those things to
| create what they have created do they deserve to capture all
| the products value?
|
| Some things are inevitable products of new mediums and
| regulatory environments (regulation is just cultural byproducts
| of a nations citizenry) wasn't wikipedia always going to exist?
| Wasn't Craigslist? Wasn't IMDB? Weren't social media networks?
|
| What amount of profits should a group who scrapes all their
| core data from other people be morally obligated to hoard for
| themselves?
|
| If a small number of people profit off this and replace the
| very people whose data they scraped so that those people lose
| their jobs and homes and any financial security they have and
| deaths of despair skyrocket...
|
| Does this seem morally good or just?
|
| What is the dividing line?
|
| If this concentration of wealth is so extreme it destabilizes
| the very culture and people that birthed it, do we just
| consider the family losing their homes to be this kind of
| hilarious ironic thing and fuck 'em the losers?
|
| I think UBI and Paid Re-Skilling needs to be talked about now,
| I think the mechanisms and evaluations for rolling out UBI and
| Paid Re-Skilling should be discussed now.
|
| It seems very likely thay paid re-skilling will only exist in
| the transition and isn't the final state of LLM much less AGI.
|
| Unless we do this in earnest we are effectively committing to
| cultural decohesion and chasing down the misfortunate
| "Redundant" with murderbots and culling the now irrelevant laid
| off to stop unrest.
| mistrial9 wrote:
| there is enough capital circulating now among certain groups such
| that there is no practical driver to change anything about the
| current economics.. "let them eat cake" basically? Appearing to
| look cultured while arguing for the effective economic demise of
| the artisan, at scale, is the hot take today here
| TheOtherHobbes wrote:
| It's not even a hot take. This is the straight neoliberal party
| line. There's some original research here, but no original
| thought - down to the debunked "Don't tax rich people or
| they'll leave" line, which even economists will tell you is
| objectively untrue.
|
| Ultimately it's about freedom and values - not the capitalist
| take on freedom, which is all about property rights and the
| "freedom" to consume and pick one of a number of performative
| lifestyles, some of which are labelled dissenting, but the
| _freedom of imagination_ to reinvent culture and technology in
| completely original ways.
|
| You might think the AI companies are doing that, but they're
| doing the opposite. They're automating privilege - reinforcing
| and amplifying the existing system of economic
| disenfranchisement, value enclosure, and diminishing
| accountability.
| riskable wrote:
| From a copyright perspective the only question that matters is
| this: Do we treat AI models like (Xerox) copiers or do we treat
| them like artists?
|
| If we treat it like a copier it's the end user that's responsible
| when they tell it to produce something that infringes on someone
| else's copyrighted work. No different than if someone walked up
| to a machine and copied an entire book.
|
| Furthermore, if the end user never even distributes the result of
| a prompt the question is moot anyway: Copyright only matters when
| something gets distributed. No distribution == No violation of
| copyright.
|
| If we treat AI like an artist it is the owner/creator of the AI
| model that's responsible when it produces something that violates
| another's copyright. Since it is literally impossible to maintain
| a database of all copyrighted works that exist (in order to check
| if something violates copyright or not) this option is untenable.
| It's not possible to implement unless we go back to requiring all
| copyrights be registered (and provide that database to anyone
| that asks--thus, distributing all those copyrighted works which
| would defeat the purpose).
|
| I very strongly believe that the courts will ultimately settle on
| treating AI like a copier. It's a tool/machine and should be
| treated as such by copyright law.
| altruios wrote:
| A good argument. however to compare an AI model to a Xerox
| machine is reductive and not a sound metaphor...
|
| It can not be treated just as a Xerox machine, but it can be
| treated as a Xerox machine that has within it all the
| copywritten works (that a user can inventively request
| combinations there within) which it has trained on (and saved
| in the form of weights/bias). In this case the AI model itself
| is the distribution of works under copyright.
| Encrypting/transforming copywritten works and transmitting it
| is a violation of copyright (afaik; ianal).
