AI-generated art - name reasons why it is so bad and hated

as you may know Stereolab recently released a music video for a new song, and like a lot of new music videos, it is entirely AI slop.

I found the video unwatchable, though I assumed it was mostly because I just hate this stuff so much, I hate the implications of it, I think it's the antithesis of art and the first technology that I can call "evil" which has really been foisted upon us in an unavoidable way. to watch this stuff is to validate it. but there's a part of me that feels bad about it - is it just me rejecting technological progress? I mean I admit this stuff can be very impressive, is it really that different from CGI, auto-tune, sampling, things of that nature? why do I hate this so much to the point where I can barely look at it?

it got me thinking about the purpose of art in general, how it can be used to express complex feelings, things that can be felt but not said. and even if it fails at that there's at least the knowledge that a human being tried something. that to me is what it's all about; a person trying to express something, for better or worse. AI I think takes all that away and relegates the decision to a complex and impenetrable algorithm that can approximate the human experience but never quite capture it.

for example, yes, the people in AI generated videos can look good. through the stills alone you could be fooled into thinking these are real people. but when you watch them move, they move a bit unnaturally. they bend their elbows and knees in ways that regular people don't. their expressions are way too fluid. from one scene to the next, the people often look different, like it's a stunt double in every scene. even without actively thinking, my brain recognizes this stuff and finds it impossible to form any connection with it. even heavy CGI-d stuff, like say the end of Michael Jackson's "Black or White" video, doesn't provoke this reaction. you know the people are real. Tyra Banks actually made those faces in front of a camera at one point.

in a broader sense I've been kind of lamenting the death of pre-CGI cinema and TV, when there were puppets and monster suits and meticulously painted sets. yes, CGI looks "better" in that it much more accurately resembles a real city but as art I feel it's often worse, since your brain obviously knows they're not blowing up densely populated city blocks. when this stuff is done with practical effects, even hilariously cheap ones, you at least get a sense of how it was done, and even if it doesn't look remotely 'real' I feel like your brain still follows it more naturally than it does with CGI. like maybe movie studios think that "suspension of disbelief" works better the more "realistic" you make something look when the human brain doesn't quite work that way. I thought about this a lot when I took the kids to see the Mufasa movie, in which the characters really do look like actual animals, to the point where your brain sort of treats them as such. you don't really care about the characters at all, like you would in a much less technologically impressive cartoon.

I realize this is maybe two separate topics, but AI seems to be supercharging this, and I worry that it's going to ruin art as a whole for me if it keeps going unchecked like this. anyone else have any thoughts? y'all are probably better at expressing this than I am.

it's ok to have a particular taste, and there are different politics coded into one's aesthetic preferences, but i do think it's erroneous to think that there's something inherently immoral or wrong about the aesthetics of AI art (as opposed to the means of its production).

in that way, the comparison to sampling/electronic music or photography feels really apt to me: yes, it's alien or different; yes, you might prefer the cello to the synthesizer; but the validity of this artform will be founded on whatever great art gets produced by the people using it. (and it's too young to have had much of that bubble up yet.)

maybe it's a foolish (and even revolting?) act of faith, but i firmly believe there will be "good art" made using these tools in the next 20 years.

i prefer traditional SFX to CGI (generally) as well
and i don't generally like shitty MIDI music
but i recognize the way these are taste preferences, not something inherent

(again - speaking solely on aesthetics, not environmental impact, labour implications, etc!)

. I hate AI with a passion but I actually found the Stereolab video to be a best-case-scenario of its use, deliberately deploying its uncanny, off-putting quality to aesthetic effect. Still would have preferred a "real" video.

I think that the thing that I find most unappealing about AI is that there's like a meta-assumption behind its users and proponents that the audience doesn't care about craft or if the end result isn't as good as what could have been made deliberately or by hand. The "slop"-ness of it all.

I also think that more "traditional" forms of art are more rewarding to the artist and that AI artists cheat themselves of this. Endlessly refining prompts seems like such a boring and unrewarding way to make art.

that thing Brian Eno is always saying about how the limitations technology used to make art in a particular era is exactly what people a generation later will fetishize - like intentionally making stuff that looks like blurry vhs tapes in the 2010s etc - that's totally going to be the case with this era of AI in like 20 years, right? People in the future will be nostalgic for all the stuff that we currently hate about it.

maybe to appreciate current AI art you need to imagine yourself as someone born circa 2020 watching it in the 2040s

but maybe identifiably AI generated art could be become a signifier of 'authenticity' in the future, if something looks like it was done by a human then people might just assume it was done by a more modern, advanced AI?

https://www.instagram.com/purestnostalgia/

i posted this on the dystopian thread, it's some admittedly compellingly weird uncanny valley approximation of memory from decades past, it looks the part but then you'll examine the details and there's a city that looks like L.A. but it's not quite right, there are band posters on the way which are completely imagined, there are shows being watched on TVs which are mutated versions of a faint remembrance, there are few humans in these, mostly empty rooms and cityscapes, it's eerie stuff.

