Annndddd.... we're live. In the first episode of 'Authored by AI', the team introduce themselves, the project and start to delve into the complexities of AI-powered creativity.
Annndddd.... we're live. In the first episode of 'Authored by AI', the team introduce themselves, the project and start to delve into the complexities of AI-powered creativity. https://authoredby.ai
Hosted by Stephen Follows and Eliel Camargo-Molina
Guests (in order of appearance)
Edited by Jess Yung and Stephen Follows
Music by Eliel Camargo Molina and GPT-3
Mastering by Adanze "Lady Ze" Unaegbu
G46PUiGigCepW8Ld0R0D
Episode 1 - AI WRITES MOVIES
Stephen: As a pedant uh, this is before we start, but as a pedant, one of the things that has always amused me is the very first word ever broadcast on radio, in commercial radio, in the UK was the word "and" (laughs) - which is grammatically incorrect, cause it was like: "AND WELCOME!" (both laugh). I love details like that. Anyway, so we should probably go through what we've been up to the last few months, right? Actually, we should start by introducing ourselves, I guess. Uh, do you wanna go first, Eliel?
Eliel: Yeah. So, um, I'm Eliel, I'm a theoretical physicist based in Uppsala University, and I guess that's a description of my Monday to Friday job, but I'm also very interested in the intersection between art and science. Not just what we do, but who we are. Because we are two communities of people that have gone through the journey of trying to do what we love. So I'm on one hand a theoretical physicist understanding the very early universe, but on the other hand, I'm also always trying to find intersections on how to do cool stuff with art and science.
Stephen: Yeah. And hi, I'm Stephen. I guess I would identify at the moment as a film data analyst. So for the last decade or so, I've been studying how the film industry works and then trying to share that, uh, through education and writing and things to help filmmakers get their films made. But I also run a, a story agency where we raise money for charities, using strategic stories, working out what to tell and things like that. Eliel and I met a couple of years ago and I think we connected cause we're both outsiders in our field and are interested in… I don’t know if we wanna mess things up or just play with things. I don't know how directed you think it is, but for me, I just wanna try new things and see what's coming.
Eliel: Yeah. And, and I think when you say, you know, those two options that you give for me sound very similar, more or less like the same, cause, you know, bringing that kind of intersection of two people that share the same pains but are in some different ways (and we explore that)… creative, is always fascinating and you end up in places that you didn't expect. And I think that's what we have been trying to do. And in particular, this podcast and these projects surrounding this podcast is a lot like that and we are very excited about it.
Stephen: And so I was gonna say that we stumbled into this, but thinking back, we were very deliberate every stage. It just seems very unlikely that we've got to where we've got to. And we should say also, we are early in this journey, so we we're gonna catch you up with what was happened so far, and we're gonna introduce you to some of the people that we've met and talked to. And then very quickly, you're gonna be at the same speed we are. And so we don't know the ending of this yet, we don't know where this goes.
So… I should say what this is. Eliel and I have got a deal with a Hollywood producer who we cannot speak their name. That is part of the deal. The actual deal we're doing with them is top secret, and the work we're doing with them is top secret, but we're allowed to talk about around it and use it as a launching pad. So, this deal is to write a professional Hollywood feature film script using – entirely using – an AI, an artificial intelligence. And so… I don't actually know the status that we, our lawyers had to figure this out, as the first time this has ever been figured out. But technically, Eliel and I are the writers, and we are the people who would be credited and we are the people who can get the copyright. But as far as I'm concerned, the AI is going to do all the actual writing. So I don't know whether I, we are the first example of people stealing credit and focus from an ai. I don’t know whether that's gonna bode well in the future.
Eliel: Yeah. And I think it's fascinating and hearing it like that, spelled out so succinctly, it's interesting because it sounds like “yeah, of course we got a deal with a Hollywood producer to write a film and that's what we are doing!”. But the actual journey was very interesting and much more organic and almost, you know, magical, but… I think, yeah, that's very exciting. We're trying to write the first feature film using an AI.
Stephen: Well, the way I like to think of it is that you've gone the traditional route of becoming a Hollywood screenwriter through theoretical physics, which I… I quite like that! That's been your end goal, right?
