One-on-One With Artist Eunsu Kang

By John Stran

Media artist Eunsu Kang will be giving an art lecture on Sept. 21 at 4:00 p.m. at the Judith Rae Solomon Gallery in Bliss Hall to speak about her current exhibition at the gallery titled FACE.

FACE is an exhibition which creates the human face through computer generation of a collection of people’s actual faces.

Q: Where did the idea of morphing hundreds of thousands of faces together come from?

A: There is no morphing going on at this show. The neural networks study about 200,000 faces then create new faces. The outcomes are not directly using the dataset face features.

Q: When did the idea to do this come to you?

A: The idea of using machine learning for art came many years ago by observing the development in the machine learning field. The idea of generating faces and creating portraits that are artistically beautiful and meaningful came after playing with the GANs (Generative Adversarial Networks) and realizing their bias-free nature and ability to produce aesthetic results.

Q: How long does it normally take to create a digital art piece such as the ones in the Judith Rae Solomon Gallery?

A: Running the GAN may take hours or weeks. The ones in the gallery took about two months or more to get those specific results. I built the computer for that about a year ago and there have been tests for preparing this type of process since then.

Q: Exactly how does the machine generate these faces?

A: A general explanation can be that lots of layers of artificial neural networks learn the dataset and generate results.

Q: How much of the work is done by you and how much is done by the A.I. system?

A: It is collaboration. Visual generations are all done by A.I. and I guide it through its steps and decisions as though I where its teacher.

Q: Is this a common art form where you’re from or in the places you’ve been?

A: It is not common yet, but it’s definitely something people are paying attention to at this point. It is relevant to the advancement of machine learning in recent years.

Q: How do viewers usually respond to your work?

A: I’ve always wondered about that. I focus on creating visually artistic results at the end of the process and I hope that creates good language that connects the works and the audience.

Q: How does this compare to more natural forms of art?

A: All artworks are artificial unless we are talking about the sunset or roses. Between human intelligence and artificial intelligence, the results are not distinguishable. I don’t think any audience can tell that this is not done by human or not as good as other artworks done by only human.

Q: Do you feel that this will be the art form of the future?

A: Yes, it will be one form of art creators in the future.

Q: What will your upcoming art lecture pertain to?

A: I’ll give an introduction to some machine learning methods such as GANs as well as what it is like working as a collaborator and the meaning of these faces.