"i admire its purity"

Endlessly refining prompts seems like such a boring and unrewarding way to make art.

To an artist, sure, but to the many people without the talent, resources, or even just the inclination to be an "actual" artist, I can see the reward in being able to express a concept and have them turned into visuals.

I don't see prompt refining as all that different from the grunt work of art - darkroom work (or digital equivalents) is not particularly exciting creatively, there's a lot more "goddamit I didn't burn that long enough I just wasted more paper" or "I've been in this smelly little cubby for 8 hours trying to get this right" than feeling like you're Jackson Pollock flinging paint at a canvas while listening to jazz.

Some points

- The environmental impact is not well known enough, should be brought up every time, this is the scariest thing by far. I worry about why it isn't brought up more often.

- I hate it.

- There are some arguments against AI that I think are bad (that it takes no skill) or slightly dubious (focusing too much on how bad it looks). Some people are clearly much better at using it than others and that taken skill and taste. I think the look of AI is based on statistical averages and people who know how to use it better have to get deeper into specifics to drive it away from looking completely generic, I've seen prompts based on pretty advanced knowledge of art history and it gets better results. These people know who they're stealing from.
All that said, even when it's done well it often has that waxy homogenized look that I dislike. I seen a fake film trailer yesterday that was one of the most realistic things I've ever seen with it but I think there's still something about the look of it that leaves a bad taste in my mouth.
I've been surprised by who uses and likes it, people I respected, people who are very knowledgeable and skilled and I'm thinking "you've got to be fucking kidding me". If they can't see how bad it looks then I'm not sure how much I could say to change their minds. Why not get off the computer and build skills at something else?

- We should take any excuse to get as much time away from computers as we can.

- I have a generally bleak view of digital art, cgi, photoshop collage. There are indeed people who can do great things with them but I think it's a vanishingly small group and most of these people do better work on paper and canvas. This might sound crazy to some but it's made me appreciate more that colored pencils and crayons might be super common but they are extremely under-utilized, some people can still do quite startling work with them and that makes it seem like there's still a lot of unexplored territory for them.

- It's sad looking through Deviantart and it's completely clogged with AI. I really hope Cara gets nsfw content soon because I miss a good art platform. Some really good artists are on there now and I'll probably join eventually. Deviantart's inbox system is the gold standard of social media to me, such a shame they had to embrace AI, the place is in sorry shape.

- I've become more thoughtful about how we criticize art, I've heard that a lot of people stop drawing after being told their noses or hands are wrong, and I see a lot of artists make increasingly conventional stylistic choices and I think they are afraid. I wish I could make them unafraid.

I don't consider AI "art" to be art.

https://www.instagram.com/laurentaskienazy/ - this is the director of the Stereolab video.

I do find "alien Atomic Ranch" compelling as an aesthetic, but none of these images take the leap from being a mildly interesting thing to scroll past on the gram.

This might be good?
https://www.bloomsbury.com/us/im-not-an-artist-9781350418776/

There's a weirdly gentrified tastelessness in the stereolab video vs the explicitly tasteless jesus lizard one, but for either I feel like any dedicated fan could use the same tools and come up with something more compelling.

Is there a list? I've seen Tears For Fears, Voidz, Gavin Friday, Dinosaur Jr and Residents do it. I don't assume they all knew it, some people think they're paying a traditional artist and can't tell the difference.
I seen someone say the Aphex Twin caterpillar video is AI but it looks like it could have been done on much earlier technology.

I was also thinking how much less expensive the video for Black Hole Sun could be today, and their 90s use of CG now makes the video read as slop.

AI was so much better when it was worse.

