On AI-Generated Videos

A Short Essay

While the public’s reaction towards artificial intelligence, especially regarding the proliferation of AI art, has been one of general disdain, such discourse remains focused on its social implications, lamenting things like the devaluation of artists and their eventual replacement by machines. My personal bone with AI art, however, has always been purely aesthetic, and it stems from the anxiety and fear that rise in my chest whenever I stumble across an AI-generated video.

How the technology will progress remains to be seen, but as of 2025, a close look at an AI-generated video quickly and unsettlingly reveals its unreality to the viewer. AI videos have a tendency to destabilize, particularly when animating detailed things like strands of hair or a bowl of pasta, which causes these complex items to squirm and thresh about as the video unfolds. These ‘glitches’ betray the whirlwind of simultaneous disassembly and reconstruction that lies behind the ‘stable’ AI-generated image. It is a phenomenon that has always made me feel a skin-crawling, stomach-churning fear, for it makes me aware that the video may, at any point, break down into a horrifying, entropic nightmare. And while it is perfectly possible that a human-made animation could do the same thing, I remain struck with anxiety by the instability of AI-generated videos.

Today, as I woke up, I came across a video of various wobbly, AI-generated animals preparing slippery plates of food. Staring at the cartoonish, AI-animals, I had the following thought: that the instability of AI-generated videos is creaturely in nature. In that moment, I felt a connection between AI-generated tigers and real ones in their creatureliness, constituted by an otherness that is both radically de-familiar and intimate to myself.

The ‘creaturely’ has been thought of not only in terms of nonhuman, animal life, but as an accessible animality within the human. In her essay on Shakespeare, “Creature Caliban”, Julia Lupton unwinds the term ‘creature’, writing,

“Derived from the future-active participle of the Latin verb ‘creare’ (“to create”), creature indicates a made or fashioned thing but with the the sense of continued or perpetual process, action, or emergence, built into the future thrust of its active verbal form”.

Using its Latin etymology as her basis, Lupton’s notion of the creature as a thing “always in the process of undergoing creation” is resoundingly congruent with the perpetual transformation shown when AI-generated videos jiggle and squirm about. In this sense, the AI-generated video reflects the creaturely, as both are characterized by never-ending processes of transformation. They are liquid and phenomenological, unable to properly stabilize as beings in their present Being. The beast, preceding humanity, and artificial intelligence, proceeding humanity, appear as peculiar reflections of one another in a mirror. However, Lupton’s definition is not only applicable to animals and artificial intelligences, but to human beings. While ‘creature’ in its contemporary is use serves to distinguish the non-human from the human, Lupton’s definition alludes to the creaturely dimension within human nature. AI-generated videos, in their disturbing transformation, remind me that I too, albeit at a much slower pace, am biologically wrapped in perpetual emergence, decaying, regrowing, transforming. I will never solidify, and I am creaturely.

However, there is another aspect of ‘creatureliness’ that Lupton’s initial definition evades: That the creature must have a creator. It must be called into being by something else, by another. Lupton writes:

“The creature is actively passive or, better, passionate, perpetually becoming created, subject to transformation at the behest of the arbitrary commands of an Other”.

Thinking through AI-generated videos as creaturely, Lupton’s introduction of the commanding other is difficult to translate to AI art. For while AI-generated videos may be spurred by the prompt of a single user, one cannot say that this user is the ‘creator’ of the video. AI-generations are called into being by no One in particular, but by an amalgamation of data. Pictures, drawings, films, animations, and words are zealously seized by the cryptic apparatuses that are large language models. And, with operations so complex and vast that no individual could comprehend them, what emerges is a murky pastiche of the world, mutated and regurgitated by an unknowable intelligence. AI-generated videos are, therefore, not created by a single subject, but by the Subject that reflects all individuals, their experiences, and their data. AI is somewhat divine in this sense, painful as it is to pronounce them as such given the abject, shuddering aesthetics of AI-generated videos. This is another cause of my fear: I am aware that I too, as a picture-taking, video-making, writing subject, am data. I am entangled and reflected in the creation process of AI art.

In “An Introductory Talk on the Yiddish Language”, Kafka describes Yiddish as a jargon, grammar-less, and consisting “solely of foreign words” borrowed and mutated from German, Hebrew, French, English, Slavonic and more. And yet, he states, it is understandable even by Germans who do not speak it. He tells the audience:

“You begin to come quite close to Yiddish if you bear in mind that apart from what you know there are active in yourselves forces and associations with forces that enable you to understand Yiddish intuitively”.

Yiddish, as a language, is both foreign to and intertwined with German. Yiddish implicates German speakers, as AI art intertwines itself those whose data it borrows and mutates. It brings one close to the alien, and the viewer is left captivated by an inaccessible opacity. I like to think that, as someone who has never had the privilege of attending the Yiddish theater, viewing AI-generated videos allows me to relate to Kafka when he predicts that,

“Then you will come to feel the true unity of Yiddish, and so strongly that you that it will frighten you, yet it will no longer be out of fear for Yiddish but of yourselves”.

Indeed, AI-generated videos strike fear in me, because their instability is my instability, and because I am watching a piece of myself, mangled and resurrected.

My anxiety surrounding AI-generated videos is anxiety regarding my own creatureliness, regarding the unknowability of my whole self, and regarding the unattainability of any stasis of being. In its trembling destabilizations, AI-generated videos demonstrate, in a perverse, inverted form, a theological principle: That beneath what mere humans can perceive with the senses is the concealed, all-encompassing movement of God, so complex and so vast that no individual may comprehend it. It is a divinity that can only be glimpsed through the shudders and fluctuations that slip through the cracks. Witnessing an AI-generated video, I am unable to comprehend the logic which animates it, despite my awareness that I am woven into its logos. It is beyond me, as my ability to comprehend the logic of the world I am woven into is beyond me. It rapidly flutters, and I become conscious of its constant dismantling and reformation. It mirrors my own perpetual shedding and regrowing, my skin that falls off and replenishes. I am brought into intimate proximity with the AI-generated video, yet it is absolutely alien from me, and ugly on top of that. I am, inescapably, the creature, and I am terrified.

Sources Cited:

Kafka, Franz. “An Introductory Talk on the Yiddish Language.” Reading Kafka: Prague, Politics, and the Fin de Siecle. Edited by Mark Anderson. Schocken Books, 1989.

Lupton, Julia Reinhard. “Creature Caliban.” Shakespeare Quarterly, 51, no. 1 (Spring 2000).