Opinionated software: AI and the arts, revisited

33:42And then the second part is just concentration. You know, when you do have such a small group of actors who hold all the keys and they understand the tools better than anybody in the position of authority or a position of informing the public, you also have this risk, particularly with the legal concerns of concentration. And so that's why ultimately, I mean, we've put a bunch of time into trying to support open models, you know, because I mean, to use a more down to earth example, like if you saw what ended up happening with the consolidation to streaming and music, where a bunch of people got together to conspire to make an economic model that worked in a particular way, you know, and it's, and the streaming model is very clever, actually.

34:21It's quite an elegant model, but the results of it are, you know, something I've been quite critical of and I don't think has been necessarily great for culture. I have a concern that that will happen again. You know, I think that absent a strong ecosystem of open models, there's not going to be a lot of economic self-determination in terms of how people make a living from this next substrate, how people get to experiment with this next substrate. And that's concerning to me. But I think number one really is this kind of latency engulf between public information and elected people and what happens in the industry because it moves so quickly.

34:56And if this is just something you casually follow or you don't properly understand it, I don't think you're a player, you know, sadly.

35:06If you like what you're hearing on this episode, there's a place where you can join the conversation. It's called Cujoplex, and it's a private online hangout zone on Discord where smart, cool weirdos just like you, you know, people who like talking about the evolving state of independent culture, can congregate, share links, and talk about the news of the day. We've had some great conversations lately about creative economies, media theory, AI, politics, the future of journalism, and even the records and TV shows we've been obsessing over this year. Some of our chats have even inspired episode ideas.

35:41You can get access to Kujoplex along with our monthly bonus episodes by signing up for a paid subscription over at our Substack. That's theculturejournalist.substack.com. Shifting over to perhaps more frivolous aesthetic questions. What do you make of like the so-called slop problem in culture? Is it as big of a problem for culture as people seem to be making it out to be? Where do you fall on the slop question? I definitely, I mean, I have a different definition of sloptimism than my friend Dan Keller. I welcome it because I guess as a critic of the attention economy, I want for those systems to be overwhelmed.

36:27because ultimately the idea there is I really don't care if a bot or someone acting like a bot produced bait for YouTube. I could not care less. And ideally, I would hope in those conditions, there will be a flight to quality where what ultimately is important, at least from my media diet, is that I know where it comes from. I have some context about the person who made it, right? And that has been a problem with the attention economy for God knows how long. So in my mind, And I'm like, cool. Okay. So these structures are going to be overwhelmed that people are going to get impatient and then they're going to look to, and I think this has already happened to some extent, you know, there's issues too with sub stack or whatever, but I'm like, I've noticed a huge uptick.

37:09The sloppier Twitter has become, the more people are being pushed into this direction where they actually want to know where the words come from. And I'm like, that seems cool to me. Right. Holly got into a bit of trouble with her bank and comments, which by the way, she was completely correct about. But you've seen a similar pushback where I might find a lot of the anti-AI kind of sentiment in that world a little hyperbolic and kind of unsophisticated. But if it pushes more people toward bands on Bandcamp, then that's great, right? I'm actually very supportive of that as an outcome. And that's kind of happening.

37:38So I think, if anything, actually, those kind of independent structures where the value of the work, whether it's a piece of writing or a piece of music or a piece of art, is entirely qualified by its context, right? Like, that stuff's golden. like that becomes even rarer like the more content slop whatever you want to call it there is that stuff becomes this like precious jewel that has to be sought out and becomes super valuable you know so i if anything i'm less concerned about those things and i'm actually more interested in thinking okay well how do you boost that deviation more so slop i mean you know whatever but it's also a judgment call it reminds me of like hipster or something right where it means something you don't like in particular and like at the same time i don't see it as necessarily being a bad thing that like when you look at the state of the film industry or whatever I don't think it's necessarily a bad thing that individuals or small groups of people have the tools to make interesting things now like people are going to make really interesting stuff with this that's inevitable and so kind of like the hipster slur or whatever like it's slop until it isn't and there's going to be all kind of things that people make and they're like oh yeah but I like this one thus invalidating the binary distinction the truth of the matter is I mean this is closer to to our universe is like we use the hell out of these models but like we have authorship we orchestrate them it lets us do stuff we couldn't do before it's still really hard to make anything good and i think that's where it's going to go i'm i'm just interested in how people use them you know and i think that there's lots of opportunities there that could be very cool but of course yeah if people are expecting media interactions to be the same as they were 15 years ago then i don't know what to say i just don't think that's where it's going but you see it as potentially just an accelerant on things that were already kind of like and shitifying in the first place yeah yeah exactly but again i think that that flight quality already kind of happened right i mean like what are we doing right now right like we're having a kind of parasocial like exchange between people who come from specific contexts and have a track record of doing stuff and that hopefully makes this exchange valuable to somebody in a way that giving notes to notebook lm or whatever to automate a podcast is different you know and like and i think that i mean we we wrote in the book it's like we too easily confuse media for art you know and it's like yeah like you can fake an mp3 now or like you can generate an article or whatever and it's like that was kind of like a part of the art like a pretty critical part of the art but that's not the art right so i don't i don't think that these kind of exchanges or timely context or a gesture using whatever material or whatever is going to lose its value.

40:10But I do think that 100 years of a particular media industry is imperiled by the fact that everyone will be able to automate stuff that looks kind of like stuff that was hard to make 20 years ago. Does that, I think I've seen you speak about this a bit, but does the fact of being able to automate things that used to be difficult to create change what we value in art or in an artist or what it takes for an artist to succeed, do you think? A little bit. I mean, I don't want to be too relativistic about it but it depends what you mean by artist right like there's a hundred years of contemporary art that say the artist gesture can be whatever right because it's basically just a signature right so like post Duchamp or whatever you're like okay I choose to do this at this moment in time or go it is valuable or you get to debate if it's valuable or not there's that kind of art which is you know a lot of people push back against because contemporary art is a dirty word for some people but like I find that compelling and I think there's actually a lot of lessons in contemporary art as to where we're going because it's all about intersubjective consensus i mean that's the whole point of it is like a bunch of alleged experts get together and say this is good this is bad and then some of it gets canonized and valued by markets and such and institutions collect it and you know i'm like well that is all kind of this strange context game you know that has nothing to do with the the piece of paper that's being exchanged or whatever you know and then there's art as in making records but but as i said yeah like the vast majority of records or records are trash that you never want to listen to.

41:33And it doesn't matter if human made it or whatever, or it matters to them, but to you from your personal subjectivity, most stuff you don't want to listen to, right? You need a reason to, you need a context. And it's like, I think that there's plenty of room for in a particular context, someone to use these tools in adjacency with other stuff that they're doing to captivate you. And I think that that's going to happen. I mean, and then the other side too, which is also, I think from my interest, particularly interesting is that of course, this is happening on the code side too, right? There's a guy called Jeremy Howard.