A friend of mine went from making very good vaporwave to putting out unremarkable AI music

it's ok to have a particular taste, and there are different politics coded into one's aesthetic preferences, but i do think it's erroneous to think that there's something inherently immoral or wrong about the aesthetics of AI art (as opposed to the means of its production).

in that way, the comparison to sampling/electronic music or photography feels really apt to me: yes, it's alien or different; yes, you might prefer the cello to the synthesizer; but the validity of this artform will be founded on whatever great art gets produced by the people using it. (and it's too young to have had much of that bubble up yet.)

maybe it's a foolish (and even revolting?) act of faith, but i firmly believe there will be "good art" made using these tools in the next 20 years.

― sean gramophone, Wednesday, April 9, 2025 10:49 AM (two hours ago) bookmarkflaglink

this is what I struggle with; is this just me not accepting advances in technology, the same way my parents' generation railed against sampling, CGI, and auto-tune? or is AI inherently different?

I would maintain it is, because I really do feel my brain struggle to parse the 'stories' told by AI. a lot of this is because the people in AI videos just seem to look slightly different every time you see them. even when it's based off a real person - like the fake Tom Cruise movie scenes or whatever - he just looks a little bit different in each scene, in ways that are subtle but your brain picks up on them. it's not like even crappy 3D animation where you can at least tell you're seeing same bad render/model from different angles.

I have a generally bleak view of digital art, cgi, photoshop collage

I agree with you, but would point you to my wife's website. All of her art is collage and created in Photoshop and other digital apps. She starts with "real" images (often photos of distressed metal surfaces, like the orange flatbed carts at Home Depot) and then layers and combines and manipulates them.

My problem with AI art is that it's unimaginative because it's a tool for unimaginative people, whose ideas run the gamut from "Lois Griffin with giant photorealistic breasts" to "a scene from Lord Of The Rings but with current Democratic politicians". Does it have the potential to be uncanny/breathtaking/whatever? Yeah, but that would require an entirely different class of users.

(Honestly, some of the really weird "Facebook AI slop" that often gets made fun of, like shrimp Jesus or children building incredibly elaborate sculptures out of bottles? That shit I kinda like. Because it is so fucking weird, it makes you wonder why someone would do it. So much other AI art just looks like masturbation material, but the genuinely baffling shit is what makes me think that there's actual potential there.)

this is what I struggle with; is this just me not accepting advances in technology, the same way my parents' generation railed against sampling, CGI, and auto-tune? or is AI inherently different?

All the people most impressed by AI I know are older fwiw, it doesn't seem like a generational thing at all

Doesn't AI use a whole bunch of electrical power? I'm not gonna waste time with that stuff

I feel like either it is already or will get to the point where serving you a netflix video uses similar resources.

I do like the idea of people dumping their netflix subs for a locally generated AI approximation of their favorite programs.

I feel like either it is already or will get to the point where serving you a netflix video uses similar resources.

Can you elaborate?

There's a lot of practical visual uses for AI that I use fairly regularly in multimedia. Just recently I used it for the following:

Fixing a wrinkled shirt in a photo
Adding a plant to the background in a video interview
Fixing one inaudible word in a voice over
Removing noise in audio (it's amazing for this)

Personally, these tools have saved my hide a few times when I just had to "fix it in the mix".

I would never use AI to "create" anything, but it's great for saving time repairing things you missed during a shoot/recording/technical errors, etc. Saves me hours of time and has zero impact on passing along work to freelancers, contractors or FTEs.

yeah I agree there are a lot of AI tools that seem very useful, the way it's been able to upscale old concert footage is very impressive for example, I'm talking more about the generative slop that seems to be everywhere these days

I use AI transcription software. I always have to go back, re-listen and fix some small errors (and occasionally large ones — the software recently couldn't tell my accent from the interview subject's, so it assumed it was one person speaking the whole time and I had to break it all up manually), but it's pretty good.

locally generated AI approximation of their favorite programs

FLENDS Season 3

There's data centers and a whole lot of infrastructure devoted to streaming this or that. I don't know if it's true still, but for a while, netflix was even using Amazon's servers for this (which among other things points to just how difficult it is to boycott an Amazon, but it was a little weird since Amazon Video was also directly their competition).

You can already do slower image-gen experiments on your laptop, so you can maybe conceptualize the electrical cost differential based on how much your laptop uses to spit out Jesus Shrimp with photorealistic boobies versus decoding Black Mirror's latest season? I feel like if they try to move into streaming VR-type media or remote gaming stuff that gap narrows even more.

AI upscaling and colorization of old footage is a crime against good taste (especially when they bump the fps like it's a video game) even when it's not a complete disaster like the I Love Lucy remasters that brought out the celluloid demons.

