Shaping the Art of AI: Emily Spratt Curates Art Exhibition for Major International Forum on the Responsible Use of AI

By Benh LIEU SONG - Own work, CC BY-SA 3.0, Link


By Robert Florida

President Emmanuel Macron will host the upcoming Global Forum on AI for Humanity in Paris, for which Emily L. Spratt is curating the art exhibition, Au-delà du Terroir, Beyond AI Art. As a direct outcome of the previous G7 summit, the forum will bring together artificial intelligence experts and world leaders to discuss the major challenges presented by Al and how they can be overcome to responsibly benefit humanity.  

The exhibition, in the form of a digital projection, will run throughout the course of the forum, scheduled for October 28-30 at the Institut de France in Paris. Spratt, an accomplished curator, selected seven projects by leading artists, scholars, and cultural-heritage specialists, all of whom use Al tools to critique, create, or analyze art and architecture. The contributors to the exhibition are impressive. They include: Hito Steyerl, Mario Klingemann, Refik Anadol, Robbie Barrat, AICAN (Ahmed Elgammal), ICONEM, and one project including the famous Parisian chef Alain Passard.

Photo of Emily L. Spratt by John Palisay and with permission from 20 Exchange Place

The art exhibition accompanies three days of lectures, workshops, and panel discussions about how the international community can work together to support the responsible development of Al technologies. The discussions will touch upon cybersecurity, safety, and privacy, how Al is affecting the workplace and possibly eliminating jobs, and how governments can regulate data and the ethical, political and legal challenges involved in such regulation. In this context, Spratt is shaping the visual tenor of the forum.

“This exhibition explores the use of Al tools that defy our expectations of what we have designated as the reach of human-cultivated creativity, and opens debate on the changing use of images in our society today,” says Spratt, who is both an art historian and a data scientist whose research merges philosophy, ethics, and AI.

The exhibition takes a critical approach to the emerging area of AI art. In the following interview, Spratt, Postdoctoral Research Fellow at the Data Science Institute, discusses the exhibition and the forum. She also talks about her pioneering research at DSI, which bridges the arts and sciences. 

Photo of Emily L. Spratt by John Palisay and with permission from 20 Exchange Place




Florida: Can you talk about how the exhibition brings a critical perspective to AI and “AI art?” Why is it “beyond AI art?”

Spratt: First of all, the term “AI” is problematically vague in its meaning. Secondly, “AI art” has accrued some negative associations in the art world as a monolithic contemporary style. My exhibition is about the use of deep learning techniques to create images and videos that are appropriated as art, and it is particularly focused on the creative uses of Generative Adversarial Networks (GANs). Discussing art produced with these methods also requires theoretical positioning, and it is my hope that the exhibition will encourage discourse on this subject that is beyond “AI art.”

Florida: Which artists in the exhibition do you see taking the most critical approach to the uses of deep learning techniques for visual media?

Spratt: The film stills that I selected from Hito Steyerl’s video installation “This is the Future,” which premiered at the 2019 Venice Biennale, bring attention to the philosophical concerns that GANs raise in predicting the immediate future. Steyerl is well known for her critical stances on the uses of technology and she brings attention to the insufficiencies of computer vision science to mediate or replicate human experiences.

Hito Steyerl, This is the Future (film still), 2019
Hito Steyerl, This is the Future (film still), 2019
Video installation, environment
This is the future, 2019: single channel HD video, color, sound, 16 minutes
Environment: raised walkways, projection screens, Smart Glass panel, Dimensions variable
Image courtesy of the Artist, Andrew Kreps Gallery, New York and Esther Schipper, Berlin

By contrast, Mario Klingemann embraces the creative possibilities offered by deep learning methods. His art showcases the richness of the image fields inherent to the GANs universe, and leaves viewers convinced of their need for further harvesting. One of Klingemann’s videos that I chose emphasizes how “deep fake” faces may be animated, not only to create “fake news” videos, but also as a tool of artistic reflection. In another, an excerpt from “Hyperdimensional Attractions: Bestiary,” he explores how digital objects can be rendered as organic, evolving life forms of indiscriminate earthly origins. In this regard, Klingemann’s art brings attention to the very soil within which GANs-produced visuals develop: our very humanistic datasets of images.