|
| This is all to say, copyright - as it stands - needs heavy
| reform. I'm rather copyleft. Because all of this is vestigial
| nonsense from an age where printers from the 1800's setting the
| rules, and our thinking hasn't updated yet.
| parpfish wrote:
| What would happen if you made a lossy image compression
| format derived from tons of scraped, copyrighted images.
|
| There's no generative ability, but anytime you
| compress/decompress your image the model uses weights and
| biases learned from copyrighted works.
|
| Is that a violation?
| pona-a wrote:
| We treat them as models. We allowed them to be fitted on
| copyrighted data, arguing the research is an inherent public
| good. But now that these companies are directly competing with
| that material's copyright holders, it makes sense to reevaluate
| that assumption.
|
| A good first step would be to mandate AI labs share their
| weights and methodology before commercial release or lose that
| privilege. This would spare universities and non-profits, while
| requiring commercial labs to contribute something back, be it
| in licensing fees or usable research.
| deadbabe wrote:
| Treat them as a search engine.
| ljlolel wrote:
| This is wrong on so many counts, you should not be giving legal
| judgments in comments. As one example, "no distribution == no
| violation of copyright" is incorrect.
| adamnemecek wrote:
| Here I thought it would be about denizens of Bohemia
| https://en.wikipedia.org/wiki/Bohemia
| samanbb wrote:
| I don't think existing copywright law matters with respect to AI.
| AI generated art is unprecedented and I'd argue there is a moral
| imperative to distribute the wealth generated by AI to those
| whose creative work enabled the technology even if we don't have
| a legal (or technical) framework for doing this in place yet.
| Hilift wrote:
| "The Baroness Kidron amendment would require companies to provide
| a log of all of the URLs their models were trained on, and keep
| this up-to-date every month."
|
| https://www.bbc.com/news/articles/c39xj284e14o
|
| MPs rejected that amendment and sent the Bill back to the Lords.
|
| "Kidron's amendment will not provide certainty of anything except
| for certainty of more uncertainty, of continuous regulations
| stacked one upon another in a pile of instruments...
|
| "Jumping the gun on one issue will hamstring us in reaching the
| best outcome on all the others and especially because this is a
| global issue and we cannot ringfence the UK away from the rest of
| the world."
|
| However, Kidron said her amendment "does not challenge the
| primacy of the Commons" and pushed ahead.
|
| The result was a decisive defeat for the Government, with 287
| votes in favour of Kidron's amendment and 118 against - a
| majority of 169 - and the Bill will now be sent back to the
| Commons.
| scarmig wrote:
| What does Mondrian's estate owe to Oliver Byrne's estate, since
| he basically stole Byrne's color scheme and style from his
| version of Euclid's Elements? (And what does Byrne's estate owe
| to Euclid's estate?)
|
| IP law is a purely practical matter: what most incentivizes
| intellectual production? If a culture errs too far toward
| restriction, that country ultimately falls behind. The US itself
| jumpstarted its industry by mass IP theft and industrial
| espionage to steal the hard work of British inventors and
| entrepreneurs; its entire publishing industry got its start by
| copying British works wholesale in what was then the world's most
| flagrant IP violation. As an angry sputtering Dickens wrote:
|
| > You take the uncompleted books of living authors, fresh from
| their hands, wet from the press, cut, hack and carve them... all
| this without permission, and against his will; and then, to crown
| the whole proceeding, publish in some mean pamphlet, an unmeaning
| farrago of garbled extracts from his work, to which your name as
| author, with the honourable distinction annexed, of having
| perpetrated a hundred other outrages of the same description.
|
| Americans mostly shrugged, or attacked him as a mercenary
| scoundrel, before proceeding to build the world's largest
| publishing industry.
|
| The US needs to make laws governing how copyright applies to
| generative AI; when it does, those laws should grant expansive
| fair use rights. Otherwise, the US will be left behind.
| parpfish wrote:
| Something I keep thinking about is that nobody would think twice
| if some young musician gave an interview that said "{older
| artist} was a huge inspiration. their songs were fundamental in
| shaping how i think about the artform", nobody would be up in
| arms claiming that they have stolen from the elder artist. it's
| art inspiring other artists the way it always has.
