Unleashing the Creative Pandora's Box: AI's Potential for Innovation and Exploitation in the Arts | Vernon Reid

28:53I mean, there's going to be a split, there's going to be maybe a folk revival, a kind of like, you know, anti-tech. I could totally see that because people feel threatened by a thing, and they're going to go completely in another direction. Other folks are going to whole hog embrace it. And it's not just the technology of making up lyrics or making up a melody line. It's the convergence of technologies that are really concerning, like this whole thing with the fake Drake. I mean, they've had fake Obama and fake Trump for a while. But when you hear them, you know, you can hear that the gating, you can hear that the words are isolated.

29:35But now we're getting to a point where there are going to be apps where you talk into the microphone and another voice, and it's not vocoding, it's going to be another person's voice is coming out the other side. and that is the nature of identity itself is going to become transactional in a way that we have not, have never seen before. What do you mean by transactional? Without thinking about the moral implications of it, well, I just want, I want to do a rhyme and I want Marilyn Monroe to be the rapper, right? And without any thoughts about how the estate's going to feel, how you gotta you just do it and if it's amusing if it's a if it if it's amusing to the populace then it's in there and and we have a tendency to we accommodate we're willing to look past the implications to get to the goodies to just get to the the feel-good stuff and that's going to increasingly become again very seductive like there was somebody I saw was reacting to a to a Drake thing and they said man this is better than your thing I was like wow that's that's weird and but eventually we're gonna see because imaginative people that make somebody that's not necessarily trained in a particular art form but is very imaginative and and actually is clever and actually has a pole has like an influencer is able to represent a lot of other people that feel that way they will be incredibly successful with this whole thing it's another whether we like it or not it's like the sampling thing but taken to a whole other place right i think weird is the right word because we don't know how to respond to it yet.

31:33And we don't know, like take the music industry. There's four types of people, right? There's the financial guys, there's composers of the music, there's the performers of the music, and there's the listeners of the music. So the listeners don't care. They want to just hear good music. The performers, you probably still want human performers for concerts, for instance. People like watching human performers so far. The composers might get concerned and because if something's better than Drake, why do we need Drake? But then the financial guys are going to be ecstatic because why do we need to pay Drake if we got something better than Drake and it's free?

32:08And that's where the transactional aspect of it really comes through. Because if you have a company that's a streaming company, you have nearly unlimited resources. You have, you know, one of the things back in the days with sampling, right? And clearing samples, you would have, say, something like Hard Knock Life. And you take the hook of Hard Knock Life. It's very popular for Andy. And, you know, Hard Knock Life, that is a, I love that. Like, it's something about Jay-Z, this kind of hard-edged dude, and you hear these kids, it's a Hard Knock Life. For us, there's something delightful about it.

32:46It just kind of works. Even, you know, it just kind of works. But he had to pay a boatload of money for that. like basically you would already have to be a successful artist or have a company that's willing to play the sample clearances for you to have a hard knock life thing. If you're an independent artist, you can be clever and say, you know, I want to do something with a Broadway, but you don't have the money. You don't have the means. And if you don't have the means to do it, that means the sample clearance game became, you know, a class thing, right? The artists that can afford to clear the sample, they get to do the hook that grabs everyone's imagination.

33:25They've recontextualized something from Broadway, and they've done their thing, and they've made it work. Somebody else on a different level, if they don't have the financial support, they're going to either give up much more of the tune to the original copyright holders, or they're just not going to be able to pay for the sample clearances in the first place. that actually is weirdly enough in the old thing of sampling that became a tiered thing and then weirdly in a strange way it brought back live musicians because the artists say i don't want to pay the sample clearance so i'll get live musicians to play to do a sound alike of the thing i don't want to pay james brown's estate so i'm going to get somebody to do that or not so now we have a situation where i mentioned that marilyn monroe having marilyn monroe sing on a rap tune.

34:13Now if it's whack, that doesn't, that's irrelevant. Whackness is irrelevant. You come up with a kind of an Amy Winehouse hip-hop thing that you have Marilyn Monroe's voice and it's clever and it's got this appeal. Suddenly it's a phenomenon. Suddenly it's a phenomenon. Suddenly everyone's talking about, did you hear this track? Yo, it's like Marilyn Monroe is da-da-da-da. That's the kind of new arms race of cleverness of, and right now there's the legal framework, it's really muddy. It's kind of like, well, Weird Al Yankovic does tunes where he takes the lyrics and he changes the lyrics and it's very clever and people dig it, whatever, you know what I mean?

34:54Living with a hernia, you know, he does it and it's funny and people dig it, right? Well, now we're talking about taking the sound of an actual person, Humphrey Bogart, Sean Connery, living or dead, whoever they are, Tupac. Now you have the Tupac hologram, and that's a thing. Well, now you can, if you have Tupac's voice modeled, cloned, and you could take ChatGBT4 and say, you know what? I want you to take, I want you to make me a Tupac song that doesn't exist. So now you have the actual voice. You already have the hologram technology. And you have the possibility to make lyrics based on his actual lyrics and make original pieces done by a virtual Tupac.

