#idk how sophisticated this algorithm is
Explore tagged Tumblr posts
Text
'Saying things in a funny way' is an op skillset for real though.
Latest example 'when you're late coming home and I don't know why I get worried about you and think perhaps there has been a car accident and this is why I send check-in texts' = distressing concept, explains the texting as something other than controlling behavior but still sounds neurotic and may increase any aversion felt toward it
'When I don't know why you aren't home yet I start to think maybe you got eaten by a car' = funny, sympathetic, likely to make such texts less burdensome to receive.
Why is communication like this.
#hoc est meum#car accidents#communication#communication skills#why is tumblr's first idea when i start a tag with 'commu'#bd/sm community#i know they stopped autofilling from our own past tags to facilitate greater homogeneity across blogs in hopes of making tags more communal#but is that really the most-used 'community' tag on tumblr#or just like#most used on blogs i interact with?#idk how sophisticated this algorithm is#i know i have not typed that tag before because i did not know they used that forward slash#commedy#comedy
470 notes
·
View notes
Text
The idea that amity is causing an economic collapse for the rest of the nation is hilarious and i love it, but idk how likely it would be. I think you're underestimating just how big america is, and how fucked its economy already is. (Though perhaps things are run better in a world where being a corrupt politician/elite comes with the risk of being investigated by the justice league, whistle blown by lois, and then hospitalized by batman just for good measure.)
Im autistic and have managed to keep a credit score of zero, so i dont know a lot about credit cards, but my understanding is that credit cards here have a limit thats determined by your credit score, and if you go into the bank to get a credit loan (slightly different from a credit card, but usually called the same), they also look at your income and assets. As far as im aware, like debit cards, if the card goes over the limit, subsequent transactions fail (and you get a big fee). Or maybe they dont fail because debt slavery is a concept that lots of americans don't understand, and credit companies ruthlessly exploit? Idk.
Now, i dont know if this happens in other countries, but here we have an epidemic of credit companies mailing out filled forms for extremely high interest rate credit cards. Entire lawsuits have been had by people being threatened by creditors for debts in their name they had no knowledge of because mail thieves got the forms, and literally all they had to do was forge a signature for free money.
This whole system is automated. So, while individual cards have a limit, amity parkers could theoretically get an endless supply of cards. Though i imagine the amount of junk mail they get will taper and fall over time as automated systems are increasingly updated and amity's addresses are corrupted, lost, and forgotten. That transition to move everything to the APDC address probably caused a huge decrease as well. Its also possible that the computer program running the mail-scam credit cards is sophisticated enough that while people have forgotten amity, the algorithms still know the location and is tracking transactions, seeing that payments arent being made, so less cards are sent as financial risk (for the company) increases. Again, Idk.
In this situation, anything Parkers buy IS getting paid for - the credit companies are the ones losing the money because they aren't being paid by the card-holder. This WILL start an investigation, and with how crazy the dc world is, i wouldn't be surprised if magic-based private investigators are a legitimate niche field for particularly mind-whammy cases. They would recognize and counteract the perception filter, at least for themselves, and figure out what was going on. Then, they would probably just advise the companies to take all Amity Park addresses off their systems and cancel all Amity Park cards. Still, that'll take time, so AP probably has a few months or years of free reign before being cut off. Perhaps its what allows them the money necessary to set themselves up as self-sufficient?
Why can credit companies get through but not the IRS? A number of reasons. First, my handwavy excuse that danny's reality rewrite was targeting the government, so the government gets hit the worst. Second, the responsibility of taxes largely falls on the person paying the taxes, which, again, out of sight, out of mind. Third, in the cartoon, the GIW wield the threat of auditing taxes for comedic effect. But if the entire town is declared non-sentient, why expect them to pay taxes?
