2) Two of the maze walls appear to be smudged, but nothing else in the image is.
3) The floral pattern on the plate is messed up - it repeats, but berries randomly change positions and the leaves change shape.
4) The milk carton has a splotchy texture that's really common in ChatGPT-generated illustrations.
I suspect it's not 100% gen AI - for example, the sharp outline around an out-of-focus xmas tree feels like a human-made artistic choice - but I'd bet it's at least partly composed from AI-generated elements.
I hadn't even looked at the floral pattern. It's awful.
I wonder how pissed Apple was when they realized they had paid him a few thousand (going by his posted rates) for "art" that used gen AI to a significant enough degree to have nonsensical components.
>for example, the sharp outline around an out-of-focus xmas tree feels like a human-made artistic choice
If I had to guess, it looks like he did the components of it separately and then just pasted them all in together. For example, you'd expect the light on the plate to brighten the milk carton near it. And the shadow on the plate doesn't line up with the bright light shining on the carton's side. These are just classic sloppy bad artwork mistake.
Notwithstanding your previous points, these sorts of AI detectors flag many false positives, they're not worth relying on. The only one that could actually work is specific watermarking in the images themselves, such as what Google does with SynthID in their generated images.
The structure of the carton is off. The spout. The opposite end of it.
Being a trained artist myself I don’t think many artists would deliberately make that kind of design choice considering given the overall style of the artwork.
Thing is, from a commercial standpoint, who cares? It's an image used to promote a TV show on a social media account. There will be a tiny, loud community of AI haters who make a stink on social media, but the vast majority of people will be oblivious, and the quality issues with the image will have zero bearing on whether anyone watches the show; whereas the fact that the post exists probably WILL drive a few to watch who wouldn't have otherwise.
The fact is we're in a world where algorithms consume more attention than ever. There's more content than ever. We're more sedentary and glued to useless shit on a screen than ever. We consume content like we're pigs, and what do you feed pigs? Slop. I don't like this, and I try to keep my diet as healthy as I can, but it's probably still worse than it was 20 years ago, and most people don't care at all. It's all consumption for the sake of mindless distraction. The slop exists because the demand exists. People will watch the show anyway.
> There will be a tiny, loud community of AI haters who make a stink on social media, but the vast majority of people will be oblivious
Not only oblivious but actively for it; I know many people who even watch fully AI generated content on TikTok or Instagram Reels. In fact, they know it's AI yet still like it, probably because some can look pretty cool or funny.
I actually started listening to AI music unknowingly, and was merely impressed when I later realised it was AI. However, I don't think AI videos have reached that point yet.
What's worth mentioning is that there has always been large quantities of "slop" even with human-generated art; the good stuff has always been a minority.
The "non-denial denial" seems like a denial to me if you take him at his word that he is unable (presumably due to legal reasons) to comment about a specific project.
"I always draw and paint by hand and sometimes incorporate standard digital tools" would apply if I were to make a 30 second scribble in MS Paint and then upload it to ChatGPT or similar tools and ask them to make a version of it in whatever style I wanted.
Most likely, he did exactly that. Made a rough sketch and then achieved the final product either by just handing it to a model (which I'm guessing based on the maze) or he used gen AI tooling in his digital painting software to polish it into a finished painting.
It's not an accident that he's being so cagey and using exactly that verbiage.
These artists who do work for tech companies are literally given a day to do this work. Of course they are using AI. They have to. The days of having time to make something great are mostly over.
> The days of having time to make something great are mostly over.
Such times never existed, you just failed to notice. Flat design took over not because it was pretty, but because it was cheap and versatile. That's why I love the furry community - you can see those people value art as a goal itself, rather than a mean to achieve other (often monetary) goals.
It's labeled "whole milk" in small text then below in bigger text it's labeled "low fat milk". The fact that this was not caught and likely would not be deliberate by the author seems slopy (sloppy?) to me. Just my two cents
The prop doesn't seem to have the weird intermittently blurry and inconsistently spaced maze, or at least it's not on the same side. What a weird unforced error it would be if that was the only part of the image to actually be AI-generated.
