Chapo Trap House - 878 - You Will NEVER Regret Listening to this Episode feat. Max Read (10/21/24)

Episode Date: October 22, 2024

Journalist Max Read joins us to discuss his new piece on the proliferation of “AI Slop”: unwanted, low quality, often surreal content flooding the internet and degrading its various platforms. We talk about the dystopian quality of the trend, the economic factors encouraging it, and how it portends poorly for the future of online. Max’s AI Slop piece: https://nymag.com/intelligencer/article/ai-generated-content-internet-online-slop-spam.html Subscribe to Max’s newsletter READ MAX at https://maxread.substack.com/ Order Matt’s Book (and check out the new merch!): https://chapotraphouse.store Come to our 11/4 Election Eve show in LA with E1: https://link.dice.fm/b1eb3de54f54

All I wanna do is hit the drum All I wanna do is hit the drum Hello everybody, it's Monday, October 21st and we've got some Chapo coming at you. I would like to kick off this show by reminding you about Matt's book. And I think some of you may have gotten the wrong idea about this book. This book is an exclusive offer that is literally only available to order for the next 10 days. This is how we match the printing to the orders. But your opportunity to have a copy of Matt Christman's book expires on October 31st. So if you have not done, if you've not pre-ordered Matt Christman's Civil War,

no, Matt Christman's Spanish Civil War, if you have not done that by now, please do so or you will be stricken from the book of life. Also, inserting this a little later from the edit because I just confirmed we are going to release 250 more signed copies of No Pasaran, Matt Christman's Spanish Civil War. They will be on sale at ChapoTrapHouse.store next Tuesday, October 29th starting at 9 a.m. Pacific Time, noon Eastern Time. So I know a lot of you have been asking for another opportunity to get a signed copy of the book. That will be the very last opportunity. 9 a.m. Pacific Time, noon Eastern Time, Tuesday, October 29th,

chapotraphouse.store. The book will be available until October 31st. Get your orders in now because after that there will be no further chances. Okay, back to Will. I would also like to remind you that our live show in LA with E1 on November 4th, tickets still on sale and we'll provide the link to that in the show description. And last but not least, I would like to, before we officially start the show, note the passing of a great man, a man that influenced me a great deal and certainly was instrumental in the creation of this program, Chappo Trap House. I am speaking of course about Fatala Gulan who passed away today and I'd just like to say were it not for the tens of thousands of fake bot accounts and Patreon subscriptions

that his movement created for our show to get the word out about his movement. I don't think we would have the success that we have today and I just hope his death won't mean that he's, you know, cutting off all those fake subscriptions that fund the show. Yeah. What else can we say except, you know, fuck 2024. Can you believe we still have to go to school even when this shit is happening? If anyone would like to join me in a Tuesday school walkout to memorialize Mr. Goulain, you're free to join me. Unless you're in one of Fetella Goulain's network of schools, please do not walk. Please do not walk.

Stay in school. Stay in the Goulain school. In fact, if you're in the Goulain school, you need to do extra school. Like when you would usually go home, like just take a sleeping bag. School's out for summer, unless you're in one of Gulan's school networks. Please keep going to school,

and please keep getting the word about the fraud that is Erdogan, the watermelon seller. Yeah, as always, fuck Erdogan shit. Erdo fraud. Erdo fraud, Erdo shit, the watermelon seller. As always, fuck Erdo shit. Erdo fraud. Erdo fraud, Erdo shit, the watermelon seller. But in light of the death of a legend, I think it's more crucial than ever to tell people, please test your Molly. This could have been avoided.

And that if we take anything away from this senseless event, I hope it's that. Make sure you have nor can with you at all times. Yeah. Okay. So to know we've got the business out of the way I'd like to officially start the show and introduce our guests for today. Today's episode we will be talking about a recent phenomenon that I've become fascinated with. It is the phenomenon of slop. Let's slop, bye bye!

Of AI generated content that has colonized a great deal of the internet and is opening up strange new frontiers in human consciousness and perception. And joining us to talk about this phenomenon is Max Reed, who recently wrote a great article for New York Magazine about the rise of AI generated Internet slop. Max welcome to the show.

Thanks for having me. So just to kick things off I mean slop is an umbrella term but like what are we talking about when we talk about like the kind of slop that you're writing about like are there different varieties of slop and like what encompass what is encompassed by this term? Yeah, I think the the best and like quickest way to define it is to call it weird fake AI crap. It really depends on the platform. And so, you know, if you're on Facebook, and

this is probably the sort of most famous weird one, if you're on Facebook, and you scroll for a while and you click around for a while, you're going to start getting served up. These absolutely insane images of Jesus made out of the bodies of shrimp, like really muscular policeman holding Korans in the in the surf on a beach. Morbidly obese women on mopeds driving through villages in India. I saw one that's like a African woman with chains around a huge orb of dead babies dragging it over like a cracked like a cracked ice sheet, like over ice flows. This sort of like, there's like a sort of Lovecraftian aspect to the Facebook stuff, which has like really grabbed people's attention. But you know, if you're on Twitter, you can

see it all over the place in like any reply to any popular tweet. There are now thousands of guys either like restating the premise of the tweet in like very obvious chat GPT language or having weird conversations with each other or often like there are sort of news, news Twitter accounts that are just pulling links and summarizing them with ChatGPT and throwing them up. Spotify, there are fully AI-generated country bands, electronic artists with AI-generated images, AI-generated biographies. Kindle, if you have a Kindle, you may have seen advertised on your Kindle lock screen these absolutely disgusting AI generated books called things like Healthy Children's Quest for Eating or, you know, Norseman Saga like Viking Revenge Story with these like really bizarre AI generated images.

The way AI guys define it, I think, which is a little bit friendly to the AI itself is unwanted AI generated content. But, you know, I think it doesn't really get at the sort of like lizard brain revulsion that most people have to this stuff. You will never regret liking this AI generated image of a baby covered in fire ants with that holding a sign that says, today is my birthday. Right, exactly.