Eliel: Well, apparently it's easier than becoming a full-time Professor!
Stephen: (Laughs) Oh my God, that's a depressing way of looking at it. You've spent the whole time trying to become a professor, but accidentally become a screenwriter.
But yeah, I think the thing is that this is a fun, real world experiment for us, but not fun as in, I mean, we're enjoying it, but it's not light. We understand that some of the issues here are really kind of deep and meaningful, and so rather than just go ahead and do this behind the scenes, we decided to use this as a launchpad to talk about these issues and to meet people and to interview people and to try and have a big public conversation – partly about what it's like to write a feature film using AI, which is something that very few people have done. And we are sort of discovering our own methods, our new methods, but also to talk about the near future and what this means as audiences, as screenwriters, as writers.
You know, there'll be many writers out there thinking, “wow, you know, I remember when I used to take yellow cabs or black cabs, now I take Uber. I'm a writer. Where is that going?” But also just what it means for audiences. You know, AI-generated art, AI-generated feelings, it’s all quite new, but I suspect in the next 10 years or so, you know, my niece and nephew are 5 and 8, I suspect by the time they're teenagers and into their early adulthood, they will be in a world surrounded by AI-generated creative works. And their understanding of it... they'll be maybe “AI natives”. I don’t know what term we'll invent for that generation, but that is a world that we are all going to enter. And I feel like you and I are seeing 10 minutes into the future and trying to share that because it's, um… it's exciting, but it's also kind of terrifying. How are you feeling about it at the moment?
Eliel: Yeah. At least we're trying to do that. I also feel sometimes like when I was a kid and my parents bought the first computer and after, you know, a month, I already was so familiar with it. Cause as you say, I was a “native”, but now I'm in the other position, right? I'm in the position where I'm just trying really hard to keep up and that there's gonna be, like you say, people in 5, 10 years, this is gonna be part of their reality. And, you know, we've talked to around 50 people from many different areas relating to this, and the question always comes up like, why are we doing a podcast? So we say “yeah, we're writing a film using AI, we got a Hollywood deal”. “Yeah, but why are we doing a podcast? And, and this is interesting because it's such a powerful drive that I think we have to do this and also so many different dimensions. To echo what you just said, for me it feels a little bit like with every new technology, but also in an accelerated manner. It feels like we are on top of a train that is going forward and there's no tracks. And we build the tracks as we go. And these tracks are: how do we use it? What are the ethical considerations? What's gonna happen? How is this gonna affect, uh, human life, jobs, people… or what is it gonna be possible with it?
And with AI, I've had for a while this feeling that the people or the entities in charge of building these tracks for the train is made mostly of big corporations with a little bit of maybe a decentralized community of computer developers in the internet in some very weird way to understand, but just a little sprinkle of that. And it feels like there's a big gap in that conversation and in that work of building those tracks that we need to fill, which is people like us, people that are insiders-outsiders, people that are creatives, people that are thinking about how is this gonna affect my life, but also my work and what I'm gonna create… And you know, what better way to do that than by trying to explore how this is gonna affect the film industry. So yeah, well, we are trying, I guess the podcast is a way of sharing our journey, sharing our conversations with top thinkers from, you know, AI, the film industry, creatives, neuroscientists, and the like, just to delve into this and, and maybe start getting our hands dirty and building the tracks for this train that is accelerating faster than we can even comprehend.
*Music transition*
Stephen: So, hello and welcome to the first episode of Authored by AI. So we should probably actually talk a little bit about the history of AI writing scripts and things like that. And also what is AI and what the hell are we doing?
Eliel: Um, yeah, and I think it's, you know, we've gone through a journey and I think part of our contribution is accelerating that journey for other people. So I remember like, GPT-3 – which is what people call a leap, a new generation of AI models – comes out and we start hearing noise from our sources, what we read on the internet, the people that we know, but we were very skeptical, like, “yeah, yeah, this is gonna be just another one of those parlor tricks with AI”.
Stephen: Well, to be clear, I had no way of understanding it. You and I were chatting anyway during the pandemic on Zoom and you were like, “well there's this new model called GPT-3 and everyone's talking about it. Certain people are talking about it, but they're saying no one else can have access cause it's too powerful.”
