6.18 Appupen & Laurent: What If AI Falls Into The Wrong Hands?

This is the story of an entrepreneur who almost sold his soul to AI.

In this episode, Michelle speaks with Appupen and Laurent Daudent, the illustrator-scientist duo that came together to create 'Dream Machine' - a graphic novel on AI! The book follows Hugo, who is on the verge of selling his AI tech to a large corporation. But what if the motives behind this purchase are not exactly ethical?

Appupen and Laurent discuss the evolution of AI and why the hype and fear around it might be misleading. They also address their differing opinions on AI, working together from France and India, and why they regret making AI write a chapter of their book!

Books and authors mentioned in this episode:

The Circle - Dave Eggers

Movies and TV shows mentioned in this episode:

I, Robot - Alex Proyas (Director)
Her - Spike Jonze (Director)
Teri Baaton Mein Aisa Uljha Jiya - Amit Joshi, Aradhana Sah (Directors)

‘Books and Beyond with Bound’ is the podcast where Tara Khandelwal and Michelle D’costa uncover how their books reflect the realities of our lives and society today. Find out what drives India’s finest authors: from personal experiences to jugaad research methods, insecurities to publishing journeys. Created by Bound, a storytelling company that helps you grow through stories. Follow us @boundindia on all social media platforms.

Transcript

Michelle D'costa  00:08

Hi everyone, welcome back to Books and Beyond my guest today Laurel and apapun are a creative duo, who has brought alive the world of AI like we have never seen before. apapun has been one of India's leading voices in comics. And Laurel is the CEO and co founder of light on a startup at the forefront of large language models, which is the new generation of AI. And both of them have worked on a very cool graphic novel called Dream Machine, which is all about AI and the real world. Everyone seems to be talking about AI these days, right? With the introduction of chat GPT, you know, which is an AI tool where you put in prompts commands, and then you get out. Right now being in the publishing industry, we have heard stories of how so many books written by AI are now flooding the market, which is a cause of worry, but is it really, you know, how far will Gchat GPT go, how fast will the next generation of AI go? t I'm really curious to know what's in store for us regarding AI? Are there suspicions bases, you know, do they uncover something more sinister? Let's find out!
So, you know, I recently read this novel, the circle by Dave Eggers, I'm not sure if you know, both of you have read it. But the novel talks about a corporation, Okay, who wants complete transparency from the world? Right now, of course, in its ideology, and you know, sort of how it begins, it sounds really positive. It sounds like this world where, you know, people don't lie. People don't cheat, you know, you can't hide anything from anyone. It sounds great. Until, you know, you read further and you realize, okay, there is something sinister in this, you know, we might not want to be, you know, transparent all the time, or we might not want everyone to be transparent, right? It's, I would say, there's, there's also free will, there's choice, right? And this is something that would, you know, stayed with me when I read your graphic novel, Dream Machine also, right, because while we know that AI can be, you know, really, I would say really helpful, really convenient. It's sort of, you know, you know, makes us lazy people it does the things that you know, we might not want to do, but I do think that your book sort of makes us think, you know, it sort of makes us introspect and wonder what AI has in store for us, you know, so do so through this interview, you know, we'll obviously delve deeper into your book. But to start out, I wanted to sort of, you know, just decode what is artificial intelligence in the first place? Right. So this is something that that this is something that I understand about AI is sort of, it's this science of making machines, you know, where we train machines to think like humans, right, and that's the usual understanding of the term. But through your work, both of you know, and up. And, you know, have y'all arrived at a different definition of artificial intelligence? Or what is your understanding of the term?

Laurent Daudet  02:25

Yeah, well, I think there is a misunderstanding it just in the world, the intelligence because these machines are by no means intelligent, they are just automating some some of the tasks that we do on the web, with a book to a machine, we are talking about the latest generation of AI models that are called large language models, a bit like char GPT. And the GPT is automating tasks with texting input on texting output, on that, that just what it does mean. So it's, I would be very careful with the notion of intelligence, because it's not mimicking human intelligence, it's just making some task easier for us. So they are more like, assistance for us.

Appupen  03:14

I think it's been made to like, look like some solution for us, because we don't want to solve any of our problems, really. So we like this godlike figure in the sky, which can come and like solve things for us. And it's been marketed very nicely in that way, so that we don't ask more questions about what's behind it. And we play with the fun end of things, you know, we are obsessed with, oh, I can change his voice, I can change your face. And I can put big eyeballs to something like that. We kind of missing the point there. So when we met also, the idea behind the book became something like that. Everyone is going to be asked an opinion about AI Oh, they will have to form it. The way it's coming into everyone's life right now. So can we inform the people about what it is? And I'm from advertising, I know how these narratives are made, we made some lies for people, which everybody believes. So can we like just take off that facade and show what it is? And for that we need this kind of like real technical stuff, also to explain things and so people believe what we're trying to say. Laura likes to say that it's a statistical model, predict, which can only predict the next word and look at what what kind of like reference, we're giving such a model right now, you know?

Laurent Daudet  04:37

Exactly. This is something we try to explain in the book. It's what's what's behind the salon language model. They are just as a as a group and we're saying they are just statistical models that are just trying to figure out the statistics of words how they appear in the in language, and they have no notion of truth. They have no real they have no notion of what's fake They are unfair and what's biased on bias on this one.

Appupen  05:03

It's programmed to make you happy. It's programmed to, like, please you, it lies to you, you know, and nobody's calling it out. Like, or making that a real problem, they were just giving it all the solutions or expecting it to come up with things like that. It doesn't know what's good or bad. Like in art, it's predicting the next pixel. So basically the same idea, you know?

