Artificial Intelligence is taking over the world one step at a time. Amid the marching footsteps of AI in almost all spheres of our lives, graphic artist Appupen and French professor of Physics Laurent Daudet teamed up to create a novel – Dream Machine – that began as an explainer, but slowly turned into a story of epic proportions. Excerpts from an interview taken by Kunzum Review:
Bhavneet: Your book – Dream Machine – is based on artificial intelligence. Where did the idea to write this story come from?
Appupen: I was in Paris doing a residency in Angouleme and met Laurent by chance there. And we started talking. He had some interesting perspectives about AI to give, which got me interested in it and I started asking him whether he wants to work on a graphic novel with me. So that’s how the project started. And then after he said yes, we thought we’d make a short explainer sort of thing and then slowly, it grew into a book. And then we got a publisher interested in it and grew more.
Bhavneet: Who did the storyline come from, you or Laurent?
Appupen: The storyline is based on the information that is there in the book. The information of the book is like a blueprint. From LLMs, how they started and training and all that right to the end, is basically from Laurent. That is the teaching module kind of thing. And we fit it into the story. So, I make the story and the art, and I write it also.
Laurent: I provided some reading material for Appupen, and I think he did great work in translating that and simplifying it up to the point that it was understandable by anyone. And then make the grid drawings that go with them.
Bhavneet: Talking about the drawings… You’ve used three very basic colours for this book. It’s not a very colourful book. It’s just whites, greys and teals. How did that color scheme come into play?
Appupen: We wanted to do a two-colour thing. It was also a decision with the publisher because we’re looking at a bigger market and a better price, so we settled on the two-colour idea. I’d been wanting to work with a good two-colour set for a while after reading all these D&Q (Drawn and Quarterly) books and I like the feel of it. It was only what kind of blue to use. I wanted a more greenish blue, so I pushed for that. That’s how we have it here now. So that and the grey and the black complement each other very well. And it’s also, you know, the book has information which has to be taken seriously. I didn’t want to blast it with colours and make it like a Disney animation or something. I wanted to keep it a little serious, so the story and the information in the book is taken seriously. That was the attempt.
Laurent: We discussed a lot about the exact variant of blue to go with it over many meetings.
Appupen: Yeah, my publisher was not understanding why I was not for that Facebook kind of blue because it gives you an idea in your head. It’s a very corporate blue. That is probably the idea of AI that you have. If you just tell AI to make a book on AI, it will probably put it in that colour. I wanted to angle it a little different, so that when you open it, you’re struck with, ‘OK, it’s not what I expected’.
Bhavneet: In terms of artificial intelligence, you know, the way the story line progresses, do you think there is a possibility something like this might play out in the real world?
Laurent: I like you. You’re using the ‘real world’ because it’s a pun on REAL, the villain company in the book. But no, I think that’s what people are working on at the moment. It’s AI that is able to do planning and reasoning, and that’s the next challenge and who knows what will happen in the next 5 to 10 years. I think we don’t want to dwell into AGI (artificial general intelligence) or like super-intelligent AI, because I don’t really know what it means, so I don’t even really know what is intelligence really, but I think at least what we’re going to see is AI helping us, being more personalized. So it will go into the corporate world, go into the government world as well to first to help us be more efficient, but then being more efficient means maybe replacing us or at least some part of the work that we have been or that we are doing. I don’t think there is really a possibility of AI running a whole company or a whole country altogether. To me, that’s way too science fiction, but some people might want that.
Appupen: That’s the thing. I got a good reality check on where the technology is and I can always ask him what he thinks is a possibility, but that goes into the fiction realm. So, when we are warning people or when we are trying to inform people about it, this is the best we could think of as like probable situations which, sometimes the worst situations also that can come up, because the business side of it is so full of highlighting the happiness and the great things that can come from it.
I’m from advertising, so I know how you don’t highlight the weaknesses of companies, so this seemed a lot like that. It seemed like we have a clear agenda of popularizing AI and getting people to play with it with these generative things, you know, move somebody’s head or change somebody’s voice or something… all easy to do things to make AI friendly to users across and not show them or not make them ask deeper questions about AI.
The point of the book is to inform people about AI before you’re asked to make an opinion on AI. Right now, nobody knows really what AI is. I think the book is trying to tell you as much as we know and clear your head about what it can do, can’t do, what are the problems, things like that.
Bhavneet: So Laurent is a professor of physics and has a deep interest in AI. How much did you have to study artificial intelligence to write this book?
Appupen: I studied more than I did for my 10th standard exam, I think. He sent me these modules which I had to go through, and I can always go back to him and ask him to give me more if I’m interested in some areas, explain those things I didn’t get, I asked for additional material on things like some of the environmental aspects, and things like that. Then I did a little bit of research on my own also to add things to it because we want to like, look at it from as many perspectives as possible. This topic is going to affect everybody. Everybody’s interested in it. And the book also tries to serve that.
Bhavneet: When we talk about artificial intelligence, where do you see artificial intelligence taking us in terms of the quality of our lives?
