Barely Detected: Reflecting on the ‘Aura’ of Generative AI Art

Nettrice Gaskins
5 min readOct 5, 2024

--

SightEngine’s GenAI image detector

In January 2023 I wrote an article titled “The Aura of AI-Generated Art” that received a lot of attention. I referenced Walter Benjamin via Hans Abbing who defined “aura” as the quality integral to an artwork that cannot be communicated through mechanical reproduction techniques. According to Benjamin, the mechanical reproduction of a work of art voids its cult value, because removal from a fixed, private space (ex. a temple) and placement in a mobile, public space (ex. a museum) allows exhibiting the work of art to many spectators (exhibition value). In my opinion, this qualification is elitist and exclusionary. The social value of a work of art changes as societies change their value systems.

We are in the age of “intelligence.” Courtesy of ICT Works.

Computing and the World Wide Web disrupted earlier notions of “aura” in contemporary art. Intelligence, especially artificial intelligence or AI, is challenging what we label as art and is creating a new space for artistic (aesthetic) practice and discourse. Since the emergence of generative AI or GenAI there has been a rise of deepfakes, fraud, spam, memes, and image manipulations, as well as the creation of tools that detect AI-generated images. Anyone with access to GenAI tools can generate images but it takes effort (practice, skill) to fool the detectors.

Out of curiosity I recently tested one of the GenAI detectors using a recent image I created with Midjourney (see above). I was surprised when the result was “Not likely to be AI-generated or Deepfake.” GenAI was at 5% and Midjourney was at 3%. I thought this first attempt was lucky but then it happened again:

Fooling the SightEngine detector… again!

The second image I uploaded was based on an early 1990s oil painting I made when I was an undergraduate art student. In Midjourney you can use images as part of a prompt to influence a job’s composition, style, and colors. Images prompts can be used alone or with text prompts and you can experiment with combining images with different styles. When I uploaded the second image into the SightEngine detector I got the same result: “Not likely to be AI-generated or Deepfake.” I fooled it again! Next as a ‘control’ I uploaded a photo of my original artwork (painting):

Using original artwork (oil painting)

I was surprised that the result wasn’t 100% ‘not likely to be AI-generated’ and I tried other Midjourney images, some that were detected by the SightEngine tool and some that were not. One Midjourney image was not detected and the tool didn’t even include Midjourney in the result (below).

It finally got it right!
This one IS from Midjourney.

With the exception of the first image I tested, the other AI-generated images that were not detected by the SightEngine tool all had one thing in common. All of them included uploads of my artwork (as image prompts). They all had sufficient elements of the artist’s hand to fool the detector. According to Walter Benjamin, mechanical reproduction devalues the aura (uniqueness) of a work of art. I disagree with this assertion. Generative AI is producing unique imagery that does not diminish the aesthetic value of original works of art. Rather, GenAI expands and enhances the aesthetic, cultural, and political authority of some original works of art. According to French philosopher Paul Valéry, the understanding and treatment of art and of artistic technique must progressively develop in order to understand a work of art in the context of the modern time. Valéry writes,

In all the arts there is a physical component which can no longer be considered or treated as it used to be, which cannot remain unaffected by our modern knowledge and power. For the last twenty years neither matter nor space nor time has been what it was from time immemorial. We must expect great innovations to transform the entire technique of the arts, thereby affecting artistic invention itself and perhaps even bringing about an amazing change in our very notion of art. — Paul Valéry, La Conquête de l’ubiquité (1928)

Nettrice Gaskins. “John Amos from the ‘Gilded Series’,” 2024. Created using Deep Dream Generator.

An AI-generated portrait of recently deceased actor John Amos has joined a series of works that I titled ‘Gilded’ to reference the way I apply specular reflection or “shine” through neural style transfer as well as the way this AI-powered process pays homage to the subjects. In traditional art, gold or gold leaf was often used to honor important figures and show devotion. I shared the image via social media and Amos’ daughter responded:

The use of generative AI to capture the essence of a person is something that speaks to the cult value of ‘intelligence’; the cultural and social significance of GenAI artwork, based on its time and place of creation, and the people involved. Instead of ‘gilding’ and locking the artwork away, social media allows the immediate dissemination and viewing of the work.

My “Afro-Surrealist Carnival” series is on view at the Ulrich Museum of Art in Wichita, KS.

When AI-generated images are printed and put on display in museums or galleries or on walls this speaks to the exhibition value of the work. Exhibition value refers to value of an artwork when it’s displayed to the public, and the functions it can serve in that context. Exhibition value can be political, a display of beauty, or a way to convey the artist’s intent. Based on the examples explored here, generative AI (in the right hands) does not diminish the uniqueness of an artwork and its cult/exhibition value.

And sometimes this artwork can fool GenAI detectors (*wink).

--

--

Nettrice Gaskins

Nettrice is a digital artist, academic, cultural critic and advocate of STEAM education.