LLMs Will Never Make Art With Caleb Wilson and Arif Hasan

Call zone media. Bonjour, I am Ed Zitron and welcome to this week's Better Offline. I'm back at it. Download the newsletter, subscribe to the premium. God damn, I need the money. After all, I need to make a big gold anvil to drop on my head at the end of the AI bubble. And of course, if you're not on YouTube, please subscribe on the YouTube. Please, I need you to do that immediately. Give me some emodium because I have the diarrhea of a CEO. And today I am joined by Kill the Computer's Caleb Wilson. How are you doing, Caleb?

Doing fantastic. Thanks for having me.

My pleasure. And how are you doing, Arif Hassan of White Left?

Doing terribly, as always.

Wonderful. Today we have Two illustrious subjects, because Caleb delivered us a paper. I was going to say, I'd call it that. It's literally an academic paper called Why Slop Matters. And these are some of the names. Cody Comers, Ye Mon Du, Julia Gordon, Normal, Ari Holtzman, Normal, Tess McNulty, Spencer Stewart, Lindsay Thomas, Richard Jean Soan, Hoit Long. Hoit Long.

A bunch of people who work at university. Hoit Long. We love Hoit. And it's this, it's a fascinating paper that I'll link to in the notes because it's very academic, but it's also the single most accommodating piece I've ever read about slop. It loves slop. It like genuine, it's like an indefensive slop.

It does seem like that. I read it pretty briefly and I, we'll talk about it more in depth, but it definitely just takes some things as a given in order to mount its defensive AI slop.

Yeah.

AI slop has aesthetic value. Cultural critics of history have initially dismissing low culture only to later admit its significance. This is very fucking stupid.

I think it's like trying to use like the example like kitsch in the paper is like an aesthetic form that eventually did have value. For example, like the famous dogs playing poker painting that people, you know, famously love and don't make fun of. Right.

But the And I get what they're saying. They're saying it has a social function, it's fulfilling a need for more content. And it makes the argument that human beings want to make more and consume more, but also assumes that slop is the actual thing filling that, rather than just spraying so much slop out there that it eventually does so. Right.

I think, for me, the paper is engaging in a couple of rhetorical tricks. One, I think, is to try and accuse people who disagree with it subtly of like being elitist when anybody who does any social analysis of pop culture will take pop culture seriously in its analysis, but that doesn't mean, like they'll argue that it matters, right?

So the title works, but they'll try to extract what it is telling us about the people who produce and consume this content, the culture, that allows or encourages this content to exist. And that could be a critical eye, it could be a supportive eye, but it tells us a lot about the culture that we're in, which is what sociologists do. And the idea that the paper is presenting that, like, sure, like in culture broadly, we dismiss it.

I don't think there are many academics who are, quote, dismissing AI slop. I think it is an object of, quote, unquote, serious study because It says a lot about where we are, you know, for whatever reason, however it says it. And it's mainly just making the argument that, like, if it's so bad, why does it exist? Which is the stupidest fucking argument you could make.

I love that as well, because there is actually a valid discussion to be had here, which is, and this is going to sound mean, and this is dismissive. I don't think most people using slop and making slop actually should be allowed to create stuff. Like, they don't want to make the effort to make something good. So they just, they're like, oh, every idea I have must be brought to life. I don't think that that should be the case.

I don't think everyone should be able to have every idea legitimized.

Because it goes to, like, painstaking efforts to sort of, like, define what slop is in that way, right? It's asymmetrical by their own admission because of how easy it is to make and how fast you can kind of crank it out. Yeah, the examples that they used was kind of strange. I can remember if it was in this one or one of the responses, but it was like, well, slop only matters to the person who makes it in the sense that it's hyper-personalized or whatever. Right. So it kind of discredits itself in that sense because being able to hyper-personalize your content doesn't necessarily matter. The amount that you can crank out doesn't matter if the whole point is to hyper-personalize it, I guess. You know what I mean?

Yeah.

It's also probably worth noting, if you haven't heard of this paper, I don't know why you would if you're listening to this, these are not computer scientists. These are humanities people, so they're having a humanities door conversation about this subject.