42:03He's an Australian software developer I was listening to. There's a great podcast. Actually, I can recommend there's a podcast called machine learning street talk. If anyone wants a good podcast, listen to about ML AI stuff. It's great because basically now, I mean, outside of high level engineering, writing software is more or less automatable. And he was creating this distinction where he was like, he doesn't believe agents can currently engineer, but in terms of actually manually writing the code he's like yeah they're awesome and anthropic famously i mean they would because they're kind of pushing called code which is unbelievable by the way anthropic famously is saying you know we're hiring the last software engineer at the company you know kind of as a news story but they they mean it and i'm like well this is interesting because by my estimation i think that the opportunity for artists generally was to publish to the app store that's like the most insane development in the last 20 years being like okay you don't have to express yourself through these 20th century mediums, like you can express yourself through a network with literal software.

42:58I mean, wow, that's a new thing that we can do now. But of course there's a huge technical leap to make for most people in order to do that. And I'm like, well, that's going very quickly, right? And of course the vast majority of software that people vibe code or whatever is fucking terrible. But that doesn't mean that a really smart person who is like, hey, I have this crazy, you know, psychedelic project that I want to do that is this all-encompassing world-building art project that happens to be deployable through whatever, you can build that now. So for my purposes, at least my very kind of wide-open kind of definition of art, which definitely incorporates opinion-aided software, there's all kind of opportunities to express yourself now in ways that would have been very, very difficult before.

43:41And that gets me pretty excited. And so I can say that. I mean, like, we're doing projects at the moment where I'm like, fuck, like, I kind of half-made something today with Claude that would have been like a six month research project. And like, I went outside for a cigarette and it was done when I got back. Yeah, no, it's fucked up. Like my best engineer friends like have a whole orchestra of agents now where they're basically walking with their daughter. And the thing is, the problem still persists that if you come in there and say, build me something better than Facebook, it's gonna suck, right?

44:15But if you have all of the ideas and you orchestrate it like an artist would or like a director or whatever. And you engineer this system where you're like, no, I thought about everything here. And you build that schematic and you figure out how to do it. I mean, there are still some things to consider, but you can get very close with these tools. And for what, like 40 bucks a month or whatever? I mean, come on, like this is crazy. So I mean, that to me as well, in terms of progressive politics, wherever you fall on this, I'm like, how do you feel about building your own stuff? Like that's fucking crazy.

44:46That used to cost a lot of money. Yeah, we had a little episode recently. We were talking about Yuri Ryback and his vibe coding journeys and the idea of like, oh, people can actually build these alternate internets now. We have a lot of friends who seem to be tinkering with this. What I will say, though, is that like the majority, for people like us, it's still a fringe pursuit. Of course, but it'll change. I think what will likely happen from this too is that, again, talking about people being dissatisfied with the attention economy, it's like these models are going to run on your phone pretty soon.

45:18The possibility of offline experience being augmented by this stuff is really enticing to me. Like talk about a golden age of like hardware or whatever, where I'm like, okay, I want minimum viable intelligence and networking capability on my person to do wild stuff and I don't want to sit at a desk anymore. That sounds liberating, honestly. That would be cool. Yeah. And honestly, that's not a million miles away from where I anticipate OpenAI is taking this, whatever you're thinking about them. But I'm like, honestly. You mean with their collaboration with, what's his name? Joni Ive. Yeah. Yeah.

45:47Their new, what's it called? Like, is it screen free? Yeah. I don't remember what. Yeah. But I mean, that's where it's going. And I'm like, is that worse? Okay. There's privacy considerations and all kinds of things. Like, of course, there's all these considerations there, but I'm like, that sounds good. I mean, like, what does that look like? You know, self-driving is apparently working now. I haven't tried it myself, but there's all these urban questions to ask, being like, what if you didn't have to be at your desk all day? Does that mean there's going to be more demand for art and music because people are out more?

46:18Does that mean cities change because you don't need a car? I don't know, but there's all these interesting questions, I think, where the answer isn't necessarily the worst thing in the world. And sometimes I think the scale of that realization when there's a lot of half-truths, and of course there's a lot of marketing language. but like I spoke to this like biotech person recently who's like very experienced I'm like they're not lying about medical therapies they really genuinely aren't I just started a fellowship at Oxford and I was speaking to the guy who runs it congratulations thank you very much yeah I know yeah my mom thinks I'm really posh now I still need to go I need to actually go but one of the people there was explaining how a friend of his who works in like life sciences two years ago looked at AlphaFold, you know, the protein folding system that DeepMind did and was like, oh yeah, that's cute.

47:06But this is kind of like low hanging fruit, whatever. And he's like, recently he told him that he's actually thinking about changing his profession, not because he thinks that he wouldn't have a role to play, but because AlphaFold and the work that they're doing is so sophisticated that he doesn't think his career would be intellectually rigorous enough to keep pursuing it. So he's interested when he's still young to change. And I'm like, that's a signal. That's not me like believing marketing hype. That's someone who knows someone who's at a very sophisticated level. Who's like, no, no, no, this is real.

47:34He's like, I don't want my life to be just inputting data for this model. AlphaFold just does it. And as a result, I'm going to think of something else I want to do. And I'm like, okay, that's a, that's a big signal. Like that's not bullshit. Right? Like I want personalized cancer therapies, please. I'm sure it's really complicated. It's been in the news this week, right? It's more complicated than a one-liner, but like, yeah, let's go. That sounds good. Let's get that rolling, right? Like, that sounds good to me, right? And I understand, like, you can get so jaded and so tired of constant hype, but a lot of it's kind of qualified, you know?

48:05And maybe it's coming from the wrong mouth or something. But I'm like, what if, to the best of my knowledge at least, a great deal of this is somewhat true? Like, okay, well then, I mean, again, people in that world talk about new social contracts, but they're really poor at publishing what they think the world should look like. Like, okay, there's room to step in here. Like, who's going to be the one who's like, Like, yeah, like free cancer therapies for everyone. That sounds good. That would be cool. It would be, you know, and I don't want to sound hyperbolic there, but I just think that like, yeah, some of this stuff is actually kind of amazing, you know?

48:34And I'm like excited by it. I'm like, we're in a new century. Like, let's go, you know, and there's going to be problems, you know, and I might not have a job, which sucks. Maybe, I don't know. But like, yeah, but it's happening. Yeah. So we actually touched on this a moment ago about the rise of kind of like more analog stuff in response to this. We're already seeing a fetishization of the organic and the analog and the Luddite in certain areas of culture. The return of, you know, cassettes and vinyl and all that stuff. But even now you see sort of Gen Z, it's popular with flip phones. You know, zines are really making a comeback.

