Does AI Film Need an Avant-Garde?
A New Look at "The New Sensibility"
Literary men, feeling that the status of humanity itself was being challenged by the new science and the new technology...are inevitably on the defensive. They know that the scientific culture, the coming of the machine, cannot be stopped.
At a glance, this quote would not seem out of place in a contemporary article about generative AI and the arts. Yet, these words were penned nearly 60 years ago by Susan Sontag in her classic essay One Culture and the New Sensibility (1965).
In that essay, Sontag rejected the pessimism of these "literary men," advocating instead for a new sensibility merging science and art, a sensibility she saw as capturing modernity’s essence. Her ideas, transgressive at the time, celebrated New York's avant-garde scene, which blurred boundaries between "form and content, the frivolous and the serious...and 'high' and 'low' culture" (Sontag, 1965).
I first encountered this essay in 2022, shortly after DALL·E 2 unleashed a flood of uncannily realistic AI-generated images. Like Sontag’s literary men, I felt a profound sense that "the status of humanity itself was being challenged." At the same time, there was something infectious and unexpected about her enthusiasm for the aesthetic promise of the scientific culture. Her call for a new sensibility felt particularly resonant in our technological moment, leaving me to wonder: Does AI film need an avant-garde?
In 1967, Robert Rauschenberg launched Experiments in Art & Technology with avant-garde artists like Andy Warhol alongside scientists from IBM and MIT. These partnerships were representative of Sontag's "new sensibility".
Around the same time, I attended a talk by a respected media theorist who argued that we should leave the historical notion of the avant-garde behind. They pointed out that a concept so deeply embedded in the culture of the past might no longer serve us well in the future. They are, in many ways, correct: the avant-gardism of the 20th century, championed by Sontag, carries with it the political, cultural, and social baggage of its era. Today, the term “avant-garde” ironically signifies institutional respectability rather than radical innovation, revealing itself as a backward-facing concept largely depleted of its once-revolutionary force.
And yet, I can’t shake the sense that the avant-garde tradition remains relevant, especially in the context of contemporary filmmaking and generative AI.
Illusions of Realism: An Old Problem Returns
To explore this further, we can consult A.L. Rees’s A History of Experimental Film and Video. Rees traces avant-garde filmmaking from Georges Méliès’s early spectacles at the turn of the century to digital video experiments of the 1990s. He argues that these filmmakers not only participated in broader avant-garde movements like surrealism and futurism but critically engaged with the rapid commercialization of cinema and its spectacular “illusions of realism" that dominated the 20th century. This friction between mainstream narrative realism and experimental film practice, Rees suggests, defined the avant-garde as much as its relationship to broader art trends.
A still from Sora AI’s promotional film, highlighting the realism enabled by generative technology.
The persistent question of realism and representation returns to us in a new form with generative media. Today, AI video technologies like Google’s Veo3 and OpenAI’s Sora promise to bring forth a new kind of hyper-realism—one so advanced it is marketed as "a world simulator” (OpenAI, 2024). However, such realism raises profound ethical and political concerns, from inherent biases to potential misuse. The U.S. government even recognizes the possible weaponization of deepfakes as a national security threat (House Committee on Oversight and Accountability, 2024).
Ironically, these anxieties imply that the boundary between reality and illusion has, until recently, remained relatively clear. However, as Rees and many avant-garde filmmakers have sought to reveal, the artificial representations of cinema have long blurred this boundary, bearing real societal consequences.
In light of these generative pipelines that promise to consume our attention, the call for an experimental film movement feels as necessary as ever. Such a movement, rooted in the avant-garde tradition, could expose and subvert the mechanics behind these AI-generated illusions, acting as a necessary counterbalance to AI video’s inevitable proliferation. This aligns with Rees’s definition of experimental film as a challenge to mainstream cinema through critiques of “drama, visuality, identification, and non-linear thought.” Rees extrapolates this stance into the 21st century, anticipating the transformative impact of digital innovation:
It might even be said that the role of media artists in this environment is, at last, to be an avant-garde.
— A. L. Rees (1999)
Film still from Ballet Mécanique (1924) an experimental Dadaist film by Fernand Léger and Dudley Murphy. In contrast to narrative studio films at the time, the work presents a radically different expression of being human in an industrial world.