I have no drawing skill whatsoever and enjoy farting around with AI. Some uncanny valleys are new and unexplored and that can be interesting. A couple of days ago I wrote out a dream and asked GPT to recreate it in the style of a Chester Gould comic. The ways it fucked the brief were quite funny.

It’s also very good for theasuarusing, copy editing (“rewrite this para but with 30 less words”) and finding useful quotes from PDFs. When I’m writing an essay I’ll happily drop $20 on GPT just to help with the essay. I do have to ask GPT to give me “verbatim, real quotes only” or else it makes them up, even when it’s scanning a PDF. I know I’m being lazy but I don’t care.

But yes the energy wastage is immoral.

AI upscaling and colorization of old footage is a crime against good taste (especially when they bump the fps like it's a video game) even when it's not a complete disaster like the I Love Lucy remasters that brought out the celluloid demons.

― papal hotwife (milo z), Wednesday, April 9, 2025 3:18 PM (forty-three minutes ago) bookmarkflaglink

I agree it's shitty to do that with old films and TV shows but it can be a godsend with things like old concert recordings. a lot of 70's bands didn't really get any good footage through their classic runs but AI can upscale the stuff that does exist and make it look and sound nearly professional. every once in a while you'll notice something weird but as a whole it's a good trade off.

So much other AI art just looks like masturbation material, but the genuinely baffling shit is what makes me think that there's actual potential there.)

― Instead of create and send out, it pull back and consume (unperson), Wednesday, April 9, 2025 8:15 PM

There's a lot of totally baffling AI porn of course.

Adding a plant to the background in a video interview

― bookmarkflaglink (Darin), Wednesday, April 9, 2025 9:10 PM

A plant? Why?

yeah I agree there are a lot of AI tools that seem very useful, the way it's been able to upscale old concert footage is very impressive for example

I haven't seen these but upscaled photos is one of the most infuriating things about AI, it adds odd things and creates unpleasant textures. Extremely annoying that some places are flooded with this stuff and it will circulate instead of the originals.

And what's the energy costs for all of this?

A plant? Why?

It was a drab background with no color.

I'm just pulling these possibly bogus stats from google search:
"As of 2019, streaming one hour of a video with a 4K quality on a mobile phono using a 4G network consumed approximately 1.15 kilowatt hours."
"A single ChatGPT query consumes roughly 0.0029 kWh"
"A single image generation can consume as much as half of a smartphone's battery charge, approximately 0.011 kilowatt hours of energy."

So you can make roughly 400 text queries, 100 image generations, or watch 1-2 episodes of "IS IT CAKE" ?

But what if all the cake was AI?...

i was watching an "AI-upscaled" My Bloody Valentine concert from 1988 on youtube, i had to double check the year because the quality was so crisp, like "wow this reunion show is amazing, cool that they're playing just pre-loveless stuff, they still look so young!"

I’m sure this will be ignored but I’ll keep shouting this from the rooftops until people listen: this entire thread is pointless until someone can actually define “AI.” The difference between the method used to create the Stereolab video in the first post and the upsampling method in the most recent post are wildly different.

okay sorry that's why i used scare quotes

Likewise, if you can’t define “AI,” none of the claims about climate impact make any sense.

brimstead, I apologize if that was too harsh! It’s been a frustrating few years for anyone in this area of research since there’s so much deception and Silicon Valley bullshit.

no worries :)

Darin, what do you use for removing noise in audio?

Unperson: I like your wife’s art very much. Especially the “paper” works.

A lot of AI art will turn out pretty good (already seen an example or two) and I reckon people will miss it once the ecological impact becomes impossible to ignore and a lot of it is cancelled.

Although if its true you don't need that much of a scale up via data centres (and its accompanying water and energy consumption) it will be here to stay.

Its more of a local thing but the 'slop' I get irritated by is how every high street in London has the same Pret/Leon/Five Guys etc etc. Food options while really nice shops/restaurants close.

So I'm thinking that if the world is this much of a fucking slop the internet mirrowing it kinda follows.

this is what I struggle with; is this just me not accepting advances in technology, the same way my parents' generation railed against sampling, CGI, and auto-tune? or is AI inherently different?

