Gugusse and the Automaton
- Trevor Pronoso
- 1 day ago
- 13 min read
The Work of Cinema in the Age of Algorithmic Reproduction
WEB EXCLUSIVE Written by Trevor Pronoso | Contributing Writer
Last February, the Library of Congress discovered what was thought to be a lost film by renowned early cinema illusionist Georges Méliès titled Gugusse and the Automaton (1897). What makes this film uniquely special isn't just limited to film historians and cinephiles (though it doesn't hurt to see more recovered films lost to time). What makes this film truly worth your attention is that it is considered to be the first depiction of a 'robot' on film. The Library of Congress has a nice, succinct summary of the film that I will just quote in full below:
Gugusse the clown appears to control the actions of Pierrot Automate, a child-sized automaton standing on a pedestal. By turning a crank, Gugusse makes him march and wave a stick. As Gugusse turns the crank, the automaton gets bigger until it is the size of a grown man. Suddenly the automaton is controlling his own limbs. He hits Gugusse on the head with his stick. Gugusse pulls the automaton off the pedestal and picks up a large hammer. As Gugusse pounds the automaton on the head, he gets smaller and smaller. At the final stroke of the hammer, he disappears.
There's something so prescient about this film resurfacing 129 years after its initial production, now made publicly accessible to a modern audience. Automatons were all the rage when Méliès made this film, yet in the year of our Lord 2026, what's now keeping us preoccupied is the rapid adoption of Artificial Intelligence and their uncanny ability to produce photorealistic representations of daily mediated life—of humans—and whose depictions follow the directives of language prompts to generate its life-like animations based on large databases of material collected across the internet. From FSMs (finite-state machine) to LLMs (large language model), from humans recreating robots to robots recreating humans, the century-and-a-quarter gap between the film and today can't help but make its subject matter feel providential for today's discourse of technology, especially when considering cinema—including the use of the term 'cinema' denoting a certain prestige and privilege—as something separate from that of any other visual media deemed 'content' or 'AI slop' that exists outside the pitch black theatres of orthodox sacramental cinema-going.
You're not gonna see Gugusse and the Automaton screened at Academy Cinemas anytime soon. You're gonna watch this film—if you're curious enough to watch it in the first place—on your PC, laptop, or smartphone. You're gonna watch this film on a device that seemingly 'robs' you of the essential experience of seeing it on a 20 feet-tall projector screen, right? Are you going to watch a 'movie', or a 'video' permitted by the Library of Congress' host website's video player? Or are you gonna download the movie from the website as an MP4 file and watch it locally on your device, mirroring the viewing experience as closely as possible to that in a theatre? Is 'cinema' this obnoxious, conservative term that denotes a gatekept, intimate authenticity divorced from—or to distract oneself from—contemporary existence, or has its form become synonymous with a certain visual phenomenology, in which its mediation results in a singularly digital mimesis?
The existence of Sora 2 makes André Bazin sound like a misled ideologue: cinema's capabilities for objective recording of reality à la Lumière figures the possibility of a 'total cinema of immersion', a one-way mirror. You don't need a camera to present and represent—or even interface with—reality because our eyes can't tell the difference anyway. The indexed image lacks the reverse index of entropic thermodynamics of the material world, the reference without the referent. One could say photorealism has kept up with reality, or that reality has kept up with photorealism. The ontology of the camera and the image is no different to the ontology of its captured subjects (i.e. social media influencers being perceived as 'shallow', all human activity in service of the image, the need to 'capture the moment'). The path towards the 'authentic' has turned out to be 'synthetic'. The idea of the image as a 'mummification of change', as time halted to be revisited again and again in the future, speaks to a certain anxiety of change itself, an anxiety that the end of an image—or the end of a shot past a certain duration—will represent the end of existence, the end of thought and pathos. Reality ends the moment the shot has run its course.
This misguided sense of realism and 'relatability' is what informs big blockbuster franchises such as the MCU to keep pumping out films that constantly operate on a never-ending mode of self-referentiality, of encouraging its devoted fans to rewatch the past films of their beloved cinematic universe to remember and reassure themselves, to measure the next installment's belonging and logical consistency within that said universe. The authentic image divorced from reality, the finitude of repetition begetting a Pavlovian dogmatism towards tradition, towards standardised narrative storytelling. With such a mindset, taste becomes relative, red is blue, and some trash is better than other trash because it was produced by thousands of 'real people' rather than a cold, unfeeling algorithm. One ceases to think past the image, beyond the 'cinematic universe'. The image cajoling us into a 1:1 recreation of a reality we never see, a Byzantine Bazinian backwash of 'seeing' replacing 'being'. The movie studio and AI are simpatico.
