__ __ _ _____ _ _ _
| \/ | ___| |_ __ _| ___(_) | |_ ___ _ __
| |\/| |/ _ \ __/ _` | |_ | | | __/ _ \ '__|
| | | | __/ || (_| | _| | | | || __/ |
|_| |_|\___|\__\__,_|_| |_|_|\__\___|_|
community weblog
Something that nobody ever asked for
This all breaks down the second you are placed in the driver's seat, because you do not actually know what you want. How could I make such a proclamation so confidently? I can't, but I will anyway: what you want most, more than anything else in the world, is stuff that you never realized you wanted. from Human art in a post-AI world should be strange [Owl Posting]
posted by chavenet on Jan 24, 2026 at 2:03 AM
---------------------------
Oh gosh, I really liked this even though I disagreed with one of its premises. I think his criticism of art-via-filter is spot on. Does Netflix have ten thousand movies? A hundred thousand? I can't know, it only shows me the same 20 or so, over and over, because it believes in its filter, and does not want to expose its catalog. Are there really more novels printed every year than I could read in a lifetime? Maybe? I can't find anything to read at the bookstore because everything is laid out on theme tables and all the books look the same, and I don't understand how to do physical searches for books anymore.
Filtration enforces genre, which enforces a certain amount of anonymization: The thing you create is about its category, not about you. That's a bad way to approach art, and it's not really any good for readers or listeners or viewers, although it's presented as offering them a firehose of exactly what they want.
But! To bring AI into this feels...incorrect. Not morally wrong, just counterfactual. We still feel no closer to the AI-written novel, the AI-directed movie, than we were five years ago. Billions of dollars have been spent, and there has been progress (I hesitate to use the word "progress" when it's to a goal I find kinda abhorrent, but you know what I mean), but the basic structure of generative AI still makes it a lousy tool for creation. How many times do you see a reddit post or blog and pick up on that AI feel, and kinda groan and scroll past? It's just not there yet, it's odd that it's not there yet, and while I don't want to reiterate my theories on why it won't get there at all, I think there's a plausible case to make that we are not ushering in the age of the AI auteur.
But no, living in corporate-algorithm world is bad enough for the artist! "Am I writing this thing well enough within genre boundaries that it's legible? Am I writing it weird enough that I am legible?"
posted by mittens at 3:33 AM
---------------------------
Humanity has been producing art for somewhere between 45,000 and 100,000 years
@least ~70k [nature, 21 January 2026:]The newly discovered Liang Metanduno hand stencils are the oldest archaeological evidence revealed so far for the presence of our species in Wallacea. Notably, the dated rock art presented here is the easternmost in Wallacea and is located along the northern route to Sahul, as is all of the dated Pleistocene cave art reported thus far from Island Southeast Asia. This strongly supports the view that the initial peopling of the Sahul landmass involved maritime journeys through northern Wallacea. It is also evident from our findings that the first modern humans to reach Sahul about 65 ka brought with them a sophisticated artistic culture, with implications for the likely age of early painted rock art sequences in northern Australia. Given the extreme antiquity of the Liang Metanduno stencil, it seems reasonable to anticipate that evidence for rock art production of a similar age will be present at other locations along the northern route to Sahul (previously)
> I really liked this even though I disagreed with one of its premises
substack (archived)
Why is the chimpanzee in the film?
posted by HearHere at 4:21 AM
---------------------------
not only will weirdness become more in demand, so will the sherpa-guides to the far reaches where weirdness may be found. i've seen little niche oddities turn into whole subcultures in my lifetime, & the craving for nonstandard flavors (while i remain convinced that mango-macadamia nut coffee is an invention of AI) will only increase. but right now such collectors & discoverers are still, in the main, amateurs. this will change. the great geniuses of aberrant taste await. they will be celebrated, not doomed to poverty & obscurity, only to be rediscovered after it's too late to help them.
in the meantime, let's try not to stop young artists from taking huge chances (which is the basic teaching of most "creative classes"). don't "kill your darlings". they belong to the future.
posted by graywyvern at 4:48 AM
---------------------------
The thing I love about this piece is that it's really wiling to swim in the space, rather than stay in the instant-reaction realm that is doomerism/boosterism.