49:08Do you think that's just going to be the case forever as we see AI encroach more and more and make the baseline of culture more abundant? Or what happens next? Maybe, yeah. I mean, I don't know. I mean, in a sense, like, if you think about the record industry or whatever being this blip, it was a pretty significant blip, right? Where you had this kind of purple patch in the, let's say, the mid to late 20th century where you have this explosion of subcultures and so on and so forth. It's pretty hard to compete with that, right? So, like, you could see nostalgia, like a perma nostalgia for that stuff, for sure.

49:41I personally don't feel it a little because I, you know, I mean, I'm 41 now, right? Like, I had that stuff when I was younger and I love it. I try to think back at like the artists or the thinkers that I admire. And I think, you know, there's a reason that like Stochasm wasn't playing chamber music. You know what I'm saying? Like there are new challenges. And I think some of the responses to those challenges could be more analog, as I was saying before. I mean, I think the idea of getting off a screen and finding different ways to coordinate yourself with other people, somewhat enabled by this tech is really fascinating, you know?

50:15But yeah, but some of the Luddite turn, I mean, people do what they want. That's cool. I'd like a lot of that stuff but for me I just don't find it all that interesting I actually feel a little bit kind of suffocated by the 20th century where I'm just sick of it like I genuinely I'm just like there's a certain period of time where like there's an ambient revival every year for fucking 20 years and I'm like well no because it's true and I mean like with all the love in the world to like Apex Twin or whatever you know like Guy made some great records like I shouldn't use his name I got into trouble using his name before but I think he's a perfect encapsulation I saw this where like the internet flattened time in such a way that like you'll never be able to compete with people who were famous in the 90s.

50:55They're just perma-famous. Like the meme of things that happened in a time where there was this concentration on these big cultural moments and that forged these identities and deep resonances with people like millennials. Like that just stuff like repeats over and over and over again. I'm like look I like some of that stuff. Like I respect some of those people but like I just don't care anymore. Like I don't want that. But I can also understand if you're 21 and like you know you're oversaturated with TikTok or whatever I could see why that's very appealing because it carves out you know a different way of living but I went through that when I was younger so now I'm just kind of like not I'm not looking for that like it's not the high mark of credibility to me like how much something resembles like a punk zine or whatever you know like that's not like it's cool I love that stuff too right like I have a history of that stuff but it's not what I'm searching for I'm like what's what's happening now so I don't think any of it's illegitimate but i think some of it like there's a lot of very easy culture stories that people will tell yes of being like you know the kids of course the kids know better they're rejecting this ai crap and they're getting back into zines like i did when i was a tina and you're like i'm like i think it's more complicated than that you know like yeah it's a very like binary way of looking at what's happening yeah i mean i think like the greatest rebellion of zoomers is that they don't care like that's the overwhelming feeling is i'm like you'll talk to people who are younger and you're like oh like what about discord records and they're like what is that why would i care about it you're like oh i'm like actually yes that is the most offensive you could be to a millennial sensibility is just being like okay like whatever i have my own preoccupations that matter to me and my friends you know like i don't need i don't need for young people to like validate my tastes from 15 years ago you know like i don't know i mean maybe i'm being too kind of dismissive or something but i just i think a lot of that can also be kind of cope or like i'm skeptical about how popular that stuff actually is you know because like a lot of the mechanisms we might use i'm being presumptuous of you now but i might use is like i still read the press and i'm like and then the press will tell me oh the kids are all right they're into this thing that the press understands and i'm like are they really because i don't know like when i walk around like i don't see people like they have their own fucking thing they're doing and it's like right and like and it's not my thing and it's their thing and i'm like okay and they're probably cool i don't know yeah so i i i think this luddite turn is kind of like blanky for millennials you know it's like comfort where it's like oh no things will be okay every and it's like i don't think that that's true i think that like a lot of that stuff is very special and the meme of it is very strong because it was a very special time and i have admired it and i lament in some ways that a lot of the special things about it have gone but i'm just not naive to the fact I think I think everything changed I think everything really changed permanently and I think grappling with that is more intellectually profiling for me than yeah you know I will say I think what what you're saying about it being kind of a blankie is definitely real yeah and I do wonder if some of it maybe more recent iterations of it like I'm talking about even the past year is kind of maybe a response to both the attention economy like a preference towards listening to a full album, like a CD versus just like a stream.

54:07And here's what the algo is throwing at you. And also, I think security, privacy, anonymity concerns is that's part of why you're seeing this uptick in a re-embrace of analog. That makes sense. I think it might be interesting to see how that kind of then with those who are more progressive thinking about what's happening in tech, how elements of that analog might get incorporated back into the tech for that reason. I think that's why some people are getting back into thumb drives, for instance. Yeah. Totally. Yeah, no, I think that makes a lot of sense. And it's like, yeah, I mean, I taught for a while at NYU, a lot of very young people.

54:40And like, yeah, getting a bit of a window into their world was an eye opener for that, where I'm like, yeah, I can, it's totally rational. If you kind of grown up in the status games of social media, like heaven knows what that would have been like as a young person, then yeah, figuring out ways to live your life offline with some discretion makes a whole bunch of sense. And all I'll say with that is, as we talked about before, like, I think that that's incredibly compatible with these incredibly powerful models. I saw earlier someone put together like a, you know, they're running like a deep seek model or something like this on a piece of hardware.

55:13And it's like a survival kit where he's like, you know, I don't need the internet. Like I can go into the wilderness and learn how to, you know, this thing's solar powered. I can, I'm like, okay, like, cool. And I think that's, that's the more interesting question on the AI thing is that there's all these assumptions kind of baked into what like the WALL-E future looks like. you know where it's like again like a millennial preoccupation i think of like scary like the obvious kind of low-hanging fruit narrative is we're all just going to be like drones kind of watching personalized slop at our screens and i'm like i think maybe in 30 years time that's actually what people will look at the attention economy as that's the best possible outcome is they'll look back in 30 years time and be like oh god that was that was humiliating like remember when And in order to like get anyone to care about your music, you had to like post thirst traps online or like how humiliating was that, you know, that weird period in history.

56:05And I'm like, yeah, like, but there's room for AI models in that future. In fact, I think that like AI models more likely move us toward that future of people just being like, oh, I don't need to go on Twitter to stay up to date with things. Like I, you know, me and my agent can navigate a world of information together and I'm fully networked with my friends and whatever. Like there's so much that can be done proactively there that is privacy preserving and kind of pro-social, you know. I'm far more interested in that future than, you know, Matrix kitsch or whatever, you know. Another interesting point around this, a few episodes ago we had on W.

