In a rural Hawaiian backyard, a Filipino man’s steps rustle through the grass amidst tropical bird songs. He nears a starfruit tree, stopping at a shrine to gaze upon a framed black-and-white photo of a woman with a 1950s hairstyle.
A sudden strong wind disturbs the tree, scattering the shrine’s items. The man stumbles, falls, and hits his head. He regains consciousness in a dim, foggy forest, where a clay-masked woman stands over him, sword raised.
With a sword at his throat, she demands in Ilocano, a common language among Hawaii’s Filipino population, “Who dares sleep beneath the sacred tree?” He claims to be disoriented and attempts to escape. She pursues him, moving with a mix of running and levitation. He falls once more. As she closes in, sword aloft, he hurls a rock, breaking her clay mask and exposing part of her face.
“Mom?” he questions.
This scene introduces “Murmuray,” an independent short film by Brad Tangonan. Its signature blend of vivid nature imagery and ethereal, desaturated aesthetics echoed his prior projects.
The unique aspect? It was produced using artificial intelligence.
Tangonan joined nine other filmmakers in Google Flow Sessions, a five-week program providing artists with Google’s AI tools, such as Gemini, the Nano Banana Pro image generator, and the Veo film generator, for short film creation.
The films varied significantly; Hal Watmough’s “You’ve Been Here Before” creatively combined hyperrealistic and cartoonish styles to illustrate the value of morning routines, whereas Tabitha Swanson’s “The Antidote to Fear is Curiosity” offered a deeper, philosophical look at humanity’s interaction with AI.
These short films, presented at Soho House New York last year, avoided the impression of being generic AI-generated content. Every independent filmmaker interviewed highlighted that AI empowered them to realize stories they couldn’t have afforded or managed within traditional budgets or timelines.
Tangonan expressed after the screenings, “I view all these instruments, from a standard camera to generative AI, as means for artists to manifest their inner visions.”
Google is clearly pushing the narrative that AI is merely another creative tool, a point that holds merit as AI video generation technologies rapidly advance and become more integrated into artists’ workflows.
By 2025, firms such as Google, Runway, OpenAI, Kling, Luma AI, and Higgsfield significantly advanced AI video capabilities beyond the somewhat awkward, prompt-driven experiments of the previous year. The AI video sector, backed by substantial venture capital, is transitioning from experimental development to practical post-production applications.
While this explosion of AI tools promises to “democratize access” to filmmaking, it simultaneously poses a risk to jobs and creative expression, potentially burying them beneath a flood of uninspired content. This high-stakes situation has created divisions among artists: those utilizing AI face accusations of complicity, while those who abstain risk irrelevance.
The debate is no longer about AI tools belonging in the creative arsenal; their arrival is inevitable. The real inquiries are: What cinematic artistry endures when the industry prioritizes speed and volume over excellence? And what emerges when individual creators leverage these same tools to craft truly meaningful work?
But is it slop?
Numerous arguments against AI in filmmaking have emerged, many from prominent figures within the industry.
Last October, director Guillermo del Toro declared he would sooner die than use generative AI for filmmaking. Similarly, James Cameron stated in a recent CBS interview that the concept of generating actors and emotions via prompts is “horrifying,” arguing that generative AI merely produces a homogenized composite of all prior human creations.
Werner Herzog criticized AI-generated films he had viewed as lacking “soul,” asserting that these productions merely represent “the common denominator, and nothing beyond this common denominator.”
Both Cameron and Herzog contend that AI usurps human creative control and is incapable of genuinely portraying unique personal experiences.
Watmough explained to TechCrunch, “While it’s simple to react negatively to AI in the abstract, it’s more challenging to criticize a creator who has used it to produce something deeply personal.”
Tangonan, characterizing “Murmuray” as a “family story,” shares this perspective.
“AI acts as an enabler,” Tangonan stated. “I retain full creative control. The ‘AI slop’ seen online often represents the lowest common denominator. Indeed, if you relinquish control entirely to AI, that’s the outcome. However, with a distinct voice, creative vision, and style, the results will differ.”
Employing AI in filmmaking isn’t simply about generating a movie from a prompt. Tangonan, for instance, crafted the “Murmuray” script independently and collected visual references for his shot list. This material was then processed by Nano Banana Pro to create style-consistent images, forming the basis for video generation.
Keenan MacWilliam also meticulously crafted her short film “Mimesis,” a simulated guided meditation, to genuinely reflect her unique visual style instead of merely combining elements from other artists’ works.
MacWilliam authored the script and narrated the fictitious meditation herself, achieving a balance of tranquility and humor. Against a dark, aquatic canvas, surreal visuals of flora intermingled, dissolved into smoke, transformed into seahorses, and vanished.
All the visuals originated from MacWilliam’s personal archive of scanned plants and animals, a testament to her habit of carrying her scanner everywhere.