And you brought it up as, like “this is the kind of bullshit that scientists say to try and make us scared”.
Eliel: Yeah, I was, I was just “Stephen, we have to do something with this. We have to get access to this!” I sent an application, I started thinking, how do you get access? And, and, um, you know, at the same time, very skeptical. I was so, so eager to just open it up and just discover very quickly that it was just a bunch of bullshit. Um, so I did get access. I opened it up and… it wasn't bullshit. (laughs)
Stephen: And you were terrified! And I was scared on your behalf cause I hadn't had access at that point. But for you to be scared, and I know you and I know how smart you are and I also know that you didn't want it to be, um, what it actually probably is. You wanted it to be another chatbot recycling what people are saying and you know, whatever… an impressive parlor trick. But the seriousness in which those few days where you were sending me quite a few WhatsApp messages of essentially – oh my God, oh my God, oh my God – was quite terrifying for me. Because it's one thing for you, like if this is a horror movie, you're staring at the monster. I'm staring around the corner staring just at you. And your reaction to the monster is a little more terrifying for me than if I'd saw it myself, cause I didn't know what it was.
Eliel: Yeah, and I would say that this originally was just like, “Stephen, there's this AI thing. I don't know what it means. I don't know what it's gonna mean for the future.” And then starting to show you stuff and then suddenly we converge on the incredible, now it's obvious in hindsight, but, you know, “can this AI write a film?” A very important motivation for us was that we very quickly understood that models like GPT-3 and every other model that is gonna come after that are of a whole new kind. They are in many ways, very different from every other model that came before them. Step one then is to try to understand what people did when they tried to do this before.
*Music transition*
Eliel: So the speed at which AI develops is staggering. It's really, really fast in particular language models. We only have to go back 6 years to get to the point where the first experiments in writing films using AI were being made. There's this film called Sunspring, a short film that was written by an AI for a 48-hour film competition.
Stephen: It stars Thomas Middleditch, and it was quite a successful film in its time. It was this idea that Oscar had been working with an AI scientist to try and generate dialogue for his movies, for short film. Oscar's made many successful short films in the past. And actually I know Oscar from personally working with him on some previous films, and so they fed it on a bunch of dialogue and the dialogue they got out was coherent sentences, but not coherent between the sentences. And there's no, sort of, plot coherence to it. So what Oscar then made was this art short film.
Oscar Sharp: I'm Oscar Sharp. I am a filmmaker, I suppose. Um, I've never quite felt comfortable with any of the terms I'm supposed to use – director, filmmaker, writer. I have made a number of films and… some of them were written by robots!
Stephen: So how did Sunspring come about? What was the genesis of the project?
Oscar: Uh, so that was myself and marvelous Ross Goodwin, uh, mysterious Mr. Ross as I always like to call him. We met at NYU, I was in the graduate film program. He was in ITP, which is their technology arts program. And we met on a course that was about using technology in the process of trying to make a documentary. They taught us on the first day how to hack into people's webcams. That was, that was disturbing. Um, it wasn't act- to be clear, it wasn't technically hacking. It was, um… find webcams that people had foolishly left open. But unfortunately that's a lot of them. So it was a bit uncomfortable, but we went on a lunch break together and quickly got talking about various things that we wanted to do. And he was doing a lot of generative text work at the time. He'd invented a camera that would, uh, when you pointed at something and pressed the button instead of a picture coming out, it would write a poem about whatever you had pointed it at, which I thought was rather brilliant.
We eventually end up at this conversation, you know, can Ross use the artificial intelligences that he's playing with to generate a screenplay? Cause there I am in the graduate film program and there he is in the technology arts program. It seemed like a good overlap. And before we got to AI, actually we were trying to be sort of more hand tooled about it. He was asking me to describe to him various paradigms of story generation, sort of Vladimir Propp and Joseph Campbell and Sid Field and so on. And none of those really got us anywhere, but once he got his hands on LSTMs – long short-term memory, recurrent neural networks – it just became a different ball game, a different way of approaching the idea of generating something, which in the broadest sense is much like something like GPT-3 today, in that you give it a large training corpus, you give it a bunch of scripts to read, as it was back then. And you know, in the case of GPT-3, the entire internet, more or less, and you say: okay, this is what the humans did. Here's a seed, here's a little bit of something that a human has written. In the case of Sunspring, the opening line and a couple of other things that it had to include according to the rules of the 48 hour film contest we were doing it for – Go! And it would do what we're all familiar with now as sort of auto complete. It would guess what the next word that human would say is, and then the next, and the next and the next until you had an entire screenplay. And that's how we got the screenplay for Sunspring, which, you know, then becomes the “first film ever written by an artificial intelligence”, although, you know, a much more “stupid” one than the ones around today.