Michelle D'costa  05:27

Right. So that's actually very interesting, because, like, you know, anyone who has spoken to about AI, they usually never, ever describe it in this way. So now, we're always gonna think of AI this way, where, you know, it predicts the next word, right? Like you said, it predicts the next statistic. But this is sort of really interesting, because it brings me to my question, then why is sort of AI so threatening? Right? So let's say for example, if it is, if it's a model, which literally just predict something, you know, which is like something, which is just upcoming, right? Why is it sort of why is it posed as a threat to us? And and this is a trope that we have seen in lots of stories, right? Like, not just in your book with movies with other books. So there was this recent Bollywood film, where there is a female robot that's there. And, you know, we see, you know, for most of the film, we see a very positive side of her, obviously, I mean, there's, there's no room for character growth over there, as usual. But at the end, we do see a negative side, we do see a spin where, hey, if this doesn't work out, you know, it could always lead to destruction, right? So I want to understand, you know, if, as both of you are saying, if the model literally just predicts the next thing, then why is it such a threat for us?

Laurent Daudet  06:36

Well, the, the, the AI is not a threat by itself, the AI is just a tool. So it's like a sharp knife, I can use it by cutting of my vegetables, so I can use it to stop by neighbor as well, right? It's just a tool. So it's not the AI is not threatening is the way some people might use these AI that may be threatening, like, launch massive disinformation campaigns, or, because you can do things at scale. It allows to, it allows people to do yeah, as I said, buy CDs information, or spam or any type of virus on the computer system, it makes it much easier to to deploy this.

Appupen  07:29

AI is a threat, because we're doing it at this large scale. And we're going to hand over a lot of responsibility to this thing, purely because of some commercial interest or some profit motive. One of the main things in this the hurry, the race we are in the race is only because of some somebody decided they want to be the monopoly in this. And we are all running that race for that person. There is also the environmental side, which they don't like to talk about. If you check about AI, you will hear that AI can help solve climate issues or like global warming and stuff like that. But that is after AI has reached a fantasy level of performance. We are not taking into account how much energy or resources it takes to get to that level. So there is a big like, propaganda going on about AI. The fact is, we're going to give all this control to this thing. That is where the danger is. And as Laura said, it's not a definitely, it's not AI. It's the people behind the AI. And we have enough case studies to prove that the people behind such powers don't really work. They never like give you the right one. And suddenly, when it comes to a new field, like AI, we have such faith in ourselves, you know, as humanity, we're like, oh, we're just trying to make things more convenient for people and the people whose jobs are taken by AI, they will be free to follow their passion, you know, like birdwatching. So, you know, there's a whole narrative being built there. Because now the people who can afford to go birdwatching, it's their jobs that are under threat, not the guys who are like at the conveyor belt level, you know, so that's, that's why the narrative has changed. So when the narrative itself is so convoluted, I definitely don't trust or I have a problem with how it's being pushed on me. It's like NFT or something, you know?

Michelle D'costa  09:19

Yeah, totally. I think I think taking it with a pinch of salt, taking it and sort of like using it responsibly. I think that is something that that we have seen time and again in stories and, uh, you know, as humans, we know, that is really, really difficult to see, I think a lot of science fiction, a lot of, you know, interesting stories, which sort of cover the good versus evil is something that we have seen, and I think that's what I enjoyed, you know, in your book as well. And for me, though, the aspect of technology, though, the aspect of AI was definitely the hook of the book. I was drawn to it for the characters. I thought it was brilliant. I thought the way you have carved the relationship of Hugo, who's our protagonist here, and I know who this partner is something that sort have made me stay till the end. So I would love to know a bit about, you know, how did you all come up with this dynamic, you know, like the characters and the relationship that you're living?

Appupen  10:11

Well, you know, I needed a LM writing this thing, I needed a character who can do multiple things, because we have a lot of explaining to do about AI. So my task was first to understand the text stuff that Laura was sending me, which sort of gave like a spine of the whole plan, like a blueprint, because we know that information moves in a linear format. So I have to weave the story and the writing around this, this info module that has been given to me. So we have to start from a pretty basic level and go to a pretty advanced level. So I'm thinking, Okay, I need a character, who is probably teaching AI, and who's probably a CEO of some company, and has been like a scientist to something, you know. And I found that all these things like, coincidentally match with what Lauren is doing. So, you know, you had to bring in a lot of this with that character. So that's where, like, we I sort of, like convinced Laura that, you know, I get based the character loosely on him. Look, waist, yes, but otherwise, no, I'll just make it up from my head. But like, you know, there is there is because I went and stayed in this house that we've drawn. So, you know, I know how the breakfast layout is done, you know. So there are many things like weather helped me because I have to show like life in France, not like life in India, where I mostly lived. So we have to bring it to that sort of level. And I think it was, like, I had to do a little bit of convincing, but like, I think afterwards, it was okay. For every part of it. Yeah,

Laurent Daudet  11:48

I think this the main character who is has much more doubts. He, he has, he sees both sides of the technology, the good side on the bad side, on its, I would say that, who is maybe 10% of myself 10% of apple pie and 80% of all imagination.

Appupen  12:14

Because because he was he's too much of the negative side, right? Yes. Yes. Yes.

Laurent Daudet  12:24

More the optimistic person? Test, you know. And

Appupen  12:28

also, like, when I was writing, I wrote the character Anna, and I sent it to Laura, this is before I went and stayed with them. Okay. I sent it to him. And he's like, how do you? How do you mean the name Anna? Bye, wife's name is.