Isnt it an intentional choice? The show Pluribus has a not-so-subtle analogy to an LLM that has access to the world’s knowledge, doing its best to give whatever we ask it for, and not being the brightest tool in the shed. I think the artist accomplished what they wanted to in the context of the show.
EDIT: I stand corrected, the show’s analogy to AI is coincidental.
Interesting. Still, original intent aside, the audience will draw their own comparisons and develop their own take aways.
IMO that's fine because art isn't a one way street. Audiences also play a role in how it's interpreted and what happens to it in private and once copyright and trademark expire.
Doesn't the word "coincidental" imply Vince Gilligan is OK with the anti AI take? He's not denying it's there, he's saying it's not intentional, it's coincidental.
I agree with "art isn't a one way street". But, it's also up to the artist whether people's interpretation is "right" or "wrong". Some artists love when people find meaning in their work that wasn't intentional, and some don't.
One story that's burnt into my brain, is about Ray Bradbury giving a guest lecture. This is meant to be a quote from Bradbury. It's hard to know what's real these days.
From "Listen to the Echoes: The Ray Bradbury Interviews"
"Weller: have you encounted academic misinterpretation of your work?
Bradbury: I was lecturing at Cal Fullerton once and they misinterpreted Fahrenheit 451, and after about half an hour of arguing with them, telling them that they were wrong, I said, “Fuck you.” I've never used that word before, and I left the classroom.'"
I think it's fine to read Fahenheit 451 and have your own opinions about its main theme, but, it's another thing to get into an argument with the author about it.
Bradbury said Fahenheit 451 is about the effect of mass media on society, if it was written today, it might be about the effect of AI on society.
If it was intentional, they wouldn't be as shy about admitting it.
But as far as your "correction" goes, even if the anti-genAI overtones at its inception are a coincidence, it wouldn't be too crazy for someone making related art today to play up the coincidence anyway. So I think your original idea was still a reasonable guess.
"It either is AI-generated slop or it looks like AI-generated slop for no artistic or thematic reason whatsoever."
The artistic and thematic reason is obvious. It's a commentary on the show and AI art in general. To ignore this message because it "looks like AI" devalues the entire concept of human art.
Everyone in the industry who is trying to find a way to weasel and deny using AI is using the same "incorporate standard digital tools" bullshit phrase. I think Epic said the exact same phrase when the Fortnite community caught them using AI art in the latest season of the game.
It's not that it's ugly--it's that it's obvious AI slop. We are, for the time being, still in that "uncanny valley" phase with AI art where a human who knows what to look for can still easily discern AI slop from art. For now, there are still telltale signs.
In a lot of cases the telltale signs are like neon lights, like that “hypercandy” style that a lot of generated images have. I know what those are within milliseconds of seeing them.
I think there must be an "uncanny valley" for artwork. It is somewhere between norman rockwell and aunt tilly's nature painting. Closer to norman rockwell in detail.
Surely the fact that "It says low-fat milk and whole milk; no one would do that" is followed by "oh the real props do that" should cause everyone who thought the former to down-rank their own ability to tell what's AI and what's not. If they don't then they aren't incorporating evidence about the world and their skill.
To have a separate conversation from everyone else (who is talking about whether it's real or AI), I think it's interesting to see people's epistemology. If you thought something was X because of A (i.e. P(X|A) > P(X), maybe much greater), your posterior for P(X|A) should be different from your prior in response to the evidence "it was X, but not X also has A" and I think the directionality should be obvious.
For those who don't do that, I should update my adjustment factor to their claims of fact and not in their favour.
It's funny that this comment is also an example of a confidently wrong judgment. The props are 100% accurate to what's going on in the show. Not really possible to explain why without spoiling it, but the milk cartons are supposed to be suspicious.
His response is confusing to me as well. I didn't edit it that drastically. I moved one or two of the clauses around but the fundamental thrust was: if you think it's an AI image because of the whole-milk/low-fat discordance, but real-world designs for the show also have that discordance, then you should consider that the discordance does not mean it's AI.