One major aspect of audience experience I do think it overlooks is the desire to be part of an audience of more than one. That is, even if this is not often a conscious part of their choice of what to consume (I know how that word is despised, but how else do we say "watch or read or listen to"?), people want to be able to discuss what their friends are also consuming. This aspect of the popularity of sequels and spinoffs gets easily overlooked in favor of assuming it's just about personal familiarity and wariness of the personally unknown. It's nice to be on the same page: that's basically what cultures are by definition.
A lot of AI prognostication assumes that everyone would want to tailor their experience, type a prompt and get a personal film, if the result could be good enough: the boosters (and some doomers, the "yeah and people make bad choices for themselves" type) tend to assume it could be; the anti-boosters scoff that it couldn't, in a "we can always tell" sort of way. But I think it's very possible that one could get really interesting bespoke stuff (like, actually bespoke, because it really does know what you like) and yet prefer only whatever stuff a consensus has formed around, due to a genuine preference for consensus in itself.
All that said, it's equally possible that the human desire I'm talking about is in fact easily overruled. Maybe the norm will be to personally generate all of one's own art and hence be unable to fully discuss it without sharing the result. On the other hand, I've been considering it likely that non-customized AI art could be the future, in other words that things won't be all that different from now except that AI did much of the work behind the scenes instead of CGI.
In such a world, copyright will start to seem silly, because so much of what was "officially" made by this company or that person will be more accurately attributable to the sheer luck of stumbling upon that particular output first, assuming it wasn't a particularly creative prompt (it might have been, I'll grant that) but instead was "Okay, computer, give me the next big hit!" Realizing this is a bit weird because I know that one of the biggest current arguments against AI is that it runs roughshod over intellectual property. What would it be like if, having ostensibly done so, AI couldn't the reap any benefit — the economic rent, arguably? Having assumed it will be a challenge to suppress AI through regulation, are we instead headed to a future where that's going to be preferable — where we'll all want the Mouse to lose its ears even if that spites a lot of faces?
(I have genuinely mixed feelings about copyright as a friend to artists, but not in some half-dopey "piracy actually helps the little guys" way so much as "IP monopoly steers audiences away from the little guys toward the big ones" way (I try not to pirate in the same way I do tend to tip even as I oppose tipping-based economics). A world truly without copyright of any kind would be one with a playing field so leveled that the overall result is very hard to imagine, and whether it's better or worse is hard to say.)
posted by InTheYear2017 at 5:04 AM
---------------------------
mittens:it believes in its filter, and does not want to expose its catalog.
Oh yeah, this aspect of streaming drives me up the wall. They even dropped the ability to sort one's own watchlist, the thing I've curated for myself! It feels like in the executives' dream world there wouldn't even be a search function, just an endless autoplayed stream with like/dislike for feedback.
posted by InTheYear2017 at 5:13 AM
---------------------------
It's nice to be on the same page: that's basically what cultures are by definition.
I need to tell you how much I love this sentence.
posted by mittens at 5:24 AM
---------------------------
what you want most, more than anything else in the world, is stuff that you never realized you wanted
I've been saying this for probably most of my adult life, it's nice to be vindicated.
posted by JHarris at 7:24 AM
---------------------------
It's an interesting piece. But I don't think the way he focuses on the unique concepts in "Being John Malkovich" quite works. In a world where anyone can put any concept that flits through their head into an AI and get a reasonably finished art-like product out, you'd expect to see a whole lot more material where the appeal is a unique concept.
In a way, it's a lot harder for an AI world to produce something like "F1", where there is a near total absence of original ideas, but the appeal is an incredibly slick production including in-car footage and celebrity cameos.
But even with works like "Being John Malkovich" or "The Velocipastor", I think the appeal is less about the unique concepts, and more about the difficulty level of turning a unique concept into a watchable movie. In Malkovich it's the steady escalation of absurdities to greater and greater levels that makes the movie great. In Velocipastor it's the deadpan yet emotional delivery of lines like " Dinosaurs never existed, and even if they did, I don't transform into one!"
The AI "Shrimp Jesus" image series is a wonderfully original concept, but it's neither well executed or developed into anything layered.
I don't think the nightmare world where most art is AI because that's the cheapest to produce isn't one without cool concepts, or where you can't find the stuff. It's one of badly developed concepts, where ideas that are potentially good are executed badly or in a boringly predictable way.
posted by TheophileEscargot at 8:02 AM
---------------------------
There will always be an audience for dreck. There will always be an audience for truly creative work. AI might or might not steal the careers of everyone producing dreck. Most of the people making truly creative work already can't make a living from it.