Gugusse and the Automaton in comparison doesn't stand a chance against these warped standards: "it looks old and it comes from a time where even my great-grandparents weren't yet born", "the special effects looks fake and amateurish compared to 'modern classics' such as Everything Everywhere All at Once (2022) or Parasite (2019)", "there's no music or dialogue that will immerse me into the film", or simply, "it's just plain boring." But for me, what makes Gugusse and the Automaton truly outmoded and outdated—if not unrealistically naive—isn't its antiquated post-Napoleonic milieu, its relatively deteriorated celluloid quality, or the exaggerated theatrical movements of its actors; it is the film's denouement: Gugusse/Méliès wielding a large hammer and smashing the automaton down to its miniature size until the final blow makes it disappear altogether. Gugusse delights at the novel liveliness of the automaton until it hits him in the head with a stick. The ending implies as such: humans have successfully defeated—or at least tamed—the encroaching dangers of technological progress. It's about as cruel a joke as playing "We'll Meet Again" atop footage of atomic explosions in Kubrick's Dr Strangelove (1964). Ted Kaczynski would have relished the unintentional irony of this film.
Calling it 'outdated' registers a bit too defeatist, but neither is calling it 'prophetic' any more comforting. Watching a deepfaked video of Donald Trump dressed up in star-spangled Nike Tech—deliberately a reference to Nicolás Maduro's capture by the US being co-opted by AI—feels like we're still in the initial stages of indulging as much human-like behavioural spectacle from robots as possible. The stick hasn't landed on the heads of the general populace just yet. Or maybe I'm wrong, and it's already struck us even after Grok allowed users to make sexually explicit deepfakes of women and minors without their consent. Maybe that is what makes Gugusse and the Automaton outdated: it shows us history through a linear, chronological series of events dictated by causality; one event follows another, 'this event' is caused by 'that event' (Gugusse operates the automaton, the automaton hits Gugusse, Gugusse destroys the automaton), etc. The 21st-century digital landscape, by contrast, decisively throws us into the chaotic, disorganised soup of history, where every single event is happening simultaneously, a constellation without any guiding lines to help us see its pattern. Maybe we're just circling the same star over and over, giving us the illusion of deviation. One Chrome tab on X, another on Reddit, another on TikTok, another on YouTube as one rewatches Jerma985's "Dollhouse" VODs for the nth time; spectacle is just one—or several—commercial applications away. Is this considered cinema, the omniscience of a determined fate via going back and forth between several entities? Cutting towards the same image after another?
D.W. Griffith built his entire career on cross-cutting between disparate characters in fatalistic situations to achieve that sense of Darwinian "omniscience", an omniscience that spans across multiple stories and multiple generations. By the end of his movies, one tends to forget whether or not we should take his images seriously or not, since they reek of either full endorsement of prejudice and white supremacy (The Birth of a Nation, 1915) or contemplative resignation at the cyclical nature of violence, social unrest, and heteronormative hegemonic reification within humans (Intolerance, 1916, and every other film he has made). If a single Chrome tab were no different to a character's entire psychology and way of living, to what end do these intertwined tabs and characters reveal to the director, the user, the viewer? If Mulvey can speak of a gaze that imprisoned cinema's discernment to what was pleasurable as also patriarchal, what type of gaze characterises seeing the same type of image across different modalities of engagement? Can one even call that a gaze, or is it some pre-cognitive hallucination, some foggy, atavistic libido for whatever endorphin-inducing stimulus appears on screen, a 'content brain'? Whether it be livestreamed genocide or livestreamed 'let's plays', you can always switch the channel, close the tab, write off the character as 'shallow' and 'lacking narrative depth'. Griffith was willing to change the content of his films if the test audience reacted negatively. One need not remember when and why they took the blue pill; Griffith and social media algorithms dare not make you aware of its gentrified montage. Cross-cutting merely stitches together the status quo's rehabilitation, making evident the Van der Waals forces between two shots that eliminate their spatial distance and imply their unity, their unchanged trajectory towards a predetermined fate.