Well, it's also– It's customized for the user based on the AI slop. And I must repeat, I don't get AI slop on Facebook. It's either that I have someone in Facebook who is like, we can't let him see this. We can't let him see this. We must mask the slop. Or there's someone in Facebook who's like, my man, I'm going to make sure you have Facebook premium where you don't have to be subjected.

It's like in C.S. Lewis, the demon screw tape. You have your own screw tape demon that's hand sorting your algorithm for you.

But the slop I see is bad. Like, it's just, it's slop slop. Like, it's not like someone has lovingly put together something using these tools. Not to even say that that would be super valid, but it's not. The things I see are not the creation of someone thinking. It's a Grammy who's like, I'm going to make King Kong fight my husband. Yeah.

I think in order for this kind of personalization argument to really have any sway in terms of like the way we should primarily understand slops aesthetic or cultural value is they would have to make the argument that most of the AI slop people consume is curated or created by them with that personalization in mind instead of fed to them through an algorithm like through social media like Facebook or TikTok.

Right.

Like if more people are consuming. uh, fruit love Island, uh, then people are consuming like fan fiction where they get to insert themselves into the hunger games or whatever. Like if that's, if, if it's more fruit love Island consumption, then self insert fanfic, then the personalization argument doesn't really matter.

And it's really a conversation about, like you said earlier, the sheer volume of content that this being blasted onto these platforms is and the algorithmic biases that put this type of content in front of people, unless, Hey, we get to create this thing. I, I don't know if anybody follows, uh, the YouTuber Sean, but at one point he says, this is the only, uh, human created, uh, endeavor that I would deny the title art.

Um, yeah.

And, and I think, I think we're at that point here. Um, but for the purposes of like talking about this content as content, we would have to separate out kind of our ability to personalize it versus what ends up getting fed to us through these systems. And that changes the conversation about the cultural or aesthetic value. Because I think, and this part is not unique to AI slop, the thing that AI has done has, it has turned down the quality, it has turned up the volume that's being produced.

But what it effectively does is it recreates the scenario where people mistake aesthetics for message, right? Or they consume the aesthetic of the food that they want. Right. And so they're very enamored with the aesthetic of, you know, and by aesthetic, I don't just mean like the exact picture. Otherwise, you know, advertising with like gross looking food wouldn't work.

But it looks like what a graphic, what you think a graphic designer will put together for an advertisement of food. This reality show is, has the trappings of what we would call a reality show, which is like a very funny reflection on the fact that reality shows are scripted, right?

Yes.

That they're not really reality shows in the sense that we think of them anymore. Very kind of like a Somali Chrome has consumed itself sort of thing. I'm probably mixing academic terms here. But like it feels like a reality show. And so I'm consuming it because I enjoy the feeling or the genre of reality in the same way that if you were like really into, I don't know, like 10 years ago, if you're really into like YA dystopic fiction, you'd start consuming a bunch of really shitty books because you like the Hunger Games and you like the feeling that you got reading it, a genuinely good series of books. And then you start reading like Divergent or whatever, and it has the trappings of it. And you get, and you remove yourself further and further from

like what, we think of as what art transmits to us and what kind of effect it has and closer to the comfort of the familiarity of the form that we enjoy.

Yeah. And I think the thing is people, the vacuous argument of like, can AI make art? The only way I think I could ever consider AI art is if someone was just They were doing a very, very, very specific thing on device with a real understanding of the underlying weights of the model. I think that's where it really lies. If they actually have a hand over the technology, because if you think of it as a thing with a billion brushes, if you're not the one tweaking the brushes, then who gives a shit? The paper even does make this kind of point that this has the aesthetics of art, of what you expect art to be, But is not art itself. And I think that, I still get back to my point that there are lots of people that

believe that they are denied access to making art because they're incapable of it. When what they actually mean is they don't want to learn. And it may be more difficult for them to learn. Because like, Caleb, you can play music. I can't play music. I'm not sitting here being like, I'm not allowed to play guitar. I'm not allowed to be.

In the secondary of the Seattle Seahawks. It's not fair because I'm a 5'10.

