ARTificial Intelligence

Speaker 1

when it comes to artificial intelligence, AI, the question being, can I be replaced? Phil, do you think I could be replaced? mean, as special as I am, as wonderful as my comedy is, my jokes, do you think that I could be easily replaced? I don't think the computers are quite there yet, although we could get a likeness of you and get it to do some funny things like dance the mariachi band stuff or put you into humorous art situations, but that's about it. You know, because I have to admit, you know, I'm a I'm a comedian. But I also I'm a procrastinator for sure. I put things off a lot. I have a lot of projects that I'm supposedly working on right now. And sometimes I'm lazy. And I do kind of wish that I had, you know, kind of a magic box that I could just type in. OK, give me a new five minute stand up set on. immigration or any controversial issue and I want it in my voice. I want it the way that I would sound and please spit it out for me. And something like that somebody attempted a couple years back when it came to George Carlin. A person put a prompt into AI to write an hour long George Carlin special and wouldn't you know they spat out an hour of George Carlin-ish stand-up comedy. I'd like to start off with a heartfelt apology. I'm sorry it took me so long to come out with new material, but I do have a pretty good excuse.

Speaker 2

You know, I think it does capture the essence of some of his jokes and his style of humor, but it's not quite a perfect replication. There's a little lacking originality. And you think about it, part of this is because of what AI is. It's an algorithm trained on a body of existing work outside of the AI. And if you don't have that, it's not going to be able to generate something above and beyond the material it has to work with without it. taking great liberties and inventing and going in a new direction. I would love to hear basically an updated version of that, in particular with the voice, because the voice is pretty off when it comes to George Carlin. I think what put the kibosh in that is the George Carlin estate. They either sued or they gave a cease and desist. In a way, I'm wondering, is that sort of a protection that artists have in the age of AI? Can the court system basically help artists to stop from being exploited by AI? Yeah, part of this comes into the copyright issues around your likeness as your image. And as both a living person, a performer, as an actor, as a singer, someone like that, you have rights to your licenses and images under the copyright systems and the current intellectual property systems that we have in place. And if someone is recording like an album in your voice that's not really you, that's kind of seen as borrowing from you. It's distinct from being an impersonator. It's actually trying to replicate what you are doing in your style. Although the flip side of that is we want some room for artistic expression and AI itself can be exact.

Speaker 1

And you have artists who I think are kind of getting ahead of the game. In particular, you have actors like Matthew McConaughey and Michael Caine, have signed these voice deals with AI companies. Michael Caine is a little up there in age. Yeah, and he's got an iconic voice. That's one of his best assets that he has that he can record a bunch of clips and he'll be with us for a while. Is this immortality? Is this the way to become immortal? And also, is it a way to just make sure that your family and your loved ones are making some more bank while you go to the other side? There you go. I mean, I have to think like back to the 19th century when they start inventing the wax discs that recorded voice for the first time. you these people like Thomas Edison was one of the major contributors to this. He went around with the discs and he tried to record famous people. There are early clips of presidents around to the turn of the century that were being recorded. Our famous opera singers, Enrique Caruso was one of the major recording artists of the early 20th century and they got him his voice down on.

Speaker 2

early records just so they would retain that. And the idea was that, you know, after this person passes, they're no longer able to appear in the opera hall and deliver these stunning performances. But now we have a recording that kind of makes it timeless. you

Speaker 2

So this is kind of like a new evolution into that. Not only do we have a recording, we have a computer algorithm that can manipulate and apply and bend that recording to get it to do new things. I do wonder when Michael Caine passes, hopefully this episode gets out before he passes because I don't want anyone to think there was a conspiracy here. Actually, that would be great. People would be like, my God, they got Michael Caine to do the interview on Happy Hour Econ? That's amazing. But kind of with that, I do wonder where will Michael Caine's voice? We get AI Michael Caine on.

Speaker 1

be popping up? Well, Michael Caine's voice eventually becomes something that's kind of in the public domain. And you'll get to use him for commercials that he might not otherwise have done or wanted to do. Or, you know, some, you know, weird Michael Caine, kind of fans out there who kind of want to produce Michael Caine fan fiction, you know what, it seems like there's so many possibilities for the future of AI that could that could be fun, but also weird. Weird, very weird. And you know, we just saw an earlier iteration of this come through as the copyright started to expire. on cartoons that were made in the 1920s and 30s. The famous one being the Mickey Mouse, the Steamboat Willie cartoon. You know, as that moved into public domain, well, the first thing they did is like Disney would deploy its lobbyists to try to extend the deadline on how long they maintain full control of that. But then as it moves into the public domain, you start getting some really interesting things like there's slasher films that are done in the style of Steamboat Willie. budget and it's probably completely antithetical to the family image that Disney liked to project for itself, but now it's public domain, they can no longer exert that kind of control. It enters into a new type of a marketplace. you wonder about this, an actor who has recorded his voice into AI being brought back 100 years from now. in a film or maybe as a hologram or whatever technology we have at that time and deployed in ways that are antithetical to that actor's image as they would like it preserved. The question is, do we have a legal structure that allows their estate or their family to exercise some commentary on?