|
| so why does it _feel_ different when LLM ingests some music to
| "fundamentally shape how it 'thinks' of the artform"?
|
| i can create loads of speculation about what that difference is,
| but to be honest it all feels pretty post-hoc and hand-wavy.
| bell-cot wrote:
| > so why does it _feel_ different when LLM ingests ...
|
| Most-ish people are biased in favor of both "the little guy",
| and in favor of "artists" (whether authors, musicians,
| painters, or whatever).
| jazzyjackson wrote:
| Because an LLM is not an artist, an artist is working to
| improve themselves and putting something new into the world
| that aligns with some desire or vision they have.
|
| The LLM lacks all desire and vision, it is a machine that can
| produce a trillion variations of something that sounds like
| Miles Davis, for example, without an ounce of intention to what
| it wants to do, what art it wants to make.
|
| Fundamentally the LLM is not a living thing that interacts with
| other living things and has an idea of how it wants to
| influence itself and others. it is just a printing press
| capable of flooding the marketplace with lookalikes and slop.
|
| [Despite my tone, I appreciate the prompt, it's a subject I'd
| like to write on at more length than an HN comment]
| Pet_Ant wrote:
| Because it can produce millions of variations, the value add
| is in separating the wheat from the chaff. The art is in the
| selection. That has intention. If the first result is bad,
| respin.
| kelseyfrog wrote:
| Copyright exists to formulate art[1] as capital.
|
| Let's take a step back and recognize this for what it is -
| juridification - the replacement of social relations with legally
| enforced economic relations. It assumes, incorrectly, that
| everything made by people is subject to exchange. It's not, and
| for the most part, that's not how artists themselves conceive of
| art.
|
| The artist community mostly operates on attribution. That is,
| influences are appropriately stated and attribution is the social
| currency. Unattributed works violate social norms. The problem is
| when these social relations are codified as economic relations.
| It's weird, like when Milton Friedman argued that children owe
| their parents for raising them[3].
|
| As much as some people believe that this would imply the creation
| of rights, it misses the point. Freedom to choose does not
| include freedom from having to choose. The infinite variety of
| rich human relations are flattened into one dimension - economic.
| That erases what's most important to the human experience.
|
| 1. Insomuch as art is literary, musical, or artistic work
|
| 2.
| https://www.sv.uio.no/arena/english/research/publications/ar...
|
| 3. https://periferiaactiva.wordpress.com/wp-
| content/uploads/201...
| AlexandrB wrote:
| You're not wrong, but without copyright how do artists eat?
| While artists are happy to operate by attribution, commercial
| entities don't give a shit about such things and would be happy
| to use or distribute popular works copyright-free without
| attributing anything. For example, Spotify is a terrible deal
| for artists, but in a copyright-free world Spotify would just
| pay nothing to artists - an even worse deal.
| zzzeek wrote:
| I'm not ripping off your song / photo / artwork. No, I'm
| _democratizing_ it. You should thank me !
| CuriouslyC wrote:
| I can make a collage from art/music/writing/etc under fair use,
| and it's protected by copyright. What AI does is less direct
| than that, and most of the people who use AI for creative
| purposes are themselves using AI output to create a new
| collage, or using it like a filter to tweak the style of an
| input or make it more aesthetically consistent.
|
| If you have a problem with people vomiting unedited low-effort
| AI slop all over the internet, call that out and we can have a
| conversation, but your stance against making tools that can
| enable a new generation of creators to focus on the things that
| really matter (aesthetics, composition, structure) rather than
| things that don't so much (brush technique, instrument skill,
| word repertoire) smacks of elitism.
| zzzeek wrote:
| I like bgwalter's post above. I'd like to "democratize"
| Facebook. Can I download the entire set of profiles and
| photos and make my own mashups? Let's see what the facebook
| lawyers have to say over that.