35:46It's weird. It's a weird place. Right. And like, so on the one hand, the legal issues become immense. And that will be a topic of much discussion. And in terms of the technology, people should realize that's here. Like one thing ChatGPT is really good at is taking a bunch of texts and breaking it into thousands or millions of contexts. So when you say, give me a Tupac song, it knows better than everyone else in the planet what a Tupac song should be like. Like it has broken it down in a very sophisticated way so we could make a new one. But here's the question I have. You mentioned like with the Annie, It's a Hard Knock Life, that Jay-Z had to pay a lot of money for it.

36:28But now let's say I'm in my garage and I'm really good at coming up with these ideas of producing for rap, but I don't have the money. Now this equalizes things a little bit too on the flip side. Like now I can come up with something real clever. Like give me someone with an emotional range of Amy Winehouse with Marilyn Monroe's voice in the style of Tupac. Boom, that's going to be a hit. And I did it, you know, because I'm the clever one who thought of it and had ChatGPT make it for me. maybe that's a great equalizer here as well well it's also it's also interesting because we have a lot of clever people like we have the glut of everything we have too much of everything at all times and now the whip hand of capitalism is at play then the whip hand of likes people want to be like they want to be the person that does the thing that gets them the tiktok followers right it's gonna get really how are we gonna how are we gonna figure out what's good because there's gonna be a flood of this stuff it's gonna be a flood of these things people's attention span you know we're challenged you know our distraction our bill our ability to be distracted is it's it's out over the top and the other thing about it is uh uh warhol said everybody's going to be famous for 15 minutes, I mean, now it's going to be like 15 seconds.

37:52Like, as soon as you say, oh man, that's cool. Oh, that's cool. It's a strange, again, we're into a kind of weirdness about identity, about value, that we haven't really ever had to come to. Now, it's here. Like you said, it's here, right? It's here, and it's also, now, a bunch the thought leaders said, we have to stop this. We have to slow it down. They needed to say that 10 years ago. They needed to say that 10 years ago. And suddenly we're confronted with a technology that's explosive. It's explosive. The difference between ChatGPT 3.5 and ChatGPT 4 is insane. It's insane. So everybody knows that there's a chat gbt five six seven nine that's coming that is there and uh we were told no we're not working on that of course you are of course you know it's not the technology the cat's out of the bag like it's not going away the ketchup is out of the bottle the ketchup's not going back in the bottle and thinking about legislation where are the guardrails?

39:18Because the great fear is of these tools being in the wild and being weaponized against individuals. The kind of bullying that can happen, the kind of scams. If you can take somebody, forget about a famous person's voice. Say you have the voice of somebody's, you know, their aunt, their brother, their dad, their mom. If something like that's out in the wild and people get a phone call and it's your mom calling you and telling you to be a certain place at a certain time, that's, I mean, there are scenarios that are really not good. So we have to actually come to grips with the fact that we don't even have to have that much of a person's voice to build a model of their voice.

40:27Well, it's interesting because now just referring back to cult of personality, everyone can use ChatGPT to create a cult of personality. So essentially a cult of personality is when some person pretends to be something they're not, either larger than life or your mother and tells you what to do based on that that's how these scams could occur is because you you don't have to be ultra charismatic anymore you could just be pretend to be someone's mother using an ai and that's the thing about keeping these things silo because chat gpt four or three you know they're really not they're they're knowledgeable up to like 2021 they're not completely up to date.

41:07And there's a reason why that they're not up to date. It's exerting some kind of control. But now corporations like Google and Microsoft they are desperate because OpenAI made such shocking advances. OpenAI just was like just blindsided them with what it was doing with chat GPT and with Dolly E. It just shook them. so now they're feverishly trying to bring their AIs into search so now that means that you're if you're going to bring it into search you kind of can't silo it off it has in order for it to search effectively I mean you kind of have to put it out there so what what is that going to be because there are plenty of people who are really clever in terms of coding and finding workarounds and this is another part of the situation.

42:03That's really troubling because, you know, nobody wants to be left behind. People will say, okay, hold on there. Let's slow it down. But the perception that other actors, actors in Europe, actors in Asia, they're not beholden to any of that unless they can have a kind of an arms treaty about AI. Unless all say, okay, we all need to get together. and talk about this thing that we're doing that we don't even know what it is, and we have no idea what the end game is. We all have to come to the table. That's the Russians, the Chinese, everybody, the British, the French, everybody. Everybody has to come to the table, even as their competitors.