MacWilliam informed TechCrunch, “I dedicated considerable effort to developing applications based on my unique dataset, which then served as references,” also noting her collaboration with her regular composer and sound designer. She emphasized, “My aim was to discover fresh avenues of expression for my existing artistic themes and style, not to supplant the contributions of my collaborators.”
A recurring theme among the Google Flow event filmmakers was a preference for using AI exclusively when human input was impractical or when the unusual qualities of AI-generated content enhanced the narrative.
Sander van Bellegem’s “Melongray,” for instance, depicted life’s acceleration via psychedelic imagery. A scene where a salamander morphs into a balloon, though unplanned, demonstrated how AI could extend his creative and physical boundaries.
To be [efficient] or not to be?
Modern film studio budgets face intense pressure due to escalating production expenses, the shift towards streaming, and cautious corporate mergers. Consequently, significant investments are reserved for reliable moneymakers (e.g., another Marvel film), while innovative mid-budget features are largely neglected.
Introducing AI could intensify studios’ cost-cutting tendencies, potentially leading them to replace elements like actors, sets, and lighting, sacrificing artistic integrity for efficiency. Conversely, AI’s efficiencies could also reduce obstacles, enabling film studios to more easily create original content.
Even Cameron conceded in his CBS interview that generative AI has the potential to reduce VFX costs, possibly paving the way for more inventive sci-fi and fantasy films – genres typically demanding high budgets and often limited to established franchises such as “Avatar.”
Tangonan explained that the scene in “Murmuray” depicting the woman flying through the forest would have required costly visual effects or elaborate on-set rigging, both beyond a short film’s typical budget.
However, even creators who acknowledge efficiency gains recognize the inherent dangers to artistic integrity.
MacWilliam observed, “Broadly speaking, efficiency isn’t always conducive to creativity.”
Empowered and isolated
Independent filmmakers find the abundance of potent tools both advantageous and detrimental. While it undeniably “democratizes access,” it also promotes solitary work. The greater one’s self-sufficiency, the less incentive exists for collaboration.
Watmough conveyed to TechCrunch, acknowledging a friend’s voice contribution to his short, “I recognize I’m a solo operator, having created this myself… but that’s not how storytelling or filmmaking should ideally function. It ought to be a collaborative endeavor, as greater involvement enhances accessibility and fosters deeper connections with audiences.”
While directors guide creative choices, they don’t encompass every aspect. The filmmakers interviewed discovered they were unexpectedly assuming roles like set designer, lighting director, and costumer – positions demanding skills they lacked. This proved frustrating and exhausting, diverting them from their core creative passions and sparking concern over the rapid disruption of an entire industry.
The filmmakers also expressed a preference against replacing human actors with AI, although some conceded that AI-generated performers are likely unavoidable for smaller production houses. The technology to create actors, their expressions, and movements already exists and is continually improving. AI video startups, such as Luma AI – which secured a $900 million Series C funding last November – are even developing systems that allow a single actor’s performance to be captured, then modified by AI to alter character, attire, and setting.
Tangonan articulated, “Ideally, I would collaborate with live actors, a cinematographer, department heads, and a complete crew to create something extraordinary, using AI to supplement and achieve elements impossible on set due to budget or time constraints.”
If artists don’t define AI, studios will
MacWilliam commented, “I believe that creating any artistic piece involving new technology inherently demands a thorough self-assessment and openness to discussing the work.”
She continued, “These are simply instruments. How will you employ them? Will you use them ethically? Will you critically examine their implications? Will you be open and share your understanding?”
Many, however, do not perceive AI tools as impartial. Beyond concerns of job displacement, copyright issues persist. The AI video generation company Runway has allegedly used thousands of hours of YouTube videos and copyrighted studio materials for training. Other entities, including Google, OpenAI, and Luma AI, have faced scrutiny regarding similar practices or the unauthorized use of copyrighted films and stock footage. (Conversely, some tools, such as Moonvalley’s Marey, rely solely on openly licensed data for training.) Furthermore, environmental concerns are significant; estimates indicate that producing mere seconds of AI video can consume electricity equivalent to hours of streaming.
Predictably, several filmmakers I interviewed reported encountering disapproval for their AI experiments.
Tangonan remarked, “When I share content online, many of my filmmaking peers immediately react with the sentiment that we should collectively resist using any of these tools. I simply don’t concur.”
Should filmmakers hesitate to engage in discourse about AI’s appropriate use and ethical limits, the dialogue risks being shaped without their input. The decisions would then fall to profit-driven studios prioritizing efficiency over artistic merit, rather than to artists committed to responsible application.
Watmough asserted, “The film industry struggles due to a lack of innovation and prohibitive costs. Tools such as AI are vital for its endurance. I believe active engagement is crucial, as non-participation risks transforming the industry into something unrecognizable and unsustainable.”
Correction: A previous iteration of this piece inaccurately described Ilocano as a Hawaiian dialect of Filipino. Ilocano is, in fact, a language originating from the northern Philippines, widely spoken within Hawaii’s Filipino population.