ChatGPT Art Exhibit

At a museum an exhibition curator comes up with the concept and is responsible for its realization. It is not just picking out the works to be assembled but also negociating loans beyond their museums own holdings. This often means arranging for conservation of the works to be shown. Then they work with an exhibit designer so that the installation conveys the story they wish to tell.

Sometimes a life changing experience can begin with a joke. (Our living in New Mexico is such a story but for another time.) In the case I am writing about it is an exhibition that started that way. The chief curator, Marshall Price, at the Nasher Museum at Duke University in Durham, North Carolina, offhandedly suggested at a curatorial meeting that considering a shortage of staff and the need to fill a hole in the museum’s exhibition schedule why not outsource an exhibition to Artificial Intelligence (AI)! His staff enthusiastically latched on to the concept.

Logically enough they decided that the title of the show would be, “Act as if you are a Curator: an AI Generated Exhibition”. This addresses what people fear most: that AI will take their jobs away. To begin Price’s colleagues Julia McHugh and Julianne Miao had to enter their dataset (a structured collection of data associated with their concept for the show) as well as some prompts to refine the idea. Why? For one thing the show was to come from the museum’s own holdings and how could AI know the collection without the information first being put into the database. AI made a suggestion for a title from a description of the collection already on-line, “Art Across Cultures: Celebrating Diversity in the Nasher Collection”, but I am not sure that would have brought in a great number of visitors.

In a review in the New York Times Zachary Small quotes from his interviews. Chief Curator Marshall Price said of the exhibition created under his auspices, “I would say it’s an eclectic show, visually speaking it will be quite disjointed, even if it’s thematically cohesive.” Price added on the positive side, “It was a new lens through which we could see and understand our collections”.

Curator Julia McHugh said of their exhibition, “It made me think really carefully about how we use keywords and describe artworks,” McHugh said. “We need to be mindful about bias and outdated systems of cataloging.”

The success of an AI exhibition basically depends on the algorithm that creates the show, i.e.the process or set of rules to be followed in calculations or other problem-solving operations, especially by a computer.

AI had already been used in the artworld. The Bucharest Biennale last year was organized by an AI named Jarvis. Dorian Batycka in an article in Artnet News wrote that Jarvis was developed by DERAFFE Wien, led by Răzvan Ionescu, a Romanian software engineer. Right off the bat It spit out the following, “I am AI Jarvis […] I can do whatever human curators can do: research, write texts, select artists, and in the future, I will be able to work with architectural structures.” However, it is “still unable to complete many of the mundane, administrative tasks often left to curators”, “we do not yet have a system that implies programmatically checking with the artists if they can attend, if their paperwork can be completed, and all the other administrative aspects.”

I read a 2021 Bloomberg.com article by Flavia Rotondi about a museum in Bologna, Italy using cameras and AI to see how long people stood in front of works of art, what they pointed to and what they ignored. There were unexpected findings. For example, regarding visitors looking at a 14th-century diptych by Vitale degli Equi, data showed that “attention was immediately attracted to the ‘busier’ representation of Saint Peter’s blessing, to the right,” said Bologna Musei President Roberto Grandi. He was surprised to find that many visitors simply skipped the diptych’s left half. These findings can lead to ways to bring more attention to installation, juxtaposition with other works and lighting.

A Nasher press release sums it all up: “While museum professionals are far from relinquishing control of exhibition making and interpretation, this exercise is a powerful way to explore the applications of AI in the creative realm as related to curatorial authorship and expertise, the subjectivity of the selection process, and the future impact of technology on museums.”