Speaker 1

For those of you out there, you're all probably hoping and wishing, like, can we finally get a Jaws five? For those of you who don't know, Michael Caine was in one of the Jaws. And I remember listening to an interview he was doing. He never saw the movie, I don't think, but I think he was able to buy his mother a house or something like that. Which I think, you know, being in a Jaws movie and being able to buy a loved one a house is a pretty good trade off. Or you get things like unfinished works and one of the big talks, the Game of Thrones series, George R.R. Martin has been sitting in writer's block for the last 15 years to finish the book that's supposed to come out on that, to wrap up the series. And he's getting up there in age. Do we upload all of George R.R. Martin's works into an AI machine? turn it loose and figure out what kind of a story it comes from. That's an intriguing possibility. It also has a question of authenticity about it. Although I see this as an evolution, it's an age-old problem. There have been many, many instances of books and compositions that were not finished at the time that a masterwork had died. Mozart's Requiem, for example. And he's working on this up until his death and it actually takes other people to come in and they fill in the bits and pieces to get this unfinished masterwork into recording. AI, I think, becomes an interesting tool for that because it takes away some of the element of bias, whoever the interpreter is that's trying to finish an unfinished work. is going to bring in bits and pieces of their own persona to that, whereas AI is an attempt to algorithmically do it through statistical probability from an uploaded body of work. So think that's really interesting because you could get points of comparison. What does AI do with this material versus what does another author or another composer do with this material to finish an unfinished work? It gets really interesting.

Speaker 1

I think one of the projects that I'm working on is I'm writing a feature film and you know, whether you like it or not Films that work have a structure to them. Yeah, there's a three act structure and there's rising action and the denouement and all that so I've been using AI as a way not to write the script but to basically show me how Films that are kind of what I'm going for work how they work structurally. So it's sort of like, hey, can you beat out far as gump for me? You know, what is act one, act two, act three, so I'm able to see it structurally, and then kind of use that as a way where I can map on my own narrative and say, Okay, this is where, you know, by page 30, I should be here in the in the script, and all that. So I find that a tool. Yeah, it's a it's a it's a tool. It's it's really helpful. It's kind of like, you know, I didn't go to film school. So it's not like I was in the belly of the beast, reading just script after script and watching movie after movie and getting that education. So it's definitely helping me in that regard. Absolutely, absolutely. I'd say the same thing with my academic research, because I have plugged paragraphs into AI and said, hey, critique this. And what it does, it runs through its algorithm and it spits out a response. And it might say, well, your argument's a little weak here, or you could use some data here to reinforce this. And that helps improve the final product that I turn in as my finished paper. So yeah, it's a great tool. some cases it's a bit of a crutch. And I think that's the sweet spot that we have to shoot for. AI as an assistant, as a tool to do things that involve the mind is very powerful. It's efficiency improving and it's great. But if you just turn it loose and say, go generate something on its own, you get what's referred to as AI swap. Verbose but incoherent paragraphs or bad cat videos that are...

Speaker 2

completely computer generated and obviously. So one thing a lot of people notice there were sort of these early AI videos of Will Smith, the actor and musician and controversial slapper of Chris Rock. There were these videos of AI of him eating pasta. the early ones are nightmarish. They're so weird and creepy to look at. And it's like six fingers and all that. Yeah. It's, is he eating the spaghetti and his fingers and the sausage just it looks so weird. Now, it looks like Will Smith is eating pasta. And you know, I'm looking at that. And I'm like, that's really cool that we've advanced, you know, the technology where we could look at this. But other than that, I'm really not interested in it. And something that I've kind of noticed with with AI, the the slop, you know, as you put it, is even though there are some like interesting ideas like I saw one it was I think it was like a mix of kind of a mashup of Harry Potter and Apocalypse Now. Right. Okay, so we get to kind of play out these, you know, fantasies of what would happen if Harry Potter was dropped in the middle of Vietnam. But, you know, outside of like those interesting visuals or the, you know, the idea itself, I'm really not that interested. I'm not connected to it the same way that I would be with real people.