The most, one of the the most jarring physical aspects of this type of AI slop is, I don't know if this is just because of the stuff that it's fed or if it's a specific protocol, but when it makes people, and specifically children, they make the eyes so disgustingly huge. I mean, is that a thing where like they told the machine like, oh, there's, you know, it's a dumb ass like, um, evo psych thing

where like, oh, big eyes are cute. Yeah, it's possible. I mean, it's hard to say some of this is like, you know, if you've played around with an AI generated with an image, an AI image generator, then you know that it's going to insert all kinds of weird sort of formal qualities to the image that you didn't intend to because so much of what it's doing is kind of doubling down on cliches. You know, the way these work, obviously, is they're sort of regressing to the mean and they'll find whatever. There's thousands of Thomas Kinkade, millions of Thomas Kinkade images online. And so it's just going to keep kind of aggregating those into your image.

But a lot of the... So for this article, I talked to a lot of guys who those into your image. But a lot of the so for this article, I talked to a lot of guys who are trying to make a living making this slop. And they were really consistently explaining to me like, No, no, like, I'm not making this weird, because I'm weird, I'm making it weird, because that's what plays. So there's nothing that would, you know, it would be totally likely to have somebody typing in, like you said, some kind of Ed Psych based, like, oh, yeah, larger, neoteny is extremely important to like getting my Facebook dollars or whatever. Okay.

So like, I want to get into it later in terms of like the economy of the people creating this. But like, in addition to like, you know, the seemingly overnight mass proliferation of slop, drivel, or as you describe it, demented nonsense that is on the internet. Aside from providing like a sort of an amusing and bizarre phenomenon, there are knock on real world effects of the proliferation of AI generated search results

and like an AI generated content. And a couple of them you talk about in the piece is you talk about a website that publishes science fiction stories called Clark's World. Could you talk about what happened to Clark's World when he like in terms of his submission process for generated content? Yeah, Clark's World is a digital sci fi fiction magazine. It's like quite sort of important to the sci fi world. Like it's a it's really well known for publishing first-time authors and bringing them amazing tales or weird

Yeah, weird tales are amazing. Totally is yeah And and basically any if you're at all familiar with sci-fi any author you know of right now Has been publishing cultural and maybe even got their start there and I bring all this up because one of the important things about the story Is like they're really it's really important to them to accept any submission from anywhere. They read everything they get because they want to discover unknown authors. And they have a commitment to this mission beyond making money because I don't think it's a hugely money-making enterprise. So in January of last year, which is maybe a month and a half after ChatGPT finally debuted to the public, they started getting this flood

of submissions that were, I mean, like bluntly just terrible and not terrible in the, you know, in the normal way terrible in a way that we now sort of all know is really typical of ChatGPT. There would be like weird lists in the middle of them. Neil Clark, who runs the magazine, told me that all of them, basically all of them ended with the scene from the end of Return of the Jedi. Like somehow they would place it in there. So it would always be like, you know, the earth is about to be destroyed. These three scientists are going to save it. And then it would do like a bullet pointed list of the fictional scientist characters. And then it would do a paragraph that was like, and they have come up with the, you

know, the nuclear Tron annihilator saver. And there'll be a paragraph about that. And then the final paragraph would be like, and all the planets were like celebrating with fireworks and like Endor was celebrating with fireworks and Coruscant was celebrating with fireworks. And the problem here was not so much that they couldn't tell that this stuff was AI because like it was really obvious it was AI. It's just that they were getting you know hundreds of emails an hour that was just this. And there was, because of their commitment to reading everything and not just dismissing things from writers they'd never heard of or people from specific countries say, they

kind of had to read through all this stuff and just check off everyone as being AI. And these submissions got so, the speed at which they were coming in and the scale at which they were coming in over the course of January and February got so immense and Clarksworld is not a big operation that they had to shut down submissions and basically be like, we're not taking, this is the lifeblood of the magazine, the submissions. So they shut down for six weeks or something, which was a huge warning sign about what's going to come. They have, to be clear, they've come up with a kind of system that Neil wouldn't talk to me about

because he feels like he needs to be really secretive about it because so much of this, I mean, it's very similar, I think, to spam detection systems where you can't say what you're doing because you don't want to give people an edge to get in. But there still are, and I'm sure we'll get into this soon, there's still like a million YouTube videos out there of people encouraging their viewers to submit stuff to Clark's World that they have written through ChatGPT.

So this is not even a solved or an ended problem. It's just they've sort of staunched the bleeding for a while and are figuring out what to do next. Clark's World example struck me as something that was like particularly vile because like, you know, as you described Clark's World, like this is, this is a, like, you know, a digital publication that prides itself on taking, as you said, taking submissions from anyone who wrote a story and reading them. And then if they select them, paying them about 12 cents a word, which is, you know, like you're not making great money doing it, but like they are paying for writing. And they're paying for original writing, and they're giving a platform to people who would otherwise not have it. But then like, as you

said, in the article, there are like, YouTube videos from like slop influencers are like, how to make $2,000 submitting stories to Clark's world. And it is just like, you know, like, you know, hampering the ability of this like important literary outlet that is like, you know, like, you know, hampering the ability of this like important literary outlet that is like, you know, a small operation, but nonetheless doing I think very interesting and compelling work in the field of fiction. Yeah, and I mean, the other sort of big, to me sort of scary example, as I talked to a bunch of librarians who were seeing a very similar thing, where, you know, it depends

on the library and what its collection policy is and how it works exactly. But a lot of sort of big systems, you know, a person can walk into the library and request a book or request a books on a specific topic and a librarian will fill out a request form and it goes to your collections librarian who orders it or who receives it. You know, they can even order it directly sometimes and you'll have a librarian who receives it on your desk. And it's not Clark's world-level flood yet, but if you, there's a lot of librarians who will tell you that they are getting more and more AI generated bullshit sort of crossing their desks. I talked to a librarian who first started seeing this

because he saw a recipe book that was like a Mediterranean diet recipe book and he like opens it and the first page of the meal plan the recommended lunch is marinara sauce which is not you know I don't see a problem with that I mean I'm the father of an Italian American I don't want to say anything about you know what's going on but I don't think that's a healthy lunch frankly for our children for our librarians for anybody and you know it's for me one of the things that is important to think about here is like, librarians, this is not like a group of people who are like really well paid and have like a ton of resources and time at their disposal. You know, they're like just doing their best to keep these things.