Fourth, let's say a couple of magic investigators working for the IRS get through the filter. They harrass a couple of Parkers and the Parkers' response? The government doesn't recognize their town exists, has stopped paying for infrastructure upkeep, has cut off their banks, has forsaken their schools, and on top of All That, created an entire branch of the government to hunt them like animals because their people have been legally declared non-sentient. If they were to kill them and leave their dead bodies out in the open on main street for everyone to see, nobody from outside city limits would remember they even existed, let alone come looking for them. (The same warning is given to the credit companies' debt collectors, and the credit cards start getting canceled shortly after that.)
Last, if Amity Park is claimed as part of the Ghost King's domain, the US legally can't demand taxes from a foreign nation. (That would devolve into a huge argument over the sovereignty of the ghost zone and their right to claim AP, but it could only be started if enough people on the outside can remember long enough to actually contest the claim, and how much influence giw has when they discover that their sworn enemy has stolen an entire city from their nation - and even made them forget.)
Regardless of whether the credit companies deliberately cut them off or the cards naturally expire (and they stop getting new ones), afterwards, Amity Parkers have to go out of town to get loans and credit, which is a hassle when large swathes of their financial history cant be found as Amity is increasingly sectioned-off from the rest of the world. It's just another reason to remain in Amity and become self-sufficient.
However, i like the idea of various perception-filter-immune individuals (magic users, supernatural creatures, god-blessed, maybe a few metas and psychics, etc) stumbling into Amity and all its bat guano crazy. Amity Park's forged document industry absolutely explodes in growth. It's an in-broad-daylight market with competing businesses and legal regulations and protections because it's seen in Amity Park as a legitimate and necessary commodity. All these people living on the edges of society to hide their strangeness now have a way to participate without drawing attention. The black market for forged documents nationwide is collapsing as those immune to the filter introduce those that aren't, and nobody in the business can quite puzzle out why.
OTOH, crime might increase as fugitives running from the law seek refuge in Amity Park's perception filter. Sure, some of them are running from discrimination, but most will inevitably deserve their fugitive status and start causing problems - right up until phantom puts his foot down and proves exactly *why* he's top dog. Amity Park would probably treat the situation as not that different from when the portal first opened and most of the new visitors were trouble makers until phantom established the status quo.
Onto the CAT - it was absolutely a parody of standardized testing, and the idea that lancer would just have the answersheet on his on his person at all times is laughable. I supposed one could argue that clockwork orchestrated a series of events to make it happen, but it's definitely not realistic.
The two main standardized tests in the united states are the SAT and ACT, which are designed for college admission. Many states also have standard tests for their schools to benchmark progress at certain grades, which is what the CAT probably was, given that danny is a high school freshman. Here's a link:
https://educationadvanced.com/resources/blog/list-of-standardized-tests-by-state/
My understanding is that the tests, once complete, are mailed to the committee that created the test for grading, and the summary of results are mailed back to the school, but in my 5 minute quick dive, i couldnt find anything to verify that, and its very likely that different states handle their tests differently. I also discovered that this is the first year that the SATs were taken digitally instead of pen and paper, so that's another level of automation that parkers can exploit.
As for birth certificates and missing persons, i think most of that problem could be handled? Again, Amity Parkers working in neighboring hospitals know that the patients are going home safely and getting regular checkups. Idk what all kind of details go into hospital records outside of legal documents and history of appointments (for billing and health records.) Would hospitals actually record that such and such baby was assigned to such and such doctor for check ups? Idk if that kind of record would exist in order to collect statistics. If it does, then the Amity secretary is the one writing up the records and filing them correctly. Perhaps 'correctly' includes the information that the newborn was moved to Amity General Hospital, Go Look There For More Information. (After all, all Parkers are liminal, and AGH specializes in liminality.)
At the very least, a record of the baby's existence is available, and all the neighboring hospitals would have at least one Amity secretary to handle Amity related appointments, records, billing, etc. Hospital administrators are aware that a perception filter exists that complicates treatment of some of their patients, from... where was it? Oh yeah, Amity Park. Local hires are necessary to keep things running smoothly. All questions regarding the Parkers get directed to That One Secretary. They can answer your questions.