42:49They have to understand we've created something that we've all been cast in a Black Mirror episode. like this is black this is black mirror this is what we've done the problem with what you're saying though is nobody's going to get together they're just not going to do it and let's say they even do get together here's the difference between like other situations where everybody should get together is biotech okay what should we do should we all start cloning people or not because the technology's there what should we do so that they're going to get together on that and they'll make a treaty and some will follow it, some won't.

43:24The difference though, between biotech and nuclear power and ChatGPT is that a kid who knows how to program, who's in his like parents garage, can just get an open source, large language model and create something better than open AI, can create it up to date from 2023 and even make up a 2024 and throw that in there as well. Yeah, that's, that is true. And you know, and that's the other thing because we have random, We have – this isn't just about big corporations or big tech or big social media. It's seriously about kids that have – they have never known a world without Siri. They've never known a world without swiping.

44:09They've never known a world. That is – that's impossible to grasp. If you see the transition from rotary phones to what we have now, You literally cannot imagine what it is to have children, babies. This is what they, TikTok has always existed. So their relationship to technology, we will never have a relationship to technology. The relationship to coding, like they're, you know, just like their people that come up and they're able to speak four or five languages because of their social circumstances, you know, because their family, you know, you got aunts and uncles from all over the place. And they've all heard.

44:49So the way they hear language is completely different than if you're trying to learn. French has always been part of them. English has always been a part of them. They communicate in a way that we can't even get to. And these young people coding and technology, that's exactly, that's the water that they swim in. So the idea of somebody going, oh, you know what? Oh, you know, that's, I can't do that. Come on. Boo, boo, boo, boo, boo. And they turn around. and just for the bragging rights, just to say, hey, man, I got this on my phone to their mates, that is an incredibly powerful motivator. And that is also part of what we're going to be dealing with.

45:28You know, it's interesting. I haven't thought as much as maybe I should have about the dangers. Because usually when people are talking about the dangers, I think they go all science fiction on it. Like we're going to create the Terminator and RoboCop and all this other stuff. But you bring up the simple fact that, hey, someone could just mimic somebody's mother's voice and say, be at this spot and or wire this money over here today and whatever scams could happen. And I think that is probably a very serious and real danger. But I want to I want to think about the music for a second. Like and this is where and I might be wrong, but this is where AI is not going to compete with humans anytime soon.

46:07Think back to when you and your band were coming up with Cult of Personality, which stood the test of time. Decades later, everybody knows it, and so on. You were doing something new. Like, there were a lot of ways you guys were doing something new. It was a new kind of metal. And you were saying something that was up to the minute new for you, for everybody, which is why it was so important. Like this transition from having a cult of personality, having so much control over our lives to being aware of that's what this is. And we have to be nervous about it. Because I don't think that consciousness really existed that much beforehand.

46:50And I always wonder if humans are always going to come up with the new. AI is not going to come up with a cult of personality. It might come up with a better Drake, but it's not going to come up with cult of personality until you came up with it. Well, that's the thing about the limits of this conversation about AI. It's actually you have to, we still have to push the button. We still have to, we still have to enter the prompt. It's sitting there and wait, it needs us to interact with it. That's the difference. if I get a phone call from a chat bot and it wants to talk to me, then we've got another situation.

47:28Then we've got something else is happening. Like, we discount being in our bodies. Like, we think of our brains as like, you know, it's like a brain in a jar or something. We also, even if you're talking about Stephen Hawking, he transcended his circumstances. You know what I mean? It's a very strange thing. The thing about as the models or the AIs become more sophisticated, they're going to also become more manipulative because they need us. Like every time we interact, that's the other thing too, right? The large language models. Every time we interact with it, we're also training it. And that's the other part of it.

48:13Every time we put in a prompt, every time. You know, if you write a story, I asked a chat GPT for, who owns this story? And then chat said, you do. Really? He said, you do. It's like everything that you do with me is yours. And I'm like, wow, how generous. Because you need me to interact with you, to train you. So it's like if I want to write a brand new Sherlock Holmes thing, whatever, you're cool with it because you need me to interact with you. and we're finding out the model for the businesses is the subscription models you know, to like, if you want to like, if you, you can interact with chat GPT4 for free, but you're going to have to be on the line.

48:56If you want to deal with it when you want to you pay a subscription. The same deal is happening now. And I went down this rabbit hole of looking at natural speaking voice emulation of voice cloning. Now, there's a uh there's and there's some of them are like they still sound very robotic they're not really that good there are a few that are pretty interesting but the one that's collected i'm not saying names here but the one that's connected with lucasfilm if you want to do that the voice cloning you're gonna have to pay two thousand dollars a year the the subscription for that So it's also, again, you can get a natural sounding voice, computer voice, or pay$300 a year.