56:39David Marks, culture critic and theorist. And he raised an interesting point that the things that are more analog or quote unquote like organic may be increasingly becoming classed. You know, like for the people, like it's almost like the equivalent of like access to like organic produce. Like real photography is something that will only be for like the realm of those who can afford to not be on the Internet all the time. No, I think that's true. I mean, I've made a similar point where it's like it's really interesting to see where people in the Valley send their kids to school, you know. and it's like really commonly hands-on you know in-person tactile education very very cautious about giving them too much access to a lot of the tools that they themselves are developing there was like a video clip recently circulating with Peter Thiel basically conceding this you know where he was like hey you know a kid like max like maybe an hour and a half a week on the screen you know like so I mean I think I think there's a there's a real clue in that and it's true I mean like it's always been a class dynamic though I mean I remember there was like a mini story popping up about like you know discrete instagrams people who only post every now and then and then delete something from instagram as like a class marker because it right yeah that's that you don't need to go and sell yourself you know and i'm like one that's existed since the beginning of instagram you know i mean like being in the arts you see all strata right and like i know people who very wealthy obviously and pursue the arts and and know people who have to you know kind of hustle and it's like that's always been a debate that's always been kind of a a class dynamic is that, yeah, obviously if you're more secure, you don't need to go and debase yourself in whatever way, right?

58:12Which is also why, as much as I say, as I'm critical of the attention economy, it's not like a judgment thing. I get it. But yeah, I think that the prediction is useful insofar as making the point about class, which I do actually think is true, but I'm not sure I agree with the prediction. I think that if anything, the more interesting question to ask is how to push forward that vision of using these tools to you know increase people's time and engagement with the real world or the material world let's say and so you know in my mind I mean for example there's been an open consultation in the UK specifically around the AI data question like my advice god knows if anyone will take it was okay well like make some concessions on the data front but then figure out a way to apply a levy to one reduce the cost of arts education to near zero because that's actually the biggest, like the sunk costs of arts education are actually the biggest pain point for people.

59:06Like when we talked earlier about what someone's data is worth, like a data licensing market for someone who makes 70, they might be great pictures, but a model doesn't care. They're just tokens. The licensing market to that might be worth 70 bucks if you're lucky. It'll be worth less than that probably. But your sunk costs of what, like up to a hundred grand, two hundred grand, I don't know what it is in the States currently, of pursuing an arts education that isn't going to deliver for you in the market that's the pain point and so it's like okay well like make make concessions by all means for on this data question because ultimately you're going to have to because it's a competitive field and countries can't disadvantage themselves in in contrast to other but artist groups could then lobby and say well our two biggest costs are education and space to make art if you're going to make this fully automated future possible fine whatever but we want rehearsal spaces venues you know like paint the positive vision of this where it's like okay we're going to have an off-screen life where I want to go to a place and I want you know I want there to be 24-hour art whatever we're working a lot with Hans-Orik Obrist and he mentions Gilbert and George always said that museums make no sense because they close when people get off work yeah which is like such a brilliant point totally you know think about that all the time yeah yeah it makes it it makes it such a brilliant point and it's like well okay well let's just assume for a second let's just say okay all the bad things are going to happen what's the positive that can come out of this push for a screen-free future push for you know universal basic services to minimize the sunk cost of living a creative life that's that's somewhere to push because i think you actually might have some agency on that and that's actually speaking directly to to the places where people hurt because we had i mean we've lived through an explosion like elite overproduction right an explosion of degrees you know people giving loans for any fucking thing yeah that was the whole progressive push for the past 30 years or whatever and it's like that's a big part of the story you know so it's like okay well i'd be much more convinced of arguments to like relieve student debt or relieve the sunk costs of pursuing a creative life like start augmenting the value of the arts to the general public yeah that's a way more compelling i'm like there's way more that you can push on that in the context of this conversation than flogging a dead horse of let's say data licensing or something like this which it might work out for a very small minority but the vast majority of people it's it's not going to move the needle it's just not so to that point I think the class distinction is a good point.

1:01:25And it's like, okay, well, the answer to that is how do you make a full life encountering the arts more rich? Yeah. That's bigger than people posting automated shit online.

1:01:49I want to shift the conversation to music. In music, are there specific AI tools or types of tools that you're seeing artists engage with more than others? and if so why do you think that is yeah it's tricky music's the one area it's kind of tricky music has been stalled a little bit so in the image domain you have like a lot of these kind of state-of-the-art models that people use and you pay for services to use them the challenge with music is because music companies are particularly litigious the open music models up until very recently were lagging we're kind of wanting just slightly different from stuff we've done where we've kind of fine-tuned models on very small data sets or done very small voice models that we use.

1:02:30But when you're talking about these kind of like a Suno quality model, which people end up using Suno and Udio because they're kind of like the standard big kind of closed weights, centralized models that are really freaking good. But the problem with them is the ability to customize them is quite limited. This is an open debate, which someone who really knows what they're talking about will challenge me on this because there are ways to play with the weights or whatever but the general point is that music is a much more restricted area and i think what's going to change is i think this year might be the year you're going to start seeing big very capable open models come out in music and so all of a sudden the ability to go in and fine-tune and so on i mean what we do for example we've done this a bunch is we create our own training data right what that basically means is you know the distinction is a bit silly in a way, right?

1:03:19Like you could take a record that you make and you could fine tune a model on that specific record. In a sense, we've taken that a step further where because we know we're training a model, you'll create a kind of record or a recording environment or whatever to produce a very specific timbre or series of elements that you can then fine tune a model on to get special results. If you learn how to develop the data in the right way to train the model how you like it there's actually a lot of authorship you can place there there's a lot you can do we've not yet seen what i think is native to the medium which is sharing models the interesting thing about ai models is they're there for other people to use and throws up what i think really interesting questions about authorship where you're like okay you know we did this holly plus experiment years ago now where you know we trained a model with holly's voice anybody can use it and so the output of that model is in a sense both the product of us the product of the model and the product of the prompter you know and i think that that is going to get really interesting and what i assume is going to happen and i think this is good and bad but i assume what's going to happen is with all these major deals you're going to start seeing commercial fine tunes explored so you know like for example well like the music the big music industry skepticism about ai will change the second that they start doing the official suno stuff right and then you're going to be able to prompt through the you know infinite drake model or whatever it's going to be and there's going to be a period of time where there's experimentation on that being like well whichever way you look at it the good or bad news is i can tell you now that fine-tuning is very effective so you know it is already a solved problem if you want to listen to infinite drake like stuff you can listen to infinite drake like stuff and it will be imperceptible that's going to happen now again the the wall-y negative perspective of that is being like yes to the vast majority of laid-back listeners to use Liz Pelley's term you know maybe it kind of sucks you'll have a system where whether it's YouTube or Spotify or Asuna or whatever you know people will listen and then ad revenue will be split between the person and Drake will make some money from that interaction I'm sure you know but the other side to it that I think is more interesting is going to like the idea of releasing models for other people to use is actually kind of a new medium and it might turn out to be crap I just don't think that we've properly explored what that means on our side we're kind of enthusiastic about that prospect it's like an interesting thing to share with people you know like particularly when there's some means of kind of reciprocal benefit involved with it um yeah it's almost like hyper sampling as a way to think about it but i mean it's different that's a reductive way of putting it but like when you think of it as like an artistic medium that hasn't been yeah totally that's the thing we use the term spawning just from sampling because sure because it is it's kind of like sampling and it's kind of different right because it's Rather than like a one-to-one reproduction of something that has existed, you're basically creating a feature space or a latent space or whatever of possible stuff that you can prompt through, right?