In New York, they're doing their best to like keep Eric Adams off their back and keep things open. And to have to also be looking at every, you know, sort of collected book that comes in and being like, actually, this is shit. This is AI shit is not a useful way for them to do their jobs. And you know, there is like, Marineris has this funny, there are like, without being too alarmist about it, people are ordering like, you know, living with cancer books and like parenting a child with ADHD books that are just chat GPT generated or 90% chat GPT generated. There's no way that they're getting the advice that they very clearly need from something that a guy in Nigeria has like compiled from his chat GPT answers.

I mean, Max, I think the most potentially alarming example of this happening recently with AI generated books is the guide to mushroom foraging, which is like, it is very important if you're putting out a book about which mushrooms are okay to eat, that it be accurate. Yeah, I would say so. And like the variation between different kinds of mushrooms are sometimes imperceptibly small to the human eye, let alone a computer generated view of this. So it was literally giving people advice on how to poison themselves to death with mushrooms. Yeah. And you know, this is one of those like so much of this is just based on nobody on

like what's being searched for. And you know, so you kind of are literally setting up a system where the kinds of things that people are looking for guides for almost by definition, like really important things. So the kind, you know, there are things that you need to know about that you need to. and so mushrooms is like the perfect life or death example where what are you going to do if you want to go mushroom, if you want to go foraging somewhere? Like probably even I would,

would just type into Amazon somewhere like mushroom foraging guide. And you know, some of these are really obviously AI, really like unsophisticated, but some of it is not, is like not meaningfully different in its presentation from guides from like actual mycological societies and I You know would be very nervous about

Ending up with a book with a book that is telling me to eat what I mean This is we're out of my my knowledge arena here But well and a mushroom has death in the name is usually a bad one folks just stick to eating marinara sauce Don't bother with the mushrooms marinara sauce will always be safe to consume. And that's a lot of vitamin C in it. Yeah. I think, you know, like the issue that this presents for libraries is an interesting one. And, you know, look, maybe the New York Public Library system wouldn't have so much trouble

with their budget if they were giving people the books that they really wanted, like Child Hunger Quest, The Search for Snacks, Volume 1, or Is This Snake OK to Grab? Useful information like that? But you know, prior to the Internet, things like libraries were like the physical storehouse for like the sum total of the repository of like human knowledge and culture and like that the promise of the Internet, as you write, is like, you know, something of, you know, to use a cliche, the information superhighway and like, you know, the Internet, you know, still is. It's like you said, it's where we go to answer questions, to find out about new things. But at the same time, like like with libraries, the Internet now is being colonized by this essentially fake.

Like, you know, not fake news, but just sort of like things that aren't real, things that are not a reflection of the like I said, the sum total of human knowledge. And you mentioned the idea that dead internet theory, like what is that internet theory? And what does this phenomenon describe? Yeah, so dead internet theory is a really classic like schizo poster, conspiracy theory, you know, folk, you know, folktale, like, uh, I mean, I, I don't mean to be dismissive. I love it. It's like one of my favorite sort of things to encounter online. But the idea is that most of what you see online is not

made by humans, that it's some kind of really sophisticated, you know, ongoing Psyop or, uh, you know, sort of big business, uh, conspiracy is just presenting robot generated people, having robot having conversations about robot generated topics, pushing robot generated stuff on you. The internet itself is, as the titles are just dead, and that you, the person consuming it, are maybe the only person left alive on it. And you know, what we're talking about, you can see why this theory is so attractive, because it is often how it feels. Like, especially Twitter in particular has this problem right now, because there's so much, there's like so much bullshit bot spam everywhere across the

platform, uh, you know, and, and because they're promoting blue checks, it's like, you know, if you're a, if you're, if you run a spam network, of course, you're going to get a, all your, all your bots are going to have blue checks. And like somebody found a, um, a you're going to get a all your all your bots are going to have blue checks and like somebody found a tweet that was very obviously a description of a photograph. It was like a nice couple. It was just text. It's just said a nice couple staring at a sunset and holding hands or something like that. And then hundreds of replies from bots underneath it

complimenting the photograph, which obviously didn't exist was just this sort of texting. I have no idea exactly how that happened. But that was to me like the signal experience of using Twitter in 2024. I think with like exactly how that happened, but that was to me like the signal experience of using Twitter in 2024. I think with like a lot of things, especially like Twitter engagement farming and like the types of videos that were like, kind of lowest common denominator

and or geared towards children, the internet of like the latter Trump years to early Biden years was sort of like the latter Trump years to early Biden years was sort of like, it was already perfectly positioned to be subsumed by AI, at least on the lowest tiers. Like the types of things that were being made just in response to the vectors of monetization on the current internet. Like, I don't know if you remember those really weird Elsa videos. Oh yeah, for sure. That were all over YouTube.

Oh yeah, I show my son those all the time. I say a bunch of them. Kids love them. I mean, he's four. Kids love them. He needs to know about surgery. He needs to know about freaking Spider-Man. That or like, you know, the very grim economy of Twitter

even before Elon, where like, I've gotten messages from people who run those like, Oh, F girl management accounts, where they're like, how much would it cost to get you to reply to me underneath your big tweet, which is just so horrible to think about. But yeah, just everything, the entire way that the internet has been monetized in the last like four years, especially has set it up perfectly for like, I would say, yeah, the bottom 30% of content to be just done by machines who aren't particularly good at approximating human desires or even like, you know, human posting patterns. Yeah, I mean, I think one thing this article

really made me realize, you know, AI, since the first sort of new generation of generative AI stuff debuted a couple of years ago, the tech investors, like the people whose book it is, I've been really talking it up as like the next generation of tech growth and tech wealth, they want it to be seen. I mean, part of this is about sort of replacing crypto as the, as the place

where, where investor money can flow. Correct. They really want it to be seen as the new thing after platforms. But you know, writing this story, it became like, you know, the, the, the sort of short version of the story is in the 2010s, like the rise of the platform as a business model, like the software platform as a business model meant huge amounts of money flowing into the pockets of Facebook and its investors, of Google, of Amazon, even of Twitter to some extent.