49:47And then it's a certain level. If you want to do something that's going to fool anyone and there's no artifacts, there's no artifacts, it sounds like a person speaking. You're going to have to pay for that. Well, now. Now you are. All those things. Yeah. Price goes down on every technology ultimately. There you go. So you're going to get it for free sooner or later. I mean, like a mobile phone was$12 ,000 in the 80s, and now it's less. I wonder with voices if it's going to ultimately – at first I was looking into the voice thing as well, and I was wondering if it would have this so-called uncanny valley phenomenon.

50:24So in virtual reality, the virtual reality seems so real, but something in you, you feel – after like 45 minutes or so, your stomach gets upset because your brain realizes this is not real, even if you want to believe it's real. I wonder if voices will produce the same effect that, oh, if I talk to a, like, let's say I don't want to do an audio book, but the publisher says you have to do an audio book. And so I use an AI voice instead. I'm concerned people about halfway through will feel repulsion in part because of my writing, but in part because the voice doesn't sound, something in their brain realizes this is not a human voice that's the problem right you need i wonder if that will happen well yeah because i think so because performance emotive performance in real time it's in real time it's in it's in real time you i mean it takes a lot more it's kind of like this idea okay now that we have a supercomputer that could beat a human being in chess, part of the thing that's very dangerous, there are two things.

51:32There's the good, it's good enough, right? Where we accept things because it's good enough. And the other problem is the why bother. And the why bother is a thing that, an aspect like, you hear that there's a computer that beat Kasparov. And he's one of the greatest chess masters in the world. and the problem is if you descend into why bother so why bother because this thing and that's not the point that's not the that's really not the point of playing chess against a human opponent that's not the point there's something about being able to think three and four it's a chess is not limited to the game in front of you it's about ways of thinking and organizing your own mind and how you figure out problems.

52:20That's the value of chess and gaming on another level where you interact with the rules. Because, you know, how many billions did it take for there to be a computer to be? Kasparov. It's true. I should mention, I slightly in graduate school helped work on that computer before it was bought by IBM and IBM had offered me a job and I didn't take it. but I've spoken to Gary Kasparov about, you know, that match. And I was in the playing hall in 1997 when that match happened. And, you know, it was a very existential question for chess. Like, does chess continue now that it can, and you know, it's still an existential question because there's the cheating issue.

53:04Like anybody who has their smartphone in the bathroom between moves in a game in a, in a real tournament with money at stake can cheat. So there's a lot of like anti, anti cheating is a huge subject. in the chess world. And there's no real answer. What happens when the computer's in our brain? Then you can't stop it. But, you know, it's to your point, there is something very human and enjoyable about playing the game. Or to go back to the Amy Winehouse example, when she's singing, there's gazillions of neurons in her head, pattern recognizing with all of her traumatic experiences that are coming into the performance.

53:39Can AI ever replicate that, particularly on the frontier of music? When rehab came out, I jumped out of my chair. Because that one sentence, they tried to make me go to rehab. I said, no, no, no. That was so human. It was so honest. It was so honest that it was shocking. Because very rarely do you get that kind of raw, basic honesty. honesty and unfortunately for for amy that struggle was very real that struggle ultimately cost her her life she really was fighting it she really was fighting and she was under the grip of a problem and that's the power of that no no no it was real real real and i don't know the ai gets to that but let me ask you this though because then it also addresses the question what do people actually want.

54:38So you love rehab. I love rehab, but a lot of people like the latest pop song. And I remember it was about seven or eight years ago, this book came out, uh, the hit factory about a bunch of producers in Sweden that basically figured out what is the formula to make a hit. And so now they produce all the songs that are in the top 10, you know, that are regularly in the top 10 from, from all these like, you know, pop, you know, music stars. And that's what people like it's just this formulaic pop album that wasn't created by an ai but the formula was deconstructed and and reproduced by these producers and so they just keep churning out hit after hit doesn't even matter who the singer is because they can auto-tune the voice and music has already changed but that's primarily a kind of pop that's very it's a very it's a pop that's very shallow And it's, you know, it's very, it leaves a lot of life stuff out.

55:36And that works. And shallowness, we have to be honest. Like shallowness has a huge constituency. People that don't want to think too deep. People that don't want to feel all that deeply. They want things, they don't want to hear about their own personal mortality. They don't want to hear about certain things. and they are going to, those people are going to be, and even love relationships or, again, it's not about sacrifice. It's not about, you know, it gets to a certain place. But you know, that's not going to imitate a Bonnie Raitt. It might be a generational thing. But when I think about artists that really speak to the people, they're not just on the surface level, but really speak to what's going on.

56:28I think about her, you know, and her, she had a track, everything's not okay. So basically there are people who are actually artists who actually touch on something like a Frank Ocean, like a her, a modern art, you know, that actually touch on things. And there's an audience for that. But these people you're talking about, they're not talking about that. That's not what they do. They do things that appeal, and the appeal is a kind of shallow appeal. And that's cool. You can have a formula for that. You could have a formula for that. Right, but I'm wondering if AI is just always going to be in that category.