A Newer Sensibility Beyond the Avant-Garde
This brings us back to our central question: does AI film need an avant-garde? While the revolutionary call for an avant-garde can be intoxicating, the reality is more nuanced. The avant-garde once implied a radical rupture, a rebellion against ossified traditions of comparatively centralized institutions. But in today’s sprawling media ecosystem, where Hollywood's formulaic genre films coexist with TikTok’s algorithmic micro-trends and YouTube's influencer vlogs, what "guard" is there to advance beyond anyway? The very idea of an avant-garde feels increasingly antiquated in a world where cultural production is fragmented, fluid, and shaped by the pervasive logic of data-driven attention platforms.
Perhaps these are some of the very pitfalls hinted at when we are urged to move beyond the heroic myths of the avant-garde. Indeed, Sontag, writing three decades after her original essay, reflected critically on her earlier optimism, acknowledging that many of the utopian ideals she once embraced had ultimately served to reinforce "frivolous, merely consumerist transgressions." Maybe our guiding question about AI and the avant-garde is misdirected, giving too much weight to a term burdened by history and nostalgia.
But this is not to say the past has nothing to teach us. Rather, what we might gain from thinkers like Sontag and Rees is not reverence for the avant-garde as a historical artifact, but a willingness to imagine new sensibilities. In contrast to the traditional avant-garde's emphasis on rupture, a newer sensibility favors adaptation over opposition, shedding outdated modes in favor of those that reflect evolving ideas of what it means to be human. The search for a new sensibility encourages bold experimentation, creative appropriation, unusual partnerships, and even potential missteps. It seeks new forms, sensations, and intuitions necessary to construct ourselves pleasurably in an increasingly unstable cultural landscape that is shifting beneath our feet. Media theorist Joanna Zylinska (2024) reflects this sentiment in her writing about AI art, stating:
Our ethical task, the task of self-care and care for life in its multiple demands and responsibilities, is to rise to AI, the same way in which we previously rose to painting, photography, or digitality—and emerged creatively and socially altered with them. However, it also means acknowledging that this form of rising to AI may (need to) become an uprising.
Film still from Kiss/Crash (Cole, 2023). The work explores anxieties about desire and conformity in AI video models.
The challenge, then, lies in cultivating this new sensibility within the context of AI cinema: a domain of infinite mutability, targeted personalization, and insidious persuasive power. The structures of its emergence (its means of production, distribution, and consumption) are unlikely to mirror the collective avant-garde movements of the 20th century. But this departure may itself be an opportunity. I am encouraged by the artists and filmmakers already critically engaging AI, whether through playful disruptions of generative realism or works that expose the uncanny seams of synthetic images. Despite my anxieties about AI media—or maybe even because of them—I sense this moment holds a rare and urgent potential.
If, like Sontag, I cynically look back on this optimism in thirty years, so be it! I’ll be in the good company of those experimental enthusiasts who searched for a new sensibility.
References
House Committee On Oversight and Accountability (2024) Mace: Deepfake Technology Can Be Weaponized to Cause Harm. Available at: https://oversight.house.gov/release/mace-deepfakes-pose-real-dangers/ (Accessed: 30 October 2024).
OpenAI (2024) Sora (Blogpost). Available at: https://openai.com/index/sora/ (Accessed: 7 September 2024).
OpenAI (2024) ‘Video generation models as world simulators | OpenAI’. Available at: [https://openai.com/index/video-generation-models-as-world-simulators/](https://openai.com/index/video-generation-models-as-world-simulators/) (Accessed: 14 November 2024).
Rees, A.L. (2011) A History of Experimental Film and Video: From the Canonical Avant-garde to Contemporary British Practice. Palgrave Macmillan.
Sontag, S. (1965) ‘One Culture and the New Sensibility’, in Against Interpretation and Other Essays. Reprint, 2001, New York: Picador, pp. 293-304.
Sontag, S. (1996) ‘Thirty Years Later...’, in Against Interpretation and Other Essays. Reprint, 2001, New York: Picador, pp. 293-304.
Zylinska, J. (2024). ‘After AI Art.’ View: Theories and Practices of Visual Culture . https://doi.org/10.36854/widok/2024.38.2870