Is AI Replicating Human Art?

In the concluding days of my trip to Barcelona, some leisure hours remained prior to my journey back home. I chose to spend this time at Casa Batlló, a grand property on Passeig de Gràcia. The house was given a makeover in the early 20th century to suit the luxurious lifestyle of the affluent Batlló family, and is recognised as a masterpiece by the renowned Catalan architect, Antoni Gaudí. Presently a museum and a UNESCO world heritage site, it ranks highly among the favourite tourist spots in a city known for its complex rapport with its visitors. (If anyone from Barcelona is reading this, rest assured, I was visiting for a literary festival and was just momentarily a tourist, so there are no hard feelings, I am sure?)
Touring Casa Batlló provides a unique experience as it’s indeed a two-sided coin. The intriguingly built house proves to be a masterstroke of an artistic innovator who’s been given unrestrained liberties to deliver his supreme creation. However, it’s brimming with tourists busy with their selfie sessions on tight staircases and there’s a souvenir store inconveniently located at the core of it all. It felt analogous to stamping a QR code directly on the private parts of Michelangelo’s David or a maestro pausing a concert of Stravinsky’s Rites of Spring for an advertisement from the orchestra’s sponsors.

A surreal anticlimax marked the finale of our tour. Roughly sixty minutes were spent meandering through the peculiar and magnificent structure, where each nook and finer point hinted at a bizarre sense of existing within a conscious, respiring, foreign life form. We were then led down an extended aft staircase (draped in black commercial carpet; fitted with frosty glass handrails, conspicuously designed by an architect quite unlike Gaudí), heading into a designated audiovisual zone known as “The Cube”, adorned entirely with LED displays. Our senses were bombarded with a three-minute sensory immersion video named “Gaudí Dreams”. Turkish multimedia artist, Refik Anadol, crafted the piece from an alleged billion images of Gaudí’s work, refined by several AI algorithms, thus intending to simulate the feeling of dwelling inside the renowned architect’s dreaming brain. It’s mechanically impressive; the viewer is projected through sequentially morphing geometric spaces and twisting tunnels for three minutes. It delivers an intense and unequivocally psychedelic experience, though it echoes more of a sensation of being immersed in my MacBook’s screensaver than within the thoughts of one of the previous century’s leading creative thinkers.

Gaudí Dreams might have been intriguing and pleasing to the eye had I viewed it in a gallery or another setting, however, located in the underbelly of Casa Batlló, it had a similar impact to watching a stream of viral TikTok clips straight after engaging with a Tarkovsky film. It prompted thoughts on the current dispute concerning AI’s ability to generate art – a contention I habitually bypass on account that it requires an assumption that we hold an understanding of what constitutes “artificial intelligence” and “art”, when essentially, we do not.

The debate surrounding the artistic capability of artificial intelligence (AI) can be deeply encapsulated by referring to the comparison between the Casa Batlló and an AI-generated rendition. To argue that AI falls short artistically, one could contrast the profound representation of Gaudí’s imaginative mind in the Casa Batlló versus an AI-generated spectacle of light and sound. Indeed, one might be inclined to suggest AI’s value lies largely in the automated creation of trite, retro reproductions.

However, it cannot be confirmed with absolute certainty that this is a wholly bitruistic view. Dancing on the edge of this contested topic, doubts remain as to whether AI is devoid of potential to cultivate noteworthy art. Perhaps it is the overly polished, faultless nature of creations like Gaudí Dreams that renders them less than captivating. In an essay penned earlier this year titled 13 Ways of Looking at AI, Art & Music, AI-employing composer Jennifer Walshe defines AI-generated art as “gunk”: a categorisation that reflects the challenging-to-quantify yet unequivocally distinct aesthetic of such creations. The peculiarities and apparent flaws found in certain forms, such as beer cans shifting into faces or celestial violin melodies melting and being consumed by a hum of static, create a sense of intrigue. This uncanny aesthetic perplexity is of interest to artists like Walshe and can elevate generative AI as a potent tool when handled judiciously.

However, the notion that AI is devoid of artistic capacity was strongly emphasised by science fiction author Ted Chiang in a New Yorker’s piece. Chiang contends that artistry demands an array of decisions made at various plans, indicating that inspiration, the initial and significant decision involved in the birth of an artwork, cannot be detached from the myriad minor decisions required during its manifestation. Consequently, software, bereft of decision-making aptitude, is seemingly unable to create art.

The argument that Generative AI is a seductively powerful force tends to overlook how it currently serves as an instrument for making choices. The electronic musician, Holly Herndon, effectively utilises machine learning in her music in unique and emotionally poignant ways. For example, her 2019 album, Proto, was produced in “partnership” with an artificial intelligence network trained to respond to, and creatively adapt, her own voice and compositions. She suggested that an oversight in Chiang’s essay was the misconception that AI exists in a vacuum. The development of Proto entailed 18 months of prompt engineering, data collection, and model training – undeniably a remarkable artistic achievement.

A frequently overlooked nuance in this conversation is the difference between art (or “art”) crafted by AI, versus art crafted utilising AI. It’s challenging to envision an AI independently producing something as riveting and uniquely human as Gaudí’s Casa Batlló. However, it is not difficult to foresee an artist utilising AI to craft a work of such magnitude and significance. Things are clear: AI is here to stay. However, so are artists. Despite an unremarkable AI “experience” at Casa Batlló, one can easily envisage Gaudí himself finding potential in AI technology.

Condividi