Mario Klingemann, Hyperdimensional Attractions, Bestiary (excerpt), 2019
Mario Klingemann,
Hyperdimensional Attractions, Bestiary (excerpt), 2019

Robbie Barrat, Runway Still, 2018

Florida: One of the artists here, Robbie Barrat, is famous for his association with the half-million dollar AI art sale for Christie’s. What is he bringing to the exhibition?

Spratt: Robbie Barrat is the 19-year old artist and computer scientist who created the open-source code for an artwork that a commercially driven group of young

entrepreneurs took to create an “AI portrait” that did indeed do rather well at auction. In this regard, Barrat’s presence in the exhibition cannot help but invoke a conversation surrounding the questions of artistic authorship on this type of art, and the ethical issues affecting the open-source spirit of the computer science community. I should also add that the community more or less shunned the business operation that used his code for the Christie’s sale.

The reason I wanted Barrat to be a part of the exhibition, however, is because he is doing fascinating work on the use of GANs for haute couture and I expect that he will soon be recognized as a luminary for the fashion world looking to AI. For the exhibition, Barrat used images and videos of Balenciaga’s runway to generate it anew. For me, it was tremendous fun selecting the fall line from his GANs-produced reinterpretation of Balenciaga for what would be shown in Paris. His generated runway video that is in the exhibition is an unforgettable visual journey of melting pastels and a tour de force of fashion poses in their most iconic expressions.

Florida: What other forms of art are explored in the exhibition?

Robbie Barrat, Runway Still, 2018

Spratt: Refik Anadol explores architecture. Employing an enormous dataset of photographs of European architecture, he has brought what the artist terms “Machine Hallucinations” to the exhibition. In his practice, Anadol uses GANs and images in 512 dimensions to create immersive, film-based installations that defy both our experience of gravity and our understanding of the physical properties of construction materials. The video presented in the exhibition reveals the computer’s understanding of the relationships of architectural images in given datasets. This is visually manifested in the presentation of architectural forms replete with constant undulating movement. The images appear radically transforming and ultimately disintegrating into what can be described as an exploration of surfaces in space. By extension, these images exceed our physical and cultural experiences of architecture in the world as we know it.

Florida: There is also a large cultural heritage component to the exhibition. Do these projects offer more everyday uses of AI than the more abstract ones?

Spratt: Yes, definitely. In juxtaposition to Anadol’s art, the contribution to the exhibition by the cultural heritage foundation ICONEM demonstrates the use of AI in photogrammetric representations of architecture through the realistic visualization of the UNESCO-listed site Le Mont-Saint-Michel. The exhibition also presents the photogrammetric model in a manner that brings attention to the aesthetic qualities of its rendering as its own conveyer of meaning, separate from the representation of the site itself, although often overlapping.

Also, as exemplified by the images in the exhibition from the AICAN algorithm created by Computer Science Professor Ahmed Elgammal, the appropriations of generated images as art can take on cultural and political connotations of their own depending on how they are labeled and curated. For instance, “Accord de Paris” is intended to question the outcome of the 2016 climate agreement.

Florida: Can you further explain the title of the exhibition?

Spratt:Au-delà du Terroir, Beyond AI Art” is intended to bring scrutiny to the point at which we see human agency evaporate in the image-making process, if at all, and to underscore the dialectical relationship that the machine-learned and generated image has with the inherently human and cultural origins of its dataset. In this regard, the exhibition also puts to question our means to contextualize its output; it is art that is in fact anti-terroir—it demands visibility in its opposition to all that we guard as being cultivated by human hands and intelligence.

 

(AICAN.io (Ahmed Elgammal), Accord de Paris, Los Angeles, 2017

Florida: Is the reference to “terroir” also related to the fact that another project that is in the exhibition ties into your collaboration with Alain Passard, a chef that many Americans would know from the Netflix series “Chef’s Table: France”

Spratt: Yes, in part. I was delighted to collaborate both with Data Science Institute computer scientist Thomas Fan and with the three-star Michelin chef, Alain Passard. Our project definitely exemplifies the theoretical component of the exhibition.