And what became really clear reporting the story out is that the AI stuff is not actually a separate thing from these platforms. That the way generative AI works is, as you're saying, actually sort of perfect for them. That you have these businesses that are effectively infinite marketplaces for consumption. And now you've got a machine that just infinitely produces content

and can fill every single niche that exists on those platforms. It couldn't be better, you couldn't have designed a better thing for Facebook if you tried than just something that makes awful images that can be shared at will forever and ever. Yeah, in the dead internet theory, and you bring up the example of interactions on Twitter,

and I have the quote for the thing you referenced where it was a photo captures a couple exchanging vows at sunset. The emotions it invokes are love, happiness, and the memory of a special day filled with promises. Hundreds of verified accounts swarmed in the replies to praise the missing image. This picture, again, no picture was posted. This picture exudes pure love and joy.

A magical moment indeed. This photo truly encapsulates the beauty and magic of true love. Such a beautiful moment captured in time, filled with love and joy. And there is a distinctly, whether it's a dead internet theory of this idea that like you may be the only real person left on the planet

and everyone else is a bot, but it has a distinctly science fiction quality to it. And you write in the article, the rise of slop has appropriately the shape of a good science fiction yard. A mysterious wave of noise emerging from nowhere, an alien invasion of semi coherent computers babbling in human like voices from some vast electronic beyond. And this, of course, invokes to me that this is very much a Philip K.

Dick story. And, you know, Felix, I know you're a fan, but Max, I'm wondering, are either of you familiar with the Philip K Dick story? Pay for the printer? Yeah. Okay, Max, when I was reading your article, all I could think about was that story. And for those of you who don't know it depicts a sort of post apocalyptic human society where it was like hundreds of years after nuclear wars have

destroyed much of everything and humans live in these sort of like isolated colonies. But human civilization has evolved to become sort of have a symbiotic relationship with an alien species called the built on that arrived on Earth to feed off its radiation. And the built on what they what they do is that they can perfectly copy any object that you give them, that they absorb it into this like giant like amoeba. And they copy like a wristwatch or a car or a piece of clothing. And in the story, human beings have become entirely dependent on the built ongs to manufacture all of the things that make human civilization possible.

But in the story, the built ongs are getting worse at it and they're getting sick. And every object that they copy becomes shittier and shittier and literally disintegrates after a short amount of time. So in the story, everything is just filled with trash and like buildings, clothes, cars, etc. are just turning to dust as humanity finds itself in a situation where it no longer knows how to produce or manufacture any of the things it used to on its own. And I really couldn't help but think of that story in reading this is like what AI has done is kind of a version of the built on is that it has a useful quality to it. But the more it's used and the more it copies from other copies, the shittier and less useful all of this actually is.

Yeah, I mean, you know, related like, first of all, Dick is like all over. I mean, he's all over the actually is. Yeah, I mean, you know, related, like, first of all, Dick is like all over. I mean, he's all over the present moment. He's been all over the last 20 years. Like any anything that you encounter that's about the fake and the real, about decay and about like distance from the real is also Dickian. And also, like, I mean, you know, like everyone's like, oh, this is like a cyberpunk shorts. You know, this the world is like a cyberpunk short story.

And it's like, well, no, it's really like a 1960s short story by like a completely insane guy from Los Angeles who decided, you know, who named everything really weird and every other page is like completely unreadable. And that feels much more true. It's not cool. Like it's not cool. Like a Gibson story. It's just kind of stupid and awful and like really sad.

Ultimately, like all those, all those Dick, the great Dick novels are tragedies. And I was going to say like part of what the sort of, like part of what's happening with this is a, in addition to talking about it as a kind of perfect technology for platforms, is it's just the latest version of something the tech industry, the software industry has been doing now for 30 or 40 years,

which is giving you a copy of something you already like and enjoy that's not as good, but is cheaper to produce and is sort of good enough. And because of who we are, because of, you know, how we're interpolated as subjects, because of whatever you want to say, we tend to prefer the cheaper, more convenient, good enough things. And you know, like that's all that's understandable. And that's all well and good, except that eventually as as as happens in Pay the Printer, it's like the original thing that was that was quite good and like required a lot of time and energy and thought to produce it is no longer available or is hard to find. And instead, what you have is, you know, child's adventure for healthy eating, like the mind readers of sixth grade or whatever it is. or whatever it is.

At the very end of Pay for the Printer, like the story ends with the protagonist, a man he encounters, giving him a single wooden cup that he carved himself, which is like the simplest, most basic piece of technology. But he was like, this is the future. Like, we have to relearn how to do something as simple as carve a cup out of wood. But, you know, going back to the idea of like, how useful is this?

Like, obviously, as you said, AI has become a marketing term. And like, I think I think it really bears focusing on that what is referred to as AI is in nowhere near the same galaxy as what science fiction authors are like what we think of as artificial intelligence as it exists in movies and science fiction. It is a universe away from technology that is actually has for intelligence. But it does have a certain utility and I'm wondering are there other examples of things like generative image programs or chat TPT? Are there examples of things that are like that actually provide a useful function to human civilization?

My genuine answer to this is that it's extraordinary shitposting technology. It's just really good for just stupid, funny jokes. In many ways, it does remind me of early... Sometimes using it reminds me of early days of using new software 10, 15, 20 years ago, where you like the first thing you want to do is figure out how to break it and how to make it do weird things. And it works in ways that are surprising and interesting. And you know, I actually sort of think it's a much more philosophically interesting thing than it is a kind of like profit generating center. Like I don't I don't actually know how you tile it is like I don't like the kinds of uses that keep coming up,

you know from their businesses that are pushing it are like summarizing emails and summarizing PDFs and like answering questions sometimes right and often like completely wrong and none of those are to me, like top level, like productivity enhancing, whatever's. But the idea of having a computer that you can like sort of talk to and it gives you, it like actually does a really good job of imitating Tony Soprano's voice or whatever is a very weird and interesting thing. And, you know, like says something interesting,

I think about language and about consciousness that you can have something that has no reason, that cannot itself reason, but can so fluently mimic like human speech that you could fool people into thinking it's alive. But like all of these uses are like leisure uses, like the use of these are uses that are sort of like, you know, um, like it's fun to fuck around on the computer type uses. I would say the one, like the one place that it seems like really good and interesting

and important is translation. And for years now, even before ChatTPT, when I talked to people who worked at jobs, especially tech jobs where there was a huge international group of people trying to communicate with each other, they were all using, I had never thought about using Grammarly before,

even though it's advertised before every single YouTube video I've ever watched. But everybody's using it to like make their English sound, you know, like solid and fluent. And like that that seems like a good old Internet type, you know, like let's all get the big human race getting together and, you know, chatting in English type thing. But beyond that, I mean, you know, there's not quite as many

like this is going to solve everything. Yeah, I was going to say off the top of my head, exactly per your answer, that the positive examples of AI that I can think of off the top of my head is the people who took one of our live streams and then made it so that Felix, Matt and I were speaking perfect Mandarin Chinese. But even if there is no such thing as the scum of the online transaction, he married the president of the New York Stock Exchange.