AI will definitely put more people into poverty, but it might not impoverish our culture too much, at least to those who care.
posted by rikschell at 8:12 AM
---------------------------
There will always be an audience for dreck.
as a for instance, the more I hear people talk about everything that's horrible about AI generated music, the more I hear them describing Muzak, which if you dig into it has been with us almost since the beginning of recorded sound. Most people just don't care about what they're listening to, as long as it isn't too loud, or otherwise draws attention to itself.
Cool, daring, adventurous, challenging art has never been for everybody. Except every now and then, for reasons, something cool, daring, adventurous, challenging does crash on through the various filters and barricades ...
posted by philip-random at 10:13 AM
---------------------------
InTheYear2017: people want to be able to discuss what their friends are also consuming. This aspect of the popularity of sequels and spinoffs gets easily overlooked
Thank you for this. I'm reflexively turned off by franchise media, but had never really considered this angle--it kind of reconfigures how I'll think about such properties and their place among us moving forward.
also kind of abashed because I'm quite aware of how much the shared aspect of cultural production is crucial for me, but never got to that framing of big, serial IP on my own
posted by working_objects at 11:09 AM
---------------------------
Extending a little from the insightful distinction made by TheophileEscargot, I think much of the standard criticism of "franchise storytelling" is about execution when it thinks it's about ideas. That is, whenever they see a trailer for a prequel or spinoff, people bemoan Hollywood's lack of originality in its choice to make More of That... but I think the problem is more in the predictable lack of originality even beyond the premises, the boxes we expect in advance that those movies are going to check.
We don't normally regard the set of all WWII movies as belonging to a "shared cinematic universe" even though they kind of do, and not just for being based on the same history, but in sharing a vocabulary for how that history is presented. But one thing they don't entail is an expectation of the same cast of characters showing up time and again, except for the national leaders in the various conflicts. Of course, people do also roll their eyes at plenty of WWII movies, and again for the good reason of predictability, just the Oscar-bait kind instead of the "genre" kind. But few people ever say "come on, historical films are played out, we don't need any more of them", and I think one reason is the expected shape of the cliches, so to speak.
(This is something I wish the current architects behind the Big Franchises bore in mind more often; expanding a world shouldn't have to be seen as synonymous with catering to expectations, but if anything, the opposite, forcing us to fully regard elements and perspectives previously in complete shadow.)
Something I wonder about oral cultures is to what extent originality is/was ever valued in itself when it comes to storytelling. The broad impression I get is Usually Not, because the storyteller's main job is to faithfully channel the narrative that came down from the ancestors, and there's a consistent expectation across basically every storytelling tradition around the globe for familiarity. So it might be the case that we live in a weird sweet spot for the notion of the authentic self as the core engine of creativity, rather than unseen and unsung forces, and perhaps AI is going to partially revert us to the mean, to a world where culture is expected to transcend individuals more than the other way around. I'm not sure I'm looking forward to that or anything, but it's worth bearing all angles of this stuff in mind.
posted by InTheYear2017 at 12:20 PM
---------------------------
>the basic structure of generative AI still makes it a lousy tool for creation.
It's not a tool for creation, any more than a gumball machine is a tool you use to make a gumball. It's a dispenser. And it's being used to sell you stuff that was stolen from you.
posted by Sing Or Swim at 12:23 PM
---------------------------
I have mixed reactions here. First, he's absolutely right that art can be great precisely because it's surprising or weird. Being John Malkovich is an excellent example... how they heck did that get made? In books, my go-to examples is Gideon the Ninth— no one had previously thought of "lesbian necromancers in space", and since AI produces an amalgam of what everyone has previously produced, it's very unlikely that it could have written such a thing.
I'm not sure I agree with the corollary that all artists should aim at the surprising or weird. At the very least, surely not much will be surprising any more. Also, humans are not machines who have a "surprising or weird" dial— it's harder to do that kind of work than it looks, and the failure mode is incoherence or a lack of emotion.
There's times— a lot of times, I'm afraid— where we want art that hits our comfort zones, that works the way we expect it to. That doesn't mean we want slop, but you can do good work within a genre. As most artists know, limits can be inspiring— an entirely blank canvas is dispiriting.
I'm also dubious that AI will soon be able to provide satisfying genre works, especially in film. AI can do a lot more than anyone expected ten years ago, but there are a lot of skills involved in moviemaking. Do people really think that moviemaking is easier than customer service, which AI also can't do?
posted by zompist at 1:17 PM
---------------------------