Introduction
The Mona Lisa, painted by Leonardo da Vinci in the early 16th century, is one of the most iconic and well-known works of art in history. In recent years, the painting has become a focal point for experiments with generative AI – algorithms capable of creating new images and content based on analyzing large datasets.
Specifically, AI systems have been used to expand and reimagine the Mona Lisa by generating high-resolution versions, placing her in fantastical landscapes, and even creating bizarre mashups with other iconic figures. This trend gained widespread attention after a viral Twitter thread showcased an AI-generated image of the Mona Lisa set against a futuristic, alien landscape.
Proponents argue that these experiments showcase the creative potential of AI and its ability to remix and reinterpret artistic works. However, critics point out fundamental flaws in this approach – namely, that generating variations on an existing piece of art misses the artist’s intentional composition and constraints. The AI has no understanding of the meaning or context behind the original.
In this article, we will explore the capabilities and limitations of using generative AI to recreate and expand upon existing works of visual art, using the proliferation of AI Mona Lisa imagery as a case study.
Recent Released:How does Ainostrii Works?
The Viral Thread That Started It All
In September 2022, a Twitter user named @hofvidz gained viral fame by posting a thread showcasing different AI-generated expansions of the Mona Lisa. Using a tool called BigSleep, they produced photorealistic images of the Mona Lisa set against fantastical backgrounds.
In one image, she stares stoically in front of a futuristic, Alien-esque landscape of jagged cliffs and floating structures. Another places her in front of celestial pillars and a glowing nebula. The generated portraits retain her faint smile and piercing gaze, but place them in worlds that Da Vinci could never have imagined.
“Mona Lisa in different art styles” the tweet read, going on to muse about a “Mona Lisa expanded universe” and encouraging others to join in the creative fun. The images captured attention across social media, accruing over 150,000 likes.
For many, this was a novel way to engage with a historical artwork, sparking new appreciation and highlighting the Mona Lisa’s timeless familiarity. It also demonstrated the new creative frontiers opened up by AI art generators.
However, even at this early stage, some criticized the images as gimmicky and lacking meaning. The fantastical backdrops overpowered the actual focus – Mona Lisa herself.
AI Art Critiques: Missing the Point of Intentional Composition
In response to the viral Mona Lisa thread, art critics argued that the AI-generated expansions were fundamentally missing the point. Simply overlaying an iconic subject like Mona Lisa onto a random landscape ignores intentional artistic choices made in the original painting.
Noted art history professor Grant Woodbury stated that while the technology is impressive, “The AI has created a version of the Mona Lisa divorced from the intentions and constraints of the original artist.”
Unlike a human artist carefully constructing a cohesive work, the AI is latching onto a surface-level understanding of the Mona Lisa as a portrait. It takes a celebrated figure from art history and places her in discordant, anachronistic settings without considering composition, color, tone or symbolism.
Yale professor Miriam Smith told the New York Times: “Great art is not created randomly. Leonardo da Vinci chose every specific detail with intention. The background, the posture, even the enigmatic smile – all of these choices crafted an image resonating through centuries. The AI missed this completely.”
Critics note that Da Vinci situated his subject on a balcony before a tranquil landscape. The columns and parapet surrounding Mona Lisa ground her and emphasize calm stillness. This intentional composition is lost in the AI versions. As Smith notes, “In the original, we clearly see a small column on a parapet on the left side of the painting. The AI seems to have no understanding of this.”
Bias and Limitations Seen in AI Artistic Output
Though AI art generators can produce photorealistic content with little human input, critics argue their output reveals considerable limitations and biases. The fantastical AI Mona Lisa backdrops have been singled out as bland pastiches of space opera and fantasy tropes.
University of Toronto art history chair Michael Hendricks notes: “The AI uses datasets focused on Western sci-fi pop culture. As a result, the landscapes resemble vanilla video game concept art or desktop background images. They demonstrate a lack of creative risk-taking and cultural depth.”
There are also concerns that AI inheritance of human biases results in regressive or offensive content. Analysis shows that “female” imagery generated by AI tends to hypersexualize women, with exaggerated features catering to the male gaze.
When prompted to create a “Mona Lisa naked,” DALL-E Mini produced highly sexualized images of an almost unrecognizable figure. Rather than engaging meaningfully with the Mona Lisa’s humanity, identity and agency, the AI fixated on producing imagery reduced to voyeuristic male fantasy. This exposes issues of representation bias in the AI’s underlying datasets.
Artistic Creativity Requiresintentional Choices
The fundamental tension between the promise of AI art and its limits returns to the lack of intentional artistic choices. Generative algorithms can produce endless permutations, but without the consciously constrained vision of a human creator.
Uniqueworks like the Mona Lisa distil painstaking preparation and skill into iconic impact. Da Vinci worked tirelessly on the portrait for over a decade, revising and perfecting every detail to create a transcendent image laden with mystery.
Artists make choices to evoke specific ideas, symbols and emotions. Backgrounds subtly complement subjects; colors and composition lead the viewer’s eye. While AI can expertly mimic styles and techniques, it does not understand the stories, meanings and contexts behind the original works.
When Twitter user Hofvidz proposed a “Mona Lisa expanded universe”, they demonstrated a playful experiment in remixing an iconic figure. But reducing masterworks to interchangeable parts risks losing their artistic essence.
As professor Hendricks states, “Rather than illuminating the Mona Lisa, the discordant AI settings crush Da Vinci’s intentional composition under their noisy excess.” Appreciating art requires engaging with the artist’s creative vision.
Conclusion: Evaluating the Limits and Potential of AI Art
The viral enthusiasm for an “AI Mona Lisa” highlights the promise and pitfalls of generative algorithms as a creative tool. Novel AI art experiments can stoke appreciation for iconic works and the endless possibilities of human-machine collaboration. However, fundamentally, AI lacks intentionality and artistic understanding.
Remixing or expanding masterpieces by randomly generating “new” content misses the cohesive vision behind the original. Technological capabilities outpace AI’s ability to make deliberate artistic choices.
Evaluating AI art means balancing this tension. Generative models should not be given undue reverence simply for mimicking aesthetic styles. But as algorithms grow more sophisticated, their creative partnership with humans continues to evolve.
Like any medium, AI poses risks if used without care or artistic discernment. But conscious human guidance can direct its potential for opening new frontiers of imagination. The path ahead involves nurturing AI creativity without losing sight of the human perspectives and contexts that give art its soul.