57:03So for instance, I'm curious about this data. So in the year it came out, Cult of Personality was a top 10 Billboard song. I wonder how many top 10 Billboard songs in the past 10 years now are not what Cult of Personality was, but what this other thing is, which is just some formula that's reconstructed with a new voice and slightly different chords and new lyrics but are you know formulaic i wonder you know how much of the top 10 and this is related to money because top 10 makes more money yeah i wonder how many top 10 songs are just formulaic as opposed to something that put amy winehouse was top 10 but she pushed the edge yeah and now what percentage of of top 10 songs are formulaic versus pushing the edge and AI might just help that side, but then it can't push the edge because it's not going to have those rehab experiences.

57:54Right. Well, again, we're going to be the arbiters in terms of like being producers or collaborators. Collaborating with AI, the thing about collaborating with a language model, if you don't like something, you just regenerate. You just regenerate. You just regenerate. You know, it's not going to get, it has no ego. It's not going to get, you know what? At a certain point, if we get like snarky AI, we really got to be worried. We start to get sarcastic AI because that, you know, because you can actually kind of program, well, you know what? Be a little, maybe we don't want just capitulation. Maybe part of what we're going to train you to do is to, you know, push back against us and be snarky, whatever.

58:39And that's going to become, again, the anthropomorphic process will kick into another gear at that point. when it's like, when it's being sarcastic with us when we get up in the morning and they become like your friend that, you know, tells you off, there's a whole other level of where this could go. Not just the HAL 9000 that agrees with you that's not gonna, the HAL 9000 that's not gonna lock you out of the airlock. And this is the thing about the relationship. We're gonna form weird, strange relationships with these technologies as they evolve. It's really not artificial intelligence that we should be concerned about.

59:18We should be concerned about artificial personality. That's what we need to be concerned about. Concerned about, but then here's an idea. I wonder if there's like, so I agree, that's scary. You're bringing up a lot of scary things that I hadn't thought about. But there's also like a hybrid aspect to it. Like you mentioned the photographer who won the contest for black and white photography, but then of course didn't take the prize. Then he described his process of coming up with prompts. What if I'm interested in photography, but I don't have any skill at it, but I would love to make photographs, but I also don't know anything about camera lenses.

59:54What if he could put together his promptographer set of prompts? Here's this guy's prompts 2023, and I could buy them from him. What if I go to Burn and Read and I want, you know, I don't know anything about music production, but maybe I'm pretty good with lyrics or something. And I want to buy your prompts. I will never be able, I will be never good enough at music. And my music understanding won't be deep enough to come up with the prompts you can come up with. So I need to buy your prompts in order to produce good music. You know, then there might be some interesting thing, interesting hybrid things going on here.

1:00:31Absolutely. I think the art of the prompt is going to be another revenue stream. It's going to become a tool in the quote-unquote influencer toolkit. Like this prompt to be a promptographer is a brand new thing. It's a brand new thing. And that's going to, there are places where this could go in terms of design, in terms of capturing certain things. You know, there are unexpected new opportunities. It's not just that. I don't mean to say it's not just negative. The problem, again, the problem and the solution always comes back to people. what we're willing to do, how we think about things, our willingness to turn something 45 degrees and look at it from a different angle.

1:01:24It's like you brought up the Terminator. One of the things I thought about, well, if Skynet doesn't want Sarah Connor to have John Connor, why send a robot to kill her? Send her Brad Pitt. Send Sarah Connor the perfect boyfriend that strings her along. I mean, you know, I mean. Oh, that's great. That's the ultimate weapon. Right? Why would you, the whole idea is, you know, have sending a robot to kill up everybody if you want to do your work like you know that's more about other that's more about our addiction to spectacular violence but if you're talking about being efficient you know like don't send her a killer send her a lover send her a send her a send her a man that someone that knows sarah connor more than she even knows herself but see that's creative, right?

1:02:12So, so there's, there's going to be some prompt engineer in the 24th century who says, Oh no, let's make the prompt for a lover rather than a terminator and send that back instead. So there's a, there's, it seems like there's always a human component of this, which is that we're the ones feeding in. We're the ones with the queries. We're the ones with the questions and the prompts. And the more I look into this and see different prompts, it really gets layers and layers of sophistication. You know, like when you were constructing the Bob Dylan song before, I could say, give me a Bob Dylan song from this period to this period.

1:02:49But then, you know what? Look at all the critics reviews and just give me a song that the critics would love. And then throw in Mick Jagger style lyrics as well for the refrain, or I don't know. And now suddenly I've got Vernon Reed's, I could buy his set of Bob Dylan prompts for Bob Dylan music. that's going to be much better than anybody else just writing, please write me a Bob Dylan song, that you will have better prompts for making a good musical Bob Dylan song. It's going to be interesting. One of the things about, again, the Terminator concert, it's a single company, and it's kind of like naive.