1:06:17Then the challenge in that circumstance is to produce a distinct enough vibe, for want of a better term, that it's actually worthwhile other people using it, you know? Like what would it mean to produce a distinct timbre or series of possibilities for people to prompt through where the output is constrained by the artistic gesture? that's fascinating actually and like i think it's just gonna happen but all that fun stuff doesn't really get to happen when currently at least in the music domain this stuff is entirely controlled by a couple of platforms you know which is why i go back to the original point where i'm like i think that the explosion of interesting things that could happen is really being constrained by the fact that these models are currently being kind of held they're being protected with these kind of you know people are not really able to get their hands dirty with them and i think that that's a mistake, you know.

1:07:04What makes a music tool feel ethical to you versus unethical? I think this is, it's complicated. I could just say our position on it, right? So like when we first started training models, this would be in like 2017. The first question you ask yourself is like, what data do you train on? And we're like, oh, okay, see where this goes. Like if we were to train this on like Michael Jackson songs and do a like negative land style, you know, like appropriation art type deal, like this will be a new story. There's the low hanging fruit, right but because it's different to sampling because this isn't a one-to-one reproduction which i actually don't really have much ethical concerns about the thing about spawning is like i could train on all of david bowie's work and then get bowie to do anything i want and that is like that just doesn't feel right and i think i think there's interesting ways you could do that so i don't want to be like a school teacher about it or something but like for us the more interesting challenge was we put together like a vocal ensemble to train on being like no then everyone involved kind of knows what's going on and the outputs are actually something we made together and that's kind of the interesting question and we've done that since with like members of the public and so on and so forth but that's our position and it's more it's more like a research position so it is kind of a guess an ethical line to make but like for me the more interesting question is just like i want to know what you can do with the stuff that you couldn't do before like what kind of social configurations can be made from this new stuff because this kind of appropriation approach when in 2017 that would have been like a big new story and low-hanging fruit and whatever we chose not to do that but like in five years time I think that that would be rather banal you know like there's gonna be the infinite drake like and whether it's official through youtube or whatever or like it's going to happen and it's banal to me like the more interesting question is like okay what if groups of people decided to find new ways to train models together and like what are those outputs what are the affordances like what what are the new interactions that come out of that and that's kind of what we base a lot of the prices right right and how does that change the nature of taste or how we interact with music which is more than any other art form i i would argue inherently communal it's so context dependent exactly that's exactly it and like i mean i'm borrowing from a lot of holly's focus now but it's like that's why we started looking at like really early group singing and like we're like oh this is like a fucking pre-modern thing like actually like ai models get way more interesting when you start thinking about pre-modern tendencies and we're like we just did a show at cafe where like we had hundreds of people every weekend come in willfully sing to our system and then train a model together and like everyone had a really good time and i'm like this is the nice hybrid where you're like we're all here in a room together we're all doing stuff it's all being documented we're all training a model together the walls have memory and you can come in here and hear the model sing to you there's something very like evergreen about that that feels honest to music and an art because the music industry as we know it may die but like that thing won't and actually you know honestly most people i know who are very dedicated to music that's why they're in it like they make albums to play shows so i mean i think that there's a lot you can do but the i just think the ethical side it gets really slippery because 10 years ago i think we were right to raise these questions about the data thing because no one was really talking about it and now i'm concerned that it's kind of it's really devolved into like ethics are used as a cudgel like it's devolved into like a lot of ignorance and kind of browbeating and i'm like that's not that was never our intention at all actually i think like risk aversion too i'd argue exactly yeah yeah it's because i mean ultimately it's also kind of futile like i you know we have a three-year-old he obviously interacts with stuff we do or whatever like he has an expectation now like we were playing around with some stuff and he was like i want a song about a truck and he got a song about a truck and he got a song about a truck one time and now for the rest of his life he believes that you can just conjure songs from what you want that's wild yeah and i'm like we can have discussions about the ethics of this and look there it's a nuanced issue and blah blah blah but like that boy believes you can conjure media like i grew up taping the simpsons and missing the first 10 minutes sometimes right like he believes you can just speak media into the world like you don't roll that back that doesn't roll back like so it's more interesting to be like what world is he gonna live in right like and i don't want that to be wally world where he's like you know living in you know infinite thirst traps or whatever like whatever the the but that capability is here forever right like so like wow you know that's big right so there's a degree of futility too i think where people like oh no well what about that i'm like it's all gonna fucking happen it's already happening honestly but like in the next couple of years there's going to be someone who meets the correct taste threshold who meets the whatever who's going to make that thing that you really like and then there's all this complicated stuff to figure out of like how do we value things like what is valuable in this new context there needs to be a hundred ideas made but the risk aversion doesn't get you there you know the risk aversion asks us to believe that my three-year-old who now knows that he can sing a truck song and it will become a euro pop whatever thing you know the risk aversion asked me to believe that that's going to go away and i just don't think it's going to go away.

1:12:12I don't think it helps anyone to pretend it's going to go away. It's not going to go away. Well, yeah, I want to bring this to what I think is kind of the real pain or anxiety point for a lot of folks around this stuff, which is, of course, the monetization compensation question. What challenges does AI generated music pose in terms of the entire current system of how recorded music is monetized, including streaming platforms, music marketplaces, etc.? Yeah, I think I forget the exact figures, but it's like 80 plus percent of streaming revenue goes to the top 1 % or something. Right. So like, I think that there's a there's a cream of people who for one reason or another will be absolutely fine.

1:12:51Right. And I think that that ends up in the realm where actually the person making the score or the sync music or whatever is is hired for prestige. It might not be the best thing they ever did, but it's a it's a brand deal. Right. And there's a whole economy around that. And I'm like, that's not going away. If anything, that becomes more valuable. Right. Then you have that realm of like what I talked about earlier with fine tunes, where I can imagine that there'll be a lot of desire for infinite Drake. I mean, the market hasn't been tested with this, but like, I don't see why people wouldn't like to listen to infinite Drake when they listen to infinite playlists and they're passive consumers.