* Snippet of Sunspring*
Character 1: In a future with mass unemployment, young people are forced to sell blood. That's something I can do.
Character 2: You should see the boy and shut up. I was the one who was going to be a 100 years old.
Character 1: I saw him again. The way you were sent to me. That was a big, honest idea. I am not a bright light.
Character 3: Well, I have to go to the Skull. I don’t know.
Character 1: Huh? What do you mean?
Character 3: Well, I don't know anything about any of this.
Character 1: Uh, then what?
Character 2: There's no answer.
Oscar: One of the reasons that that Sunspring in particular got the attention that it did is that it's a sort of a robots falling over video. It's not the video of a Boston Dynamics spot successfully running upstairs, opening a door and disarming somebody. It's the video from before that of Boston Dynamics, tall man shaped thing falling over because it doesn't add up, particularly. Like, it adds up in certain ways, it's sort of, it's uncanny, but by no means all the way through to something that's, practically human.
Stephen: I spoke to Oscar at the time about this, and he said that he had these meetings with Hollywood producers and they'd be like, “oh my God, so an AI can write a film.” And he'd be like, “well, it can format text in the shape of a script, and I as a director can interpret it and shoot it.” And, and the people would be like, “oh, great. Could you write me one?” And he'd generate one. And then they'd be like, “yeah, but can you make it good?” And he'd be like, “well, no, this isn't a coherence machine. This is creating dialogue.”
And so that's the framework I had for AI-written film. Sure it does a simulation of it. And with a talented set of artists and performers, you can add coherence and art as a wrapper to the whole thing. But fundamentally, it's not thinking, it's not planning, it's not creating a narrative or a journey that would be entertaining to watch. It just becomes a curiosity. And what was fascinating to me about GPT-3 and the way that we’re interacting with it was that it fundamentally didn't feel like that.
Eliel: Yes, so this is the exact same mind space. I expected anything that an AI would write would be in that direction until we get access to GPT-3, which again, it's, you know, the most advanced language model so far, and language here used in the most general way. And that thing that, what was making me think right away, okay, this is next level stuff that it's not, that it can write in English, it can write in essentially any relatively major language – including computer languages, but it can also do math. And that was, I remember like one moment where I said, okay, this is something else.
When I realized and understood the very subtle but profound idea that it learned how to do math, but nobody told it the rules of math. It learned it from absorbing a curated version of most of the internet. A lot of books, and it was just told: understand language. Okay? Very broadly speaking. But it learned math and it learned to code in Python or C++, and it could also start writing stories. And the idea of coherence is central for all of this. Cause you cannot write code if you don't keep coherence from, you know, what you define in the beginning of your code has to play a role later on. So coherence, which is also very important for stories, seems to be one of the main big points of this together with a little bit of a more general approach to what language is. Um, and this is what made us believe very quickly that, okay, this whatever comes out of GPT-3 is gonna be, not like Sunspring.
Stephen: But also because the way that writing works in the film industry is rewriting. You go in and you edit and you do it again and again, and you get notes and you constantly improve. And that wasn't something that we were hearing other people doing. In our early experiments with GPT-3, we found that it was actually pretty good at knowing what was wrong with something and giving itself a note. And so by linking that back to the beginning and saying, okay, now do that note. Now, give yourself a note. Now, give yourself a note. We realized that iteratively, the AI could produce much, much better work than it would in the very first instance.