Florida: How did you end up doing research related to the culinary world?

Spratt: Honestly, it goes back to the fact that I began my studies in the Cornell Hotel School. Although I transferred to liberal arts fairly quickly, my affection and appreciation for the hospitality industry was never lost. 

The first time that I dined at Arpège, my love for gourmet food and my passion for the visual world as an art historian came together in a coup de foudre. I was completely struck by the visual qualities of the plates, which to me brought to mind Matisse. I immediately understood that Alain is as much a visual artist as he is a great chef, and when we discussed this, I was not surprised to learn that he also is a practicing visual artist particularly active in the medium of collage.

We therefore began a collaboration with the aim of exploring the creative relationship between his plates and his collages with GANs. I then was delighted to bring Thomas into the project and right off the bat we were successful in getting the AI to create collages out of the images of the plates. But, it was another experiment that had the most interesting results. Alain and I share a love for the Renaissance painter Giuseppe Arcimboldo, and we therefore wanted to try to create AI-generated portraits out of the machine’s interpretations of the Michelin plates in the manner of the artist’s exploration of the four seasons through portraiture.

Emily Spratt, Thomas Fan, with Alain Passard Gastronomic Algorithm, Collage 9fb2fpctl, New York, 2019Giuseppe Arcimboldo, Vertumnus (portrait of the Holy Roman Emperor Rudolf II), Milan, 1590-91

(Left)
Emily Spratt, Thomas Fan, with Alain Passard
Gastronomic Algorithm, Collage 9fb2fpctl, New York, 2019

(Right)
By Giuseppe ArcimboldoVertumnus (portrait of the Holy Roman Emperor Rudolf II), Milan, 1590-91 - LSH 87582 (sm_dig3224_11615), Public Domain, Link

What happened, however, was that the AI better accomplished inserting portraits into the Michelin plates! I remember Alain succinctly describing the results as being interesting but also frightening! Overall, the response of the GANs to the organic, delicately rendered plates was one that demonstrated the limits of replicating the fine lines of distinction of what we as humans find appealing when bringing the visual sense into communication with our memories of the experience of the taste, touch, smell, and even sound of food. While this delicate equation was not possible to replicate in our generated images, the AI truly brought us to another visual dimension that we would have never considered otherwise.

Because these images are truly anti-terroir, they force attention on all that we appreciate as having terroir, thus refocusing our attention on the exquisite and unreplicable beauty of the plates from Arpège.

Emily Spratt, Thomas Fan, with Alain Passard Gastronomic Algorithm, Plate 6p130fa2ffptl, New York, 2019

Florida: In this context of radical artistic experimentation on its newest frontier, AI, is it surprising that Chef Passard, who is known for his pioneering culinary innovations, is the first chef to have collaborated with an art historian and a computer scientist to use machine learning techniques to examine the visual qualities of Michelin Plates?

Spratt: I don’t think so because the project is completely connected to the analysis of the creative process and sensory experience.

Florida: How do you feel being in the role of curator for such a significant international forum?

Spratt: I am deeply humbled to have been chosen to curate this exhibition. I am also tremendously grateful to have collaborated with such extraordinary people on the exhibition. From the government leaders to the artists, and also my digital technician, Rob Kesack, who is a recent graduate from Columbia’s program in Historic Preservation, I was able to build a “dream team.”

Overall, it is my hope that my framing of these important projects using machine learning techniques will bring a critical lens to this art. My curatorial intention has been to demonstrate that we are at the dawn of a new era of creative engagement with technology, and it is au-delà du terroir, it is beyond “AI Art.”

In the same vein, this exhibition ties into my research in the DSI, where I am weaving together fields that are usually mutually exclusive: art, art history, AI, and ethics and philosophy. At the core of my research is the premise that the effect of machine learning on our engagement with images is fundamentally changing our society.
Emily Spratt, Thomas Fan, with Alain Passard
Gastronomic Algorithm, Plate 6p130fa2ffptl, New York, 2019

 


 



550 West 120th Street, Northwest Corner Building, Suite 1401, New York, N.Y. 10027    212.854.5660
©2019 Columbia University