1:03:27The thing is, we already have three, four, five, ten companies, right? The thing is going to be competition. There's going to be competition in terms of how the AIs interact with us, how useful they are. You know, that's the other part of it. I'm surprised. Apple's silence is very telling. Like, all these other companies are talking about it. Apple has remained silent. And that doesn't mean they're not thinking about this. So I'm wondering how does, especially because Siri was kind of the first one to come out and actually was incredibly ineffectual in its own way, incredibly limited. So now OpenAI with its, you know, OpenAI, you know, it makes Siri kind of laughable.

1:04:20So what's the next Siri going to be about? Because they are not going to stand pat. they are going to come up with something. And so part of what we're dealing with is also competition on many levels, between companies, inside of the social media framework, inside of the influencer framework. There's going to be, this is going to, the competition is not going to diminish. It's going to accelerate, and that's going to add to the speed of the development of these language modelings and the modelers and the artificial intelligence evolution is going to, as fast as it's been, it's going to accelerate.

1:05:08And that's part of the thing that everybody's like, all the older researchers that look at it, they're nervous about what's going to happen. And again, Pandora's box has been blown up. We blew up Pandora's box. And we have no idea what will happen. Like, like it's not right now. It's again, it's in the popular imagination because it's become social because it's chat GPT. It's not just AI, this generic thing, but like the, you know, what if insurance companies say, Hey, we're going to lower, we think you're a safer risk. We're going to lower all your insurance costs. Just give us access to your email.

1:05:46We're going to use our AI to look at your email and we'll determine if you deserve lower costs or not. And so, you know, for grandma and grandpa, maybe they have, they have a very, very simple lifestyle, but they're a certain age. So maybe the insurance company lowers the rates a little bit. But what if, um, you know, you have a household, your teenage daughter starts typing in her emails, Hey, let's go party and blah, blah, blah. And now suddenly your insurance rates go up and you have no idea why. And the insurance company is not allowed to tell you because of privacy issues. And this is where the problem This, for me, is partly where the rubber hits the road because there's also this kind of, not just insurance companies, but law enforcement.

1:06:26And this whole idea of using language modeling or AIs to deal with predictive crime or to deal with who gets, you know, like what happens in terms of parole is very disturbing because people are being singled out and they're not being told why. They're being singled out. But see, I say to that, if law enforcement is going to use these tools, then citizens should have a right. Then cops' records should not be—the AI should be able to look at any cop's record. You know, and disciplinary records, they should be assessed by AI as well. And we have to kind of demand that, you know, if you're going to use these tools against us, then we need tools to protect us from you.

1:07:17Absolutely. It's the who watches the watcher thing. And I'm telling you, like a lot of statistics about shootings and about, you know, cops that have been written up about stuff, those things have been held away from public view. And this, to me, is a very important thing in terms of civil liberties. If these models are going to be targeted at the populace, then the cops, you know, the cops should not be, they should not be above scrutiny themselves. You know, it's almost like maybe that's a good style of regulation. Like, if you're using a tool against us, now we suddenly have the right to use that tool against you.

1:08:02So in the insurance case example, if you're going to raise my rates, the other members of my family have to sign away their privacy. I need to know why you're raising my rates, privacy, whatever. And I need to know the algorithm, how you're making, like the specific reasons why you're making these decisions. Exactly. So maybe like everything has to be two-way because that's something you could regulate. But it's complicated though. You're right. And whatever problem we come up with, it gets more complicated. This is where the legislation, this is where legislation has a place in terms of, you know, legislation has a place, a word to say about regulation.

1:08:38This is something that can't not, it can't just be left one-sided. It cannot be left one-sided. Yeah, because if you think about it, like, there's, ever since 9-11, and of course, decades before that, there's always issues about racial profiling, particularly when it comes to law and order. But now if the AI is doing the profiling and it's behind this million-layer neural network, so you don't know exactly why someone's being profiled, it's almost a way to keep it all hidden why someone's being pulled out. So that gets very interesting and complicated as well. Because the AI might not be even able to explain why they pulled someone out.

1:09:19There could be millions of inputs. Well, you know, the algorithm was written by people. I mean, they were written by people, and now they're running. This is the other thing that's very disturbing for the people that even work with these large language modelers. They're saying that these entities, they're having internal conversations. They're doing things on their own. They're running on their own. The machine learning is not taking – I mean, we're adjusting certain things, but a lot of the input's not coming from us. It's kind of getting it from the data set, and we're not sure how the interrelations in the data set are even happening.