Speaker 2

Yeah. Yeah, exactly. And, you know, I've played around with AI as a tool myself. I have a little Instagram account that's kind of economically themed, that's AI generated photographs. So I take Karl Marx and like, you know, he's, he's deceased. He's out of copyright. He's also a fun target to, to poke in various ways. And the Instagram account is taking historical pictures of Karl Marx and putting them into scenes in the modern world. The very first ones that I tried. they came out a little weird. It's like the Will Smith video. I asked it to generate a picture of Karl Marx doing the laundry and it spits out kind of like this really abstract photo of Karl Marx hanging off of a laundry clothesline. So it's kind of strange. I thought it was funny enough to post on Instagram. Nonetheless, had fun with it. But if I do that now, I can get, I can say like, AI, please generate a picture of Karl Marx on a pogo stick. in the middle of the park with Fidel Castro. And it'll do that. And it'll not only do that, it'll animate it. And with reasonable precision. So I think that's actually kind of amazing and fun that it can do that sort of a thing. And yeah, it's a great humor device. But again, all built on the human prompt of what I'm trying to tell it to do. Another example from real life at the Independent Review, which is the journal that we publish at the Independent Institute. We have scholarly articles on economic content, all sorts of things, but sometimes we do thematic issues that require some cover art. Well, it turns out there are no photographs of Adam Smith.

Speaker 2

There are like two known images of him from his lifetime, but they're both kind of bad. One's a side profile and the other's a painting that may or may not be of him. So how do you make images of Adam Smith without using the same two images that everybody else uses? we can actually plug those into AI and ask it to construct Smith in a studio. So if we wanted to do a cover art for Adam Smith writing at his desk with the Wealth of Nations on it with enough prompts, I could get a reasonably accurate depiction based on those two images. And I think ultimately too, if you were to do something, a piece of work that was more expansive, like a docu-series or feature length film about Adam Smith, I think you're probably gonna have better audience response, human connection, if we're using real people. At this point, I really just don't think that AI is going to be able to replace that. And interesting enough, Ben Affleck and Matt Damon, They were guests on the Joe Rogan experience recently and the subject of AI came up and you know, obviously Ben Affleck and Matt Damon have had, you for decades, they've had really successful careers in Hollywood and making movies. And one of the things that Ben Affleck talked about is... For example, if you try to get Chachi BT or Claude or Gemini to write you something, it's really shitty. Because by its nature, it goes to the mean, to the average. It's a useful tool if you're a writer and you're going, ah.

Speaker 1

And it's pretty beautiful.

Speaker 1

What's the thing? I'm trying to set something up where somebody sends someone a letter but it's delayed two days and gets and it can give you some examples of that. Actually it's very likely that it can it's going to be able to write anything meaning particular that it's going to don't think it's

Speaker 1

for or an imp I'm gonna be making movies like from whole cloth like Tilly nor like that's bull. I don't think that's gonna happen. Whether it's you're going into chat GPT or or anything else. It's compiling all this stuff. It's spitting out the average response to it. And when we're talking about artwork, when we're talking about, you know, movies and the things that are really going to reach people, you can't just do average, you gotta you gotta go above and beyond. Yeah, so this is something. Economists and statisticians really dig into this, although the two fields are very closely related, the concept of the normal distribution. And that's as, you know, if you flip a coin, you have a 50-50 chance of heads or tails. You keep flipping that same coin and recording it over and over again. The more times you flip it, you're going to get close to a 50-50 distribution of those coins. That settles in around a midpoint, essentially. And these are patterns. that tend to play out just statistically in human life. So we cluster around the midpoint, we cluster around the mean. But oftentimes in economic life, the really innovative creative destruction events, the discovery of something new, we refer to that as a tail event. It's an exceptional outcome. It's not the normal calculation that's made. Someone has figured out or noticed something completely different that no one else in humanity has ever done before. They make an astounding innovation, whether it's technology, whether it's a piece of art, a piece of music. And that thing on the tail reorients how society looks at that item or that genre even. It's the Beatles upending rock music, that sort of thing. Those are tail events. And if AI is just all a convergence toward the mean, you know, by definition, you get fewer and fewer tail events. It's programmed to come around that.

Speaker 2

midpoint. If you want something really interesting or odd or extreme or boundary pushing, AI that would be configured to promote tail events, although I think the state of technology we have right now, it would just start turning out weird rather than something truly innovative. And that's where the human element comes in there because you actually need a human artist or a human singer or actor or performer. to really push that boundary to give AI something to orient around and it becomes the new mean. And I think so much that we are, you what we're talking about when it comes to economics and people might, you know, miss it because economics itself just sounds so sorry, nerdy. It's so it comes down to the choices that people are making the human connections and the relationships that we have with one another, whether it's as consumers or producers or friends and relationships, etc. And you know, part I do wonder sometimes if when it comes to AI, is there such a consumer appetite for it? Or is it kind of being forced on us? And one example that I would give is, there's this supposed AI actress named Tilly Norwood going around. And I don't know Tilly Norwood, I've never seen anything that I guess she has has been in and I guess I'm not alone because Chris Pratt, the actor had the same thing to say, he's like.