Working with Amity Park pretty often means it doesnt quite fall into that 'out of sight, out of mind' category, (a lot of people in neighboring cities are gonna have family in AP after all, and it was never Danny's intent to make any families forget each other) and the staff have gotten good at recognizing when the filter is screwing with them and know to seek out That One Secretary to help straighten things out.
To be honest, i was actually imagining that the babies and such are actually being born in AGH, but the hospital calls one of the neighboring hospitals to essentially outsource the creation of the birth certificate and ssin, providing the pertinant info. The calls and subsequent record creation are like 90% of the work that the amity secretary does at that hospital (any records, not just birth). The parkers that go to neighboring hospitals only do so because they need that particular hospital's specialization for treatment, and their liminality and the effects it might have for that particular treatment are encoded and treated as meta powers. Honestly, this set-up might only last until enough ghost doctors are hired and AGH is expanded enough to handle literally anything, as an increasingly liminal population means that in more and more cases, the liminality of the patient effects treatment in a way that the other hospitals cant account for.
In other words, looking at the hospital records would show a few things:
AGH stopped handling all childbirth, all amity parkers converted to at-home delivery, and unfailingly send the forms for birth certificate requests to neighboring hospitals instead of APG. (Babies are actually born in AGH. Looking up the doctor responsible for their delivery shows that they're employed at APG. I think labeling as at-home delivery would be easier than porting and forging records that they were actually at the neighboring hospital.) This is Really Weird, but anyone who is able to investigate this far already had to overcome the perception filter, and a few questions asked is enough to figure out that the situation screws with legal processes. If they can't get past the filter, they just see an increase in birth certificate/ssin creation in surrounding hospital records, and the statistical anomoly disappears if you discount Amity Park address, so its clearly centered... what were we investigating?
Amity park must have had a recent mutigen event (like dakota city, or central city in continuities where the collider activated everyone's powers, etc) because nearly every amity parker is now listed as a meta in the neighboring hospital records. If they get past the filter, they'll find the AGH records use the term liminal instead of meta. If not, something clearly changed in the greater illinois region, but nobody can really wrap their head around what could have caused the statistically higher percentage of metas from illinois.
After a few years, it becomes apparent that Amity Park's death rate suddenly dropped way lower than neighboring cities (according to the number of cases taken on by the neighboring hospitals). It isn't difficult to connect this to the suddenly meta population, but it's gonna garner interest, much of which will NOT be good. A mutigen that causes extended life/immortality is gonna make a lot of people go crazy for answers. (Maybe AP life expectancy has actually increased, but largely a lot of ghosts are just choosing to not legally report their own death.) Since Amity is under the perception filter, whoever finds out about it is able to harrass Amity without any other outsiders knowing, BUT they themselves have to have worked past the filter as well. It's definitely a double-edged sword.
There is a slow decline of Amity Park patients in surrounding hospitals (as AGH can accommodate more situations and liminality becomes the most important specialization in more cases), even as AP birth rates continue to exceed (or match, if the investigor can see than there are no birth records in AGH) surrounding towns. The ratio of AP birth certificates, death certificates, and other cases in the neighboring hospitals is wildly different compared to any other towns they serve. Birth certificates are way up, death certificates are way down, and the number of patients is declining despite apparent explosive growth in Amity.
If they can see AGH records, there are zero births (but a lot of patients refered within minutes of birth), exceptionally low deaths, and nearly all cases are associating with complications of liminality, which entirely replaces the term metahuman. Statistics for injuries (especially from car accidents), health defects, complications of health decline, cancer, etc are all suspiciously low, as liminality increases durability and decreases recovery time and seriousness of the injury (higher threshold of damage before checking into the hospital means overall fewer cases.)
A new category of health record has also been created: legally alive but biologically dead - ghost health needs are different from a human's after all. Parkers are encouraged to schedule a doctor appointment as soon as possible after they die so they can update their records to reflect their changed needs.