Laurent Daudet  12:48

That's pure, pure coincidence, but it was very fun.

Michelle D'costa  12:51

Oh, that's lovely. I mean, I can only imagine the kind of, you know, collaboration that would have I think it just exciting listening to it. You know, and what was interesting, as I read that, you all had originally thought of this book as a series of pedagogical chapters, I think, right? Where you want to call it a lot of science, technology, facts, all of that. Right. And only when I think your editor Christian had suggested you don't make it more fiction, sort of your your had to dramatize it. So could you please tell me a bit about that, about that creative process? And how did this become a graphic? No.

Appupen  13:24

It's to do with how we met actually, I met him by chance in Paris, there was no plan for either of us to make a book on AI such. And when I started talking to him, I heard things from him that I thought I would like to voice or that agrees with the way I'm looking at this point, because he started talking about the problems of resources. He said, like we are relying on better batteries and better technology now, because we are using up too much for this journey. And in my head is like, you know, it's like we have to go on this journey. And so even if we finish up all our resources, it's, you know, we have to, like, find this end of this thing or something. So I we kind of connected on many of the points and these are things that I don't hear my AI tech friends in Bangalore talking about, like, it's like some family business or something. They don't say anything bad about this, right? Like in the tech world, you don't agree like it's always more tech is the only answer to like bad tech. Like, you know, we can leave out like a million people from like, Remo, MGNREGA, things like that. It's okay. It's just the tech is like coming in place, you know, but we have to try it out before it's in place. Like before, we've proven it, we have to try it out. This is the attitude that the tech world has been giving me a lot. And because we now don't draw lines between what is tech and what is commercial. This AI thing itself. I've asked many of my friends, I've challenged them to like, you know, can you see it separately? Because earlier stuff I can see separately like that. What we're seeing now is like it's such a mix that you know, so As science people are also afraid to say anything because it's so much on the funding. It's like, you know, I can talk about it because I'm not in that business. So, you know, I think we can only breathe. So these things that Laura said, like, then I was like, okay, I can, I asked him, Can we work on something? You know, I wasn't so confident to tell him like, let's work on a dystopian novel, because I don't know where it is. So the safe area is like, you know, let's make an inform in like, sort of something about AI. Explaining AI. That's the first thing to do. Because people don't know what AI is. So it came from that sort of an idea. So it went in there. But when the then Laura came, he met the publisher.

Laurent Daudet  15:54

Or, yes, yes. So I met I met this, this publisher, who will tell us that it's, it would be much, much better on with a, if we tell a story, instead of just having this explainer. That would be abstract and so on. So it's, so as a as a proponent, we're saying this, the technical chapter of this day as as blueprint, all across the book, but we make it as a story. And we make the character on that. This idea came later in the project, but I think it ended up being also a nice, very nice ID. Yeah. Is

Appupen  16:33

the right idea, right? Nobody wants to read a textbook and like, you know, when we bring it in, and I'm very happy that you said that you liked the characters and the connections. Yeah, no, because that

Michelle D'costa  16:46

sort of elevated the book lend my sense, but of course, I mean, you know, for listeners who don't know how the book looks like, it is fabulous. I have rarely seen a graphic novel this well made, especially in India, okay. I mean, obviously, you know, abroad, you do see a lot more investment into graphic novels, comics in general, right. But in India, we it's sort of very concentrated. So I just love that. So apart from the colors apart from, you know, the production quality, I felt it was the storyline, right, because at the end of the day, as you said, Who wants to listen to facts, or, you know, like a textbook, it's like always being in the class, it's being, you know, getting a lecture from from someone, which which could be great, but I think it's the way you've written it, that that sort of has made it very interesting. So, you know, another thought that came to my mind was that I think I read it on your website, that you sort of called this the world's first graphic novel, on AI. So I wanted to, I wanted to know more on that. Because, you know, in my research I did, I think there were two other graphic novels, recent ones, I think there was one called carbon and silicon, I think, which came out in 2021. Where are these two robots, you know, from the Silicon Valley, and they sort of they're created to care for this aging, human population, you know, and then of course, you see the certain downfall, then there's another one, I think, came out in 2023. It's called the Infinity particle, where a young inventor, you know, falls for lifelike AI. And then, you know, they sort of confront, you know, as usual, there's this introspection, they confront their freedom, autonomy, all of that. So wanted to know bit bit about that. What was your vision for this book? And why do you sort of claim it's the first graphic novel on AI?

Appupen  18:23

It's a stupid marketing claim. You know, people like to say, like, India's first this thing was, was that thing and all that. I don't take that seriously at all. But actually, when we researched, there weren't any real graphic novels detailing AI. So I thought it wasn't a wrong thing to say. But you know, it's not something I really like to say or anything, where it's it works on the website, I think, because people talk about it. But

Laurent Daudet  18:52

I think it's, at least I know, I don't know if it's the first but it's one of the first to talk about the the latest generation of AI like with Chad GPT. And just this, this new, what's called generative AI, this is probably this is probably the novelty factor.