Yeah, yeah, yeah, yeah. I'm not allowed to play quarterback for the Raiders because somehow I'm not good enough. I mean, look at the current fucking guys. It's not a good time, fellas. It's interesting watching and also kind of exhausting watching an academic look at this stuff, because it suggests there's so much more thought behind all of it, as in like the literal creation of the models and the image models themselves, rather than just making things that can resemble other things.

Yeah, I think that one of the problems, and this often happens in my understanding of academia, mind you, I'm a college dropout, so like take what I say with a grain of salt. But one thing that seems to hang up a lot of people in academia is when they take a bunch of arguments like at face value, which is like kind of the whole point of academia is to be a critical thinker and to take a look at underlying systems and what might be motivating things that are not on the surface. But you often get caught up and I think this is exactly speaks to what you were saying earlier about people talking about like– they use the phrase democratizing access to art or whatever.

Yeah.

And it's like, look, I have acquired somehow a bunch of Wes Anderson films. And the really early stuff is low budget. And there was not a ton of barrier to access for Wes Anderson to start putting together some truly remarkable art. And through that, he ended up getting access to more resources and all that. And certainly we're not in a meritocratic system when it comes to rewarding artists.

you know, art, but in the same way that, you know, I strongly believe that there should be democratic access to the tools, the materials that we use to make, you know, some of our greatest works of art. I also think that the people making the argument, they're not actually honestly making the argument. We have to understand that they're not making in good faith, because when you take a look at the people who are like tweeting out or putting together videos or about, like, hey, we're going to replace film studios with AI or whatever, and you see this, like, shitty movie that, like, there's, like, three heads and whatever, you, they never answer the question of what they're attempting to communicate, like,

what message value their film has, right? A lot of times it's just meme slop, right, where they paste together a bunch of memes and and want everyone to jump on reference humor, like we're all living in a Family Guy episode or something. And there's nothing that they're trying to say, and so by virtue of the fact that we know that there's nothing they're trying to say, we also know that their arguments about democratizing art are bullshit.

They don't care about that, because they don't actually care about art, they care about monetizing content.

Yes, they want to make a job making content. That's actually a really good way of putting... Anyone who says democratizing art is just saying, I want to be able to become an influencer.

I want to get paid for pressing buttons on the computer.

Yeah.

I will say this. I think that there's an almost neo-Dadaist quality to some of the Italian brain rot stuff specifically. And if you don't know what that is, God bless you.

There's sort of like... It's like... And all that. When you have just like some horror... some broccoli head kids showing it to a man clearly who's homeless and experiencing mental distress, truly awful videos. Yeah.

But the thing is, is like they, so, so those came where they're born out of like the shrimp Jesus era of like early, you know, stuff that was like, okay, this is just really stupid. And so a lot of what happens is these kids kind of take this and they're making fun of, um, sort of just like the tech industry in general by making these, uh, characters. But to your point, Arif, like the second that they do that, like someone else will just take it and monetize it.

Like you'll see, you know, we went into a sort of genealogy of this with the fine folks of Internet Today, Ricky and Elliot on my show. It was a lot of fun. But the second that, you know, you get some of this stuff that's like, okay, this is something that people are kind of sort of claiming through memes.

Then in like farmer's markets pop-ups, there's like trading cards of Italian brain rounds. Okay. Well, you know, clearly there's the grift is just nonstop.

Yeah. Yeah. And a lot of that as well is like the Italian brain rot memes are mostly just playing on the oldest form of humor, which is funny words. The actual art itself is not really the funny part. And it's frustrating as well because I'm sure there is someone making art with LLMs, but that is not the same as AI art. And they are being drowned out by these freaks, I guess I'd call them, just because...

I don't, like, I'm not even sure I'm comfortable saying that LLMs can make art. Like, actually, I'm taking a step back now because it's so hard for you to have enough, because it gets back to the artist thing. It's for you to be able to do that, you would actually have to be able to train and understand the LLM on a technical level. So you're basically a mathematician, software engineer, what have you, which at this point you could have just learned to paint or something or draw.

