Rethinking LLMs: Artistic precedents
Duchamp, Cage and Bourriaud on reimagining creativity, uncertainty and remix
Amongst the dichotomous discourse of the AI in education debate, I've been thinking a good bit about artistic precedent. Before I was a teacher, I did some vagabonding in artistic communities, spending a lot of time studying art history, especially the 20th century from modernism. Yeah— I was that weird dude who read The Story of Art for fun (it has amazing visuals, fold outs, etc.).
I spent a lot of time on UbuWeb studying various forms of conceptual art. Founded by Kenneth Goldsmith in 1996, the site offers free access to a wide range of avant-garde works, from Jean-Michel Basquiat's hip-hop to Ulrike Meinhof's radio plays. UbuWeb's approach focuses on lesser-known artists and works, challenging traditional artistic hierarchies. This perspective changed my understanding of avant-garde art and made me reconsider conventional definitions of creativity.
This post will walk through a few fascinating artistic precedents from the 20th century that can complicate and illuminate our discourse on AI and LLMs in education. I'll walk you through authorship and provenance with Duchamp and the readymade, consider uncertainty with John Cage, and discuss post-production with Bourriaud.
Duchamp and the readymade
Marcel Duchamp was a pioneering French artist who revolutionized 20th century art with his invention of the "readymade" in 1915. Readymades were ordinary, prefabricated objects that Duchamp selected and presented as works of art, often with slight modifications. One of his most famous readymades was Fountain, which was simply a urinal that Duchamp signed and submitted to an art exhibition.
Duchamp believed the act of the artist choosing and presenting an object was enough to make it art, regardless of the object's aesthetic qualities or the artist's technical skill. This challenged traditional notions of what constitutes art. Duchamp's readymades and "assisted readymades" (where he combined found objects) influenced later artists like Robert Rauschenberg, Damien Hirst, and Jeff Koons, who incorporated found and mass-produced objects into their conceptual artworks.
In the AI/LLM context, we might think of prompts and outputs as a kind of digital readymade. The act of selecting, refining, and presenting AI-generated content could be seen as a form of curation akin to Duchamp's artistic process. This raises questions about authorship, originality, and the value we place on human intervention in AI-assisted work.
John Cage and uncertainty
John Cage, an influential American composer and music theorist born in 1912, revolutionized 20th-century avant-garde music with his unconventional approach. Central to Cage's artistic philosophy was the embrace of uncertainty and chance. Inspired by Zen Buddhism, he sought to create music that welcomed unpredictability and indeterminacy.
Cage's famous work 4'33" exemplifies this approach, featuring a performer sitting in silence while ambient sounds become the "music." He often used the I Ching, an ancient Chinese text, to make compositional decisions, resulting in pieces that unfolded uniquely with each performance. By relinquishing control and inviting chance into his creative process, Cage challenged traditional notions of musicality and composition. His approach aimed to free art from the constraints of the ego, encouraging audiences to experience the world as it unfolds moment by moment, without preconceptions.
When we think about AI and LLMs in education, Cage's embrace of uncertainty offers an interesting parallel. The unpredictable nature of AI outputs could be seen as a form of digital chance operation. This uncertainty challenges us to approach AI as a collaborative tool rather than a deterministic oracle, potentially opening up new avenues for creativity and exploration in learning.
Postproduction and remix
Nicolas Bourriaud's concept of "postproduction" in art offers a compelling lens through which to view contemporary artistic practices in the digital age. By focusing on how artists reinterpret, remix, and repurpose existing cultural materials, Bourriaud highlights the shift from creation ex nihilo to a more curatorial approach.
This perspective not only acknowledges the overwhelming abundance of information and cultural products available today but also celebrates the artist's role in navigating and reconfiguring this landscape. The notion of postproduction thus blurs the lines between creator and consumer, original and copy, suggesting that creativity in the 21st century often lies in the ability to meaningfully recontextualize and transform the vast array of pre-existing cultural elements at our disposal.
In the context of AI and LLMs, we might view AI-generated content as a new form of raw material for intellectual "remixing." The ability to effectively navigate, select, and meaningfully combine diverse sources of information – both human and AI-generated – could become as crucial as the ability to generate original content from scratch.
So what?
These artistic precedents challenge us to rethink our approach to AI in education. Instead of viewing AI as a threat or a panacea, we might consider how it can be integrated as a tool for creative exploration and critical thinking. How can we update our understanding of authorship, originality, and academic integrity in light of these new technologies? How might we design learning experiences that leverage the unpredictable nature of AI outputs? And how can we cultivate the skills of digital curation and synthesis that will be crucial in an AI-augmented world? By engaging with these questions, we can move beyond the binary debate of human vs. machine and towards a more nuanced understanding of AI's role in education.