Appupen  19:10

So that means like, when the next technical, technological innovation comes, like, if somebody makes it, they can also say it's the first

Michelle D'costa  19:19

Huh, that's interesting, actually, like with, like, yeah, like, you know, when you like, earlier, I sort of, even though, like, I'm not really well versed with software's and how it works, but I remember like, you know, years ago, I had this, you know, mobile phone, right, and obviously, I'm just more well versed with the hardware part of it. And then remember, all of a sudden, I couldn't use the phone, and I was like, but I haven't dropped my phone. Anyway, that's the first logic that Oh, like it shouldn't, you know, sort of work and then someone told me when was the last time your software was updated? Before that, I suppose there's something called software and then I sort of figured that there are no new versions, you need to have it updated and all of that right. And as you see, even with, even with our systems company Windows laptops, all of that, you know, it's just the hardware there. But at the end of the day, what's important is the application or the software in it, right? So yes, as, as you said, no, no, I think it's, you know, new generation AI or generative AI, as you call it. You know, the more versions we see the more developments we see, I'm sure we'll see different kinds of

Laurent Daudet  20:18

yes, on already. I mean, the book, we finished the writing, maybe last last June. And if I if if I were to rewrite the whole story, no, I would change already things here on there. I mean, it's, it's changing so fast that I, there's a bit of a paradox in writing a book about some technology that's evolving extremely fast. No, we have a text to video, there are new models released every week, or even by the day, things are changing so fast that, yeah, there are opportunities for new books.

Michelle D'costa  20:57

Right? And who knows, maybe maybe we'll find that both of you collab again, and come up, come up with another tale, you know, hoping for that. So we're

Appupen  21:05

just sorry, we're just waiting for technology to catch up. We know, while we're doing this, we know that, you know, it has, the things are moving very fast. We mentioned that in the book as well. So the what is moving fast is the technology or the, you know, some components in it, but the behavior of the overall thing. That's what we have to focus on to make the book like, live, you know, so what I was trying to do was that, like a earlier, we say the same things with like a previous machine, or how like, you know, companies or corporations are taking over or something. But the idea behind it, like, what will happen with it, I think that will remain the same how it's being done, maybe update. So I think that info module parts will have like more updates and all, but if you ask me my take on the whole thing, it might still be the same. Yeah,

Michelle D'costa  21:59

I think the soul Yeah, you're right. Because like, this is something that we have also asked other writers where, you know, sometimes the whatever they write seems very timely. And then we wonder how will that sort of poll 10 years later, but see, always a book is sort of a testament to the time that it was written, you know, and at the end of the day, I think the soul of the book, or what you're trying to say, through the book is, I think universal, is something that will stand.

Laurent Daudet  22:24

Hopefully, yes.

Michelle D'costa  22:25

So definitely, I really liked this, especially the spoke of versus I would say good versus evil, or just, you know, sort of anticipating the the kind of, you know, I would say responsibility humans have, we don't even really know, the kind of, you know, I would say the opportunity that we have, and it's all up to us to sort of, you know, I would say use it for a benefit or to sort of, you know, use it for a very selfless need. So, Laura, I also saw this on your website that, you know, as a scientist, you say that, you know, you're amazed by the new capabilities of all these AI models that you come across, right. So, what you said is, they can now perform tasks that they have not been explicitly trained for, I found that really intriguing. So, what exactly do you mean by this, could you give us some examples? Yes,

Laurent Daudet  23:12

for instance, as we say, this model, they are mostly statistical model just to predict the next word, but you can ask models, like, charge GPT to summarize the text or to extract keywords or to rephrase the way a child would, would speak or all these all these tasks, they are not hardwired, in the, in the way they are programmed, this is just something that that because the this task, the model are able to perform this task because it's part of the training corpus, and they have seen a huge training corpus, like billions of billions of words. And in there, there are some summaries, there are some charter stock and there are some key words on the bed. It's not the the intent of the of the people who designed the model. They did not have that in mind. It goes the the number of applications are way beyond the what the programmers had in mind. And I find that completely fascinating. So like

Appupen  24:24

making those summaries and or, okay, so just trying to that is part of this kind of stuff where it wasn't really like meant for that.

Laurent Daudet  24:33

Yeah, it's something that's what that's what, why we people at Stanford call it foundation models. They are just models that are capable of a variety of task on the DC due to their nature, and that's, that's the new models that have billions of parameters. Because they have billions of parameters and they have read, read between quotes of course they are have been trained on the massive datasets of texts, that where it comes from this disability to perform tasks they have not been trained on.

Michelle D'costa  25:12

Oh, that is really fascinating. Like it has a mind of its own. I think that's what you mean. Like,

Laurent Daudet  25:17

I would be very careful with the type of words you use. It doesn't have a mind. I mean, it's, it doesn't think he doesn't have any feeling he doesn't know what's right or wrong. It doesn't know what, what's good or bad. So it's a it's just a just a tool, just a machine. So

Michelle D'costa  25:37

absolutely. Yeah, I do i Yes, I understand. Because when, especially in this in this context, when you say it has a mind of its own, I do get it. It's almost like an independent entity. So regarding, you know, I, this is something that I read that you had set up, and you said that AI is just a tool, right, just like an engine. And you said that we don't have to relive one more disaster to sort of prove that, you know, reckless profiteering is not great for us. So I want to know, which disaster are you referring to? And in what ways do you see it and sort of bad for the planet? Oh,