Isn't that a bit strange? I mean, I want to... That was such a big hit in my family. I would say the other example is the Homer Simpson concert videos. Yes, I love those so much. Where Homer Simpson sings Radiohead and Google. Sings popular songs on the radio. I think that's a good example. I think that's a good example. I think that's a good example.

I think that's a good example. in my family. I would say the other example is the Homer Simpson concert videos. Yes, we're over. Simpson sings Radiohead and Google sings popular songs from the 90s as all of Springfield and the Transformers dance for him. I don't mean that's just incredible. I thought that was genuinely great. Yeah, and the other one I will mention that I'm never going to forget is when the Microsoft image generator launched and everyone made different favorite characters behind the wheels of the plane crashing into the Twin Towers.

So seeing like Shinji from Evangelion doing it, Homie, like everybody you'd ever wanted to see, you know, in the place of Muhammad Atta doing it. It was just a beautiful again beautiful Humanity coming together to like really bring make something great. Yeah that I mean I I definitely agree with you that there there are parts of it that are reminiscent of like other sort of like late web point Oh early web 2.0 things specifically thinking of when everyone Worked together to

make it a do Drake in the band of brothers episode, why we fight that was, that was like a huge day for me when we figured out how to do that. But like the weird thing about it is that like w w with those things in years prior, you, the thing exists for like a few months or even a few years and then you figure out it can do these things and you know along the way as you're making it do like these these joke things or these novelty things, maybe someone discovers sort of incidental to that that it has this whole other use

and it gets used for things that like nobody really initially imagined. And it's one of the cooler things about emerging technologies like that. With AI, it seems like you start out with it being fun. And then the more it gets used, the less, the less your inputs actually matter. Unless you're doing things like, I don't know, you know, those people who plug, uh, viral tweets into chat GPT and try to monetize the replies, you know, very grim things. Yeah.

I mean, part of the problem is it's just not like, I think there's a two part thing, like it's just not actually that good at the things you want to be able to use it for. I mean, I suppose there are programmers, you know, I think there's a two-part thing. Like, it's just not actually that good at the things you want to be able to use it for. I mean, I suppose there are programmers, you know, I hear from programmers who say that it's made them a lot more productive. I wouldn't, I don't know. Like math is not my, whatever they do, I don't know what it is and I can't help them with that.

And I also talk to programmers who says it isn't, you know, increased their productivity at all. But it's, I think even if you're like a dyed in the wool poster, you can play around with it and make a lot of 9-11 jokes with it or whatever. But at some point, you know how to make 9-11 jokes yourself. You've been doing that your whole life, man. You don't need, I don't need Chachi BT.

I can just type out Shinji Ikara behind the wheel of the plane driving into the thing. And that's all you need. You don't need it actualized. Get on the plane, Shinji. You have to pilot it. So there's like, so's all you need. You don't need it actualized. Get on the plane, Shinji. You have to pilot it. So there's like, so, you know, on the one hand, you're sort of like, okay, this is really funny, but like I've sort of run out of uses for it.

And then on the other hand, even though like, you know, you think to yourself, okay, maybe I'll plug this PDF into it and it'll give me a summary. And eventually you realize that the summaries are not actually that good. You know, all the things, all the sort of those kinds of like productivity saving life hacky little, you know, like get things, getting things done type things. They don't really work quite well. So what what you're left with is like creating shitty content for platforms. Like that is that is, in fact, the like most visible and to me,

the most obviously profitable like way of using any of these generative AI platforms. Let me love you the right way. I need you on this country not made. Yeah, it's a 10 minute from my place, but first I got to make it out of my driveway. Now, we talked about how like AI is a marketing term that's like it's seemingly everywhere now and like every commercial you see on TV, they're like, oh, like it's powered by AI generated insights. It's like, oh, great. Thanks. I don't know what that is. But I remember just a year ago, it was crypto and NFTs.

And it's the new concept that Silicon Valley and the people who own and fund all of it have decided this is the new thing that we all have to like and is everywhere and the most important technology. And I think a key part of that is in the marketing term, artificial intelligence, that there's this sleight of hand where like despite this nonsensical drivel and like demented algorithmically generated images or text, it gives the illusion that like,

this is just machines doing this. This is just the computer. But ultimately, whether it's the video of like, whether it's a image of like a Korean stewardess carrying Jesus Christ out of an anthill, or the mushroom foraging textbook that could kill you. The slop is ultimately the product of human beings and the decisions that they make and the actions that they take, which is at the end of the day about making

money through scams. And I think there's a connection here to this kind of hustle and grind culture and the dream of passive income But you get into like some of these people who are like making making a living creating AI generated slop like and they're like basically like young kids from like Kenya or India so like how does one actually make money create creating this drivel? Yeah, I mean one of the sort of funniest discoveries here. So one of the ways this article started is was trying to get to the bottom of all the Facebook stuff is just sort of like, what is this? Why is it here? It didn't have, you know, early on when I was like reading,

you know, like looking at these insane images, it wasn't clear like what the play was. There wasn't like a clear scam. Often there wasn't like a link. often. There wasn't like a link They weren't like linking off. They weren't selling things. There wasn't an advertisement associated with it And I should credit by the way for a for media, which is a really good subscription especially if this is the kind of thing that interests you for a for is like a Fantastic tech site that covers this stuff really tightly He's a bunch of people used to be advice and so I was following a lot of their coverage too

they were sort of similarly baffled by these things. And it's also impossible to get in touch with any of the people who do these pages, obviously. Like why would they talk to some guy? And eventually I got some, this guy who is, who I am like 99% sure is a Kenyan guy named Steve, but who insisted to me for a long time that he is an English college,