I think part of this might be its not easily possible to separate out or cordon off AI in the way you could with CGI or autotune. CGI affected films and not much else, so if films weren't something you watched much maybe you didn't even have an opinion at all. But AI isn't restricted to one domain, its in all domains, maybe its making you think about a particular type of art differently, it's probably also making you think about the continued existence of your income. And if its the same thing doing both it might make you feel more negatively about it. Autotune didnt do that

The high energy cost is from training the machine learning models. The cost of generating images once a model exists is negligible.

After playing around with some models, it's my opinion that AI gen images are largely a hoax. I separate them into 2 groups.

1. images made from short text prompts. these are all hoaxes. they use a llm to extract keywords from the text and give them weights. The pool of recognized keywords is much smaller than you would think. If you try to go beyond the keywords, they fail miserably. they work by taking a stolen base image found with the keywords then modifying it to match the keyword weights more closely. Sometimes they know how the keywords relate to each other and sometimes they don't -- keep rolling until they get it right!

2. img2img where you input an image and edit it with the help of ai. if you take a photo of your slob self and tell it to make you look like chris hemsworth, it can probably do it. it has a clear, easily defined goal with very limited options. its artistic input is minimal.

https://architizer.com/blog/inspiration/stories/ai-will-destroy-creativity-if-we-let-it/

My take on this question

The new Pulp video released yesterday is AI too while on the same day I read a press release where Jarvis commented the new album “was written and performed by four human beings from the North of England, aided and abetted by five other human beings from various locations in the British Isles. No A.I. was involved during the process."

Like every form of automation in this society its designed rn to devalue what you do and to have destructive societal effects.

I think this is correct on an objective level, though I'd say probably remove more so than devalue. But I think on a subjective level its probably more difficult to enjoy things made by the same thing thats contributing towards precarity, which I think may be a factor in the way people feel. Are they wrong to feel this way if they do? Possibly, but I do think it feeds in to the way some people feel

Nobody is wrong or right

Citation needed

But all I'm saying is that if you are faced with a good piece of work and it turns out that it was even partly made by AI that's still a good piece of work.

Like the example bought up earlier I bet it must've sucked for a painting fan to see an amazing photograph for the first time. But that photo is still good. Feels like this is what's happening here.

This may be objectively true, but for the reasons you outlined above, not necessarily subjectively so. As part of the experience of art (or anything else) is the audience or receiver. And if this person, for reasons of their own is less receptive than they might be, they might not view it such an objective way, even if purely sub-consciously

The painting/photographer example I think would be more analogous if photography threatened not just to replace painters, but everyone - this might make the photography not so straightforward to enjoy

Its not about consumerist enjoyment but evaluation, and good works go beyond their conditions of creation long after you or I are gone.

Much of the best literature was created by people with the time to do so, using their privilege. They were also one of the tiny percentage of people who could write or read at the time. No one says Montaigne is rubbish because he owned a castle.

In that case it makes sense. removing the subjective element and looking at things objectively - especially in the future with some remove as well. There's an argument to hand that over to AI as well

This is maybe where this kinds of ends up, where we can just have the works and bypass the people altogether. I guess in some ways we're already there in certain sectors, with pieces written by ai and read by bots.

David Abram has written a book arguing the alphabet is actually the technology that kickstarted the current ecological crisis. i’m not 100% sure that this is the wrong idea. here’s an interview where he talks about it https://childrenofthecode.org/interviews/abram.htm🕸

this is going to sound ridiculous but I'll never think of Stereolab the same way after this.

― encino morricone (majorairbro)

somebody on ilx a while back was saying they were, like, libertarian socialists?!?!? idk how true that is, but makes sense if so

The new Pulp video released yesterday is AI too while on the same day I read a press release where Jarvis commented the new album “was written and performed by four human beings from the North of England, aided and abetted by five other human beings from various locations in the British Isles. No A.I. was involved during the process."

― I am using your worlds

even if they hadn't used AI in the video, they sound like Queen going "no synthesizers!" problem is, it's hard to rail against AI without looking like a luddite.

i admit that there might be some viable uses for thneeds.

Always worth remembering the actual luddites were not reactionary grumps, they were workers actually quite proactive in using new technologies for their movement. Which otoh brings home how much capitalism invests in making everyone think every new technology is great and will change everything.

this is an awesome read, btw

oh good! i didn’t read the whole interview but that book (Spell of the Sensuous) is great and there’s a bunch of videos on youtube of the talk he always gives (the best version out of the ones i watched is an interview with Matt Segall)

The stereolab video is really bad. They’re going for a Lynchian eeriness but it doesn’t come off at all. Unfortunate misstep by a legendary band.

