An in-depth interview with the Egyptian creative whose AI work has gone viral.
With a specific focus on interiors and architecture, Hassan Ragab, an Egyptian creative currently residing in Los Angeles, is a prominent figure in the ever-changing world of AI art. Despite this, Ragab avoids labeling himself an AI artist, as he has a distinctive philosophy (that you will learn more about in this interview). To Ragab, art should not be confined to labels or limited to the tools used. As an experimental artist, Ragab challenges AI bias, whether it is the tool or the viewer, and hopes to learn more about how AI is modifying the way the world, himself included, consumes creativity.
١. What pushed you into AI architecture or AI art?
As a computational designer, I’m always inclined towards learning about technology and how I can benefit from it. I’ve been having an eye on AI for a while, but I couldn’t use it in my own work – it was very explicit and complicated. So I always read about neural networks over the past two, or three years out of curiosity. But then earlier, by spring 2022, I started seeing these new AI models and the imagery, it was something like nothing else that I’ve seen. I even remember there wasn’t a description on the images or anything, I just saw the image. That’s weird. I know contemporary art. I love the art world in general. But I’ve never seen something like this before. So, I tried to dig deep into it. And then I got to know there’s something that’s called text-to-image generators, and from that point on, I tried to put my hands on whatever generator that I could use.
But, the problem is, during the early days, those models like Midjourney or speed diffusion, were the only few platforms that you actually can use. They were by invitation only; that was before the winter release. So I had to apply to get permission to use them. Months passed and then Midjourney accepted my request to try it. I remember it was on a Saturday night, and I was really tired and didn’t want to watch TV. I just hopped onto my computer, and I wanted to know what’s going on with that. And I remember, by the first image that I generated – I think there were 25 free trials – I subscribed because I knew this was like nothing else that I’ve seen. I was drawn into it from that moment on, and I couldn’t stop ever since.
٢. How do you feel about AI and the copyright conversation? At the end of the day, the images that AI creates are built from millions and millions of images from different people.
It’s actually bringing the conversation back from point to what is originality in the first place. I mean, what is creativity basically…so the way that I like to think about it is that the human mind is basically incapable of creating new ideas by itself. So basically, everything that you see around you, every creation has some sort of what you call an inspiration, or, an original source or multiple sources, which you, the artists, like any artist, will see these sources, and then they will just break them down, and then connect them together in a different way. Think about mathematical series- we really don’t make new theories, we’re just discovering what’s actually in the world. So, that’s my take on what it takes to create something. And from that point, everybody kind of “steals” in some way. If you’re just seeing something and you’re just trying to copy it exactly the same way then, no, you’re not an artist or you’re not a creator, whatever you want to call yourself, that’s not art.
If you’re an artist who’s stealing using AI, you will also steal using other mediums.
٣. I agree with you about what you said that if someone wants to steal, they would seal it with AI or not. When you insert yourself in the prompt, that’s what helps create something original, because you are taking the time to write the prompt and input all those little details that you can play with.
I think AI is a very empowering tool. It’s giving people like me who were always interested in art but never had the time to either study it or to just, craft my artistic techniques. I also like drawing. I liked doing Arabic calligraphy at one point in my life. But then, because of my life and my career, I couldn’t continue doing so. So now these tools are enabling me to get rid of the technical skills that I need to create an artwork and just focus on the idea. That’s basically what’s it giving to everybody.
٤. How do you feel about the bias with AI-generated designs?
I don’t know if you read anything that I wrote before, but I spoke a lot about the bias – especially with heritage and dealing with architectural heritage, and the AI narrative. But to start with, think about AI like a child that you try to guide to learn something. As your parent or as your developer, you have to teach them some things. AI has a set of rules, and it’s still in its infancy. Really, hasn’t developed that much. The development is so fast, but I think the potential of having a bigger AI is much bigger than what we will have probably in the next couple of years, or maybe the next decade. But for now, I think AI has its own biases that come from a machine learning perspective.
But then that also creates a problem of diversity and inclusivity and all sorts of stuff. I was really concerned about the bias in architecture using AI but after a while, I realized that these tools that I’m using, are actually biased toward art, rather than architecture, and they are biased towards certain types of words. So basically, they are very good at creating white female portraits, and it’s because there are a lot of pictures of a lot of people that are generating these images based on that. So that’s something that’s not great. But it’s something that I am coming to realise that we have to deal with. The best way to deal with…I’m not sure if we’ll ever have a very diverse and dynamic AI model that serves everyone in the same way, because, again, you have limitations on the data set that you feed into the machine learning…but I think that in the future, everybody should know how to train their own AI model.
For example, in architecture, at first, I was having problems with creating images using architectural concepts from Egypt, whether from ancient Egypt, Islamic Egypt, or a historical era that Egypt has passed by, but after a while, I was starting to think that maybe the best way to address that is actually by building our own model that addresses our own culture. A lot of the products from AI that are generally in introductory architecture have a lot of Orientalist impression and I really hate that. I mean, I’m working with it, but at the same time, I really hate that because I think we can interpret that in a much better way that is understanding to the history and the culture.
٥. I’d love to know more about training your AI at this stage and how can you help your own AI build something unbiased. Is there a way that you can teach it as it is still partially public domain?