Word gets out, slowly but surely, that an american city hidden from the world exists under a perception filter. Knowing its name makes you forget about it. Most people treat it like an urban myth, but the supernatural populations are suddenly immigrating in large numbers to live in such an accomodating and friendly city. It freaks out the magic users a bit at first, and I'd imagine it only takes a couple of years at most until JLD hears about it, but idk if they would tell the Justice League or not. If Constantine is the first to discover Amity, would he even share with the others at all or keep it to himself?
Why doesn't the justice league know about Amity Park?
Okay so it's been a bit sonce I watched the show but one of the things in DpxDC is the anti-ecto acts, which I love, but correct me if I'm wrong, I THINK ??? they only show up in reality trip? SO: What if Danny, when using the gauntlet to undo everything, also got rid of the Anti-Ecto acts? but this is babys first time editing reality so he uh Fucks Up A Lil'. As a result when Danny used the reality gauntlet to wipe the AEA from existence he accidentally wiped Amity Park from perception. A big 'nothing matters over here' jedi mind trick, and now no ones looking at Amity. So, the Justice League actually WERE looking into and monitoring the situation in Amity, but when the perception filter closed them off, all of that suddenly went ignored.
This is noticed when someone (Alfred, Dick, Tim, literally anyone) realises theres just. A BIG dusty pile of case files semi abandoned somewhere in the cave when going through a (time period)ly cave cleaning.
They put it down because it's Not Important.
They come back to finish the cleaning the next day and do the exact same thing, but there's nothing to actually distract them this time and it pings as weird. Because why would case files be not important? They are by definition important, because only things flagged as important go into case files.
They try to get someone else to read it, because as long as they don't read the information in the file, they don't put it down.
That person goes to read it, gets a line in and then says something like 'that isn't important' and goes to leave. Person A pushes it and person B ALSO catches on.
Que the Batfam trying to figure out hey, what the fuck actually?
Meanwhile, how is Amity fairing? Canon compliant everything's going alright? Or have knock on effects to No One Look Here started to show?
1K notes
·
View notes
Note
Idk much about music so idk how to word this but whats your favorite musical feature or aspect of working in music? Or a fun fact about music?
We don't give jazz scales enough credit. they make everything easier, but for some reason everyones so scared of them like "ooOOoOoo jAzz!! its sOphiSTACTED and COMPLEX whats even goin ON its so maSTERful and DIFFICULt-" no its not. how did it even come to that.
Like for real. the creation of jazz was on basically the exact opposite ideal. but i digress on that point because THE BEAUTY of Jazz Scales(TM) that I think deserves more appreciation is that As SOON as you know what scale works for a song. You can Bullshit Anything and people will compliment you and think you are more musically talented than you may actual perceive yourself as. OBSERVE
I will teach you to sound like you've done nothing but master the piano for sixteen years straight using the same trick that I used to manipulate and lie to people: Jazzy Megalovania
Step 1: Left Hand
or your right hand! Fuck it! I'm not the boss of you and you are guaranteed to sound fucking superb either way because that's just how jazz work alright. it's literally impossible to sound bad ok? ok. so yeah just play these notes in the number order shown here.
no note name knowing required, just look at the shapes on your piano and compare and contrast, you'll figure it out. But if you really want to know these notes are called D, C, B, and B flat.
So basically what you wanna do is play those notes repeating, constantly, in a steady beat. or a swing beat. or just something to keep yourself anchored to a tempo and time, capiche? down the line you can even play them as octaves or is a different order oOOoOooo yeah i know, real game changing stuff, anyhow
STEP 2: THE BULLSHIT
SEE THESE NOTES DOWN HERE?