Eliel: And that relies on what I think is one of the most important advances in language models like GPT-3. Models like the ones Oscar was using for Sunspring could keep track, remember around 50 maybe something like that words. A line or two. Models like GPT-3 can to some extent have some working memory of texts that are 3000 words. That kind of new dimension of now we have a thing that can actually remember and that we can talk to it after the fact is the key of the whole process.
The moment where I was able to tell you, and answer to you “yeah, sure, it can do it” is the moment that I understood the big leap that there is between. “Give me a story” and it just prints a story. “Give me a poster” and it just gives a poster to, “okay, go back to the story, and can you please change this character that you said, Jane and, and make, make her more angry and rewrite it, please.”
But then the realization that it goes, it takes that story that it wrote, it understands the context, it understands what you're saying, and it changes it. It makes is not just a super intelligent AI, but a super intelligent co-worker. Some entity you can discuss, give feedback to. And while you know you're not training it when you're doing it, the amount of knowledge that it already has is so broad that it's more about trying to explain in a way that it understands what you think would a valuable contribution to what you're trying to do. And that kind of next step is what made me answer, you know, right away “yes, we can.” You know, it's a small thing, but it's a big change in what an AI model for language is.
*Music transition*
Stephen: Um, so what does the AI script writing do well and what does it do badly?
Jacob Vauss: Yeah, that's a good question. I think the, at least the version of GPT-3 I'm using, it tends to write the beginnings of stories or scripts pretty well. It's pretty good at setting a scene and it knows from other things what's intriguing for the beginning of a movie, whether that's just putting three characters in a room, or having a family leave on a vacation. It just kind of gets inciting incidents and that sort of thing. And then it's its own following through, I think, where it kind of falls short, whether, you know, it falls into those thoughts spirals that we were talking about or you know, in my uses with it kind of moves with dream logic where, you know when you'll be in a dream, you're talking to, like, your mom for instance, and then you turn around and you look back and it's like your brother and, so I think the AI is kind of similar to that, where it will sometimes forget where it is, forget who's talking that sort of thing.
I'm Jacob Voss. I'm one of the co-founders of Calamity AI, which is this series that explores human and AI collaboration in all things creative.
Stephen: And what kind of projects have you done so far?
Jacob: Yeah, we've done a few things. We've had the AI write short films that we've then gone and directed with actors and, you know, tried to make shorts that are actually compelling even though they're AI-ritten. We've made AI recipes, we've had AI emulate, for instance, like a Dr. Seuss book or, this movie called Lady Bird, and then we would go and recreate scenes from that, or chapters from the book. Um, so we've kind of made all these things in the spectrum of AI creativity.
Stephen: So tell us about the how, how you started deciding to use it for film.
Jacob: Yeah, I'm a filmmaker and that's always what I've been and what my passion is. And I was attending Chapman University for their film program and I was just scrolling through Reddit and I stumbled over this website, which is shortly read, which is so simple, you just put in, you know, whatever your prompt is, and then the AI will continue to generate. And so the idea was “what if we could have it do scripts?” And so I tried inputting, like the slug line of a script, like “interior building night”, and then it just kept generating. So from there it was like, you know, “we have to make this”. Because there weren't that many AI short films that seemed to exist up until that point. There were a few, but not like a ton.
Eliel: So do you feel like it was more like, I guess the two extremes would be a word processor on steroids on one side or a writer that is collaborating with you on the other?
Jacob: I would say the word processor on steroids. For me, it felt more like working with this kind of strange monkey that would go in any direction it wanted rather than like a tool that I could predict and use.
Stephen: Let's play a short clip. This is Solicitors, which is one of your shorts on YouTube.
* Snippet of Solicitors*
Character 1: I'm a Jehovah’s Witness.
Character 2: Well, I'm sorry I don't talk to solicitors.
Character 1: But I have a great story. I was a drug dealer, shootout, shootout, shootouts. In retrospect, it was a very dangerous job, but I loved it. And then I hit this cop car and they chased me. I was weaving through the city, but they were so fast. And then I saw a road that led to this alley, and I figured since I had some time, I could lose him. So I swerved. There were two women in the alley. They were arguing. I was going too fast, they didn't see me. I hit them. And the car turned over and over. And over. That's the worst pain I've ever had. But they lived. I didn't. Before I died, God appeared to me and told me that if I would come here and tell you this then he would bring me back to life. I was instructed to tell you the whole story.