1:13:22Drake will make a lot of money from that. You know, great. Like, fine. Then there's kind of context-based musicians. We probably fall in that camp, right? Where people listen for reasons other than just simply consuming a piece of media, right? It has a parasocial quality. it has a place-based context there's the whole live music economy which like we've been that was our whole deal right like we made records to play shows you know and and the and i actually think that's likely going to be fine i don't see why i mean like streaming numbers on almost all of that are terrible with rare exceptions right so like that world in my mind has already moved to live touring for the past 15 20 years anyway that like festivals and so on so forth and why should that go away that's more important or just as important what i think in terms of like value in the AI universe is the one mistake that people make, I think is, I've mentioned this a few times on the data licensing side, where the regrettable thing is when you have, let's say, a data set of 300 million songs or something like that, that's probably on the low end there, but let's just say, and yours are 10 songs in that data set.

1:14:24Unless you are named, you're kind of contributing to this very amorphous corpus where the model is learning from song structures. It might learn some timbre it's like oh you used an 808 like like every other person in the world the model knows what an 808 is now but reverse engineering the provenance of an output to your contribution is now unless things change and there's been a lot of papers about this recently sony's trying to do a lot of work most of it is a philosophical distinction that people are saying okay well what we'll do is we'll analyze an output and then in a perfect world where we have access to all the input data will run a statistical analysis to compare the output to what's in the data set.

1:15:05But AI models don't work like that. It's not so clean. So what you'll end up finding is that the song actually excites areas of the latent space that 100 ,000 songs are responsible for. And it might well turn out that the Universal Music Artists is the one that gets paid on that. There's just no easy way to cut it. So the proposal that we've made for a long time is actually if you were to see the base model so pre-training data goes into these base models like a stable diffusion they'll use the image example let's just say that you're not going to solve the provenance issue there and actually the licensing payments on the pre-training data are negligible for most people what you can do is offer an artist the ability to train a laura model or like a fine tune and you can say 100 of the proceeds that go to people who use that one so right they go to you know they go to fucking meshuggah.ai or whatever right and 100 percent of the commerce made on meshuggah.ai goes to meshuggah right and you just accept that yeah actually if you prompted birdsong in the style of meshuggah through their model the birdsong was probably recorded by somebody and there's no clean way to get that back to them but everybody let's say has the same conundrum right so you just say okay the base model we aren't going to figure this out, but we are going to figure out commercial fine tunes.

1:16:27Now that hasn't been explored. And honestly, I'm not confident there's going to be incredibly overwhelming demand for everyone's band to make a fine tune. What you might see is that there's going to be some like the My Bloody Valentine fine tune will be hit a lot because they have an incredible guitar timbre that's really hard to prompt. And you might as well just go and pay five bucks a month to get the, you know. So I think we're going to see that and the spoilers will go to people with distinct vibes or something that it offered in that space, you know? And like, I mean, that kind of is how music works a little bit like already, but like, so I think that that stuff's really cool.

1:17:00And then the thing that I hope honestly is that there's all kinds of other interactions that could be pursued. I was talking a bit about, I hate the term vibe coding, but people know what that is. The ability for artists to understand that there's going to be a new information substrate. They're like, okay, well, I do this thing. I want to find ways to package it to be interactive with people. The hurdle to offer that stuff is going to lower dramatically. But the fundamental thing that people need is they need access to open weights models to be able to experiment, right? And I think that's why as much as it is a bitter pill to swallow on the whole open model thing, even though some people are going to gripe about the data that went into those models, I think long-term, maybe there's a net benefit kind of ethical argument if you want to talk about ethics to having this stuff open and available to as many people as possible to set their own prices, to not be kind of locked into certain corporate environments, blah, blah, blah.

1:17:56I think on the whole, that is a stronger moral argument for where this goes than most arguments that I encounter. Yeah, that makes sense. Of course, I do want to ask about what Holly recently argued about Bandcamp. She was saying that Bandcamp was actually wrong to issue this blanket ban on all AI-generated music. Can you explain why this is potentially problematic? Well, as with most things, Holly was completely correct. I'm not just saying that because she's my wife, she just generally is correct on these things. What she was actually arguing is not that they were wrong to issue a blanket ban, but that the blanket ban is impossible.

1:18:33And she was vindicated because I think a journalist followed up with them and was like, well, how would you implement this? And they're like, well, we can't. And Holly was like, yeah, that's what I said. Basically, you do have analysis tools right now to tell if something was AI generated. Basically, there are certain artifacts that you can tell but what she was saying is that for example if she entirely writes a song and then uses an ai to you know make some elements in it that she's trained herself or whatever the the virtuous path is that you want to make in this best case scenario that will have the same artifacts as something that someone prompted in two seconds on suno i mean i guarantee within the pop industry they're using the hell out of these models of course they are there's interfaces that you aren't seeing you know what i mean because it's incredibly useful right so there's this other scenario where it's like okay if someone didn't write the song right like a pop producer spins up you know something and then generates a hundred songs auditions some of them which is what's going to happen you're going to have models that can audition these things for you then you're like we think that these five songs that were generated out of a million that we generated overnight are to your taste right all that producer has to do then is go pay a human to record which i mean interpolation is rife there are all kind of debates of hip producers like hiring producers to perfectly recreate samples and then delivering to the label a video of the fruity loops or whatever where some surgical mastermind has figured out the perfect way to recreate this thing that was clearly relifted but there's no case right so like right so her argument which i think is completely legitimate is like is the song that she made with these elements and her bespoke pipeline for doing whatever less or less talent which is arguably extremely original music creative output.

1:20:14Holly's top of her field. Whenever people think of Holly's music, she's top of her field with the stuff she's unfuckable with. So you're like, okay, you've got that versus a pop producer that pays someone to humanify something that Suna wrote. In the perfect band camp scenario, the humanified one goes through because that's legitimate, real human music and the other one's not. That's silly. But the other point I think people mistook with her is she wasn't even arguing. She's like, this is a spam issue. Like, yeah, you don't want people spamming you with a thousand EDM songs a day. That dilutes things absolutely within your right.

1:20:46But this wicked problem of determining what an AI touched and what didn't is going to become more and more difficult over time to the point whereby you're going to have to impose like some really draconian, weird policies that may be kind of reactionary in order to enforce this over time. And she's correct. You know, right. Because it's already things like there's like already AI plugins for Ableton and things like that. And so it's like, what do you call that? Is that AI generated music or not? yeah totally but this is part of the wicked problem again of this kind of culture wars is like i get this a lot you know holly the experiences with the band camp thing is like there's just no room for this nuance and like in that circumstance like she's telling people the truth you know and it's like you have to have a lot of experience to know the different particulars that go into it but it's like you're gonna find out sooner or later that this is actually much more complicated and so this kind of binary of like human ai good music bad music whatever the more that you push on it the more you kind of it just kind of gets into a situation that's like it's not in the spirit of like a good creative ecosystem or something what i personally think with the bank app thing is they're probably really well-intentioned people and it was probably a comment that got taken out i think it was like something that they said on a discord or something there's so much demand for ai anti-ai crap that it went up into the news you know but the wicked problem really is this kind of culture war because it's stupefying the difficult question which is like we've got to figure out how to live with this stuff is being prolonged and delayed and you you almost wonder at some point it's almost maybe that the discourse is kind of more valuable than they are at this point in time right that like actually maybe the culture war is the culture now you know like it doesn't matter what you're debating about like everything now you know every piece of media is just a scissor issue that just kind of is thrown in the middle and the explosion separates people into two sides and people just fight about things and you're like At some point, you wonder whether people have any genuine interest in resolving problems or whether this is the entertainment now.