JUST. FUCK EM UP. Do whatever. Doesn't even matter if you don't know the first thing about rhythm. play them one at a time, play them all at the same time. it will sound good, no matter what. Cause it's a jazz scale--probably. I don't actually know the name of this scale, or even if it is an actual jazz scale. It's probably a D...Minor...somethingsomething. listen just google bing fire duck go what a jazz scale is ok and the same formula works ok bECAUSE
WHO KNOWS! and who even cares, all i know, and all you need to know, is that the basic principal of a jazz scale is that you just do whatever the hell you want and it sounds great. thats the official definition written by BJ franklin himself when he invented plates.
also If you want this to sound like megalovania, just start by going from the right-most note to the left most (That's C, A, A flat, G, F, then D for anyone who's curious)
so then BOOM. combine those two steps and you don't even need to know how to play megalovania and it'll just sound like you've been analyzing it's composition from the day you could count to 2.5. Congrats. you're now a poet. an artist. submit jazz as your college or job application. do you not feel the aura of class and prodigal power rush through you? you can now say you are better than people because once you play this on the piano, who the fucks gonna replicate that? who's gonna hear shit like this and think "i can one-up that with a round of jingle bells" NO ONE. fuck everyone but you, and your jazz. you are the best now.
you cannot spell sophistication without siphoning your mediocrity into the algorithm of pre-researched jazz arrangements. or, most of the letters anyway.
notice how I have no rhythm and even played off of the designated scale a few times--such is the power o jazz scales. so do not feel intimidated cause there's a 600% chance you will sound even better than a half-asleep emotionally drained tumblr user than i am.
#lie to people about your skills all the time. that's kips wisdom of the day for ya#<3#AHA. I HAVE TRICKED YOU. INTO LEARNING LEGITIMATE MUSICAL THEORY#TAKE THAT VISUAL ART WEBSITE NOW YOU HAVE MUSICAL ART KNOWLEDGE. BITCH#i go slep now#ALSO IM JOKING KIND OF I AM ACTUAL A SKILLED AND TALENTED MUSICIAN AND SHOULD YOU BE IN NEED OF MY SERVICES I DO HAVE COMMISSIONS#I AM NOT PATHETIC. I PINKY PROMMY
54 notes
·
View notes
Text
The original poster has put out some more thoughts on image generation and its relationship to existing forms/modes/models of art and artists, and it's sparked further thoughts in me. Nothing that alters my original conclusion (that generative AI cannot be considered "art" as such because art requires a conscious mind making choices, rather than an algorithm with a theoretically repeatable output given identical input), but worth chewing on further.
One of their contentions is that AI image generators are "learning" in the same way that a human being learns: they take in vast amounts of information, distill it, and then create based on what they've absorbed. An artist's work (and, for that matter, the artist as an individual) has been shaped by all the art they have seen in the life, in addition to the socio-political, environmental, material, etc. forces that shape us all.
Which constitutes objection number one: a human being is infinitely more complex than a generative AI, at least at the moment. You could quibble about how much art a person can "absorb" versus an AI, and the AI would probably win by orders of magnitude, but an artist isn't just the art they've seen: it's their personal experiences, their existence as a social creature, their quirks of biology and chance. All of those things feed into their art (unless we're doing a particularly extreme version of "death of the author", I guess). James Joyce doesn't write Ulysses without having grown up in Dublin, Tim O'Brian doesn't write The Things They Carried without having been sent to fight in Vietnam, H.R. Giger doesn't design the xenomorph without having engaged in a particular sexual act (supposedly). An AI, by definition, doesn't have any of that; all they have is another person's art, and thus all it can do is remix the art of others without the underlying experience. This also neatly obviates the objection of "what about fanfic and fanart?" All art is, in a way, fanart or fanfic, as it's based on human experience, not just other works of art, and you don't get the Inferno without Dante being pissed at specific politicians.
(Quick sidenote: George Steiner, in his book Real Presences, argues that every work of art is a response to another work of art, consciously or not, and thus art criticism is largely unnecessary, as art as a whole is its own criticism. I think there's something to this, though Steiner was also a devout Catholic and argued that every piece of art was a reflection of God or something. Idk, religion isn't my bag. But still, it's a compelling notion.)