It's just basically the same as that now. And it's just the fact that this debate exists pisses me off as well, because even through cutting it up in the way this paper does, it legitimizes it. It treats it on the level of kitsch, because even the shittiest looking reality TV show you've ever seen, even the worst Hallmark movie, still has probably a first and second AD. Definitely has a director of photography, definitely has a producer or two, definitely has runners, the bunch of people on set. even if they're doing it in a way where they don't allow the people to sleep or whatever, like how all reality TV shows work, there is actual real intention going on.

Yeah.

And I think to kind of dig at the existence of what we would call AI-generated art, I think that there are ways for artists to, especially like you mentioned, like someone who's deeply invested in CS and machine learning and modifying the weights and stuff like that, To produce like resistive art or art that like in the same way that pop art used materials that were not often thought of as like artistic materials or depicted subjects that were not, you know, traditional subjects of art and stuff like that. Like there are ways that you can, and I'm thinking of like one of the examples of people like to go all the way back to when conservatives used to complain about modern art a bunch. One of the things that they would make fun of is the urinal.

That was genuinely a very impactful piece of art, specifically because it had opened itself up to the kind of ridicule that's what it was attempting to do about like the nature of art and what we consider art. And you could do something somewhat similar with AI art so long as you do it with like a bunch of intentionality.

And there was an attempt to kind of interrogate the way our obsessions, maybe the wrong word, but the proliferation of AI art. and AI content throughout our day-to-day life. There's something you can do about that that would probably be an interesting art piece, but you have to do it with a lot of intention, and it's always really difficult to do something like that without it being interpreted in a different way, and so it would take a really strong artist to be able to do that effectively.

But it is possible, but I just think that the parameters for something like that are fairly narrow.

Have you guys, either of you, seen... There is a creator on Instagram whose name escapes me, but it started out kind of, as these things go, one person makes it and then a million other copycats or versions of it pop up. But it's like this kid teams up with Tung Tung to destroy the AI bubble. Oh, yeah. Yeah, and then Tung Tung turns into dust like Thanos and he starts crying. It's actually like, you know, it's a joke.

It's literally based on a JoJo's Bizarre Adventure bit when Avdol... dies, I just realized I spoiled JoJo, I'm sorry anyone. But it's like that's literally based on a very specific JoJo's bit. Anyway, yes, I know the thing. but keep going i.

Thought that was very that was just my point it's like that was actually kind of funny and you.

Know that's a reference to something made by a human being where the art is a reference rather.

Than well in that part of that creator specifically also made that you know i mean tongue tongue it wasn't his creation, but he made a video with tongue tongue in it. You know what I mean? There's like, there's like hand drawn animation and his own voice acting that made it good.

Right.

Yeah. And that's the thing. It always comes back to, oh, a human being did something. And people will argue, oh, what are you talking about? Well, a human being should just be able to type a prompt in, and it's just as valid to do that. No, it's not.

Shut the fuck.

Up.

We do not consider the people who commissioned artists during the Renaissance to be the artists who produced those paintings, right? I think that that's a ridiculous argument, to argue that the patron is the artist.

Yeah, like if someone painted my house... I don't think that... I wouldn't be like, I painted my house. I might say I have my house painted. I don't know.

Yeah, and sometimes people will use that language in ways that are generally understood. Like, hey, I renovated my house. I think most people would understand that if I'm not a contractor, they know me to be a contractor. I probably hired contractors to do that. But that's within the confines of how that language is used. But when people say, I made this, and they point to a piece of art or a piece of content, the understanding... is that you are responsible for its creation by having a hand in how it was made, as opposed to commissioning it.

And I think that we should keep that divide. I think we should very much make sure that we don't make this mistake of confusing the patron for the artist.

Well, I think that this is, it gets back to the other things, like, oh, we should tolerate AI writing. Oh, we should tolerate AI coding at scale. We should do all of these things, because it ultimately comes down to These people want to crush the meaning out of creation of anything. They want it to be about the output, only the output, and you must lower your standards for output so that they can shoulder barge themselves in with whatever form of crap they swill. Also, I mean, if you ever need to really just believe that all of this stuff should burn, I recommend going to r slash grok, which is... The worst community of people alive, I would say.