1:22:39The Coliseum is the entertainment. I mean, yeah, Holly was definitely eye-rolling a little bit, but it's... I mean, the irony, of course, being like, as with this whole conversation and with everything we've been saying, a lot of context missing, like context missing to Bandcamp's point or spiritually what they were trying to say, context missing to what she was trying to say and a lot of these knee-jerk responses. and here we are in the Coliseum. Yeah, exactly. And that's, I mean, the sad thing, we did another project like a couple of years ago for the Whitney that, again, was heavily misunderstood in kind of interesting ways.

1:23:10I won't go into the project because it's kind of complicated, but like the misunderstanding actually really fueled the project or made it more interesting. But that was kind of the conclusion we came to where we're like, actually, the big concern everyone seems to have here is that, you know, human context will be stripped away from these automated systems. And yet we seem very happy to entertain these bullshit culture wars that strip all context, have no interest in the particularities of it. I'm like, the particularities of it are the thing we should be fighting for, right? That's the culture stuff, right?

1:23:38Like, yes, yes. The interesting part is like hearing a journalist or a artist or a musician or whatever, talk about these wicked, weird, new, cool problems. That's what culture is like. If we are just to turn everything into just a flame war of people missing each other, then yeah, like it might as well be bots doing it, right? Like I I could program bots to argue all day about, you know, AI is theft. No, AI is the future. This is a very autoregressive interaction we're having. Yeah. They do want to touch on, you know, labor issues. I feel like this is probably the biggest anxiety point that the creative community and folks on the left have with AI and what creates so many of these kind of like knee jerk, throwing the baby out with the bathwater sort of stances.

1:24:19The question is here, you know, ultimately, what do you think are the most important labor issues in the arts that these tools pose and how do we address them? Yeah, I mean, I think it's going to be standard with contracts. I mean, the difficult part with labor issues is, you know, there's going to be probably contracts put in front of people, giving people the ability to, you know, let's say a label or some representative body or whatever, giving them full data rights. And I think that that's a negotiation point, but that's easier said than done, right? So in an ideal world, I think that, you know, people should have a lot more kind of agency over their representation should be very well informed about the fact that yeah everything you do now can spawn infinite stuff and so i think there's a lot of work that can be done there again i don't want to also kind of sell false promises here though where you know if there's enough of a labor push back against that stuff the thing about these models is you can make infinite stuff with believable people right you know and so there's gonna be this kind of like price discovery you know of like you might want to resist that but i don't know that might work out to be implausible outside of like your natural kind of publicity rights, which I think, you know, no one should be able to make money in your name without you earning a profit from it, right?

1:25:26So I think that that stuff's all good and isn't going to go away and it only becomes more important. Yeah. In other realms, again, it's this self-determination issue. I think that like the labor issue becomes really tough when you have no negotiating power. And the traditional answer to that has been to build your own structures, context-rich structures in which you control the context right like you say okay well i'm just going to make my own publication and then if that publication is valuable with a group of my friends or whatever it might be then we have the rights over that data and we can figure out how that works or whatever which again is is another vote for me on thinking about open models it's like the equivalent of an independent music industry is possible with an open weight model world i don't think people are well informed enough to to really even think that through much less like advocate for it but to me if there's going to be a kind of independent world there that gives you more negotiating power it comes from that direction but but but i mean the thing i often feel like i get uh i'm a messenger being shot but like the activists will tell you that this is going away i believe i know more than almost all of the activists because i interact with them quite regularly and some of them are okay but they're kind of very zealous people i'm telling you this is not going away.

1:26:41And I think that clarity is better than cope, right? I don't think it's the end of the world, but I think taking this stuff seriously has been everyone's interests, you know? And so on the labor side, I think there's going to be massive disruptions to what we understood as 20th century media industries. And I don't think that's going to stop. And I think the best possible outcome is people have the ability to experiment with this new substrate freely without having to take whatever terms are put in front of them. And maybe that's idealistic, but there is a possibility of that and a lot of that will fall into these legal disputes and the general sentiment of people i think it's high time that that option is entertained because the activist stuff and the kind of infinite griping is not helping that cause and i don't i don't know how many people are convinced on that but talk to me in five years yeah i guess it sort of just comes down to the age old sees the means of production and and yeah absolutely but but that opportunity exists you know and i think it's you know that is a real opportunity and there's of course also a reality check on that stuff right like i've started a company i've built co-ops and all this kind of stuff and it's really fucking tough and you can build the best thing in the world and people don't come you know that but that's a more interesting kind of dynamic problem to deal with than sitting around waiting for the copyright cavalry you know or whatever people expect to save them that like no like take care of your own shit like that spirit will get you somewhere even if the first platform thing you try doesn't work out or whatever the people you meet will over time grow into a coalition and you might, who knows what comes out of those interactions.

1:28:10But agency or like some sense of doing something is always productive. It's always a good idea as opposed to exactly waiting for the cavalry, which doesn't come. Well put. So our last question comes from a member of our Discord, Devin Hanson, who also is a sometime guest on Cujo and they are also a fantastic writer and musician. And they said, I think the most important and least asked question about AI and art is, do we want to live in a world slash culture where the quantity of art that exists is more important than how much of the creative process we get to experience? That's a great point.

1:28:48So I agree with the sentiment, maybe have some slight departure, but there was an essay that a guy called Ben Roy published recently. He was talking about specifically writing and like the struggle and and there was a great quote by somebody i cannot remember the name of but hopefully you can find it who is saying it was something along the lines of like i write because i don't know what my point is like i don't know what my point is until i've written it you know which i totally relate to i mean that's that's my twitter right it's like a dead june dean quote we tell ourselves stories in order to live or whatever yeah but it's true i mean so the process of reasoning to get to a point is the point right and you could say the same with making a record right i mean it's this kind of very modern phenomenon that records are these self-contained things like generally speaking you know a lot of the great records were people touring songs and then they rolled up to a recording studio one time to document it and then we're out doing stuff so like the act of doing the art is the point and and also the rich context around it is the point and so the ability to automate the media is impressive but my response to the ben roy piece was pretty much around this where i was like actually on that struggle you know or on the making i find that like the best writing for example i can kind of i'm led through that journey with somebody like as they're figuring it out it's not like a one-shot thing right like i'm understanding their reasoning as i'm reading it and it's making me relate to their position and learn something or whatever and i'm like the good news is the way i use ai models is very similar i reason with them so i don't use ai models to write per se i use them to have arguments that no reasonable person would want to have with me because I'm here being like well actually Kant said this about the autonomy of art and then there's this book from Larry Shiner god and then Ada Palmer just released this book about the invention of the renaissance and I think this but does this you know and the reasoning process with the model is the joy of using it so it's not there to publish 20 articles it's like no I actually get to have this private conversation with a surrogate that over time I hope would learn more about me so that I can get and if anything again, going back to this kind of like, there is a third way.