|
| There is a social good for artists making derived
| art/music/writing under fair use. There is only social harm
| in letting billionaire-owned AI's spit out trillions of
| garbage derivations with zero social value. We can make a
| choice as a society.
| CuriouslyC wrote:
| If they're public, it'd be fair use, and I guess they could
| stop you by not letting you download them in the first
| place, but if they're out there it's fair game. WRT the
| original point you were trying to make, I agree that using
| the pile/libgen is questionable, but so are publishers
| trying to gouge companies who use the materials for
| training by charging them a higher price than they'd charge
| people with the intention of reading the book.
|
| If your problem is billionaires, that's separate from AI
| (which can be open source and run on commodity systems),
| and we can also have a conversation there. Shitting on AI
| because some unethical actors are pursuing it is like
| shitting on top hats and monocles because rich people wear
| them though.
| bgwalter wrote:
| For some reason the many obsequious software engineers never
| suggest democratizing Zuckerberg's island or nationalizing
| Google.
|
| It is always the output of individuals who cannot fight back
| that is "democratized".
| bgwalter wrote:
| Someone is now going through all my old comments and
| downvotes.
| aspenmayer wrote:
| Minor consolation is that I think you can only downvote
| recent comments. I think it's maybe 24 hours that it's
| possible but not sure.
| aspenmayer wrote:
| More info about downvoting and other features of HN
| collected by HN user minimaxir here:
|
| https://github.com/minimaxir/hacker-news-undocumented
| niemandhier wrote:
| I live in walking distance from the place the brothers Grimm
| sourced their version of Snow White.
|
| Ai image generator frequently refuse to create illustrations
| featuring the character, everybody is afraid of Disney.
|
| Similar, Disney's Winnie the Puh just looks like Magarete Steiffs
| plush bear with a red shirt.
|
| Very often those who claim to have created an original work
| themselves just produced derivatives, at least those should not
| be protected to the detriment of humankind.
| 2snakes wrote:
| Pooh. Winnie the Pooh. <3
| lioeters wrote:
| It's possible the parent comment was thinking of the Soviet
| version, Vinni-Pukh, Vinni Pukh. The drawing style is
| different than Disney's, but also really cute.
| kens wrote:
| On the topic of computer-generated Mondrians, in 1966 a
| researcher at Bell Labs used a computer to produce a semi-random
| version of Mondrian's "Composition With Lines" (1917).
| Interestingly, he found that only 28% of people could correctly
| identify the computer-generated picture and 59% _preferred_ the
| computer-generated picture.
|
| The researcher points out that "Both patterns were conceived by
| humans, although certain features of the computer-generated
| picture were decided by a programmed random algorithm. The
| computer functioned only as a medium performing its operations
| under the complete control of the computer program written by the
| programmer-artist."
|
| The paper with images:
| http://noll.uscannenberg.org/Art%20Papers/Mondrian.pdf
| Pet_Ant wrote:
| The visual arts are now having their Ikea moment.
|
| Ikea didn't kill woodworking/carpentry/joinery.
|
| I know several people that have invested into a lot of time and
| *a lot* of money into them as a hobby. It's still an art.
|
| But Ikea did reduce it as a commercially viable avenue profession
| for more than a handful of people making boutique products.
|
| But the DJ killed the wedding band, and the iPod/Spotify killed
| the DJ, if we weren't gonna a shed a tear for them (and maybe we
| should have?), why now?
|
| People will still use art for their personal expression. That's
| older than recorded history, it's just industrialised commercial-
| grade investable art that is at risk.
| Muromec wrote:
| >People will still use art for their personal expression.
| That's older than recorded history, it's just industrialised
| commercial-grade investable art that is at risk.
|
| I think the problem here, from the social petspective, is --
| how many people can enjoy creating art and what kind of a
| society we will have, where only a few can earn a living by
| doing so.
|
| Or from another angle -- how long we can create more ways to
| create art to outrun the oncoming reign of night that strips
| away all of the fun thing we had before.