Speaker 2

this Tilly Norwood thing. I've never seen her in a movie. I don't know who this bitch is. you know, like it's fake until it's something. It's all. You know, I do kind of wonder when it comes to AI, how much of this is trolling in a way? How much of this is kind of a prank on everybody? It seems like and how much of it is legit. Well, again, Tilly Norwood, ultimately it's a sophisticated mechanical Turk because somebody is having to type in and program what this Tilly Norwood person, I guess if you want to call it, is doing on these recorded, generated videos. Ultimately, there's a human actor behind that. Don't get me wrong, I'm not totally against AI. Obviously, like I've said, I've used it as a tool. Some stuff that's been done with AI, I think is hilarious. There's an app called Menace. And I don't know if you guys are familiar with this, if you've seen it, but it's basically a prank app. the one that I saw recently was... a guy was filming a he took footage of a man who was in prison. And the app made it seem like another man was kissing him, which totally disturbed the guy who's you know, the original footage was because like, did this happen? Well, what's going you know, what's going on here? So it's pranking the human race. Right? Maybe that one is the one where I'm like, Okay, that one's pretty funny. Right? Yeah. And I'm kind of happy. I would be kind of happy if if AI

Speaker 1

kind of ruined the pranking industry because so many of these pranks that I see online, especially with like streamers and stuff like that, they're not prank so much as their crimes that are that are being justified. ransacking Walmarts just to annoy people. Or I've seen ones of, know, guys on escalators, like touching women without their permission and saying, it's a prank. It's a prank. It's like, no, no, that's S.A., dude. You should be in jail for that. So it disrupts the TikTok audience, the streamer audience. I mean, there are also some real world implications that I think we're starting to see. If you can generate an AI video from a photo. And that photo is a controversial news event. We've seen a couple of these on the Minneapolis shootings over the ICE incidents over the weekend. And regardless of where you fall on that, I think both sides have put out bad AI photos that supposedly reinforce their narratives. And it turns out, wait a minute, this is AI. So it's created for the news media a whole new conundrum. How do you detect the real cell phone video that somebody uploaded on their social media feed from something

Speaker 2

that was generated off of a computer by feeding it props. Yeah, we are in a world where we don't even agree on actual footage of what's happening. You know, are shown the same footage and have completely different takes on it. So the idea that you're throwing into that AI footage that at first glance sometimes you can glance at this thing five or 10 times and it seems real. And you need someone to point out like, no, the, you know, that person has eight toes or yes, person dissipates and vaporizes, you know, in the in the back. Yeah, when it comes to, you know, kind of the titillation of AI and the use in the arts and stuff like that. Okay, it's, you know, it could be fun. through a wall or something.

Speaker 1

you know, as we we talked about, yeah, when it comes to actually misinformation, disinformation, coordinated disinformation, it is something that I hope, you know, people are able to stay ahead of but I'm not, I don't, I don't have that much faith that people will stay ahead of it. It's a very sophisticated tool for hoaxing, to put it this way. You think of the famous hoax photos of the Loch Ness Monster or UFOs or something that was done by a guy sticking a dinosaur head on Styrofoam in the lake and taking it from a blurry image. Well, now you can do that on a computer. But then again, you also have, everyone remembers the famous photo of Joseph Stalin where he's standing next to the guy in front of the river and then suddenly the other guy falls out of favor with the Soviet Communist Party. sent off to the gulag or disappeared and then Stalin has his photographer come through and re-edit the photo and basically by hand to remove the person. Now that's a click of a button I could do that and just completely alter what looks like a very real photograph. So it's a much more sophisticated way of doing something that humans have been doing for a while but that also means it's all that much more difficult to detect it when it happens. man, I just got more scared. Here we were. you know, I thought this was going to be a light conversation about art, comedy and movies. And then boom, Stalin. I'm

Speaker 2

Wait, wait, wait till politicians, they discover that they can take embarrassing moments out of their press conferences and photos with AI. What a future, ladies and gentlemen. So until that day, this has been another episode of happy our econ, your weekly dose of reality with a chaser of sanity and maybe a couple of drips of grabs of artificial intelligence. We'll see you next week. actually.