Objection number two: the poster made a separate post where they gave a remarkable amount of detail on how a program "thinks" versus how a person thinks, and how generative AI is our best attempt at making a purely mathematical system act as though it has intuition. Which pretty much gives the game away: it can't have intuition. It is, as they themselves described, purely mathematical. If you enter the same query, you will receive the same answer. If you provide the same inputs, you will get exactly the same output. There is no actual "choice" on the part of the AI, only statistics and quasi-random numbers.
(Another sidenote: this is why we shouldn't even call it "AI," as that bestows upon it a level of sophistication it doesn't deserve. Neal Stephenson's Diamond Age has people use the term "artificial heuristics" instead, as it's more accurate, and I've always liked that. If you can understand the work that went into one of these image generators, if the code is available for dissection and you can, in theory, chart the entire process of generating a single image in the form of calculations, in what way is that "intelligent" and not simply "a more powerful calculator"?)
So by the writer's own admission, a generative AI doesn't learn like a person, because it cannot. Ascribing to it the same process whereby a person develops an individual art style, saying that an AI using a quasi-random number seed to generate an image is "pretty much" the same as a human being learning how to draw, is just a rhetorical sleight-of-hand, a simplification that completely removes the necessary complexity.
I also saw some thoughts on how, if we are discounting the use of generative AI because a person was not involved in it, we are ignoring the programmers who made it. Which, again, elides the actual work of the AI, which is rooted solely in mathematics. The rules it uses to generate its works were designed by a person, to be sure, but the result isn't dependent on the individual programmer so much as the artist(s) whose works were scraped to produce the database. The AI is a tool like any other, and you wouldn't credit Old Holland for producing the paints you used in your watercolor landscape.
(Are there gray areas? Yeah, sure. Gary Gygax should probably be given some credit for his work on the system you use in your actual-play podcast. Scott McCloud should be acknowledged for designing the format of your 24-hour comic. Osamu Tezuka is credited on the cover of Pluto along with Naoki Urasawa. Some elements in some works are more prominent than others. But again, which elements and how prominent are the results of choices by the artist.)
Writing all this out also made me realize that I made a misstatement at least once in the earlier posts of this chain: I said that human-made art bears the marks of conscious choice on the part of the artist, and I think that's pretty obviously untrue. Any work of art is, at least in part, the result of unconscious choices as well; nobody is fully aware of the factors in their life that lead them to make the choices they make. (Source: Sigmund Freud.) This obviously holds true for art as well, so my partial definition needs some updating.
(One last interesting sidenote about intuition and the creative process: I remember reading a paper, years ago, about the works of Jackson Pollock, whose "drip paintings" look like completely random messes to the untrained eye, but which actually possess a remarkable fractal complexity. The really fascinating part is that exactly how complex the patterns in his paintings were changed over the course of his career, to the point where fractal analysis could be used to not only verify a legitimate Pollock but roughly date when he made it. In particular, there was one painting which Pollock disliked at the time he made it, but later apparently came around on it; fractal analysis showed that its level of complexity was abnormally high for the time when he made it, but comparable to his later works. I need to find that paper again.)
Anyway. This is more of me rambling to nobody in particular. Venting is a handy way of organizing my thoughts. If you've enjoyed this stream-of-consciousness single-player debate, why not purchase one of my fine books?
(Joking but not joking.)
Reading posts from someone I follow on why image generators are a legitimate form of art and their use should be normalized, and it's refreshing to encounter some pro-generative-AI arguments more nuanced than "I don't want to pay artists." I still think they're fundamentally wrong, but it's been extremely helpful, because it's forced me to examine my own thoughts on the matter and determine why, exactly, I believe what I do. Something something mind needs books like a sword needs a whetstone, idk, I didn't watch Thame of Groans.