an English comp-sized student named Jacob Boyd. And Steve, you know, offered to answer some questions if I sent him some money. So I paid him over the course of our conversations. He said no, he told you no free information. Yes, he said no free information. That's mindset. That's mindset. So Steve-O explained to me like right away that actually I was like, so what are you doing and how does it make money? And I was, you know, I was sort of gearing myself for some kind of like,

you know, like crazy Rube Goldberg type scam, or they were fattening up the page and we're going to sell it or whatever. It's like, oh no, Facebook just directly pays me for this. And what I hadn't realized, because it's like I'm not on the I'm not on my grindset, is that Facebook makes payments to page creators for engaging posts, basically. Not every single one. You have to be in the program but it's not that hard to get in the program and there's also like a whole network of people who will fake a you know a an American or a UK tax account for you so that you can get into the program if

you're in Kenya which is currently not part of the program. But the program is just if you make a post that has a certain number of likes, comments, shares, reactions, whatever it is, they'll send you a check. And so, there's this whole, as it turns out, there's this whole economy sort of lurking behind this stuff of not just the guys who are making the pages themselves, but influencers who are trying to teach you how to make these things. So, um, the sort of majority of these are actually, I think there's a lot of guys in the, like the countries that I think for it's actually kind of funny because I think each platform has a country that kind of dominates like what it does. So Facebook to me, it's a lot of Kenya, Nigeria, India, Cambodia, Twitter is I think like more than half Vietnamese, but there's also a lot of Indian, like I

think it really did, like there's a, there's a, but there's also a lot of Indian like I think it really like there's a there's a big There's a company in Vietnam that does that sells like Twitter bot software basically So I think that's why there's so many Vietnamese botters on Twitter Kindle is like there's a weird number of Russians or people who I think are Russian and and possibly Israeli too But I'm not it's like harder to figure out who's behind the Kindle stuff and then But I'm not it's like harder to figure out who's behind the Kindle stuff and then Tik Tok like Tik Tok slop is mostly American French English and it's especially English born Muslim kids and I don't I don't have no idea why that is

I would love to read like an ethnographic study of like why there's so many English Muslim dudes who are on their grindset trying To make money off of Tik Tok slop but anyway all these groups have discords, have telegrams, have YouTube influencers who are selling lesson plans, um, you know, or whatever it is selling access to their top secret discord. Um, that teach you how to make the engaging content to get into the plans. So Facebook's got a creator's program. TikTok has a creator's program.

Obviously we all know Twitter has a creators program. You know, all of these are these ways to just get sent money for making engaging content. I mean, arguably Spotify even has one that's just a little more encoded because it's just royalties. And Kindle, Kindle doesn't, but Kindle, it's much more obvious how you make money,

which is you just sell the bullshit that you're making. I think it's perfect that Kenya is a hub for a lot of this slop driven activity, because in your article, you mentioned as an example of one of Google's AI, AI overview search results is there are no countries in Africa that begin with the letter K. So maybe this is just like Kenya's revenge on the algorithm. They're like, we're here. We have K and R in the title of third name of our country.

But I think the really interesting thing about like, like you said, these sort of tutorials and sort of for pay seminars that teach people how to create this engaging content. I think it's fascinating about like, as you said, like, I'm not like the people you interviewed, they said, like, I'm not doing this because I'm weird. It's because weird things get attention. And you you talk to Jason Cobler of 404 media. And I just want to read here, it says, watch dozens of Hindi language swap seminars on

YouTube, many of them offering example prompts, such as American soldier veteran holding cardboard sign that says today is my birthday, please please like Injured battle veteran wore American flag an old American woman is making forest lion out of cauliflower And her neighbors are looking at it. Keep it detailed and I just want to like one more This is actually from Jason Kobler at 404 media He just talks about like I'm just fascinated by the prompts of the things that get attention. And he says, he clicks a page called Amazing Magic and says, photos of poor people are

good. Anything that touches the heart, cute babies, children, there is a lot of good engagement. He clicks over to a page of AI generated pets. People in the US and foreign countries, they love their pets and other animals. There are many pet lovers there. And that's right. He shows them an image of typing things into a Bing image creator like African boy create a car with recycled

bottle for us. A African boy create a truck with recycled Coca Cola bottle for us. Throughout his videos, he's talking about how his phone can barely connect to the internet, and that his internet speeds are maxing out at eight kilobytes a second. So I guess my question here is that like, you know, like, there's a certain template to these images where it's like, yeah, the poor child, the poor child saying, please, today is my birthday, I have no presence. Will you give me like, or a US soldier battle veteran, like, you know, saving American flag.

What is it about like this sort of this kind of images that are being generated by people in India and Cambodia and Kenya that they're playing to an American audience? What does this say about the American psyche? I mean, it's a it's a that's a it's a great question. I mean, it's sort of it's self-explanatory, isn't it? It's like this it's like the dream. I was I's like the dream. I couldn't get this articulated right, but there is something about it where you combine the way AI gloms onto cliched images and follows a particular kind of autoregressive path as it creates its images,

along with the marketplace that is Facebook, which is supposedly reading the revealed preferences of what people are scrolling through and scrolling on. What you get out of it is this dream of what the American internet or what the American internet is looking for, which is like mutilated soldiers begging for likes, puppies and kids, Jesus just constant, Jesus everywhere, nonstop, a little bit of Quran. I mean, I think that's probably not America.

I think that's, you know, other countries or whatever. But they're like, the other thing I would say though, is that it seems clear and I don't have any like hard evidence except for just the experience of going through this and talking to other people who have done it, that Facebook is pushing AI stuff, that there is, it's like, you'll get, if you start clicking on and liking AI generated images, you tend to get a lot more afterwards, even stuff that's about unrelated subjects

or unrelated whatever. And so I, you know, I don't know why they would do that exactly or what exactly the sort of point is there, but it also feels like Facebook is kind of enforcing this, like, you know, this surreal experience of What was it the guy I talked to Sivo told me that like his biggest subjects were like US military pets Jesus and Manchester United. So there's a lot of

There's a lot of hooligans out there You know, the other thing about it is like Manchester United I think probably like 90% of actual Manchester United fans are like not even in the UK. So it's just a global obsession with, you know, with the Manchester United brand. Unfortunately, there's like no images of Manchester United players amputated, you know, begging for likes on the, on the beach, holding up the Quran or whatever.