Character 2: Uh, what's your name?
Character 1: Rudy.
Character 2: Do you believe in God?
Character 1: I did once.
Character 2: Do you? Do you wanna come in?
Character 1: I don't wanna die.
Character 2: Well come in anyway. It's about time for us to do dinner.
Stephen: I mean, if we didn't mention AI, if no one knew that your shorts had been written by an ai, how would you, how would people probably describe them to each other, having watched them?
Jacob: Um, absurdist or kind of surreal… Like, I mean, I would say that's the thing is… I don't even think people would know. If you didn't tell them, I think they would think like, “oh, this is an absurd dark comedy, with like a director who, who's kind of strange and likes like outside of the box stuff.”
Stephen: How have the audiences responded to this?
Jacob: They've responded well, I mean, the YouTube crowd is an interesting crowd in general, and so I think a lot of the comments are like, whoa. They're either like freaking out, like, “oh my gosh,” or they're like, “no, an AI didn't write this.” There will be a lot of comments that are like, “this is quite scary.” You know, our short film Solicitors ends with, you know, him being shot by the woman whose house he goes into, and there will be comments that are like, “does it freak anyone else out that this ends with the human dying?” And there are also people who morally just kind of disagree with it. Like, I think my mom kind of morally thinks it's wrong, which has been a funny line to tread. But it's been a good response so far.
Stephen: Tell me about your mom. Tell me what she said when you told her about it.
Jacob: (Laughs) I think she just doesn't see how it could end in a good place over the span of time. So she's like, “why are you helping make these?” And she's a great mom. She's a very loving, kind mom, but she's like, “why are you helping make these contributions that will weed someone like you out of this?”
Stephen: I think that's really sweet she thinks you're helping. I think that's, that's, that's a really big thing that she's saying. You are really pushing this forward.
Jacob: (Laughs) “You're one of the leaders!”. Yeah.
Stephen: Yeah. she really believes in you – and that you're doing bad – but that you're doing something.
Jacob: Yeah.
Stephen: I have just one last question, how has the industry responded?
Jacob: Um, I think there are a few like my mom who just think it's, you know, strange, or that sort of thing. Um, but no, I think it's been generally positive and they like the creativity and they think it's funny. Um, so yeah, I would say, good.
Eliel: Do you think that's a function of the fact that it’s good, but still quirky and not good enough. Do you think this is just gonna be a tide that once it starts getting really good, people are gonna like, not be so positive or laugh too much?
Jacob: Yeah, you're 100% right. I was just thinking that right before you said it, of like, I think a big reason why people like it, at least right now in the film industry is part of the joke is like, “look at how wacky this script is!” Like look at how... Rather than taking it very seriously. So I think in, you know, let's say in the next 2 years, if they start actually being really compelling shorts, I think people would be a little more, you know, “I'm not super into that”.
Eliel: So I did both things. Kind of tried to keep this separately: to watch one of your shorts and read the script.
Jacob: Yeah.
Eliel: That's two different experiences. When I watched the short, it was thought provoking, it was interesting, it was quirky, it was a little bit surrealist, and you know, the dialogue is happening so fast that you hear it and… But then when I read the script, I got a completely different feeling because then I had time to read things carefully.
Jacob: Yeah.
Eliel: And then you start realizing that if you want, and this is of course interpretation, and maybe if I wouldn't know it was an AI and I would think this was an artist that wrote this, then you try to find the meaning in the things, and then it started getting deeper and deeper and deeper and way less nonsensical. So a question would be like, do you think this is not only gonna be helpful and better than humans? Is it gonna change what a movie is? Is it gonna bring a new paradigm once these things start, you know, pulling out 10 movies a second? Are we gonna, yeah…
Jacob: Yeah. I mean, I definitely think so. I have a friend who was talking to me about how he thinks AI will eventually be able to kind of, moving beyond movies, like invent or solve problems in general because it will be able to take, you know, two different solutions we didn't see and combine them. And I really do think that will be the case with movies. And I think, you know, a lot of people think the current structure we have now for movies is like the correct one or the only one, but I really do think that AI will be able to see something we haven't seen, put together like a newly structured movie and make something that's really smart, entertaining, that sort of thing. I definitely think so.