Appupen  26:10

this climate change happening right now. There's nothing we can do about it. The point is, we're given this feeling of agency, with all the social media and everything. And it's a bubble, where we just hear what we like, or we read only what we like, and we feel that things are fairly under control in our little bubble. And that is intentionally created for us. We there's like, you know, there are wars, there's, you know, genocides happening, we can't do anything about it, there is no agency or any of those things. And that's why we get to know these news, like thinks Chomsky has said it like very nicely. That's why we're getting the news because we can't do anything about it. So now we're giving more power in the sense like, we're with control with AI and stuff. It's like, because like what Laura said, like, it's controlled by the largest like business interests, there's no country other than maybe China or us that can afford that machine. So now only corporations can work on this machine or like, make it bigger. Now, their focus is to use it for profit oriented things not for like social benefit, or like educating some, you know, kids who have no food or something like that. Earlier, if we had like independent research happening, governments like doing research, then these agendas would also come in, right now they don't come in. And we are, you know, given this idea that we can say something or anything, but we got we don't have any agency and how it's gonna go. So I think that's where, like, we don't have to put a new example, we know like, you know, even say, like, a lot of the things we believed about renewable energy, has now come out as like a propaganda by like the fossil fuel companies. So the same guys who came around early 2000s, and said, you know, a brighter tomorrow future, like all the fossil fuel companies like took that branding, and they were doing it, same way. Now, you're seeing it with AI with responsible ethical AI, everybody will do only responsible ethical AI, but it will be in the hands of like these companies, and we are a country where we have cheap labor. But you know, we are ready to export left cheap labor. And we are saying we can undercut that labor with AI? whose interest is that serve? And when do these regulations come or happen? It won't really happen. So you know, it's the profit or the selfish attitude that I'm talking about, which will take over once given to businesses.

Michelle D'costa  28:55

Yeah, that's that's a very interesting angle. And thanks for explaining it that way. See, because this is something that we have seen this kind of conversation that AI can take over people's roles can take over people's jobs. Of course, we know it's more prevalent in certain countries, definitely places like India. So there was this book. Earlier that we had interviewed this author Tyrone Solanki, he had written this book called letting the machine is learning which was sort of speaks about this effect and how you know, people can overnight lose their jobs in in a lot of, you know, companies in India and you know, nowadays using AI in different forms, you know, there are these bots, there are these apps that we have on our phone, you know, even when it comes to things problematic things like therapy, right? You would you'd rather go to go to someone who's a professional, but as you know, you know, even even that has its own barriers, right? You need to be able to afford therapy to do that. So then, you know, people would rather you know, get a free app or a free board that does that, or even astrology, that's pretty big in India when it comes to your future. Instead of instead of sort of paying paying someone you will get a free you know, and And you know, I think recently I read it was just I think probably last two weeks were in Kerala, there's a first AI teacher that has been introduced a robot. So, you know, I want to know, for both of you, this is something I'm sure that you will have thought about. Do you think there are particular kinds of roles that might get replaced by AI that are more likely to get replaced by AI? I mean, you know, apart from like, the obvious where, as you said, poop, and you know, there's, you know, class and divide, obviously, where you have cheap labor in certain countries, but apart from that, are they do you think certain roles which are more susceptible to be replaced?

Laurent Daudet  30:37

Yeah, I think it's, first, it's very difficult to make, to make claims about the future because it's, it's evolving so fast. And if we, if you were to ask me, I mean, look back five years ago, on the on, you tell me that the Hollywood screenwriter would be on strike because they feel threatened by the, you would think that I'm a fool, right? It happened. But I think it's in terms of jobs that most likely I tell people you are, you are not going to be replaced by AI, you are going to be replaced by someone who uses the AI is which is, which is always suddenly, it's it's a bit different in a way, but we we already see some massive impact of jab, jab replacement in, say, customer support, in the call centers, and so on already, we can already see the numbers.

Appupen  31:41

I think, see, the advantage of AI is it churns out things faster and cheaper. And it can do this, like large learning and collate information and give it to you. So any, anybody who's in that, in the way of that is kind of under threat first. Now, it's also taking like creative jobs, and like managerial jobs and things like that, that's where the narrative change to like, you know, you can do birdwatching, you know, when your job goes and all that. But when I was in France, at that residency, I was getting friendly with the supermarket like cashiers, and one fine morning, they all left, they were no cashiers anymore. It was all like machine handled, this is before, like, AI really takes over. So it's just a continuing thing of like us doing some sort of automation to save costs. They were all old women who are the cashiers over there. Like, the whole batch is God, now only one security guy is there. So like things like that are happening across all the time. With AI, it will become much more exponential is what we are saying. And yeah, in some place like this, like, you know, and there you see the argument that you know, oh, are you saying like no to like progress, like industrial revolution also happen the same way? Industrial Revolution happened, like 100 years? You see, we are talking overnight, okay. And we're talking like, regulations so that these things hit slower? Like, if so many people are going to be displaced for a particular employment scene? Is there any other place to absorb them? Or can we develop another place to absorb part of the site,

Michelle D'costa  33:16

when you when you have a mechanism that can sort of, you know, bear, like a cushion, where you can bear the kind of, you know, shocks or the kind of effects it might have, I think, also like the COVID vaccine, when it was out a lot of people were skeptical, because they said you're speeding up the process. Exactly.

Appupen  33:33

There at least there's a, there's a pandemic, like hovering around, right now, what do we have hovering around? Now, it's just an artificially created bubble. Because somebody wants to, like get to this point of like, profit or something the whole world has to run as though there is this pandemic, like, you know, and we are to inject ourselves with something which we haven't really tested out. But yeah, okay. So this feels like that it's an artificial thing that has been created. We're playing this game.

