Art in the Age of AI
Essay trailer created by Robert Lester with Adobe tools, LumaAI, Midjourney, and a lot of good old-fashioned elbow grease.
What happens when machines do more than reproduce an existing artwork? When a model can generate a new image in the style of a familiar painter, the questions extend beyond how quickly it was made. What makes it original? Where does the artist's contribution begin? Why should anyone care about the result?
AI-assisted portrait of Walter Benjamin.
I remember the shock of first reading Walter Benjamin's essay “The Work of Art in the Age of Mechanical Reproduction” in a college film class. Twenty years later, watching AI generate images with unnerving efficiency, I keep returning to it.
Benjamin began writing the essay in 1935, as Europe approached catastrophe. His account of photography and film examined how new technologies change our relationship to art, its authority, and its political uses. Generative AI presents a different set of technical questions, but his framework gives us a useful place to start.¹
I'm interested in what happens to authenticity, creative labor, and artistic value when generating an image becomes easier than deciding what that image should mean. Contemporary work such as Ignas Kalpokas's “Work of art in the age of its AI reproduction” extends this conversation into the conditions of AI production.²
When Algorithms Create: Aura and Authenticity
Benjamin's concept of “aura” concerns an artwork's unique presence and history: its existence in a particular place and time. A reproduction can carry an image to a new audience without carrying the original object's history with it.¹
Generative AI complicates that relationship. An image produced in Van Gogh's style is not a reproduction of The Starry Night, nor does it acquire that painting's history. But it can offer an immediate encounter with a recognizable visual language. What do we value in that encounter, and how does its abundance change our expectations of art?
Jason Allen's Théâtre D'opéra Spatial made that question especially visible. In 2022, the AI-assisted work won first place in the Colorado State Fair's Emerging Artist division, in the Digital Arts / Digitally-Manipulated Photography category.³ The category matters: this was not a contest restricted to painting made entirely by hand. Even within a digital-art context, the result prompted questions about the contribution being recognized when an artist works with a generative system.
Benjamin described a desire to bring things closer through reproduction. AI offers another form of that proximity: images generated around an individual's instructions rather than selected from an existing collection. A person without years of drawing practice can explore images that would otherwise be difficult for them to produce. That changes access, even though generating a convincing image does not confer every skill involved in being an artist.
My sense is that authenticity will become more closely tied to the choices surrounding the work: who made it, what they intended, how they used their tools, and what the result means to an audience. The physical original still matters. But for work born through these systems, the story of its making may carry more of the value we once expected to find in a unique object.
Democratization and Devaluation
Benjamin reminds us that imitation and reproduction have always been part of art. Students copied to learn, masters reproduced work to circulate it, and others made copies for profit. The technologies change the scale and conditions of those activities.¹
AI also raises a question about my own role. When I type instructions into a model, select an output, and refine it, where does the creative act sit? How much is in the initial idea, how much in the model, and how much in my judgment of what to keep?
Artists working with these systems make that division less tidy than it first appears. Refik Anadol's Unsupervised uses custom software, generative AI, and real-time animation to create a changing visual experience.⁴ Calling this simply an image generated by an algorithm overlooks the artistic decisions involved in designing the system and its presentation.
Holly Herndon's Holly+ offers a different approach. Developed with collaborators, it allows people to transform audio through a model of her singing voice, turning an aspect of her artistic identity into an instrument others can use.⁵ In both examples, the artist shapes the conditions for creation as well as the resulting work.
That expanded role interests me. Direction, selection, technical experimentation, and collaboration become visible parts of the practice. Existing skills do not disappear; they help an artist recognize and develop something worth keeping.
The economic implications are harder. Easier access to image-making can open creative expression to people who lacked the training or resources to participate. It can also put pressure on work whose value has depended partly on the difficulty and cost of producing it.
I expect intention, context, and a distinctive point of view to matter more as polished images become easier to generate. Reach and narrative can contribute to value, but abundance does not make every generated image valuable, and access to a tool does not guarantee a viable living. Questions about credit, ownership, and compensation remain part of the same conversation.
Economic and Cultural Implications
Benjamin's argument was political as well as aesthetic. Reproduction could move art beyond its traditional settings and make it available to more people. It could also serve propaganda.¹
Openly available AI models extend that tension. They can give artists and smaller teams more room to experiment without relying entirely on a single provider. But openness alone does not prevent concentrations of power. Access to computing resources, training data, and distribution still affects who can build these systems and whose work gets seen.
The question is therefore broader than who can generate an image. Who sets the model's limits? Who decides what can be distributed? Which forms of expression are encouraged, restricted, or made difficult by the tools themselves?
Governments and technology companies approach these questions differently. A simple contrast between “Western” and “Chinese” AI obscures the separate decisions about model access, acceptable outputs, and distribution. The useful comparison is between those decisions and the power they give to institutions and individuals.
The same tools that help artists communicate can make misleading political imagery easier to produce. They can also give activists and artists ways to respond quickly, experiment with a message, and reach an audience. Neither outcome is built into the technology as a guarantee.
This is where Benjamin remains especially useful to me. Making art more accessible does not settle how it will be used. Creative freedom and political control can develop around the same tools, and the choices embedded in code, data, and platforms deserve scrutiny alongside the images.
Human Creativity in an Algorithmic Age
Twenty years after that first encounter with Benjamin, I'm both unsettled and exhilarated. The speed and scale of image production have changed, but the questions that make me care about art remain: What is someone trying to express? What experience are they drawing on? What happens between the work and the person encountering it?
Photography changed painting without making painting irrelevant. I see AI as another reason to reconsider the relationship between a tool, a skill, and an artistic purpose. I don't believe it makes human creativity unnecessary. I do think it will change which parts of the work are easy to produce, which skills are rewarded, and where artists have to make their contribution clear.
That requires learning. Artists and creative teams need to understand the tools well enough to make deliberate choices about when and how to use them. We also need to keep asking who benefits from those choices and what they mean for the people doing the work.
The future of art will depend on how we engage with these technologies, the frameworks we build around them, and the value we place on human expression. Benjamin gives us a way to examine technological change without reducing it to either celebration or fear. I want to bring that same attention to the work we make now.
Notes and Further Reading
Walter Benjamin, “The Work of Art in the Age of Mechanical Reproduction,” in Illuminations, edited by Hannah Arendt, translated by Harry Zohn (Schocken Books, 1969). Essay first drafted in 1935. Read the essay.
Ignas Kalpokas, “Work of art in the age of its AI reproduction,” Philosophy & Social Criticism, advance online publication, 2023. Read the article.
Colorado State Fair, 2022 Fine Arts results, Emerging Artist division, class 0005.
The Museum of Modern Art, Refik Anadol: Unsupervised exhibition checklist, 2022.
Herndon Dryhurst Studio, Holly+, 2021.
All images created by Robert Lester with digital and AI art tools.