Well, it depends. Basically, you cannot train Midjourney because you’re not a developer. You have to build your own AI. Now, there are tools basically for stable diffusion, and there are some techniques to build your own small AI models. But you can’t build something as big as Midjourney because that would require super computers, and the only people that can do that are like Elon Musk or Zuckerberg but, Hassan can’t do that [laughs]. But what I’m doing right now for some of my clients is that I keep small models and these models, but there will always be some kind of a bias.
For example, at Midjourney, they are trying to enforce beautiful images and that will make them in a way reprogram their bots to have some bias to make something that looks beautiful, in a way. So, a good model will probably not have that and it will just be an open pathway that kind of moves freely within the data set to generate some ideas. So I think, the data diversity and training are the keys to creating an inclusive model, but again, I don’t see that happening. I don’t see anybody wanting that to happen.
٦. If the bias is solvable, we would have solved a lot of social issues a long time ago.
I mean, yeah, the bias is human. It’s not a problem with AI, it’s just following us. If you have good intentions, then you will have a good product, but looking at human history, and what’s happening in the world, I don’t think this thing…I mean, I work with it, but realistically speaking, there’s just a lot of greed, bias, politics, and all sorts of things of what’s wrong with the world and AI will follow that.
On a global level, I don’t think you will have this “optimal model.”
٧. Do you ever feel, from a design perspective, that AI challenges your design instincts?
It’s actually helping me get rid of my designer ego if that makes sense. What I really love about AI, is that it’s helping me get new ideas that I’ve never thought of before. And as an architect or as a designer, I always have an insertion of my own identity or my own vocabulary. I think one of the reasons why I am jumping off and on and off from one topic to another, is because I’m kind of looking for something. And I’m not looking for something that’s inside me, but something that’s also kind of connected with the outer world. So it’s a kind of conversation that’s going on and off between me and my context.
And with AI, it’s just one big whole context. There are a lot of things that you can explore, and it’s challenging, and it’s helping me understand the world and myself more…it helps me understand my own taste. So I don’t think about it as a challenge, or maybe it is a challenge that is helpful. It’s a good conversation – very intense with good criticism. Sometimes I hate it, but most of the time, even if I hate it, I always like the output.
٨. When you see people commenting about “AI art,” do you find any misconceptions?
I think it’s a misconception to call it AI art in the first place because I understand why people need to; we need some source of differentiation, but the thing is, AI art seems like the word painting, for example. If I can say I’m a painter, then okay, which painter are you? Are you an abstract painter? Or are you a surrealist painter? Are you an abstract expressionist? What’s your philosophy?
The label is putting AI into some sort of packaging, with everything that is there. This way, it kind of has some sort of a stigma to the recipients, and I hate that. Because I know some people think, when we call it AI art, that it is lazy, not creative, or not original. It’s not about the technicality. AI is just a tool. So, think about it as a pencil or computer software. However, the usage of AI is definitely exploited. A lot of people are misusing it or using it for their own purposes. Basically, AI is starting to do social media content, and it helps influencers gain followers on Instagram. But that’s not what it’s all about. If you’re an artist, and if you’re truthful enough, this tool has a lot of potential to create something that you haven’t seen before. And it’s actually hard to do it, it’s really hard to create something that’s kind of truthful or meaningful. And I think, including myself, I don’t think anybody is even near reaching what we can do using AI in art; it’s really hard. And it’s pinnacle is so far out of reach.
I understand it’s really hard to understand that concept. Unless you work a lot with these tools, and unless you have this experience or expertise and the love and passion for art and design and philosophy, using AI kind of enables you to merge all that to something that’s really new or you haven’t seen before. It’s really hard to describe that to people, and it’s very frustrating…but it is what it is we have to deal with it anyway.
٩. It’s the first time I hear this point of view, and I respect that. So, quick question: what type of “AI artist” would you call yourself?
I don’t like to call myself an AI artist. I think I’m very experimental, but I’m mainly interested in learning things about myself. And right now, I’m really interested in understanding what the word identity means. Maybe because I live away from my home country, or maybe I’m surrounded by a lot of historical context as an architect, and as a designer, and as a father. So there are a lot of things that make me want to experiment or have this kind of experience of new things that will teach me about the world and teach me about myself. So maybe experimental, but not even an artist. I hope that I’m an artist, but it’s something that I can only aspire for. But I think I’m an experimental person.
١٠. So as you and the world continue with AI, what excites or scares you the most about it?
I think it’s really interesting to see the influence that these tools are having over the users and people, and how their styles are going to change, according to the development of the tool itself. Social media, I think, is the driving power of how AI is flooding the internet. And it’s really interesting to see how both those powers, let’s call them powers because they’re very powerful, they kind of drive and change what we know about art and design, and aesthetics for both artists and architects.
Maybe in 10 years, there will be some researcher from Harvard or somewhere, probably an AI art researcher, who’s studying the change in the fabric of the art scene. But I also I fear for it because it’s creating these false ideas in both the artists’ minds and ordinary people’s minds. It’s also creating, some global style, if that makes sense. So now I can see a lot of people do stuff that’s kind of similar to each other, and it’s because of the tool and the media, if that makes sense.
This interview has been edited and condensed for length and clarity.