Laurent Daudet  33:59

But But you're right, I put pen to talk about automation, but I think what is what is new is that there is no white collar automation, automation that like your what what you described with the cashier, and whether it has already been happening for some time. But this this white collar or also impacting people, like with a very high salaries, like, like lawyers, for instance, or, or doctors and so on. That's something that's, that's very new CEOs and

Appupen  34:32

running companies where CEOs are like, Yeah, you know, so far, they haven't made any I think, no list, but like,

Michelle D'costa  34:39

no, which is really interesting, because there's a chapter in your book, which which, as you will have mentioned, also very clearly that you know, the most of the book you will have written on your own right, there's very much human talent, imagination, creativity, but there's one chapter which you all derived from Chad GPT. And, and you know why this is specifically interesting to me. because he in the publishing world this is this is the kind of narrative or discourse that we have heard in the past, I would say a couple of years, you know, like even Salman Rushdie, he joked that he put in a command in chat GPT, where he said, Please produce 300 words in the style of Salman Rushdie. And he said, what came up was pure garbage. And we've seen a lot of books being sold on Amazon, written by Chuck GBT. Right? So I want to know, what is your take on it? What is the experience writing the chapter using this tool?

Appupen  35:31

It was a marketing gimmick, which I thought would be a great idea when we thought of it that time. But like afterwards, you know, everyone's like developing this kind of thing. So our focus became on showing what can do, what are the limitations and how little control we give it, I don't want to give it any control over the story or the book it because it doesn't know, it doesn't know what is good and bad. Even when I'm just asking you to generate an image, it doesn't know whether it's good or bad, you know, it, you can't trust it with anything. So I finished the story. We both like go through the whole thing. And then we feed the summary of their story to AI and also art, my art from the first five, six chapters. So about 600 drawings were fed into the thing, because there's a lot of like, gray area about the copyright in images, it doesn't create, it reproduces. So let's like stay with that, because we say that in the book, also, if it is trained on the work from the book, and my style of drawing from what I've produced for the book, then its original, in a failure impinges on very little outside, I would say, because, see, we are asking you to draw things that are not there in the book. So it is already taken ideas of Paris of France and or from some previous training, and is using it but in my state. So at least we can say that, you know, we visually we haven't, like stolen from anywhere else. But that is the problem with AI generated art, it is not creating on its own, it has scraped a lot of stuff. And it's like putting it together pixel by pixel. And in the US, there's a copyright issue coming up about this, which is great, because then you can publish AI generated stuff. Laura was like very involved in the training of this AI and putting it together, I felt it is something I never want to do. And the results also, we had to ask so many times, you know, it's not like getting the trip. And it it has to be prompted for text separately and images separately. So after it gives us the text, we have to think what the image should be, and then ask for the image like separately. That's why I said like, you know, making a comic is not going to happen right now with it. But I'm sure you can make a program right atomic, but it's very dead. Like we have to keep saying like add details, add gory details and things like that. Laura, how was it a training

Laurent Daudet  37:52

I think it was a it was still interesting to see her in this image generation software that we use, how the model got better and better at catching some of the graphical style of Apple pen on get them well over produced and it's interesting to see that the main character of Google is rather well rendered whereas some other characters maybe they have less training data to be trained on they are the result is less convincing

Michelle D'costa  38:30

No I think it's yeah, it's definitely interesting like a puzzle Right? Like when you feed in certain you know, key words in the more specific the keywords in the the the better the outcome and this is something that I've seen, you know, on chat GBT I think we've all had our own experimentation or fun with the app but you know, I will say coming back to sort of like the the ways in the different ways in which you know AI can be misused I think this is something that a lot of people are facing you know, these days which is what I've heard you know, like in India like let's say no recently an old woman she fell victim to this AI voice scam where anyone could pretend to be a loved one someone could be from your contacts and and pretend to be and then all of a sudden you know you lose your money from your account I wanted to know if any of you in both of you have sort of faced anything like this you know, I'm not talking about your professional capacity but in in your personal life. Have you ever encountered this this sort of flip side of AI or is it that what I saw you're sort of aware or you're so ahead of the game that you are can predicted before?

Appupen  39:31

Nothing thing like that. We've been lucky if we haven't like been hit with something like that. But both of us we heard a really nice example recently. Of CFO of a company it's a it's a deep fake video of the CFO and he struck a deal with somebody and he ran away with a lot of money, billions of dollars. I think those kinds of stories are important. Like, like just like you say the white collar jobs going that is in part because we need the people who have a voice to get affected, we don't care if a lot of people who don't have any money are getting affected, because, you know, we even our newspaper is not catered to any of them. We only want the people who are spending power. So I think this kind of thing is like sort of a shooting itself in the foot a little bit. At least it will bring about a little bit awareness and some regulations for at least these things.

Laurent Daudet  40:27

Yes, I think we are going to see more and more of this, this deep fake type of video voice and so on. As Ben was saying, you, at least that's also one of the the objective of the book is to raise awareness about about this, that we that we see more we need education. It's also like in the journalism, what you what you read on the internet is not always right. And you need to have your critical mind, maybe enhanced by this technology and not diminished.

Michelle D'costa  41:03

Yeah, no, I think it's very interesting, because recently, someone showed me this deep fake video, which was an audition on America's Got Talent that was a very calculated one where he wanted to present himself as Elvis Presley. And it felt like Elvis Presley was live on stage singing, but obviously, it's someone else. And I think, you know, even before this, this whole new sort of ways in which people are misusing it, I think earlier, what people would also do is use email, which is a very simple, you know, tool that has been going on for such a long time. And I think this this was really outrageous in especially the publishing industry, when the Booker Prize winners, instead of the actual writers winning the winning the amounts, there was someone else impersonating them, and they sort of, you know, just wired the money, and it's gone. Yeah, it had happened to quite a few until they figured it out. It is it's not the person. So anyway, so I think I think, you know, we've uncovered so much, you know, about AI so much about, you know, what could be good, what could be bad, I want to talk more about each of your, you know, like this A Day in the Life of you know, your, your sort of work life. So a lot of we know that you're a physics professor, by day, I would love to know what it's like for you to teach. You know, you teach in one of palaces, leading universities and for you up and you've, you know, made a name for yourself in the comics world. So what could an ordinary day in your life look like? Oh,

Laurent Daudet  42:26

so as for myself, I'm, I'm no, I used to be a physics professor. But I'm now on live to run this startup company called lighten. So well, my, my, my day as a CEO of lighthouse is like managing a team, trying to discuss the next product. The next the, I'm trying to find investors, I'm trying to find clients. So it's like, it's like a business, a business model. I mean, that that is still very much related to technology, because we have to catch up with everything that's happening in the world of AI. So I think this, this technology watch is still a big part of my life.