It is people using r slash grok. These members just use grok to generate porno.

Yeah.

And I'm just picking a random headline. I'm just going to read it in the tone. I think it's, tell me exactly what I'm paying 30 bucks a month for. Every damn thing I try is moderated. Every damn thing. And I'm not even doing SFW clips. So tell me exactly what I'm paying $ 30 for. What am I getting? Tell me, what am I getting for $ 30? That is just one sentence. And it's just guys trying to- It's.

Literally an entire community of people helping each other trick the computer into seeing a nipple. I mean, that is the whole subject.

It really, or making the single worst, like, space art you've ever seen. It will be like a Roman soldier flying through space, throwing a spear at a space station. Another great one I'm just going to read is, how is moderation compared to yeaster day? I saw some complain today. It's just like people who can barely speak English having conversations about why they can't generate insanely horrifying images.

I was about to say, I wonder if some of the value is, or some of the joy, I guess is one word, is it comes from getting around the moderation and tricking it to do what you want. And probably that's still the case because it's trivially fairly easy to get to find your way to like a jailbroken text generation for sure, and then probably image generation as well.

But then as you're reading out the headlines, I realized, no, actually, that would be quite the task for these people. They are stuck in grok. They don't know how to find the jailbroken LLM, and that's fine. I'm glad that they're having this trouble.

Also, just another, this is a fairly obvious one, is most artists I've met, are very open to new ideas. Like they're very open to new concepts. They're like, oh, that could be interesting. It's why, by the way, I have some affordance for the weird Natasha Lyonne thing that she said about, rip it off Casey Kagawa, a friend of the show, where she said that David Lynch liked AI.

It's like, yeah, she probably was like, what do you think of AI? Went, ah, it sounds interesting. And they hadn't actually used it. And I think most artists are curious.

Jesus was like, okay.

I've dreamed of this guy. And the thing is, but past the immediate point of hearing about it, I've never met an artist who's used AI and then gone, oh, wow. I've met a lot of artists who have said, I am horrified. Like, just like, I find this genuinely, the things it creates are horrible to look at. Because there's this weird kind of like, I realize tinny doesn't sound right for something you're looking at, but it really just has this weird- No, that's good. Echoing feeling to it.

I've just never been able to get my head around. But it's like my mate Jeremy Willis out in Greenpoint. He's a great artist. The idea of AI seems to offend him. Just like every artist I talk to, it offends. Like Arthur Adams, a great comic book artist. Artists never seen him pick up AI. In fact, I've yet to meet an artist who's picked it up. Like just, just straight up. No working artist I've ever met has been like, I'm going to adopt AI into my workflows, let alone into my artwork. And it's just, it's stark.

There's Darren Aronofsky who made that.

I don't know if you guys saw that. Yeah, I'm saying artists. Yeah. But it's also worth noting, however.

That like you, he did that, but then you look back at his career and like shot for shot, so much of it is stolen from like perfect blue and other Japanese animations and stuff like that. It's like, okay, like guys like Aronofsky who I used to, unfortunately, there's a couple of movies that I like. And it's like, oh, this actually makes sense. This guy just steals stuff that he would like this kind of thing.

Well, also, have you seen the clips of the Darren Aronofsky, George Washington stuff? Yes. It's, it is. Oh my God. I am going to link to this in the YouTube. The YouTube link. It's so fucking funny just watching like when they're handing out. I didn't learn American history. Sorry. The thing that started the American Revolution. There's a name of a document. I don't know. But anyway, if you're looking at the document when they're handing out to people, the text changes as they hand it. The shape of George Washington. It's great. And it's just you watch it and you're like, what I don't get is how they publish this. like how they looked at this and went fucking banger ship.

Somebody was like, what if we did the American revolution, but it gave you the same feeling as having a dream when you're punching underwater for 30 minutes.

What, what if you want something that made you feel like you were telling a lie? Just this really horrible. And it looked, you'll notice that no shot stays for more than a few seconds. There was that. I mentioned this in the previous episode, there was a, a film that was made with AI that they said showed at Cannes and people were like, oh, it was at the Cannes Film Festival. It was like, no, it wasn't. It was aired outside of the film festival.

