Q&A with Ahmed Elgammal on art, artificial intelligence, and the making of the Berkeley Protocol cover

Ahmed Elgammal is part artist, scientist, innovator, and philosopher—and maybe all parts alchemist. The Rutgers University professor grew up in Alexandria, Egypt, enthralled by the city’s historic art and influenced by uncles who were artists. He earned his Ph.D. in computer science at the University of Maryland and went on to found and direct the Art and Artificial Intelligence Laboratory at Rutgers. His futuristic work, which combines art and artificial intelligence, lights up the cover of the recently launched Berkeley Protocol on Digital Open Source Investigations (which definitely can’t be geolocated).

Did your love for computer science or your love for art come first?

Definitely my love for art came first. Even before going to college, it’s something I loved very much. And I wanted to study art history or architecture. I decided to go for a career in computer science, which I loved also and which was new and exciting at the time. But I never gave up my interest in art and art history. I tried to have my career and passion converge working on art and AI.

Fast forward: Tell me about your research now.

I’m working in computer vision, which is an area of artificial intelligence that looks at and analyzes images. I am a visual person, and I think of AI from a visual perspective and that makes it easy for me to link back to art. I try to analyze art with AI. My interest was, “How can we push AI by making AI more aware of cultural, human products?” I believe AI is not only about playing chess or driving a car or being a robot. Intelligence is fundamentally about being creative. That’s what distinguishes human intelligence from other animals’ intelligence. It’s not about problem solving only. It goes beyond that. I’m always interested in how to make AI creative, how to make AI understand art and also create art.

That brings us to the cover of the Berkeley Protocol. How was it created?

We wanted to create some visuals related to landscape images. AI can generate fake satellite images. You feed [the system] a lot of images of some [particular] concept, and it can generate more and more images of that concept — which leads to the topic of the protocol. I find it an interesting idea that by creating these fake satellite images, you start thinking about fake news and fake reporting and all the effects we see with social media affecting democracy and human rights.

Did you have any idea if the images would look real, or beautiful, before they were produced?

We curated the input for this process to be mainly satellite images that….had this 50/50 split between land and ocean to give this aesthetic. Basically we give [the source images] to the AI system using what’s called a GAN (Generative Adversarial Network), which is one of the main mechanisms used for producing AI-generated images. We used our platform called Playform, which is supposed to be a no-code AI for artists who want to use AI, but don’t know programming or know the technical terms of AI.

The creation of art is always changing. Art is about communication between humans. In the end, anything you create—using AI or not—a human will be the viewer. Humans are always in the room.

The cover image looks so real. Were you pleased with how some of the images came out?

Definitely. It’s very hard to tell whether it’s real or fake. It works very well. One interesting outcome for me was that [the system] also came up with this kind of fake text — if you try to read the text it doesn’t make sense at all. Which I really liked as part of the concept. It ended up not being the cover. This fake text was removed. But what really makes these images valuable to me is looking at what’s real and what’s fake. We also worked on different iterations of the cover. We have a lot of other examples of possible artworks that really speak to the concept of “fake” images.

Is there a deeper statement that you are making with this type of art, a subtext, a commentary about technology?

In general, I believe that making images using AI is a natural next step in how humans make art. From the time of cave painting, when humans made art using primitive tools, all the way to the Renaissance, when artists made art on canvas to print-making, photography, printers. There are always breakthroughs in how we make images and use it to communicate as art. AI is the next step for me. AI can now really give us images. This new ability opened new doors for what it will mean in the future to make images. It’s very natural in 2020 to look forward in the 21st century to really explore the use of AI as a tool in making art.

By training the AI to be creative, are we giving up something that is uniquely human?

Not at all. We are not giving up anything. I think of it as a way to enable artists to expand their creativity and productivity as well. You can think of it as a productivity tool. Making one work can take a day [or more]. This can make one hundred [images] in one hour. Most artists who have been working with these tools really like them. AI is their own student assistant that they can rely on to create works and assets to use later. It’s really a nice creative partner for artists.

When you talk to your students about the future of AI, what do you tell them?

There is a lot of debate these days about AI and singularity: AI is there to take our jobs and lives and take control of our world. I am always very skeptical of that. At least in the short term, AI is limited in its abilities. AI is only able to solve specific problems. We are far away from the point where we can…make a machine like the brain that can do many things all at the same time. We are not there at all. There is no reason for that worry. There are a lot of ways AI can be used to enhance our lives. It can help with productivity, help us automate our facilities with manufacturing. There is a lot of room for improving our lives using AI. It’s not a threat. There was a debate about whether AI would take our creative abilities. In the 19th century when photography came around, the same question was asked: will photography take the job of artists? Do we need artists anymore? Photography never killed art. You have more artists than ever now. It changed the course of art and what art means. The same thing happens now. AI is a creative tool. The creation of art is always changing. Art is about communication between humans. In the end, anything you create using AI or not, a human will be the viewer. Humans are always in the room.

Are you one of the few people in the world working on art and artificial intelligence?

My goal is to make [the use of AI for art] mainstream, make AI accessible for artists. That’s why I established this platform called Playform. Just plug in the images. I like it because once you give it to artists, it’s amazing what they do with it. One of our users just told me that Playform is like WordPress, which enabled everyone to make webpages. It’s a web-based platform that is free to join. But for making this AI, you need to have some powerful processing computational resources. You only pay for these services if you need them. It’s really what I’m focused on now—to make AI accessible to artists.

UC Berkeley’s Human Rights Center, which launched the Berkeley Protocol on Digital Open Source Investigations with the UN Human Rights Office on December 1, is currently exploring the possibility of creating a virtual exhibit with Ahmed’s prototypes for the cover of the Berkeley Protocol. For more information or to join our mailing list, visit humanrights.bekeley.edu or follow us @HRCBerkeley.

More about Playform: https://medium.com/swlh/playform-introducing-f59083d5b0e