Michelle D'costa  43:09

Okay, interesting. All right, like an entrepreneur, entrepreneurs life, okay.

Appupen  43:15

Complete CEO, like, you know, that's a good point. Also, like, imagine if I asked like a big CEO, in India, whether he wants to, like work with me or a graphic novel, they laugh at me and like, send me home or something, you know, so I like how we took it seriously as a project. And we like really went into it. And we discussed all kinds of things along the way. And it's, it's a really, it's, I was scared of like collaborations before, this has been like, really nice in that way, you know. And you know, that we have like, slightly differing perspectives on this topic. And that's what makes the topic or the way we deal with it more interesting. I thought it worked very well that way. My scene is like, I'm working on some book or the other. So I work at night mostly. So I'm just starting my day now. And then I work on what I have to work on, and then hope that you know, some commercial project is going to come pretty soon. Every night, I spent half an hour hoping that some commercial project is going to come. And then I have dinner and then I get back to my toys. Or like

Laurent Daudet  44:27

the time difference between the price well walking in, in a good way, because when we were making calls was we were making calls at the it was the like, the end of the afternoon for me in like at the end of the day, and it was the middle of the night for I prepared. That was

Appupen  44:49

middle of the day for me. So you know, it worked out.

Michelle D'costa  44:51

Yeah, because that was actually one of my questions, you know, with the time difference How do y'all manage now I know that it's just it's a fit. That's That's great. So, okay, I would love to know, you know, because you're, you know, while you're have this, this obviously knowledge about AI, I'd love to know if y'all have any favorite movies or favorite books that sort of explore this topic. It may not be, of course, you know, something closely related to your book. But anything to do with AI or technology that that y'all have liked seeing on screen or reading in any books

Appupen  45:26

I like iRobot the movie? In reading, yes, some as more than stuff. Always. When I go into that science fiction, that's what I like. But I didn't like, I'm not a big fan of like too much of tech science fiction. But I think a lot of them had the same kind of narrative idea going on, which was captured well, in that iRobot movie version. Yeah, probably us. Yeah. Alex, press director, but like, weird guy. So I think that, that captured a lot of that. And I see many other movies as being different takes on that kind of thing. The AI becomes intelligent. Does it have emotion? These are the questions we always hear when we're having our sessions. Also, I think the that fantasy from the movies is really captured people. And that is one of the main reasons why we're not talking about the real AI. Now, we're all considering this AGI or like, you know, when you get smarter than us, or one guy even said when we go and terraform Mars, you know, like, so very much like, bought into that narrative.

Michelle D'costa  46:35

Yeah. So, yeah, I think you're right. Yeah. Because you're so caught up with that fictional world, right? Like, even our like, let's say like a layperson, like me, my view of AI is always obviously, you know, influenced by these movies by these stories. Yeah, yeah.

Laurent Daudet  46:51

Yeah, there are many, many of these movies from science fiction that I like, but there is one that I saw about 10 years ago. That's called her early in her, are you You see, this is? Yeah, he's one of these characters played by Joaquin Phoenix and is starting a relationship with his AI, virtual assistants. And when we saw that movie, at that time, we were thinking that it was part of our way. And now it's, it's happening now. That's also something that's, it's very interesting to see how something that at that time was considered as pure science fiction is very close to the type of virtual assistants that we have now. See,

Appupen  47:35

but that is that idea of this AI intelligent, like girlfriend kind of thing coming into the mainstream. But from the 80s. We've had it in comics and manga, the same team, you know, we brought it into that realm where it can be made by Spike Jones into I think it was Spike Jones. Good movie. Yeah, it was, it was a nice way of showing it because it took away that, that that tech science fiction part of it, it was very real. So we started seeing it in a different way. But I think it is also being pushed in a way because that's the kind of like, girlfriend or boyfriend you want. You just agree to everything. And just like you know, say nice things to you. It is an extension of our social media world. So we're buying into that, and he is very good at it. That's why the rate is so high on like, policing us. So yeah,

Michelle D'costa  48:29

that's interesting. Yeah, so the recent Bollywood movie that I told you about in that as well, so it's, it's called Teddy bateau may I say old Javea? It's my Shahid Shahid Kapoor, Shahid Kapoor and Kritis. Unknown. So in that, again, she is the female body. She's a robot, and she's sort of like, you know, how Muhammad Nana, so it's more like, yes. Saying yes to everything. And like, Oh, yes. You know, I mean, and then hearing okay, you're great. And the moment you see that's the flip side of where you see that? Oh, you know, she might not be great. And and he just thought this is a spoiler for anyone who sort of not what, so tell us tell us, but what he hacks her to death. So imagine. So it's a very, very problematical hacking,

Appupen  49:12

or it's like, no.