Méliès made hundreds of films that delighted in the magical powers of the 'jump cut', where its abrupt appearance gave the impression that reality could suddenly change and transform its surroundings to the will of the magician—on-screen and off-screen—making objects appear and disappear without precedent. Even though Méliès wanted us to immerse ourselves in his pictures through their fantastical décor and elaborate mise-en-scène, their true delight lies in actively seeking to disrupt the homogeneity of the image and the stability of reality, opening them to the destabilising powers of the absurd and the unexpected. The 'cut' disrupts our expectations of the image's stasis and invites us to think with and alongside difference—or, as Derrida puts it, the différance of language that allows thinking to begin. Méliès sought to look beyond reality through the lens of magic. Cinema seeks to look beyond the image through the invisible cut. One image to the next, a dialectical type of thinking that saw Sergei Eisenstein's unofficial banishment from the Soviet Union by Stalin, prohibiting the fruition of his theories on intellectual montage in favour of 'socialist realism'. The curse of 'realism' thus reappears again a century later; an encroaching elite group of tech billionaires actively inhibiting our ability to think alongside the moving image by restricting its manifestations to mere 'language prompts', erasing imagination and dialectical contradiction by giving you exactly what you want based on the finite data it generates itself from. The 'message' of the movie was paramount in socialist realism. The 'messages' you type on Claude AI as if you're personally chatting with a 'human being'—including speech bubbles and all the graphic features of any SMS service you use—is paramount in AI image generation.
From finitude to finitude, the image becomes trapped in the meaninglessness of the familiarity of the status quo. Marvel and AI propose that everything that can happen has happened, even if the chances are one in 14 million, as Dr Strange asserts in Avengers: Endgame (2019). The world order is apparently just and needs no radical change because we are destined for our future. The film industry is purely a screenwriting exercise without a
director because the director is always replaceable; the text, the 'sales pitch', the plot, the characters are what sell the movie. The singular master shot is shredded and gentrified into a dizzying array of close-ups, mid-shots, two-shots, shot-reverse shots, and reaction shots that make up this editing smoothie called 'coverage'. AI is no different: it shreds across multiple images and video sources and aggregates a probable array of colour values across a certain pixel resolution based on the machine learning's reliance on data (i.e. language) whose content has no real-life referent (LLMs will associate the word 'tree' based on what the internet associates images with the caption 'tree' on it.). 'Tree' no longer refers to an actual tree, but the image of the tree; no trees are involved in the making of such images. Approximations of the master shot, approximations of reality. There is no wind blowing in the trees because it's all AI-generated. It's as if the light hole in the camera obscura of Plato's Allegory of the Cave has been closed shut, and the prisoners begin producing fake images of a reality through baseless anecdotes—baseless images—relying on linguistic/literary inferences without having witnessed a particular 'reality' in the first place.
Social media and its recommendation algorithms destroy this capability to think by making the transitions between images for you. The cut is automated, a 'black box' that arranges its technical images without any guiding principle except to extract as much retention from its users as possible, as much data on your viewing habits as possible to sell to potential bidders. You don't technically 'cut' to the next image so much as you 'scroll' to the next one. You virtually see the images 'stitched together' from the bottom of the first video to the top of the next video when you scroll through Instagram Reels or TikTok. You cannot move or change where this cut lies, so you are stuck with what the algorithm gives you. The only way to manipulate the algorithm is to give it exactly what it wants from you: full immersion and the refusal to scroll immediately to the next video. Stare at a video long enough, and the algorithm will keep feeding you more just like it; skip the video, and the algorithm learns to ignore similar videos. The same types of videos over and over, the same types of movies over and over. There are no 'images' in the plural sense, but instead one long vertical image that extends as far as you're capable of scrolling for hours upon end. Images refuse to escape themselves; therefore, reality refuses to escape itself. Academy Cinemas or Event Cinemas, the cinema experience too can't seem to escape reruns of lucrative titles such as Taxi Driver (1976), Chungking Express (1994), Oldboy (2003), or the upcoming Avengers: Doomsday this year. We rewatch the same movies and images over and over to feel closer to reality, not overcome it.