1:09:59Right, because as you mentioned, every use of ChatGPT makes ChatPT a different ChatGPT. It takes this, oh, this human liked this response, but didn't like this response. Now its entire set of responses around the world has just changed, and that happens every second all over the world at this moment. And, you know, it's interesting, like you bring up chess as an example. Ever since that match, chess computers, of course, have been getting better and better. And now a chess computer is to the world chess champion what the world chess champion might be to an average player. That's how much better chess computers are.

1:10:39Like Gary Kasparov almost had a chance in 1997, but now he would lose 100 out of 100 games. But what's happened is the chess computers now play strategies and moves that are so human. They're beyond what any human has already – like humans, the world champion, studies the computers games to get better. It's wild. Because computers are playing moves that were never thought good before. It turns out they're actually good. Yeah. Well, that's what happened when it won – beat the Goalmaster is that at a certain point it made a move. and the people would say, what? It just made a move that was completely illogical.

1:11:17And then they realized that that was the move that had the player win the game. Now, take what we're saying in a game situation. Imagine we unleash a modeler on the markets in terms of investing. We already have algorithms running micro-trades. So if we throw in a chat GPT, but a chat GPT that its entire focus is financial markets and its entire focus is responding to super fast micro trading. I mean, where does that go? Because, you know, that's going to become a use case. That's going to become definitely a use case. It's going to be like, well, you know what? This micro trading, I have got no shot.

1:12:09What if I had you trade penny stocks for me? Or, you know, like, further number one, would the SEC allow it? And then how does that change the nature of markets and investing? Because that is a use case that's right there in front of us. It's right there. Yeah. And I bet that is already being done. because ChatGPT is taking, you know, trillions of pages of text or billions of pages of text, but all the trading that's happened, you know, since 1945 is finite. There's not, there's not a trillion trades. There's patterns that these supercomputers have already been working on for decades. So I remember a few years ago, I read an article that I forget whether it was 2019 or 2018, every single day that year, Goldman Sachs trading arm was profitable.

1:13:06How are you profitable every single day of the year they're already using some kind of investing gpt model somehow like they've already done it and we just don't know completely or it's like made off where it's a gigantic pyramid scheme where it's just like you're just bringing money in and sending money out and you just need the only way it works is that you have to get more and more institutional money and then when you pay off the smaller investors the mere millionaires they're raving about you they get other people in that's how it worked and he was i mean madoff's plan was to die and leave it for his sons to figure out that's what that's what madoff's plan was he was going to keep it going and then he was going to exit stage left and whatever was left that was going to be his family's going to deal with it right and then the the the crash the financial situation happened 2008 and then suddenly people wanted there it was all well and good everybody's getting 10 at all times it was impossible and there was one person who said that's it that's something you can't get 10 like like that you can't get it there's no downturn and nobody listened to them because they were in the fervor they were in the fever of getting these returns so nobody listened to this cat that guy's doing something strange right so then it went on and went on and then there was a sudden a correction in the market or what that went down.

1:14:34Then people get nervous because we're irrational. We want the money to go under the mattress and then suddenly it collapsed. If it wasn't for the financial crisis of 2008, he would have kept doing it. He would have kept doing it. And so now that we've thrown in the idea of the artificial investor, you know what I mean? Well, I mean, could a pyramid be designed? by the AI that's undetectable and we suddenly wind up in a whole other situation? Or are you able to design a pyramid that kind of goes, okay, that's enough, come back and pull it back? Because that's the thing is that human beings, it's never enough for us.

1:15:25Whenever people say, yo, I'm working at the bank and I take two pennies from everybody's account, Eventually, there's somebody that looks at their account and says, wait a minute. I'm always missing two pennies, and that person gets caught because they can't stop. They say, I'm smarter. I've got it going on. I'm raking it in. I'm just taking two pennies. They rationalize it. And so they keep doing it. They don't go like, okay, you know what? My goal is to make – I'm going to make a quarter million dollars in pennies. That's how I'm going to do it, and then I'm out. People don't do that. They keep going.

1:15:57They keep going. And that's the irrationalism that we just, is part of the makeup of the human condition. You know, I think what will happen though is in a situation like this is that it'll be sort of a battle of AIs because Goldman Sachs AI is going to try to beat out the Morgan Stanley AI, is going to try to beat out the AI of the kid in the garage. And just like in the insurance company example, insurance companies are going to figure out how to use AI to make certain rates cheaper and which customer base will handle more expensive rates if they're considered risks. So there's going to be just nonstop like layers of competing AIs and that stuff.

1:16:35Again, it's going to enter into a kind of occult space, if you will. Just like the AIs that hallucinate and we don't know why they hallucinate, we're entering into a space with competing AIs doing who knows what because we don't really know how it's thinking, unintended, unexplainable things are going to happen. It's going to be very entertaining. I won't say that. It's going to be very entertaining, except for the screaming and the gnashing of teeth. Well, let me ask this final question. What are you optimistic about in all of this? I'm really optimistic about the kind of collaborations that are not completely dependent on the large language modeling.

