In the third episode of “Creative Dialogues,” an interview series produced by the film division of generative AI startup Runway, multimedia artist Claire Hentschker expresses her fear that AI will commodify the artistic process to the purpose where the Art homogenizes and becomes a variety of art derived equality.
“Are you getting this increasingly narrow average of existing things?” she asks. “And then – if that keeps getting averaged out – all the pieces will just be a blob?”
These are the questions I kept asking myself at a screening of the highest ten finalists at Runway's second annual AI Film Festival on Wednesday available upon request on the Runway website since this morning.
Runway had two premieres this 12 months, one in Los Angeles and a second in New York. I attended New York's, which took place on the Metrograph, a theater known for its arthouse and avant-garde performances.
I'm blissful to report that AI isn't heading towards a blob future… at the very least not yet. But a deft look from the director – the human touch – makes a major difference within the effectiveness of an “AI film.”
All movies submitted to the festival used AI in some form, including AI-generated backgrounds and animations, synthetic voice-overs and bullet-time style computer graphics. None of the weather seemed quite at the extent that cutting-edge tools like OpenAI's Sora can do, but that was to be expected considering most submissions were accomplished earlier within the 12 months.
In fact, it was – sometimes painfully – obvious which parts of movies were the product of an AI model somewhat than an actor, cinematographer or animator. Even otherwise strong scripts were sometimes let down by disappointing generative AI effects.
Take, for instance, Dear Mom by Johans Saldana Guadalupe and Katie Luo, which tells the story of a daughter's loving relationship along with her mother – within the daughter's own words. It's a tearjerker. But a scene on a Los Angeles highway with all of the telltale weirdness of AI-generated videos (e.g. distorted cars, bizarre physics) broke the spell for me.
The limitations of today's AI tools looked as if it would push some movies right into a corner.
As my colleague Devin Coldewey recently wrote, control is difficult to realize with generative models – particularly video-generating models. Simple things in traditional filmmaking, like selecting a color for a personality's clothing, require workarounds because each shot is created independently of the others. Sometimes even workarounds don't help.
The resulting disjointedness was evident on the festival, where a few of the movies were little greater than tangentially related vignettes strung together by a narrative and soundtrack. L'éveil à la création by Carlo De Togni and Elena Sparacino showed how boring this formula will be, with slideshow-like transitions that will make a greater interactive picture book than a movie.
Léo Cannone's “Where Do Grandmas Go When They Get Lost?” also falls into the vignette category – but still triumphs because of a heartfelt script (a baby describes what happens to grandmothers after they die) and an exceptionally strong performance from its child star. The remainder of the audience looked as if it would agree; The film received one in all the more full of life applause of the evening.
And for me that sums up the festival in a nutshell. The human contribution – not the AI contribution – often makes the difference. The emotion in a baby actor's voice? That sticks with you. AI-generated backgrounds? Fewer.
This was definitely the case with Festival Grand Prix winner “Get Me Out,” which documents the struggle of a Japanese man recovering from the psychological effects of immigrating to the United States as a young child. Filmmaker Daniel Antebi depicts the person's panic attacks using AI-generated graphics – graphics that I ultimately found to be less successful than the camera work. The film ends with a shot of the person walking up a bridge because the streetlights on the pedestrian street flicker one after the other. It's haunting – and exquisite – and it definitely took ceaselessly to capture it similar to that.
It's entirely possible that generative AI will at some point have the opportunity to recreate such scenes. Perhaps cinematography will eventually get replaced by prompts – a casualty of the ever-growing data sets (albeit with worrying copyright status) that startups like Runway and OpenAI use to coach their video generation models.
But that day is just not today.
As the screening got here to an end and the honorees filed out to the front of the auditorium for a photograph op, I couldn't help but notice the cameraman within the corner documenting the entire affair. On the contrary, perhaps AI won’t ever replace some things, akin to the humanity that we humans deeply crave.