The pulp one is a little better but I still really dislike the texture of these ai environments. Hard to describe.

I wonder what this era's equivalent of this would be:
https://www.youtube.com/watch?v=Av-gV5HAxXk

I’ve had some uncomfortable moments recently realizing that some music (maybe 5%) I enjoy is ai generated. I think over time it will become harder to tell

Just say no to ambient

Is ambient real?

Set your white noise machines to pre-laptop merzbow

just saw this on sbky

‪Regrettably, Pavel‬ ✧@spa✧✧✧.b✧✧✧.soc✧✧✧‬
·
2h
While there is a lot to say about this image, the thing that stands out to me the most - what IMO makes genAI a specifically right wing phenomenon - is the craving for approval. It is not enough to be able to generate robot lingerie warrior lady. They NEED to also be praised for doing it

Robot lingerie warrior lady promptin’

The robot lingerie warrior is also in response to a dirty lib AI hater losing their mind over Steven Spielberg and Jurassic Park. They couldn't even come up with a Coen Bros movie or something made by a Black woman?

someone on a discord server linked me to a video by this dude

https://www.youtube.com/watch?v=xMYm2d9bmEA

honestly the vibes on this guy seem a little off to me. he talks about himself a _lot_. and i'm here saying to myself "wait, i haven't ever heard of this guy. should i know this guy?" turns out i watched his video about model train synthesizers a while ago so he didn't just pop up out of nowhere. i guess he went mega-viral when he put out a video on march 9, "You Are Witnessing the Death of American Capitalism".

anyway part of the reason i'm posting this here is because if this guy is a grifter of some sort, y'all will know about it more than i do. he might not be. i mean certainly he has a _hustle_. a hustle isn't necessarily the same thing as a grift. also, like, the stuff he's saying makes sense. this is someone who to a large extent works with AI, who clearly isn't anti-AI. talking about the environmental impact of AI is a footnote in a 30 minute video, but he acknowledges it and says that he's working to minimize the environmental impact _and_ working on developing the technology to do what he's doing in a less environmentally destructive manner, which, i mean, is more than most AI people will do.

he calls what he's doing "poison-pilling music files", and it's one of those things that i'm kinda surprised i didn't think of before now. basically when you have AI sucking up everything it gets its hands on as training data, it's only a matter of time before it runs into Little Bobby Tables, from the hilarious "xkcd" comics:

https://xkcd.com/327/

i mean, three fucking words, right? SANITIZE YOUR INPUTS. all these LLMs either don't do this or don't do it ADEQUATELY. i learned about GIGO, "garbage in garbage out", thirty fucking years ago when i was 18, and this past decade i've been watching all these AI grifters talk all this shit without ever acknowledging that basic principle. again, benn jordan clearly has a hustle, but it's not the same thing as a grift. data quality correlates with result quality. the way he describes what he's doing is comes off as "hey, here's an idea, why don't we PAY MARSHA WASH?" that's kind of my moral barometer when it comes to the business of music - "is marsha wash getting paid from this?"

(by the way even though everybody and their brother stole from that tape back in '89-'90 i've never actually seen an unedited copy of that Marsha Wash demo tape in the wild... maybe it's not out there, idk.)

anyway i _guess_ i'll watch his "you are witnessing the death of american capitalism" video, but i'm doing it in a private window, cuz i'm not super into "political" youtube videos, particularly not videos telling me things i already believe.

---

i have no idea what y'all are talking about re: robot lingerie warrior. we had sexy robots when i was a kid. we didn't need computers to make them. humans make better fetish art than robots, even (especially?) when it's fetish art _of_ robots.

Benn Jordan’s been an IDM artist for a long time as The Flashbulb. He’s not a grifter but he loves himself lots.

its not just sanitizing your inputs, its that these models can't really discern context, which is important since a large chunk of the internet is shitposting. like that time when Google's AI was telling people to put glue in their burrito to help everything stick, I'm guessing someone posted that long ago on Reddit or something and it got a bunch of upvotes which led it to getting fed into the training data and now it's telling people to eat glue because it doesn't know what a joke is. now that the richest government in the world is using AI to make major policy decisions it does seem like we're headed for some gloriously stupid times

miyazaki character, pixar character, action figure - blehhhh

the animation ones are funny because if there is any text on anyone's clothes it shits the bed.

one reason that keeps coming up for me why ai-generated art is so bad and hated .... is because it sucks ass.