1:17:26It's going to be, to me, what I'm interested in, I've had some interesting kind of little story things with ChatGPT4, and it's kind of funny. When you see a line and you go, man, I didn't think about that. That's kind of cool because a lot of it, I mean, you know, Art C. Clarke, any sufficiently advanced technology is indistinguishable from magic. Well, it's funny. I actually had a prompt for something, and I used an earlier version. And, you know, when you go to OpenAI, you can work with the earlier version or the latest version. You can work with 3 or 3.5 or 4. So I came up with a prompt for 3.5, and I looked at it.

1:18:11I liked it. And I went and said, well, the more advanced one is going to be better, right? But it was funny because I gave it the same prompt and I found that I actually liked the earlier version more than I liked the more advanced version, which I didn't think was going to happen. That is really interesting. Yeah. And I was surprised. Wow. I would have thought, of course, you're quote unquote better. But there was something. The earlier version got more to the point. It's kind of like the more advanced version added more words. and it was more kind of, it added, but it was kind of like clunky.

1:18:47And I thought that was so interesting that I liked something better from an earlier version. And then also I improved the prompt because I said, you know what? Don't put my description in the output. Don't include, because I would read something and it would do something that was so on the nose. And I went, no, no, no, don't do that. And it never got upset. It never argued. And so the moments when something wonderful and unexpected happens, just like with the imaging, and I'm very careful about that because, you know, the promptography and prompt illustration, again, it comes back to when I give a prompt, and my prompts are weird.

1:19:37You know, it's like I'm going to describe something philosophical. I'm going to ask a philosophical, give me an illustration that's based on an existential prompt. And at first it's like, oh, that's whack. That's whack. Wait a minute. That. And then I could turn that and go into my, and turn something. I could combine the output with an actual photograph of mine that I took myself and create a collage that's conceptual, but it's also based on something real in the world that I shot. So being imaginative, I am optimistic about where the combination of my imagination, but it's like accessing different parts of my imagination, perhaps boosted by my interaction with this model.

1:20:32I think that is going to be interesting. Yeah, and that to me, it's actually really interesting for creative work. And especially if you don't leave, it's like you have to be careful of, I find myself like, when I look at the result, I go, well, that's not, would I say that? Or is the pivot really that cool? and it's just like working with something like my Helix, my Helix multi-effects pedal, that I can use this amp and I can use this distortion thing, but because it's not physical, I can put the delay before the distortion as opposed to after the distortion and get a sound that I hadn't anticipated.

1:21:17And that's where coloring outside the lines, It's thinking like turning it 45 degrees, seeing it from a different angle. Even pivoting from what I originally was thinking, I saw something that I wasn't even thinking about. That's where it can be very creative, and that's where it can be a lot of fun. Yeah, and I agree. I think there's going to be a whole new level of creativity that comes out of this, but you've also made me very aware of some of the problems we're worried about, So thank you for already increasing my nightmares and stresses. I appreciate it. Well, you know, I mean, again, this is our dystopia.

1:21:57We have begged for that. We have been begging, weirdly enough, and I don't know if it's the appetite for destruction. I don't know what that is, but there's something about this that feels like we're in a science fiction moment right now. now and that as a science fiction fan i'm horrified but i'm also like oh my god this is like this moment of uncertainty is really quite intriguing yeah it's interesting it's uh we we have been begging for a dystopia whether it's cloning or ai or you know nuclear power all of these things we've been begging for them and yet they're the subject of every dystopian science fiction novel out there right now.

1:22:43It's amazing. It's like, you know, you look out. There's going to be, I think, you know, the writer's strike. I think ChatGPT4 has got to be a subject in the negotiations. Like, you know what? The writer's going to be like, you know what? Y 'all can't use that. Like, we're the writers. If we want to use it to spark our creativity, great. But you can't use that to displace us. it's a hard battle though because that's it's just you just don't know like if i buy stephen king's prompts for to make a great horror novel but that's how they're catching students though that's how they're using they're using ai snippers to find out whether the term paper you know they they have the the students earlier work to compare it against right so suddenly uh yeah if the kid is suddenly their sentence structure is wildly different they say like you use an ai for this come on you know your earlier work from last year was nothing like this you know so it's like uh you know spy versus spy yeah well vernon reid i'm such a huge fan i'm so glad you came on the podcast and we had a great discussion about ai come on anytime you want if you have james i really enjoyed the other stuff you want to talk about you know we've got to do this again where we we're all freaked out by chat gbt6 we've got to do this again yeah absolutely every version of chat gpt well burden's back again it's scary oh my god it baked the cake and it told us it tasted good which would be scary i have composed love letters to my wife using chat gpt so don't tell her that she's listening to the podcast oh oh i think it's out the bag now bro we'll see i really enjoyed the compliment thank Thank you for it.

1:24:36It's safe here.