They aired it in the city during the film festival.

Which is great because all the media fell for it. They fell for it because no one thinks, but that's the AI bubble at this point. It's just like misreading headlines and hoping no one notices. Yeah. you back to the paper this isn't even an elitist thing i actually have respect for people who can put out shit that sucks again and again and they get iteratively better and i'm talking about myself if you go and read my 2019 stuff it was rough Was not great.

I thought this was about Tommy Wiseau.

Tommy Wiseau has only got better with time. But it's also watching directors get better. Watching someone find the way to articulate their ideas is kind of the beauty of art. There was someone who, I'm sure it was ripped off of TikTok, but there was someone who posted on Twitter this video of a guy who just kept drawing every year.

And it started out with his art looking like shit. Like it just, it didn't. But over the years, you saw it tightening up, tightening up, tightening up. And the problem is, I think that, to be clear, no one enjoys that. No one enjoys getting better at anything. Perhaps you do when you get your head into it, when you get a bit of the momentum under you. But yeah, getting better at stuff sucks, especially when you don't take to it naturally. And I mean, human beings will take any friction they can reduce. But this is, the friction that's being reduced here is, the art like just the actual the actual thing like uh.

Yeah accomplishment that you get from art is like overcoming the fact that you suck at it right we're talking about music earlier it's like when i make something you know what i always really love is like because i do all the uh music for my show and i'll throw up outro music and stuff like that when i've written some something and i love when people ask me like oh how did you do this thing and it's like well that's gives me a sense of accomplishment because it was the process of learning how to do something I previously didn't know how to do and the strife that it took to get there, to learn a new skill or a new subset of skills or whatever, right?

That's the whole point. If you take that out of it, what's left? You know what I mean? Say whatever you will to bring this back to a football thing. Sorry, everybody. But Russell Wilson had that quote where it's you got to love the process and say what you will about him, but he was cooking with that.

No, that's true.

One of the– so, Ed, when you were talking about how like there were no artists that you know that have like embraced AI, I was thinking of exactly a perfect example of this that separates someone who's a content creator from an artist. Caleb, you probably know this name. You know Shadiversity, the YouTuber?

I do.

Yeah.

Uh, so he, there was like, don't worry about it. He's, he's, yeah, he's, uh, he, he, he arrived too late.

In the YouTube world for Gamergate, but he would have been a Gamergator. He's essentially that now, but for like wokeness and DEI in like Mario, uh, Um, oh good. But, uh, he was originally just like this, I accurately described him. Uh, he originally was like meant to be an educational YouTuber that like taught you how like sword fighting worked. It turned out like all of it was BS, but, uh, he gained, he gained a lot of, uh, a huge following out of, out of doing content around, uh, medieval realism.

Um, And he.

Started posting pretty recently after he started ranting about like wokeness and DEI and stuff. He started posting AI art that he had commissioned, right, that he had had mid-journey generate. And a bunch of people were appropriately mad. And he was like, well, actually, you guys don't understand. I'm a trained artist. And the really funny thing is his brother, it turns out, I didn't know this and I wish I remembered his name. is somehow an even more popular YouTuber than him with, like, 2 million subscribers who is a trained artist and is a substantially better artist than him.

And they don't talk to each other, which is even funnier.

Yeah.

And... Go on.

Yeah, and so, like, he describes why him being an artist allows him to improve the AI generation. So what he does is he asks it to generate a really sexy picture of, like, Supergirl.

Uh-huh.

And, and then this picture with like two different sons in the background, but on earth of super gold gets generated.

She has an 11 pack. Yeah, exactly.

And it's a very bad picture. And he's like, well, as an artist, I understand proportions. So he puts it into Photoshop, screws around with it, uploads it back to mid journey and was like, Hey, so I did a bunch of shit in Photoshop. Can you incorporate my changes? And then mid journey does it and makes it worse again.

And he's like, yeah, I go through this process. It takes me hours and hours and hours to do it. I put a lot of time into it. And it's my artist training that allows me to produce this. And it's just the shittiest image.

In the world.

Yeah.

