Photography is a medium that is constantly transforming with advances in technology. It is not surprising that more photographers are embracing artificial intelligence as an artmaking tool. The artists in Perspective Gallery’s exhibition “Magical Realism—AI Photography”employ artificial intelligence in a variety of ways to create photographic works. It was curated by William Harper who teaches Alternative Image Capture at the School of the Art Institute of Chicago. Although AI can generate an image to appear like any medium, all the works in the exhibition have a photographic element to them. The photographic style creates a sense of realism that can be warped to the whim of the artist.

        William Harper curates but also contributes to the exhibition. His series Dangerous Games uses his AI photography to depict children playing in dangerous scenarios. Barbed Wire Capture the Flag is full of feral-looking kids, some screaming and running, others bloody and entangled in barbed wire. With a massive budget, these pictures could have been created using child models on a set with makeup and special effects. By using artificial intelligence, Harper can explore lofty scenarios without needing thousands of dollars to create them. He considers AI to be a democratizing tool, however the low-cost aspect of AI artmaking is also part of what makes it controversial.

        Some of the artists in the exhibition approach the use of AI with cautious skepticism. In Doug Rosman’s series Leisure Time, AI is used to envision a dystopian future where human enjoyment and leisure are performed within large datacenters. Crowds of people mingle inside a large datacenter with billows of cables lining the ceiling like curtains in a theater. In Leisure Time 4, the people are sitting on lounge chairs as though sunbathing at a pool, but there is no pool. Rosman used text-to-image prompts in an open source program called ComfyUI to create this work. In the simplest terms, he types a description of the image he wants, and the AI creates it. For the Leisure Time series he prompted the AI to combine the setting of a datacenter with a leisure activity like sunbathing. He doesn’t make any edits or changes except for upscaling the resolution to be suitable for printing. For Rosman, asking the AI to create a datacenter is like asking for a self-portrait. These images represent the way human life may become shaped by AI rather than AI being shaped by humans.

        The results of a text prompt in an AI image generator are often unpredictable because it is the AI’s interpretation of what you have asked it to create. Navigating that unpredictability for some artists can be collaborative. The role of AI as a collaborator is examined by Alan Perry whose Dreaming Machine pictures are accompanied by the text prompt used to generate the image. For the prompts, Perry uses text from an ongoing writing project called Gross Gloss, a stream of consciousness style narrative that has a fantasy role playing game feel to it. Perry’s images produced by AI have a dark mystical tone to them, enhanced by the inclusion of old books and antique objects. They do not look photographic but instead feel like Renaissance paintings, a stylistic choice that was made by the AI software.

        On the opposite side of the spectrum, Michael Meyers uses AI as more of a tool than a collaborator. The work is created with no text prompts and instead he uploads two pieces of his own artwork such as paintings, drawings, family photos, etc., and combines them using Midjourney, an AI image generator. He then works back into them “like a painter” using Photoshop. What Fit (also called Babel)is comprised of several smaller images and drawings organized in a grid. To Meyers, the individual pictures were like people and their relatives. By organizing them in Photoshop, the grid becomes a structure that he likens to the steerage on a ship. The collage style of Meyers’ work makes it feel less AI generated, as it is not attempting to be realistic. However, it does have text that, upon closer inspection, is nonsense. In human made art, text often has more purpose than just to function as a visual element, so recognizable words are used that add to the meaning of the work.

        The artists in Magical Realism approach artificial intelligence from different angles. It is interesting to see AI generated images in a gallery rather than on a screen, although it would have benefitted the exhibition if more care was put into the presentation of the physical work. The track system used to hang the work detracts from it because many of the works don’t hang well on them. For some of the artists, the final presentation seemed to be an afterthought with mismatched framing styles or missing glass. Because it is more uncommon to see AI art in a gallery, the decision of how the work is shown should be deliberate.

        The conceptual framework for artmaking still relies on a person to build it. There are also still many limitations to AI image generation, such as its difficulty in producing coherent text or realistic hands and limbs. These “tells” can’t be unseen and serve to undermine the work. Artificial intelligence has a long way to go before it can truly take the place of an artist. However, just like photography, the technology will advance and eventually become engrained into our everyday lives.

Rebecca Memoli is a Chicago-based photographer and curator. She received her BFA from Pratt Institute and her MFA in Photography from Columbia College. Her work has been featured in several national and international group shows.