Just wait until it gets really good though! Honest, it's happening soon. Really.

yeah they keep saying that but some of the issues in AI artwork seem tricky to solve. with a lot of images it's apparent right away that it's AI thanks to the glossy hyperreal sheen it puts on everything, but sometimes it really does get things right visually. even so you can figure it out by looking around. for example if you do a kid's room, since it just takes the "average" of all kid's rooms it will put a lot of items together that don't make any sense. stuff from different eras, multiples of the same object, no consistent interests, stuff like that. and for me sometimes the unconscious mind registers that stuff first and you just get that off-putting feeling from it. maybe a version of this that cross-references ChatGPT and knows what belongs with what, but idk ChatGPT struggles with stuff like that a lot too. also it seems like the generation of sixty trillion AI images over the last two years might screw with the next batch of training data. I mean I dunno what'll happen if it thinks the real world just always looks glossy and nauseatingly plastic like that. it'll get even worse. though I guess it already thinks that

btw to further define what I mean by "evil" I don't mean AI in general is evil, it's like any technology where it just matters how you use it. like ChatGPT I think has a lot of nefarious uses but as a research tool or helpbot or summarizer of long boring meeting it's fine. good, even. but the stuff that appeals to our senses - stuff that generates art or music, tools that can accurately depict people doing or saying things they didn't do - that I think is inherently evil. even though there are lots of innocent/funny/cool/helpful uses of it ("fixing a wrinkled shirt" as posted above is a good example).

but any Sci-Fi movie that depicted that sort of technology would almost certainly be dystopian. it seems way way easier to imagine uses that are nefarious and immoral and a disturbing violation of personal boundaries than it is something that actually makes our lives better. its so clearly Black Mirror shit not Jetsons shit. I guess one hopeful thing though is that people genuinely seem to hate it, obviously there's a part of me that wonders if future generations will accept and embrace this stuff like mine did with the Internet, but it feels like people are hating on it more now than ever which is not generally how successful technological revolutions go.

I see the Ghibli images have already been replaced by action figure packages.

AI art becoming the equivalent of Facebook personality quizzes and Farmville might kill it faster than anything.

even though there are lots of innocent/funny/cool/helpful uses of it ("fixing a wrinkled shirt" as posted above is a good example).

thinking of gyac's post in your other thread about the holding pattern where she said women are having procedures to fix their facial expressions so they look wide awake the instant they crawl out of bed, idk how innocent these uses are actually. i mean photoshop already dealt the damage ofc but more of this can't possibly be good.

AI art becoming the equivalent of Facebook personality quizzes and Farmville might kill it faster than anything.

I always think about lasers being absolutely the most high tech thing ever achieved at one point, and now you can only find them in the cat toys area of a grocery store.

Tech has always been psychotic. All of it needs to be nationalized and laws made so that its run for the public good.

i’m the guy at Apple who makes sure all your featured photos are your exes and your dog that died

— trash jones (@jzux) April 15, 2025

thinking of gyac's post in your other thread about the holding pattern where she said women are having procedures to fix their facial expressions so they look wide awake the instant they crawl out of bed, idk how innocent these uses are actually. i mean photoshop already dealt the damage ofc but more of this can't possibly be good.

― doe on a hill (Deflatormouse), Wednesday, April 16, 2025 1:21 AM (nine hours ago) bookmarkflaglink

yeah as I mentioned above I've used it to put my face on album covers and then through Photoshop replaced all the text so it was a promo for a DJ gig. one of them was the Q: Are We Not Men? cover and someone asked if I changed the face to be Ryan Gosling. it was kinda flattering but no, I don't look like him, but maybe someone with Chi Chi Rodriguez's bone structure and some of my facial features might look like that. if I got a few surgeries maybe I would. I can definitely envision a future where photos are touched up not with Instagram filters but much more subtle AI face tuning. and between that and all the AI slop flooding the rest of the internet I can see it distorting people's view of what people are supposed to look like, I can only imagine the body dysmorphia my kids' generation is gonna be feeling by the time they're in high school.

I remember that flyer! Or it might have been your face on a different album cover that you posted here. Was great.

https://architizer.com/blog/inspiration/stories/ai-will-destroy-creativity-if-we-let-it/

My take on this question

― treeship 2, Thursday, April 10, 2025 7:54 AM (one week ago) bookmarkflaglink

These are machines that boil down a vast amount of data drawn from the Internet in order to perceive statistical patterns. They then use these patterns to predict likely outputs for user generated prompts. In essence, they show you what they think you want to see based on averages. As the artist Hito Steyerl writes in “Mean Images,” her brilliant essay for The New Left Review, “They represent the norm by signaling the mean. They replace likeness with likeliness.”
Whose subjectivity is expressed in a work generated by A.I.? In one sense all of ours — a hive mind.