One of my, one of my favorite like, um, recurring image genres I saw from a group like that is like, you know, obviously anything with soldiers or poor a group like that is like, you know, obviously anything with soldiers or poor kids, like that, that the poor kids is like a whole genre on the internet, even pre AI, like everyone loves a picture of like a poor kid with a toy made out of garbage caption is like, be thankful for what you have. Yeah. But what I thought was the funniest thing, and especially now knowing who's making this,

was that they kept generating, they kept developing real life Jesse Pinkmans. They kept developing sort of like trashy mid-30s guys wearing, they didn't get the official brands right, but trashy white guys wearing FUBU holding up signs that are like, I'm six months sober, can I get a like? And it's so funny to me that people in Kenya and Bangladesh and all these like far-flung places are like They they got the data and they're like what are like these dumbass trashy Americans like

Just they invented a guy who yeah like steals copper wiring He's a military worshiping pet lover who is also an addict and is an amputee probably. Well, yeah, I mean, it's like, you know, it's our it's our love for fake humility and compassion and stories of redemption and uplift. And then on top of that, we love our pets, Jesus and US Army soldiers. Yeah. Yeah, I mean, some of the soldier, like the military imagery is particularly, if you want it, because what AI does is just sort of expands it, so you get all those like goony,

like Navy SEAL images of guys with big beards and big biceps standing with their rifles or whatever. And then the AI just makes the rifle twice as big and with a million attachments, and makes it sort of cyberpunk. And the guys even more masculine like they already looks like gigachad kind of it's it's a real It's a look. It's an interesting thing. So I guess like, you know In thinking about how like AI has been, you know tapped to be the new trillion dollar bubble That's going to be inflated and you know, even if no one likes it or wants it or understands it, I guess it's like, where, where is this going?

Like in terms of the proliferation of this kind of thing, is it going to, is there going to be like a, is the wave going to break at some point or is this just the future of information and human identity in the sort of digital age? I mean, I, I am not feeling optimistic. I would say, I think that this is something we're going to be contending with like forever. And I think that it's the kind of thing. Um, I mean, I hate to adopt the sort of like

framing of the tech guys, but I do think this is the kind of toothpaste out of the tube thing. Like this stuff is here. We know what it is. We know how to use it. Is it, you know, generative AI is increasingly a sort of commodity product. There's models that you can download and run on your computer.

It takes a lot of energy and effort to make things at home. But that's going to that's not going to something that's going to leave. And I also don't feel optimistic about our ability to like shore up our institutions, to support local libraries, to, you know, make anything like better in any way. And I certainly don't feel like Facebook is going to turn away opportunities to make money if this is something that is apparently making the money. I do think the last thing is the kind of like cynically is sort of the best hope I have is I think

I think the next few years there's going to be we're going to see whether or not the sort of digital ad market is as ever expandable as has been assumed for a long time. I think that some of these companies, Instagram is not growing in the way that it was hoping to. Facebook has completely stopped growing in the US and is in fact, I believe, shrinking in the US slightly. Twitter is obviously like a total disaster.

TikTok is huge, but may or may not be gone in the US or owned by Bobby Kotick or something within a year. So there's all these, like, I think we're reaching a kind of, I don't know what the word I'm looking for is, like a point where we're gonna figure out what the future of these platforms actually looks like. Cause they've won so thoroughly over everybody else online, over every other form of media, over journalism, over Hollywood, over everything.

And I'm not positive that they're actually going to get the return they're seeking from that. Like part of the thing, and this is this is something we're thinking about with the AI stuff, is like we Americans are the we make more money and we spend more money than anyone else. And we're more susceptible to advertising and love advertising more than any Other people on the planet. This is one of those things We're number one and we're great to advertise to we we love that shit and we want to be advertised to and we want to buy the shit we get advertised to and If they can't make other markets do that

Then they're gonna have a lot of trouble growing in the way they're supposed to grow the problem is like, okay So they won like who's gonna nobody's gonna come around and like, you know, say, well, I've got this non AI, you know, bullshit that I that you can see. So I think that, I mean, look, when I keep telling myself, I have a I have a subscription newsletter and I keep being like, okay, but this is a great business opportunity for me, because I'm going to be the last like the dead internet theory, but good for me, like profitable for me because I'm the last human here or whatever. But you know, I think there is some hope that there will be like a kind of backlash, but I can't, I can't game it out.

I mean, it doesn't feel good. I'll say that much. Just speaking very generally about how these platforms broadly succeeded at all their goals in forming the internet and forming like just the way that everyone receives content and consumes it. It follows a broad model that we see in all of this type of shit, whether it's like Uber or DoorDash or like really any of these companies that popped up in the last 10 to 15 years, which is that like on their way up, their advantage over their more traditional competitors is that they are allowed to just hemorrhage money or otherwise not make a profit

while doing business by deluge. Whether that's like DoorDash losing a little bit of money every time a delivery is made, but it's so comparatively cheaper to having a servant who just picks up meals for you. The same thing with Uber. Or just the content spigot represented by Google and Facebook. The logic with all these things always was like, okay, like on the way up, every single transaction will lose money and we have these like massive piles of debt, but we have these big valuations because like once we achieve total market share, then we

get to set the price and that's when we become profitable. The thing with like web platforms specifically with Google and Facebook and everyone is like, you know, now we'll set the price for fucking what? You know, like, like, okay, you have total market share. Like you said, they won, but like, okay, are you going to have to swipe your credit card to like see a video of jelly roll? It was not even be real jelly roll. It's going to be a roll. Yeah, it's going to be. to swipe your credit card to like see a video of Jelly Roll. It was not even a real jelly roll. It's going to be a roll.

Yeah, it's going to be a premium for real jelly roll. It's going to be jelly roll, but he has two peg legs. I just got back from my deployment in the military. I'm sober now. I was going to say, like, in terms of like my pessimistic attitude about where this going, I have like two recent examples of like one was Elon talking recently at a Trump rally or answering questions that his biggest concern was whether or not we're going to train A.I. to be woke or truth seeking. That's a big concern for him.

