When is an AI image actually art?

AI-produced images are now leaking in advertising, social media, entertainment, and more, thanks to models such as Midzorney and Dal-E. But visual art with AI has actually been for decades.

Christian Paul Digital Art Curates in the Whitney Museum of American Art in New York City. Last year, Paul cured Displayed On British artist Herold Cohen and his computer program Aaron, the first AI program for art manufacture. Unlike today’s statistical model, Aaron was built as one in the 1970s. expert systemTo follow the decision making of a human artist.

IEEE spectrum Talked with Paul about Cohen’s iconic AI program, digital art curse and the relationship between art and technology.

How do you cure digital art?

Christian Paul: Curateing digital art is not different from any other art. Whether painting or photography or print, we all look at the sophistication of a concept and how it is translated into a medium. Therefore my curatorial options are not powered by technology. If you are a curator of painting, a work selection for an exhibition will not be powered by a specific paint or technique for brush stroke.

In 2001, Herold Cohen produced Aaron Kate as part of its experiments in the construction of data with the AI ​​model. Cohen taught the model how to handle overlapping objects in a composition, which he did by filling the model in objects from the foreground to the background.Whitney Museum of American Art

It is being said, of course, showing about poinlism as a specific technique in painting. And, as an artistic medium, AI can be an exhibition focused on technologies. But the normal criteria will still be refined to concept and its implementation.

Do you collaborate with engineers as part of your work?

Paul: yes of course. Many artists also have a background in engineering, especially when it comes to the older generation of digital artists. When there were no digital art programs or school, digital artists often have A Engineering background Or programming. So you work with developers and software engineers, and many artists themselves are programmers or coders – I would say that most artists I am working. They sometimes have to be outsourced, simply due to the amount of work, but most of them are also very deep in mourning.

What are the challenges to collect and preserve digital art?

Paul: For art institutes or collectors, it is important to have standards and best practices to monitor and keep in mind technologies, as computers and systems change at such a rapid pace. In the 90s, people began to pay more attention to implementing conservation approaches, and there are many strategies. One of them is storage and hardware protection. It is used for pieces that depend on hardware ideologically. And then migration, simulation and re -construction.

There is no silver bullet. One has to see personal artwork which approach can be the best. For example, at the Herold Cohen Exhibition, we originally re -created one of the pieces before the scratch based on the notebook of Cohen and printed the code that we got, and his son actually resumed it in the python. We rebuilt the original basic, but then also rejuvenated in the Python.

What did the Cohen Exhibition inspire?

Paul: I knew Herold Cohen for a long time. We worked together on an exhibition in 2007, and Aaron is a prestigious work. Everyone who studies digital art knows it as one of the original pieces.

We brought some of his work in the collection of the Whitney Museum, so showing that there was a point. But I also thought that it would be especially interesting to see the first AI software to create art in the light of the current text-image model. Their processes are fundamentally different, and writers and cooperation play in a very different way.

Aaron learned how to generate images of plants, as seen in Aaron Gijon since 2007, using rules that Herold Cohen provided about its size and pattern, which controls the formation of their branches and leaf.Whitney Museum of American Art

Herold Cohen wrote to Aaron with scratches. He was completely in charge of the creation of the software he developed in five different languages ​​during his lifetime, so the creation of an image was completely under his control. He moved from developed forms, in a romantic phase, a plant-based phase, and then returned to abstraction. Later in life, he taught software color composition and also created drawing devices that would execute Aaron’s work. He actually considered Aaron as an ally, and Aaron explained Cohen’s sensitivity and aesthetics.

Today’s AI software is essentially based statistically, and the corporate black box has a lot of writer and agency. The artist has no control over him, even if the artist trains and makes his own model. Artists working with AI are greatly invested in manipulating software and working with it, but there is always an ingredient that is made by corporations that they do not have control.

Can AI-Janit images be art?

Paul: Text-image is not all visual art created by the model. It’s amazing that people can do Use AI to generate images And play with it, but I will not say that final result art.

AI Art uses artificial intelligence as a tool and medium in an ideological and practical way, seriously attached to those technologies and interrogates them, one is from one Moral Or beauty perspective. Most of today’s AI artists are getting entangled very deeply with these techniques. They are keeping their own training dataset together. They train the model. They question embedded prejudices in AI. So it is quite complex and involved process, and it is not just a text indicating an image.

This article appears as “5 questions for Christian Paul” in the May 2025 issue.