And like real artists, there's a bunch of really great YouTube videos about real artists, like breaking down what he does, what his misunderstandings are. and why that doesn't make sense. And also, you're in mid-journey. Just ask it to make the changes that you're spending hours in Photoshop. You're doing this to show off that you're spending time, which ruins the point of this AI shit anyway.

The craziest thing.

But yeah, a content creator obsessed with creative content loves AI. His artist brother, who he is not on speaking terms with, despises it.

Well, this reminds me of something that goes to a... It's a greater term that I found from this guy called Rob Sloan. I don't know who this guy is. He seems like a decent guy, a writer maybe, but he has this term called Jarvising, which refers to how AI is really good at creating a thing that you work on, like a tooling system that you're constantly dicking around with, you're constantly tweaking, you're constantly messing with, because, and this is a paraphrase of Atomic Habits, which, I mean, pretty good book, kind of a dickhead, the guy, but makes this point of, There is a way to spend a lot of time doing nothing, but feeling real productive.

And I think these models really give it to you, especially with like nano banana, one of the worst things to ever this, just the worst names of anything in that you can be like computer, make the booby bigger computer, make the eyes red computer, give her a sword. And you feel like you're doing something, even though you are not like you could not make that move yourself. And indeed you are not actually doing, intentionally tweaking exactly what you're talking about because, I don't know, describing what you're doing in words is not the same as actually physically doing something. And people will say, oh, there are all sorts of automations. We don't totally know what's happening in the background. Art isn't one of those things where that's okay, though. Like,

art's not something where, like, not really knowing why you're doing stuff. That's not really part of the picture, at least when it's just When the actual process of doing... Because to be clear, yeah, there are people who probably blindfolded themselves and painted. Your hand is still fucking moving.

Or they throw something down a wall or something, yeah.

I think for a lot of these, like, human endeavors that are so difficult and complex to pick up that by the time you see the finished product, it's very opaque how to get there.

I think.

And obviously, I'm, like, very football-minded, but, like... Actually, film may be a better example. There are a lot of directors who... will break the traditional rules of film that you might get taught in your first year at film school about composition and the rule of thirds and guiding lines and stuff like that. And when you watch one of these directors, one of these outstanding directors, they'll break those rules, right? And somebody who's not invested in or ensconced in that world of film analysis or is an artist...

might conclude, oh, the rules don't matter, when really what we've learned is that a lot of these directors only break these rules because they know what the rule is there for, why they learned it, and how they spent a lot of time.

Protecting their mastery of it. You have to understand the box to think outside in order to work outside of it.

Exactly.

It's like a musician playing accidental notes or whatever, right?

Yeah, yeah, yeah. Half of what... Half of my bit is that I do deep financial analysis, but I say fucking shit. And I, I speak in it all. I say no, it loads refuse, come dump and stuff like that. Like you generally don't, you generally don't say stuff. I know it's so funny. They gave me a show, but it's just it, but it's good because it's good because I'm working within the boundaries. Cause it's not entirely me saying stuff like that. It's just those pop up and you go, oh, what a surprise. And then we go back to the analysis. It's the different way of doing still having a purpose. I mean, Lynch, a couple months ago, I watched Twin Peaks to Return. Tons of things that break the rules in that, but they work because they

were done with intention. He actually knew why he was doing them. Every bit was done with intention.

Yeah, and you know what the impact is of breaking this particular rule in this particular way. You know it even intuitively when you're a masterful enough whatever, right? And the example in football I was going to use is these quarterbacks that break traditional rules of quarterbacking do so because...

They know what those rules are. They can accomplish things within the lines and then they go outside of it. Like Patrick Mahomes is an extremely good traditional quarterback. Sorry, I know both of you hate it.

Yeah, this is still going to be great for Caleb.

But the way that he breaks the traditional rules of quarterbacking infuriates people who teach high schoolers how to play football but works for him because he can play really well within the confines of the system. Then he breaks those rules. Because it'll work in that particular moment to accomplish a specific goal that he wouldn't have been able to do within traditional football. But just like a bunch of kids learned how to shoot 26-foot threes from Steph Curry or scramble and do insane stuff from Patrick Mahomes, it's not actually helping you to try and replicate that end process of the rule-breaking. It is much better to learn what those rules are so that when you have the capability to break those rules, you know what you're doing,

what you're sacrificing, and what you're accomplishing.