That would still be true if:
* it was indeed an AI poster
* it was an AI image poster made to look that way
* it was a human-made poster accidentally made that way
* it was a human-made poster intentionally made to look that way
The truth of the show itself could have no bearing on what I was saying. The only thing it does rely on is whether or not the real-world designs did not correspond to the poster image.
The reality is that humans suck at telling AI. Sure, there are obvious tells for certain things, but if one really tries, they can make AI generated content indistinguishable from human made. You even see this on Twitter, where actual human artists are sometimes subjected to a modern day witch hunt by others saying their art is AI, and the artists literally have to prove that it was made by them, sometimes by pulling up various stages of the drawing in progress (and what is even funnier is that now Google's Nano Banana Pro can even generate that sort of progress compilation images).
My point was that the mistake didn't happen during prop creation. Those aren't milk cartons, those are HDP cartons, so the props are correct.
As to the content of your post: It doesn't make sense. Thinking something is not human created when it turns out that the real reason was that it wasn't created by a human in the show is not a valid reason to stop applying that as a useful discriminator between AI and human art. It's a Gettier case, but the J part of JTB knowledge still stands, and there's a reason grappling with the Gettier problem is so gnarly in epistemology.
How much evidence do we actually have that AI wasn't used for these "real props"?
(Personally I don't care about my ability to tell the difference between what's AI and what's not; I care about my ability to tell the difference between well-crafted and not, and that seems to be functioning fine)
I have a very sensitive ai-radar and I don't think this illustration is AI. It has an extremely uncanny look to it, but totally unlike the uncanny sameness of AI slop. I think it's highly unlikely it's an AI image, especially coupled with the denial by the artist.
False AI accusations are very harmful, especially when so much actual AI slop goes unnoticed.
I don't know, I think we're just finding that it also applies neatly to a lot of stuff we were already seeing before genAI. Edit: "blogspam", for instance, is almost the same phenomenon.
That photo is horrendous, so I sure hope for the artists sake it was a 30s prompt job.
I will push back slightly on the idea that slop is only slop because it's bad. AI art will always be slop because of its very nature, it's not even possible to describe ai art as "good" because that's not a quality it can even possess.
1) The maze on the carton can't be solved.
2) Two of the maze walls appear to be smudged, but nothing else in the image is.
3) The floral pattern on the plate is messed up - it repeats, but berries randomly change positions and the leaves change shape.
4) The milk carton has a splotchy texture that's really common in ChatGPT-generated illustrations.
I suspect it's not 100% gen AI - for example, the sharp outline around an out-of-focus xmas tree feels like a human-made artistic choice - but I'd bet it's at least partly composed from AI-generated elements.
And for what it's worth, the image pretty consistently trips gen AI image detectors (e.g., https://hivemoderation.com/ai-generated-content-detection).
I wonder how pissed Apple was when they realized they had paid him a few thousand (going by his posted rates) for "art" that used gen AI to a significant enough degree to have nonsensical components.
>for example, the sharp outline around an out-of-focus xmas tree feels like a human-made artistic choice
If I had to guess, it looks like he did the components of it separately and then just pasted them all in together. For example, you'd expect the light on the plate to brighten the milk carton near it. And the shadow on the plate doesn't line up with the bright light shining on the carton's side. These are just classic sloppy bad artwork mistake.
Notwithstanding your previous points, these sorts of AI detectors flag many false positives, they're not worth relying on. The only one that could actually work is specific watermarking in the images themselves, such as what Google does with SynthID in their generated images.
Being a trained artist myself I don’t think many artists would deliberately make that kind of design choice considering given the overall style of the artwork.
The fact is we're in a world where algorithms consume more attention than ever. There's more content than ever. We're more sedentary and glued to useless shit on a screen than ever. We consume content like we're pigs, and what do you feed pigs? Slop. I don't like this, and I try to keep my diet as healthy as I can, but it's probably still worse than it was 20 years ago, and most people don't care at all. It's all consumption for the sake of mindless distraction. The slop exists because the demand exists. People will watch the show anyway.
Not only oblivious but actively for it; I know many people who even watch fully AI generated content on TikTok or Instagram Reels. In fact, they know it's AI yet still like it, probably because some can look pretty cool or funny.