Laurent: That, I think, is linked to the question of privacy and how much we want to give away in terms of our privacy for convenience. I think AI is going to help us discuss with machines, discuss with maybe other human beings through the AI and that’s really an open question really.
Appupen: I think there are definitely many good points about AI, how it can help us in many ways. We keep saying that AI is going to make your life more convenient, but we have to ask ‘convenient for who’ and ‘at what cost’ because it’s going to be different for many people. There will definitely be very good for a section of the population. But is it the same for the other side? Or maybe somebody in India who’s going to lose their job or something? Is it going to be the same? No. So it’s completely held by business hands right now is one of the things we talk about in the book. The research and the largest machines are all owned by these large corporations now, not any research facility, colleges or universities; independent researchers are limited to do the research. So primary focus of the research seems to be on more profit-oriented things, while the many actual good things about AI are just highlighted and not enough money spent on it, it seems. Am I right?
Laurent: Yeah, absolutely.
Bhavneet: You’ve written about artificial intelligence. How much of artificial intelligence did you use for this book?
Laurent: About five per cent.
Appupen: I’m very wary about letting artificial intelligence do anything in our book. But we had to, in a way to show its limitations of what is possible, I thought it’s a good idea to bring it in in the end, sort of a marketing gimmick. So, the last five pages we have let loose with artificial intelligence. Of course we have controlled the prompts. We had to prompt it many times. We are pretty sure what we want. We don’t want to have a free run of AI in the book. We completed the narration, and we fed it to the AI and we pinpointed and asked for five separate instances or takeoffs from the end like how it can go off. So, there are sort of like 5 endings we can play with.
Laurent: The text method, including the graphics also have been generated by the AI by fine-tuning the model on the chapters and the images of the chapters. So now we have an AI that draws in the style of Appupen more or less…
Appupen: …From this book.
Bhavneet: Which application were you using?
Laurent and Appupen: Stable diffusion
Appupen: We trained on 600 drawings or so from this book itself. The beginning parts, and it’s basically to show what we can deliver at that level of training. But maybe if there’s 6000 images it will draw much better.
Laurent: For sure.
Bhavneet: Very interesting. And what was your learning from it?
Appupen: My learning is I don’t want to do anything with AI. I can make a book but like not let it do the stuff inside. I experiment very little with it right now.
Laurent: I think it was interesting to see the fine-tuned model getting better and better as it was trained with more and more data, more and more images and originally it just quotes some of the style patterns that you have and then the face of the main character is I think rather well-rendered at the end, but maybe not as well in the secondary characters because there’s just less training material for them.
Bhavneet: Stepping away from your book. Nowadays, when people are busier with their cell phones, spending more time on social media. Where do you think is the future for books?
Laurent: That’s a good point. I mean future for books… when TV appeared, the people say the cinema is going to die, it didn’t. It just changed the nature of cinema. And the people are doing film for cinema and film for TV that are sometimes different, and I bet books will still be there, but maybe they will be different and some of the books will be maybe written with the help of an AI, for sure. We already have some of them. I don’t think they are great books, but they will get better and better.
Appupen: They will get better, and AI will start creating movies and stuff also, but I think even in that case, human-made movie, will have its own stand. And books. Yes, I think it’s more about the attention span because of the social media and the amount of stuff we get for free. We are looking for instant gratification mostly. I think it’s something that should be reversed. I think it’s something I would personally try to get people to spend more time on one subject and try to go deeper into it. Even in the art that we do, we’re losing a lot of the ideas because people just look at the surface, you’re not looking beyond and understanding what is behind that or what the artist or writer is trying to say through this. You just see it as a surface-level story. That is a problem and that is why we have politics like this and industry like this and commercialisation like this. It helps that agenda. I think it’s a much larger thing that all artists and thinkers would like to do to… Make people try to stay and focus on things.
Bhavneet: What do you think is the importance of reading?
Appupen: I think the importance of reading is again learning and being comfortable with yourself. You have to be able to sit and spend time with yourself without fidgeting or being feeling like alone or something. I think that’s the space from where creative thought comes also, or any thought, for that matter. So if we are comfortable with ourselves then we are better off. We’re not like looking for the next escape or the next distraction to get us through the day. I like that because that’s when I feel better. I try to put that in my work also to show people, ‘OK That’s why I’m making physical books’. So you want people to sit and, like, read it. And that’s an exercise in itself.
Laurent: I think that’s also the question of what is your relationship with the truth? I think this AI, as it says in the book, they are just stupid statistical machine. They have no notion of truth. They have no notion of what is fair. I think when using this AI, you shouldn’t lose your ability to have your critical mind about what you read, when you read some information is there any source that I can go to that can prove that this information is correct and so on? So I think this can be a great tool, but we should not forget that AI don’t tell the truth. That’s not what they are designed for.
Pick up Appupen and Laurent Daudet’s Dream Machine from any Kunzum store or WhatsApp +91.8800200280 to order. Buy the book(s) and the coffee’s on us.