Michelle D'costa  49:16

So very, very disturbing. But anyway, but yeah, but as you said, it's more like, you know, making sure that we are sort of, you know, just pleasing, pleasing the sort of the user, you know, of the of the software and all of that. Anyway, so yeah, I think I think her was also one of my one of my favorite films. But what what I like is that, you know, this trope of being, let's say, maybe a partner or the sort of love story that you've seen, we have seen in a few stories, but there was one that I'd seen, which was an adaptation of a book. I'm forgetting the name. Now, I don't know if y'all have seen this, but it's basically the concept was that there would be a sibling, you know, sibling is the robot. Okay, where, where it's like, you know, the concept was that Where you when you're adopting children, the children would feel left out, right? So let's say let's say, you know someone who's in the West, they are adopting an Asian child, right? But how do you make sure that the Asian child still feels at home? How do you how do you make sure that they have, you know, company, so they would then hire a sibling or a robot who will then you know, sort of give them the experience. I really love that story I that it's not coming to mind. But but it is a book. So if you are interested in this, I would highly, highly recommend that.

Laurent Daudet  50:30

Nice, interesting. Yeah, I think it's also something that we, we mentioned in the intro machine, but the way people want to have the loved ones after after their best to have virtual via virtual people like, Yeah, this is all on the same IDs. Yeah,

Michelle D'costa  50:52

yeah, exactly. I think it sort of builds on this vulnerability or this, you know, this feeling of validation and very, very human feelings. I think it just sort of, you know, exploits all of that. Anyway. Okay. So that brings us to the last round of the interview. It's our rapid fire round, where you have to answer in one word or one sentence. Okay. Okay, now this is for both of you. One trait that you both share.

Appupen  51:25

Good working at.

Michelle D'costa  51:30

Okay, nice. All right. One habit of each other's that y'all have inculcated through this collaboration.

Appupen  51:41

Looking at things very objectively, and judging it for what it is. It'd be tough for him of course.

Laurent Daudet  51:54

Finding the right word.

Michelle D'costa  51:56

Oh, okay. I love that. One thing that you love about AI

Appupen  52:06

opening up the possibilities to so many stories.

Laurent Daudet  52:13

The it's everyday novelty.

Michelle D'costa  52:17

Oh, nice. Okay, all right. Which is your favorite or I think this we already covered the favorite story about AI. Okay, um, one thing that you love about graphic novels

Appupen  52:32

as a creative control

Laurent Daudet  52:37

Oh, the I think it's the best way to handle complex. Complex things like science.

Michelle D'costa  52:48

Okay, your favorite comics growing up? 

Appupen  53:01

Restaurants at the same time? Okay. Awesome.

Michelle D'costa  53:03

All right. Okay. If you could pick one chapter from your book Dream Machine. If you had to redo it, which chapter would you choose?

Appupen  53:13

The last chapter the AI generated one can be read and rest of it is thought out. Its proper.

Laurent Daudet  53:20

Yeah, I think I think same, maybe, maybe expand the chapter or maybe do it differently? I don't know.

Michelle D'costa  53:28

Okay, awesome. If you had to pick one color palette, one alternate color palette for the book. Right now. It's a mix of blue, white, black. What would the color palette be?

Appupen  53:42

Open ish. Because I have the option of only one color with the black and white. So we're doing is two color printing. That's what brings the cost down. So I have just one color that I can pick.

Michelle D'costa  53:56

Yeah, so what would that one color be? Yeah, purplish,

Appupen  54:01

purplish, bluish.

Laurent Daudet  54:03

Some type of Brian's brown, orange type of things a bit dirty as well. Nice.

Appupen  54:12

It's too warm. You know, I would, when I go to France, or when I think of like Czech and all these metal things. I can go with that color. That's why this there was a thing about this green, the blue also, I've been making, I've been putting in a little bit of yellow and making the blue a little more greenish and warmer. Because otherwise it looks like a very corporate like Facebook kind of blue. And I don't want that because at least just that color will give you this. There's a slight like mismatch in your head. This is not what you expected the color to be. And that will that will position it in a way that you're you're okay to read into it as a new thing. I don't want to go with the existing idea of that. Totally AI feeling you know?

Michelle D'costa  54:56

Awesome. Okay, this is the last question. What's next? What basically what's your next creative project?

Appupen  55:05

He's making money and like big, like business deals like he's busy with that. When he's done enough change in the world, then we'll make another book for sure. I have two books with parallel Murugan then started working on. So that's that's on right now. And I have a children's book thing also called Evil boy. Wow,

Michelle D'costa  55:26

lovely. Okay. Yeah,

Laurent Daudet  55:28

no, no. Yes, we we shall see. We shall see the future. I'm not making any creative project outside of maybe your follow up of Dream Machine? Who knows?

Appupen  55:39

Yeah. We got to. It's on the gods anyway. Like, you know, we've been asking, we know,

Michelle D'costa  55:44

ya know, maybe we'll see an adaptation. Who knows? And then y'all could go right. Go write the script, or the screenplay.

Appupen  55:52

Example adaptation. You mean, like into a movie or something? Yeah, I

Michelle D'costa  55:56

think. Yeah. Like in India, we're big on Netflix and movies. And this every every writer is dreaming of this, you know, adaptation. So I do hope that we see something like that. But anyway, I think I've had a lot of fun discussing this with both of you, I It's rare that you find this kind of rapport between, you know, between artists who have worked on something I really like the mix of, of the knowledge fun, and it just it sort of made it very entertaining. And that comes across to your book as well. Thank you. Thanks.

Laurent Daudet  56:24

Thanks a lot.