But I thought that was fascinating because I think he's kind of giving up the game here that AI is not artificial intelligence. It is something that is designed to reflect the values and beliefs of the people who fucking created. So, yeah, it is a conscious decision whether it's going to be woke or not. And the other example, I, Max, I assume you saw the comments made by former Google CEO Eric Schmidt the other week. Yeah. When he was speaking at some business conference and he said, look, we all know that like the developed nations of the world have no shot

of reaching their carbon, like limiting goals that they've been outlined years ago. We all have no shot at that. It's fake. It's not going to happen. So we should in fact just burn more carbon and use more energy to power even more powerful AI that will quote, solve the problem for us. And this to me was so frightening because I think similar to things like the economy, the market that become described like their gravity or something and not the net result of just human action and decision. And I'm terrified about the idea of like, AI just becoming the word

that we use to essentially outsource our own, you know, conscience or moral decision making so that the people who create this AI and own it will be like, well, the AI is simply just decided rationally who should live and die in the coming water wars. Yeah, I mean, this is this is the problem of the software industry as long as it's ever existed is the long term goal has always been to to like, obviate politics by putting the computer in charge. And like, you know, what's the famous like, you know, don't let computers make decisions

because computers can't assume responsibility. And it's just funny that this is where we're still happy is 2024. We're having the same kinds of conversations that people were having in 1975 about this stuff. And there still has been no sense that actually like politics is a necessary means of like determining resource distribution, that there's no, there's no like, we're not even talking about like, the calculation debate or something that might might have a positive end point or whatever it's literally like oh If there's a doomsday clock for catastrophic irreversible climate change, we're two minutes out and we go to the machine and it's like deep blue.

It's like, no, it's less like deep blue. It's more like the computers from Evangelia. These Charmian obelisks, these estimates to humanity's greatness and combined knowledge and sort of You know cleverness and know-how and we tape it in we're like, what do we do? What do we do? How do we prevent, you know the super storms and the extinction events and it goes I'm jelly roll Going to happen. I don't know.

No, no. If you've used Google in the past year and like you ask this simply like, you know, how long is fucking leftover chicken good in the refrigerator? They will spend extra time to first give you an AI response that is always unconditionally, not just wrong. It takes your very simple question and rephrases it to be a new question that it also gets wrong. It's so shitty.

It's insane. And I kind of wonder if like that type of AI that is supposed to be the most practical use case and presumably would be helping solve climate change one day, it seems like it's collapsing under its own weight of shittiness. That is to say, everyone's talked about those Indian or otherwise subcontinent websites you get when you type in like how to fix something with your air conditioner or like movie show times, you get a website that's called like I, Tonya, movie show times, movie, film, good time, happy, or like chicken thighs, delicious,

fun recipe, easy, great time kids. That like all the things that are fed into the machine to help like get results from AI quicker are, you know, it's like an animal eating its own brain. It's just making it less and less sensical. Because it can't think. It can't like see these results and go, oh, this is bullshit. It just goes, oh, these are the right permutations of words. And so the more and more of that it gets, the fewer like direct problems, much less like incredibly complex ones it can solve.

I mean, one of the things that really frustrates me about this is like, to me, I mean, I can still feel sometimes like optimistic about the internet. And to me, like a beautiful thing about the internet is like, is in fact, the interplay of sincerity and irony, like across the whole thing, everything you do. And old Google, like the example that everyone was giving when they were when they were distributing the when they were talking when they when they were introducing this new Google AI answers product was the one where it was like, how to get your cheese to stick to pizza. And the Google answer was, oh, just put some glue in your tomato sauce and that'll do it.

The thing is that it was like wrong, but it wasn't like a hallucination. It had a source. It was some guy's joke comment on Reddit. And I'm not going to, it's not a very funny joke. I'm not going to defend it. But in old Google, you open up, you search for how to get the cheese stick on pizza and you, like a thinking human being with like a relatively sophisticated ability to discern like intent and tone, are easily able to see like what are real answers to this question and what are the ironic answers to this question. What are the jokes? What are the different, like what's a movie quote? What's real? Like what's, you know, what's part of a list

and what's not? And Google likes to protect, like assumes that you have, it has so much contempt for you as a company. It assumes that not only can you not do that, you don't want to do that. So instead it's just going to serve up whatever fucking answer. But the irony is that the fucking computer can't do it either. In fact, it does it way worse than you. So you end up in this position where you're like, I can't even, I'm not even being allowed to use my, my like, my, my like higher ape thinking skills to determine how to keep cheese on pizza

Like I'm just told to put fucking glue in the in the tomato sauce I mean I like I fully agree it's it's it's infuriating and it feels like it's sort of ruining everything that Might once have caught good might once have come out of the internet I'll take it one step further than Felix's vision of the future where we ask AI to save humanity and its response is jelly roll saying I'm not normal in perfect Cantonese. I would say the more frightening the more frightening thought that I have when I think about this is that like the people will create this immaculate purely rational Godhead computer and they'll be like, oh, AI. How do we ensure that human civilization prospers and survives for another thousand years? And the A.I.'s response will be to reduce the human population to a manageable 20 or 30 million people who are connected to the tech industry or are useful to them.

That's that that's what I imagine A.I. is going for in the future. But look, it doesn't have to be this way. AI is going for in the future. But look, it doesn't have to be this way. AI in the future, or really right now, could generate videos of manuals talking to each other with Felix and I's voice coming out of their mouths. And that is what we should be using this technology for. And I'll reiterate, if there are any tech companies or AI people out there who can

create videos of manuals doing an entire episode of Chapo Trav house. That is what I think me and the rest of the human race would like to see. That's easily worth two degrees of warming to. Just shoveling the last seven rhinos into a coal furnace to produce. Not even a question. You don't even have to ask. I got to leave I got to leave it there for today. But I want to thank Max Reed for joining us today.

And Max, if people want to check out your newsletter, where should they go? Maxreed.substack.com. The newsletter is called Read Max. It's a very clever pun on my name. Leave it there for today. And now for everyone who has finally listened to this episode, I can say with utmost confidence, you will not have regretted listening to this episode. It's my birthday. Please help me. We're gonna leave it there for today. Thanks again to Max Reed.

Cheers, everybody. You got a fast car. And I got a plan to get us on a heave and working at the convenience store. Man, it's a little bit of money money won't have to drive you far just across the border ending to the city and you and I can both get jobs fine to see what it means to be really bad