1:30:49Like I look at it and I'm like, there could be a very near future where the reasoning process behind something is something that you exchange where like the act and the value of writing is understood not as words on a page or whatever, but the communion between different people who want to reason together, you know? And I'm like, in that universe, if I hook my, hopefully open source, blah, blah, blah. If I hook my Claude up to your Claude, right? Like there's a world where I might get a richer understanding of how you came to a conclusion and that might help me come to my own understanding. And that's the reason why I participate in writing.

1:31:24And I'm sure there are formalists who participate in writing for different reasons. But for me, I'm like, there's a lot there. Like there's potential in this medium through this reasoning. We're working on a project kind of in a similar vein at the moment for Venice. Like that's the reason why I care about any of this and I don't think the ability to automate and produce like a buttload of media changes. I'm like, that's fine. That will exist. Let's just assume all the bad stuff happens. And like our old media substrate is overwhelmed by crap. I want to know, can these tools better connect me with people and better help me reason through work I want to make to communicate to people or to understand the world myself?

1:32:00And I think that these LLM environments, to be honest, are kicking the ass of social feeds in achieving that and i'm like how do we hook these together how do we you know what i mean there's embers in there of something incredibly promising i think that that maybe satisfies the demand of what people who actually really care about the arts which let's be real is a minority of people like vast like everyone in the arts really vastly overestimates how much other people care about the arts like which is fine and we're not gonna be able to avoid that but like there's a lot of potential in these new exchanges that can be pulled out So I agree in principle.

1:32:34I think that like Devon, I'm inferring that we care about these things for the same reason. I just think that there's an alternate, very achievable path through these tools that is not being properly articulated, that hopefully will be so absorbing that you're not concerned about the fact that a gajillion pieces of crap were published that hopefully your agent filters through and make sure you never see. Right. And I think that's where it's going. someone at this fellowship thing i'm doing at the moment is is working on kind of constitution so like claude has a constitution and we were talking about it and i was like well the constitution is really interesting kind of like a pro i talk a lot about protocols i'm like what if we had the same media diet you know the ability to customize i don't want a customized algorithm of crap feeds i want a customized filter right like i want it so that you know me and all my friends who care about actually thinking about things aren't distracted from bait and i can program my agent to say, hey, we've all signed into this deal.

1:33:27We're never going to be presented with a logical fallacy again because it's a waste of our fucking time, right? Like we have tools to do this now, right? The big problem is that we're not, we don't have our fingers in the weights or we don't have the orchestration tools currently, but that will change, I hope, to carve out whatever that is, you know? And it's like, I genuinely believe in the next couple of years, just as, you know, everyone like fucking follows Andrew Huberman or some Luxemax or whatever. It's like, this is my stack, you know? here's my protocol thing and it's like no here's my media protocol right like if you trust me like this is my media diet the stoics will be all over it right it's going to get fucking strange and it's going to get bespoke and it's going to be tailorable and all i ask for honestly is that more concentration goes into that tailoring and the hard question of where to take it and what kind of creative communities or whatever political communities you want and how i mean you know how to deal with that kind of customization you don't want to live in a fucking echo chamber or whatever like these are the big tricky questions that are unavoidable with this new substrate right yeah so spiritually we're in the same zone i just think that like that's where we need to go you know and there's so much rich stuff in that that there needs to be a thousand people working on it you know what i mean and and there will be a gajillion eye rolls i know you know but but i think i genuinely believe what i'm saying is true and that's the interesting question is like what art do you want you know whatever bespoke arrangement you want like try and figure that out because actually, you know, I think that this is going to make it a lot more achievable.

1:34:53Yeah. And it sounds like we actually do have the tools available to start at least start doing that now or taking steps towards it. Yeah, try it. Well, the hardest part is I go back to that Jeremy Harrow point. I really recommend watching that interview. I know he's kind of on the like reasonable skeptic line, but it's like it's the difference between coding and engineering. Like it's the difference between like tweets and literature. You have to architect and engineer these outcomes. And to do that, it requires a lot of thought to be like, okay, what do I actually care about with this? You know, what do we actually need?

1:35:24These are hard, tricky problems. And yeah, like all the coding software in the world or whatever is not going to solve that social problem for you. But groups of people or very kind of prescient individuals will figure out that. And it's going to be faster and it's going to be, I genuinely believe, I think we're going to see like speed runs of these kind of, like over the next few years, you're going to see more and more and more speed runs of different alternative configurations but again i go back to this basic point of like that the likelihood at the moment is that that all happens on the servers of like two or three companies yeah first that being open and available to everyone and i think that the genuine progressive position should be the latter and you know copyright yeah cool you know copyright i think it's genuinely a distraction um you know the bigger question is what you want, you know, and that's as hard to answer now as it was a hundred years ago or whatever, you know, it's like perennial, it's hard to make something good and it's hard to know what you want, right?

1:36:23I hope I don't sound like too much of an evangelist here, but I genuinely believe these are the stakes, you know? Yeah. Yeah. Well, Matt, Dreyhurst, it's always a pleasure to have you on. Hopefully it will make a point of not waiting four more years before we have you back on again. and you always bring so much great arguments and perspectives and clarity to the table. For folks who want to follow your work or yell at you or ask you questions, where's the best place to do so? I mean, Twitter is still the only one I use. So it's just Matt with 1T Dryhurst at Twitter. And then, yeah, I mean, you can email me too, but Twitter's like my communications thing.

1:37:05And yeah, and then we have a bunch of exhibitions coming up. So if you're going to be, if anyone's in Venice, We got a show at Venice and then we'll be, we have a big show in New York next spring. So that'll be fun. So yeah, if you do want to throw things at me, then I'll be in New York. That's amazing. All right, Matt, thanks so much. Thank you. It's a pleasure.

1:37:29This episode of The Culture Journalist was produced and edited by Emily Friedlander and me, Andrea Dominick. To check out more, head to our sub stack. That's theculturejournalist.substack.com And if you like what you're hearing Leave a rating or review on Apple Podcasts Or share us with friends to help support independent journalism