(Of course I'm not going to mention the person by name or tag them or anything, and for that matter I'd appreciate it if people didn't reblog this. I'm not writing this to get into an argument; I'm mostly just consolidating my own thoughts into a semi-coherent form.)
There are a number of arguments that have been leveled against generative AI, some of which I find more persuasive than others. The energy usage and financial waste, for instance, are significant, but not relevant in the case of one person using a locally hosted image generator and not selling the results. The question of the quality of the work is similarly unimportant to the bigger question; AI-generated images are going to continue to get better-looking as the technology improves, and "good" versus "bad" art is impossible to define in a meaningful way for the purposes of this argument.
The copyright argument is much more important, particularly to me as a creator. Another user pointed out that copyright cases against LLMs (on behalf of creators whose works were included in the learning sets) could have a potentially deleterious effect on fair use and transformative artworks. I'm not hooked into the legal scholarship on this, so I can't respond to that point. I do think it's somewhat short-sighted of that user to say "AI is a tool like any other, its use can be good or evil, our true enemy is capitalism," and then turn around and attack copyright as some kind of uniquely evil legal technology, rather than a technology that can also be used for good (making sure artists are recognized and paid for their work) or evil (large corporations shutting down parodies). And yeah, the revolution would fix all this, but how long do we have to wait for that? And what can we do in the meantime?
Anyway, the one argument that made me genuinely examine my own beliefs was "what is art anyway, and can you define it in a way that does not disqualify large swaths of what is widely recognized as human creative work and also excludes generative AI?" Because that's the meat of all this—not whether image generators suck up too much energy (because it's not about the specifics of the technology, which will change and improve over time, just as new types of paint do not fundamentally alter the nature of painting) nor copyright (which is a whole other legal mess), but whether we can call this "art" at all. For that, you need a definition, and that's the sticking point.
The original poster named a couple common ways of defining art/not-art (the "smell test," i.e. I Know It When I See It, and the "quality test," i.e. Can You Hang It In A Museum, which are largely the same but from different perspectives), and points out that they and other definitions would exclude quite a lot of human endeavors that most people would describe as art (graffiti, calligraphy) as well as fields that are more difficult to define but could constitute art (mathematics, programming).
(They also ascribed to anyone who attempted to make such a definition the motivation of not just gatekeeping but unadulterated fascism, which is an argument I think holds no water and wins them no friends, but. Let's just leave the paranoia aside and concentrate on the argument itself.)
So what is art? How do we define it, and why do I fundamentally disagree that anything that comes out of an image generator can be considered "art"?
I don't think this is sufficient for a full definition, but after talking it over with friends, I think, in part, art requires a perspective, which is to say that it must be the result of individual human decisions about non-trivial components. Another way to state this would be that the artist (if indeed they are an artist) must be able to make conscious choices about the work that are beyond what is strictly necessary for its completion.
Should the background be blue or green? Would this sentence be improved by an adjective? How large of a flourish should this letter have? What if I carve the gargoyle's snarl more deeply? What color should the hair of my halfling rogue be? These choices are indicative of a product that would be widely recognized as belonging to the category of "art."
Obviously, there are still gray areas. Certain fields have both a creator and a performer; can we say that one is "more of" an artist than the other? What about commissioned works? What if the artist is creating something within a strict limit or form—for instance, the 14-line sonnet, or a novel without the letter 'e'? What about Duchamp's Fountain, or John Cage's "4'33""? What about works with a large number of creators, such as films or collaborative writings? What about works where there is a level of interactivity with the audience, such as video games or certain theater pieces? Those and other questions are certainly open to debate, and should be debated! But to my mind, they do not challenge the fundamental principle, which is that the artist is an artist because they exercise choice in the process of creation.