Yeah. And I mean, people really don't like the sports stuff, so I'm going to keep going. Yeah.

I didn't say anything because not the time, right?

Oh, I don't give a shit. Oh, these people will take what I give them. I love them, but they will enjoy my show on my terms. But Steph Curry, for example, he's not just shooting threes arbitrarily from the middle of the court. He has a proficient understanding of how defensive formations might look. He understands what his own body can do. There is a fundamental understanding of the structure of humanity and human beings and basketball players and the rules of the NBA and so on and so forth that allows him to break those rules. These LLMs, when creating anything, art, words, whatever, don't understand rules. They may understand consistencies within stuff.

Yeah, they understand relationships as they've been, like, historically defined through averages of text, yeah.

Yeah, but they don't understand why decisions are made, just that the decisions have been made, which I think is significant and also is what ultimately makes creating anything frustrating. Like, I think that the frustrated... Like, for example, when I began this show, I could not read a balance sheet.

I had never learned it. Every time I looked at one, I felt like my head was full of sand. And I eventually just, with the help of Matt Hughes, broke through it. And actually, being able to understand these nuances and not forget... a traditional finance background actually allowed me to do some really interesting shit because I don't come into it with the predispositions of the system. Nevertheless, the process of learning that stuff was what made me good.

It wasn't anything to do with anything other than having understood the system, built my own system, and then doing things with that knowledge and the process. This is kind of a fucked up point, I realize, but it's very much about actually doing stuff to get to a point. And to be clear, I get it. Learning is fucking annoying.

It's not fun sometimes, especially when you don't understand something. It's very frustrating when you sit there and you look at a concept, you're like, what do we do? I don't get it. And smashing your head against it sucks. The same with making your annoying meat suit do stuff. Make it do thing that you want it to do.

When you can't make it do the thing that feels so crystal clear in your head. But that is creation, Jesus Christ. Yeah. Yeah. Well, you know what? I think that's actually a good point to wrap us up at. Where can people find you fine fellas?

You can find me on Twitter at Arif Hassan NFL. On BlueSky at Arif.BSky.Social. And my newsletter, WideLeft.Football, where I talk about– sometimes I talk about football, but a lot of times I talk about this shit.

And Caleb?

You can find me at birdrespecter.bsky.social or killthecomputer.com if you're interested in sort of like more of the first half of the discussion we were having. That's kind of what we do over there.

And Kill the Computer was, of course, the actual progenitor of OpenAI. Of OpenAI, Jesus Christ. Of Better Offline. They're going to have my ass in the comments for that. It's been a long week, fellas. It's actually genuinely Robert Evans heard me on Kill the Computer. And that's how he found me. And not OpenAI, Jesus Christ. They're going to really get me for this.

If we created OpenAI, they would be impressed.

You created OpenAI.

Yeah.

No, they really would have our ass for that. Everyone, thank you so much for listening. I'll be back with the monologue this week, of course. Thank you for listening to Better Offline. I'm Ed Zitron. Good night. Thank you for listening to Better Offline.

The editor and composer of the Better Offline theme song is Matt Osowski. You can check out more of his music and audio projects at mattosowski.com.

M-A-T-T-O-S-O-W-S-K-I dot com.

You can email me at ez at betteroffline.com or visit betteroffline.com to find more podcast links and, of course, my newsletter. I also really recommend you go to chat.wheresyoured.at to visit the Discord and go to r.

Slash betteroffline to check out our Reddit. Thank you so much for listening. Better Offline is a production of Cool Zone Media. For more from Cool Zone Media, visit our website, coolzonemedia.com, or check us out on the iHeartRadio app, Apple Podcasts, or wherever you get your podcasts.

LLMs Will Never Make Art With Caleb Wilson and Arif Hasan

Kapitelklicken springt direkt hin

Shownotes

Transkriptvom Podcast bereitgestellt · Zeitmarken springen hin