What then to make about Gugusse and the Automaton in the present day? Martin Scorsese's Hugo (2011) may tout itself as a reparative counter-measure to Méliès' historical neglect. Yet, it is riddled with the pervasive nostalgia and wishful thinking native to the movie brats of New Hollywood and fails to embody the innovative and imaginative spirit of early cinema's practitioners. He recreates the period's aesthetic, but not its formal ingenuity. A Scorsese picture is a Scorsese picture: a stream of images rushing past your eyes that gives the illusion of momentum. The 21st century has left him morally adrift: criticise Marvel in one hand, be a guest voice actor for the upcoming The Mandalorian & Grogu in the other. The pretence of prestige and pedigree in his
ambitions for 'cinema' has disappeared; he has contributed to the simulacra that is 'modern cinephilia', where movies not only represent escape, but also a potential way of living. As Louis Skorecki puts it, the new generation has had their entire lives depicted on-screen already and requires no further exploration past the frame for the viewer. It's no surprise that the plot of Hugo revolves around repairing an automaton whose main function is to draw the iconic still of a rocket/cannon shell hitting the Moon's right eye from Méliès' A Trip to the Moon (1902). An image of an image, of mechanical reproduction. The wondrous display of spectacle fills in the chasm between reality and its perception. If the copy makes up for the original, then the fruits of Kiarostami's Certified Copy (2010) on the nature of cinema read more 'instructive' than 'constructive'; the simulacrum is indeed "true".
I see two possibilities for Gugusse and the Automaton's integration in the digital age. The first is one of cruel subsumption: it becomes another film publicly accessible on the internet for AI to wring out what Benjamin calls its "aura," producing more algorithmic slop catered to nothing but your base need for spectacle, for its eye-catching, audience-retaining qualities. The second requires a bit of context: There is only one scene that's worth remembering in Hugo, and that is the seven-minute montage flashback detailing Méliès' start as a filmmaker and his ruinous professional and financial
decline. Near the end of this sequence, Ben Kingsley, playing Méliès, shares his experience of the French Army in the First World War seizing his studio and his films. During the war effort, his films were melted down into silver and celluloid to cast shoe heels for soldiers. I immediately think back to Louis Lumière calling cinema "an invention without a future". I don't interpret this the same way Lumière did, that cinema was a fad, a novelty that would disappear as we move on to other, much better things. In one respect, I totally agree with him: cinema as solely a carnival attraction is slowly declining, or, more specifically, this "cinema of attractions" refuses to die, and its vanguard remains big studio companies like Disney and Universal repurposing their outdated utility as theme park rides. While cinema-going attendance ever so dwindles, income from theme parks steadily increases as Disney reports raking in $10 billion in domestic and international revenue during the last three months of 2025. Méliès' cinema isn't exempt from this, either; history had also moved on from Méliès and witnessed cinema's further artistic advances through Griffith, Vertov, Eisenstein, Chaplin, Rossellini, Godard, Brakhage, and Markopoulos, to name a few.
Cinema thus has no 'future' and is finally free to exist eternally, shifting and changing its forms throughout the present. If the future implies destiny—of existential certainty—then the history of cinema has shown itself to have a bad habit of near-death experiences. While Méliès' films melted into shoes is rightfully a tragedy, I see this as a metaphor for cinema's malleability and its formal utility. Cinema truly becomes what you make of it. If Jean-Marie Straub and Kōji Wakamatsu treated their cinema like bombs in their lifelong commitment to the class struggle, then political power exerts itself through the aperture of your camera. When television became more ubiquitous during the 70s-80s, Woody Allen made Hannah and Her Sisters (1986) while Peter Greenaway made his declaration that cinema truly 'died' with the introduction of the remote controller. Godard, on the other hand, went in a completely new, cinematically richer direction with Here and Elsewhere (1976) and his multiple television series Six times two/On and under communication (1976) and France/Tour/Detour/Two/Children (1979). The television became a site for Godard to engage in long-form documentary investigation and ideological exploration, a glimpse into the very real revolutionary potential of television decades before sanitising itself into satirising reality TV, as seen in The Truman Show (1995) and in all your favourite mockumentaries made in the 2000s onwards. AI-generated images and movies thus prove another addition to the long history of cinema reckoning with itself, of trying to make amends with the likes of Gugusse and the Automaton. But instead of taking the hammer and beating the automaton into submission, I'd like to imagine an alternative in which AI merely wears the cocoon of cinema's next groundbreaking transformation, where the reliance on language to communicate images ceases to exist, and AI thinks and communicates only in images. The latter will be the biggest challenge cinema faces if it ever reaches that far.
Alas, Gugusse and the Automaton is only a one-reeler. We have to make the next image—the next cut—ourselves.
Comments