| Pet_Ant wrote:
| > from the social perspective, is -- how many people can
| enjoy creating art and what kind of a society we will have,
| where only a few can earn a living by doing so.
|
| 1) Only a few people can make a living out of doing it today
|
| 2) It's already the case with woodworkers, knitters, and so
| many other arts. How many people can make a living making
| pottery?
|
| > how long we can create more ways to create art to outrun
| the oncoming reign of night that strips away all of the fun
| thing we had before.
|
| You can still create the art. When your girlfriend cheats on
| you and runs away with your best friend _and_ your dog, your
| still gonna want to write your own country song. Not because
| it 'll sell, but because you want to tell your own story in
| your words.
| bgwalter wrote:
| The article is a bit lengthy, which makes it hard to distill any
| message. Perhaps the vagueness is on purpose, since it ends with
| the classic "But what about China?" FOMO.
|
| Clearly "AI" is plagiarism. There are YouTube voice translations
| now that are trained on the original speaker's voice. How is this
| not a derivative work?
|
| If you train an AI on 1,000,000 voices so it can pretend to
| "create" an average voice, that voice is still the derivative
| work of all input voices.
|
| The EU making "AI" scraping opt-out is disgraceful. It should be
| opt-in.
|
| Vance and Trump (mentioned in the piece) are irrelevant. Watch
| Vance in his latest meeting with von der Leyen and Meloni, where
| Vance suddenly was very friendly when von der Leyen mentioned a
| total trade volume of $1 trillion.
|
| Von der Leyen also called Meloni (who was previously designated
| as a "fascist") by her first name. They all were best friends
| forever.
| aspenmayer wrote:
| > Watch Vance in his latest meeting with von der Leyen and
| Meloni, where Vance suddenly was very friendly when von der
| Leyen mentioned a total trade volume of $1 trillion.
|
| I haven't seen this, do you have a clip?
| dhosek wrote:
| What doesn't seem to be addressed is the _name_ , Artificial
| Mondrian. The works may not be infringing on Mondrian's copyright
| (and I don't think they are or should be), but the use of
| Mondrian's name would almost certainly be infringing on
| Mondrian's moral rights, if not legal ones as it implies an
| endorsement by Mondrian or his estate (although if Mondrian
| didn't have a trademark on his name, which seems unlikely, he
| doesn't have legal recourse).
| energywut wrote:
| Scale _matters_. Reducing an example down to a single artist
| reproducing a single style of a single other artist isn 't theft,
| it's a fundamentally different thing. Kawano is not making a tool
| that can predict and reproduce all art produced by all artists
| with a reasonably high degree of success.
|
| If I use a single sample from a single record (say, drawing from
| the Amen Break) to build a small number of songs, that's
| fundamentally different from being able to wholesale lift the
| musical stylings of every artist across time.
|
| Remix culture is important, and copying small elements out by
| hand with intention is an important part of remix culture. But
| the scale _matters_. The industrialization _matters_.
|
| A kid taping the radio is different from Spotify selling every
| piece of music for a subscription and sending (almost) nothing on
| to the artists. Sure, both cut the artist out, but the scale
| _matters_.
| Pet_Ant wrote:
| > Scale matters.
|
| On what grounds? Very rarely does the law see things that way
| [1]. Where is the tipping point. What if it did Mondrian _and_
| Rothko? Still good? Okay, now add Damien Hirst. Also add
| Jackson Pollock. Then Roy Lichtenstein. What is your grounds
| for delineation? If it's just a feeling, then the problem isn't
| the law, it's you. Monet could just be a Photoshop filter,
| Pointillism is. Warhol?
|
| [1] Newton v. Diamond is an example I can think of
| https://www.quimbee.com/cases/newton-v-diamond
| kleinmatic wrote:
| I got about halfway through before giving up on this article. I
| thought it was just taking a long time to get to the point then I
| realized one wasn't coming.
|
| Thinking it was all a Duchamp-esque drollery, I checked to see if
| it was AI-generated and alas, it isn't.
___________________________________________________________________
(page generated 2025-06-02 23:01 UTC)