*Music transition*
Stephen: Alice, do you wanna introduce yourself for the podcast?
Alice Fraser: Uh, yes. Hello, I'm Alice Fraser. I do comedy and various other things. I write jokes for myself and for other people and a whole bunch of other things.
Stephen: So if the AI did write a script that was meaningfully emotionally connected to something in your life, like some, whether it's a happy moment or a bereavement or something like that, would you feel as connected to that piece of art as if a human writer is an AI? Like, does the way that, because it connected with you, is that enough of validation or does it need to come from someone who's lived or bled or…
Alice: I think part of being human and part of being an audience is that, um, we are very good at filling in gaps ourselves. So we are very good at, and I think this is again, part of the reason that people panic, is you, you fill in the spaces with your own ideas of what might be there or what would be there if you were saying that. In the same way you look at the clouds and you see a rabbit, you know, you can get found poetry from fridge magnets or, you know, newspaper blowing down the street. We are very good at seeing beauty and importing our own ideas in, so I don't think knowing it was written by an AI would change it if, change it for me at least, if I saw beauty in it.
Stephen: Well, one of the things that we're interested in this journey is: how do you feel about this future as a writer? If we project a few years forward where these tools are much more powerful and freely available, does that fill you with hope or fear or…
Alice: I mean, it fills me with interest. I think, possibly arrogantly, part of what I sell as a performer, certainly, or as a comedian or as a personality, or as a voice. Part of what I sell is my own point of view and my perspective, which is always adapting in dialectic. So it's always going back and forth. It's always moving in conversation with my audience, in conversation with other artists, other work that I see, other things. So I am interested as much in having a conversation with an AI as I am in anything else, or certainly having a conversation with what AI does to the culture. For me it is just, I don't have a scarcity mentality about comedy, I guess, unless everyone in the world is going to see a comedy show every night of the week, there is still room for more voices and more people and more, you know, AI-human double acts. And I think that's a really exciting and cool thing. I don't feel like that's taking anything away from me.
Eliel: It's, I think, very important to preserve some kind of, not preserve or even like, you know, create a healthy environment where this AI change is gonna be positive. Cause otherwise big companies are gonna bring a bunch of engineers that are gonna use the Ais and gonna forget about the humans. Whether if there's a bunch of creatives with AI competing with a bunch of engineers with AI, the creatives with AI are gonna win.
Alice: Uh, it's the job of an artist to take the tools and find uses for them that are unexpected to respond to all the art that has been done before by doing something different. You know, there are things that artists did with paint that had never been done before, even though paint has been around for thousands of years. You know, it's one of those sort of features of the artistic imagination is you see what's been done before and what you want to do is something that's new and different. So the ways in which artists will find to use this, I think are going to be really interesting.
Stephen: I guess my last question is, do you have any advice for us? I'm just opening it up to like the thing you've been thinking, you know, positively or negatively or whatever. But any advice for us as we go on this journey? We're not very far into it, the screenplay.
Alice: I think my advice would be to bring as many artists in as possible, bring as many writers in as possible, and act as, uh, or teach writers to act as the farmers of AI, as the users of these tools. As the first person who gets the laptop, or the first person who gets the home computer, that let that, let that be a writer. And see what they can do with it.
*Music transition*
Stephen: In the next episode, you'll hear what happened when we did exactly that.
Eliel: We sat down with Dave Reynolds, who, among other things, wrote Finding nemo, and we just jumped with the AI and had a development session with him.
Dave Reynolds: Man, that's it. It, I don't know. I don't know how I feel. I'm like, I'm excited – I met you guys and you're really cool. But do I hate you? Do I also, do I also wanna sabotage you?
AI Voice:
Authored by AI is brought to you by Stephen Follows, Eliel Camargo Molina. Isadora Campegher Paiva, Bob Schultz, Jess Yung and GPT-3. Music By Eliel Camargo Molina and GPT-3. Audio Post Production from Adanze "Lady Ze" Unaegbu and with thanks to Rob Cave and Ina del Rosario. Find out more at authoredby.ai.