A Space of Possibilities: Interview with Ahmed Elgammal
Andrea Blanch: Could you take a moment to introduce yourself?
Ahmed Elgammal: Hi, my name is Ahmed Elgammal. I’m from Alexandria, Egypt. I have been in the United States for the last 22 years.
Andrea: How did you get involved with A.I.?
Ahmed: I came to study computer science and the most fascinating thing to me about it was how to increase the machine’s intelligence. I’m intrigued by the idea of how to make a computer understand the visual world around us the same way we do.
Andrea: Did you have any art education?
Ahmed: I studied art history in my undergrad but I also loved math and science. I had to decide between the two and I had to go with science but I never lost my passion for art. I realized that my real passion was to do something at the intersection between A.I. and art.
Andrea: So nobody else was doing this at the time?
Ahmed: There are people who have been doing things in that intersection for some time, but it was usually very small. It is very hard to get funding for things at this intersection because the funding is organized in such a way that the sciences and arts don't talk to each other. In the 18th century, Kant basically organized science as something objective and logical, and art as something subjective and illogical. Although, if you go back to the Renaissance you find that most artists of the time were also scientists, like Da Vinci.
Andrea: What value do you think there is to A.I. research and advancement by making it able to understand art?
Ahmed: I established a lab called ‘Artery’, an A.I. lab at Rutgers. I saw that a machine could not be called intelligent unless the machine could also understand and create cultural products; not just visual art but also music, literature, or jokes. There are layers of understanding and context there that humans go through, but machines cannot do anything like that.
Andrea: I read a criticism that discussed the issue of intent and A.I. For example, when Francis Bacon did his deformed faces he intended them to be deformed, but for the A.I. that same intent was questioned.
Ahmed: Francis Bacon intended to make these faces deformed for a reason, but the machine doesn’t comprehend the same intentions Bacon had.
When we started, we were trying to look at art at a macro level because humans are usually very good at looking at the details, but it's impossible for a person to look at all of art history at once. However, machines can do this well. A machine can look at someone's artwork and can figure out how the art evolved. We could give the machine images and the machine would start generating more of the same, but that isn’t art, because art is about creating something new, not emulating the past. That is my motivation: how can you give the machine art and have it create something new?
Andrea: The world is changing so quickly that you have to start changing the way you think about things like this because otherwise, you're not going to be able to appreciate the value of what’s coming and what’s already here.
Ahmed: That's why in Miami Beach next week we have a booth and exhibition and we’re showing two things: the art done by the machine autonomously and two pieces that artists collaborated with the machine to make. I see this as a very rich process that will allow people to step into territories that they haven't thought about in terms of aesthetics and possibilities.
Andrea: What are you trying to achieve in Miami?
Ahmed: We're trying to show the world two things: first, what the machine can create by itself. Second, that these are creative partners for artists in the future. I think this is analogous to the creation of photography in the 19th century, because when it was invented the definition of art back then was depicting the world on canvas, but then you have this device that can capture the world for you with the click of a button. So, what's your job as an artist? The definition of art changed as it was influenced by photography. Art focused more on the conceptualization and abstraction of the world rather than just depicting it. We now have a tool that can create things for you. It won’t take the jobs of artists away. It can explore a space of possibilities for you as an artist. You're framing it in terms of what details to feed to the machine, what you want to do with the data. Your job as an artist is the same — to control the process — but now you have a partner.
Andrea: What's the biggest risk you see with this process?
Ahmed: There are many risks; the legal issues, copyright issues, and how society will deal with this. This will be a debate until it becomes clear what role this can play in making art and what the role of artists will be. This will take some time and that's a risk by itself. Art is about the concepts, the ideas. Art is about communicating something. The risk here is not in the definition but in the execution of this kind of concept. How are you going to explore this with the help of the machine and will people accept that or not?
Andrea: Do you see A.I. relinquishing human creativity?
Ahmed: I think it is expanding human creativity to totally new levels. You have to have imagination to have creativity, but the human imagination is limited because we are constrained by our world. The machine is good at exploring possibilities, so if we can frame the machine to explore possibilities for you in the art space, then the machine can give you lots of new ideas.
Andrea: When you're going around and looking at art, does that give you ideas?
Ahmed: All the time. I get most of my inspiration from going to see galleries and museums and all the ideas come from there.
Andrea: Give me an example.
Ahmed: This whole process came to me when I looked at art and tried to understand what art is and what the differences were between styles. Can you teach a machine the difference between Renaissance and Baroque art like you do to students? As an A.I. scientist, my goal is to make the machine do these kinds of tasks that we do all the time. It will tell you that there is a person there, a flag, a couple of people, but nothing beyond that.
Andrea: So, it wouldn't understand a narrative?
Ahmed: Nothing deeper than first level iconography. It's like a little kid going to a museum and looking at art and all he sees is just a woman, a dog, or a cat. I was in Italy earlier this year with my kids and they became very good at looking at things and telling me the story. So, we humans do that well, but machines have no clue how to do that. Pushing machines to do these things is my job as a scientist.
This interview has been condensed and edited. To read the full interview, check out our latest issue RISK