What's worth mentioning is that there has always been large quantities of "slop" even with human-generated art; the good stuff has always been a minority.
Breathtakingly beautiful AI-generated videos, with undeniable artistic merit. (Images generated with Midjourney, music generated with Suno).
It can if you just go around the maze.
The first is there can be 'non-general' cases, and this could be one of them.
The second is a lot of 'standard digital tools' include AI features these days. A person can use generative AI without leaving photoshop.
Most likely, he did exactly that. Made a rough sketch and then achieved the final product either by just handing it to a model (which I'm guessing based on the maze) or he used gen AI tooling in his digital painting software to polish it into a finished painting.
It's not an accident that he's being so cagey and using exactly that verbiage.
Such times never existed, you just failed to notice. Flat design took over not because it was pretty, but because it was cheap and versatile. That's why I love the furry community - you can see those people value art as a goal itself, rather than a mean to achieve other (often monetary) goals.
EDIT: I stand corrected, the show’s analogy to AI is coincidental.
IMO that's fine because art isn't a one way street. Audiences also play a role in how it's interpreted and what happens to it in private and once copyright and trademark expire.
I agree with "art isn't a one way street". But, it's also up to the artist whether people's interpretation is "right" or "wrong". Some artists love when people find meaning in their work that wasn't intentional, and some don't.
One story that's burnt into my brain, is about Ray Bradbury giving a guest lecture. This is meant to be a quote from Bradbury. It's hard to know what's real these days.
From "Listen to the Echoes: The Ray Bradbury Interviews"
"Weller: have you encounted academic misinterpretation of your work?
Bradbury: I was lecturing at Cal Fullerton once and they misinterpreted Fahrenheit 451, and after about half an hour of arguing with them, telling them that they were wrong, I said, “Fuck you.” I've never used that word before, and I left the classroom.'"
I think it's fine to read Fahenheit 451 and have your own opinions about its main theme, but, it's another thing to get into an argument with the author about it.
Bradbury said Fahenheit 451 is about the effect of mass media on society, if it was written today, it might be about the effect of AI on society.
But as far as your "correction" goes, even if the anti-genAI overtones at its inception are a coincidence, it wouldn't be too crazy for someone making related art today to play up the coincidence anyway. So I think your original idea was still a reasonable guess.
The artistic and thematic reason is obvious. It's a commentary on the show and AI art in general. To ignore this message because it "looks like AI" devalues the entire concept of human art.
To have a separate conversation from everyone else (who is talking about whether it's real or AI), I think it's interesting to see people's epistemology. If you thought something was X because of A (i.e. P(X|A) > P(X), maybe much greater), your posterior for P(X|A) should be different from your prior in response to the evidence "it was X, but not X also has A" and I think the directionality should be obvious.
For those who don't do that, I should update my adjustment factor to their claims of fact and not in their favour.
> is followed by "oh the real props do that"
That would still be true if:
* it was indeed an AI poster
* it was an AI image poster made to look that way
* it was a human-made poster accidentally made that way
* it was a human-made poster intentionally made to look that way
The truth of the show itself could have no bearing on what I was saying. The only thing it does rely on is whether or not the real-world designs did not correspond to the poster image.
As to the content of your post: It doesn't make sense. Thinking something is not human created when it turns out that the real reason was that it wasn't created by a human in the show is not a valid reason to stop applying that as a useful discriminator between AI and human art. It's a Gettier case, but the J part of JTB knowledge still stands, and there's a reason grappling with the Gettier problem is so gnarly in epistemology.
(Personally I don't care about my ability to tell the difference between what's AI and what's not; I care about my ability to tell the difference between well-crafted and not, and that seems to be functioning fine)
False AI accusations are very harmful, especially when so much actual AI slop goes unnoticed.
Shark is just a shark. So smart, me!
:)
He’s reaching to try and find any leeway to say it was AI based on the words the artist used.
But then he ends with “well it sucks anyway”
Dude, just say “my bad” and “it’s not to my tastes”. You don’t need to double down so hard.
I will push back slightly on the idea that slop is only slop because it's bad. AI art will always be slop because of its very nature, it's not even possible to describe ai art as "good" because that's not a quality it can even possess.