Thus, by my (admittedly partial and underdeveloped) definition, I don't regard AI-generated images as art. The algorithm does not choose in a meaningful way; it merely calculates the most statistically likely next word/pixel/frame/etc. based on the database and the prompt with which is has been provided. (If you want to claim that this constitutes a choice, please submit a 5,000-word essay on whether free will exists and how we could possibly know if it does.) The remixer samples a specific beat; the collage artist cuts a particular image out of a magazine; the parodist deliberately draws in a specific way. The computer computes. It uses a mathematical operation—which, by definition, is repeatable and will produce the exact same outputs, given the same inputs. (Yes, the results have elements of randomization. We all know that true randomness is impossible for a computer, so they produce quasi-random numbers using things like the system time and so forth. I don't want to split hairs on this specific point. You get what I'm gesturing to. Don't look at the finger, look at the moon.) A prompt limits the database to certain specific sets, which the algorithm assembles according to its internal logic. The input is disconnected from the outputs; anyone could input the same prompt and receive the same art. (Even The Policeman's Beard is Half Constructed required an editor.) Generative AI is no more "creating" a piece of visual art than turning a radio dial to a specific station is "composing" the music that plays. The purely mathematical nature of its generative process makes it no qualitatively different from assembling a Lego set according to the directions.
The first obvious challenge to my partial definition is to say that it just restates the premise and shifts the goalposts: art is something that must be made by a person, and thus cannot be made by a computer. Which is fair! This is a verbalization of a belief I've always held about art, and which caused me to immediately (instinctively, unthinkingly) reject the idea that an AI-generated image could be "art." That's how I got into this discursive mess! It's why my brain recoiled when I heard someone call these images "art"!
But it also helps me understand why I instinctively categorize other acts and works as either "art" or "not art." A photograph was taken by a person at a specific time and a specific place, its elements arranged and its moment chosen according to the photographer's visual logic; it is therefore art. A hamburger put together by an underpaid worker at McDonald's is not art; a recipe by a chef that combines existing ingredients in a new way or using a new method is; a meal created by a person who tweaked a recipe might be. (That one might actually run counter to current copyright law, I'm not sure.) A mechanism assembled by a worker on an assembly line, identical in every way to another mechanism made by a different worker, is not art, because there was no choice on the part of the worker. (Could it be art because the designer of the mechanism exercised choice? Depends on the nature of the mechanism and the industry! Venmo me $20 for a debate.) A dance choreographed to produce a specific visual effect is art; an exercise designed to stretch certain muscles in the most efficient and painless way is not art. And so forth.
AI-generated images are not art. (They are also not a medium, which I saw several other commentators claim; an image is an image, regardless of where it comes from. I'm already knee-deep in linguistic debate, let's not cloud the matter any further.) Generative AI is a tool, and there are and can be creative and ethical uses for it! But to claim that it is capable of making art is giving agency to a thing that cannot have it, and claiming that someone who writes "sexy anime girl" in a prompt field is an artist is to expand the meaning of that word to the point of nonsense.
More than one person has brought up Jorge Luis Borges' "The Library of Babel" when talking about the potential of AI-generated works. It's got some bearing on the question, sure, but I feel like the more apropos point of comparison is his story "Pierre Menard, Author of the Quixote." In that work, Pierre Menard is a friend of the author's who is attempting to become the author of Don Quixote—not in the sense that he is trying to plagiarize the work, or time-travel and replace Cervantes in history, but that he is trying to make himself into a version of himself that could have independently written Don Quixote. It's partly a critique of elements of literary criticism, in that Quixote would become a far more interesting book (according to the narrator) if it had been written by a 20th-century Frenchman rather than a 17th-century Spaniard (it was written some 28 years before Barthes' "The Death of the Author," for context). But in the context of the current argument of generative AI, and specifically to my fumbling attempts to defining what is and is not art, it's an illustrative example of what I think it all boils down to: any work of art is the work of an artist, who inevitably brings to the work perspective/knowledge/experience/an individual understanding of the world. Ascribing any such perspective to an algorithm is just fetishism. (And not the kind that generative AI is most often used for.)
Or, to put this way more succinctly and directly:
11 notes
·
View notes