So it seems like there’s at least a potential in that to satirize the popular imagination, or to work in the medium of popular imagination in other ways. I’m sure there’s a lot of this kind of stuff already, i haven’t been paying attention.

https://www.buzzfeed.com/daves4/european-ai-american-list

I'm going to admit I got tricked by this one when I didn't fully read the description; it's by an actual artist (a sculptor), and I read this much of the description: "Video by Mike Bennion
This piece is a visual poem, attempting to complement the exotic and sacral nature of the music. It is heavily inspired by the 1969 film ’The Colour of Pomegranates’ by Sergei Parajanov. It uses a similar minimalist, old school film grammar, real locations, no camera moves, no special effects, no actors just ordinary people, symbolic gestures whose meaning has been lost over time"

https://www.youtube.com/watch?v=yIEM8BQThO8

I sent it to my wife and kid (kid is studying film, wife is armenian and has seen color of Pomegranates) and my son said "it's AI" and I said "no it isn't". "yes it is, look at the liquid, also, no scene is longer than 5 seconds". Then I read the rest of the description: "A combination of my subconscious and the whims of AI has produced something we don’t understand, and that is what hopefully holds our attention, to figure out the puzzle.
Mike Bennion is a British sculptor and painter living in the USA who discovered AI in the autumn of 2024 and now can’t stop making films with it."

and felt stupid

that said, it's an arresting video which, while very obviously displays it's influence, actually creates some great visuals. How much of that is down to Mike Bennion's direct prompting? I dunno.

I feel like there is no way of knowing except if you have seen the film (fairly recently) and can conclude that probably nobody took the time to recreate such painstaking visuals just to do a pastiche. People into art naturally want to pride themselves on seeing through artifice, but just think how fast computers learn (translation, chess, etc) and it's pretty clear that this is futile. In a few years, probably AI can produce an unknown Beatles hit or a Coltrane solo and only experts with impeccable memory will be able to tell.

https://www.ft.com/content/24218775-57b1-4e9f-ba64-266a3239cf27

On "slopaganda" (warning = much Ed Zitron outrage).

I don't think this was raised yet: I wonder how much of a liability it might be because sometimes It ends up plagiarizing one specific piece of art, which opens you up to lawsuits (see the bit in that Beato-Fantano video about AI music making an obvious Drake ripoff), perhaps it can be programmed so it never takes too much of anything from one source?

like ChatGPT I think has a lot of nefarious uses but as a research tool or helpbot or summarizer of long boring meeting it's fine. good, even.

in fact, these are all very unreliable use cases for llms! please, no one rely on llms for summarising anything - they can't actually do it and will at best get all the details wrong while superficially appearing to have done what you wanted.

Fuuuuuuuck this. I really like Natasha too, this is so disappointing on multiple levels:

https://www.theverge.com/film/657990/natasha-lyonne-uncanny-valley-asteria-marey"> https://www.theverge.com/film/657990/natasha-lyonne-uncanny-valley-asteria-marey

Natasha Lyonne to direct and star in a new sci-fi film created with generative AIThe AI-focused studio behind the project says its models are wholly trained on licensed content.

maybe it'll be like 3D movies where they tried again and again and again and nobody really likes the experience

Worldcon 2025 in Seattle uses a LLM to select who will appear on what panel and the fallout is furious and righteous. They should just cancel the con at this point
https://seattlein2025.org/2025/04/30/statement-from-worldcon-chair-2/
https://bsky.app/profile/seattlein2025.org/post/3lo2jzs32tf2w

"Ethical AI" to me, currently feels a whole lot like when they outsource your job, but make you train the new guy before you go.

man natasha makes this movie sound fun: “Coming together as a trio in cahoots with the astounding imagineers at Asteria, to worldbuild this film at scale, has been a synergistic dream come true.”

really sounds like they're drilling down on their core competencies in granular detail to scale up the churn rate of entertainment for all stakeholders. can't wait to circle back to this movie and make it an action item on my to-do list

Don't know who I'm more disappointed in right now: Natasha or Stereolab

Stereolab. You would think at least one of them would have a clue.

I thought the revive would be about the BBC Agatha Christie nonsense