AI or not AI? That was not the question during a session at the Industry@Tallinn & Baltic Event portion of the Tallinn Black Nights Film Festival (PÖFF) on Wednesday. Instead, the focus was on opportunities that AI can open up in the film industry, along with thoughts on best practices needed as guardrails.
“Moving Forward With AI: Beyond Constraints. Future Scenarios for European Film Productions & Business” was the title of the panel that brought together industry experts to discuss how to work with AI to enable creativity and strengthen production, finance and other processes “within the boundaries of best practices, guidelines and regulations,” as the program summary highlighted. And it noted that the goal was to “demonstrate how to produce outstanding new content and business by leveraging emerging technologies despite disadvantages.”
The speakers were director-producer Matt Szymanowski, who is currently in post-production on Captive Mind, a “hybrid dystopian feature film shot with actors” that is integrating AI technology, and directing A Human Future, a documentary on the global AI boom shot on three continents in more than 10 cities; AI expert and script consultant Maciej Zemojcin, who has worked on the Picture From Auschwitz initiative; and screenwriter Katri Manninen (Codename: Annika, Shadow Lines). The moderator of the session was Kristen Davis, the former IT and innovation director at the New York Times International, who founded the consultancy CinqC in 2016. Oh, yeah, and Charlie Chaplin made a voice appearance courtesy of an AI agent. But more on that later.
Davis opened up the event by highlighting that AI will require “significant” changes to skillsets, mindsets and processes to get the industry to a state of “augmented creativity.”
Manninen is currently studying how developing drama series with AI differs from purely human work. Poems and other texts written by ChatGPT 3.0 felt like material written by kids or teenagers, she argued. “I really want to ideate my own stories,” she said. By their very nature, large-language models (LLMs) generate “the most likely answer,” or “generic shit,” she concluded. “I use AI for everything but writing so I can focus on storytelling — and rise above the generic shit.”
That means she never uses AI to come up with ideas or be a co-writer. “But I did learn to use AI to keep track of my diaries” and “take notes” as she speaks about ideas, Manninen shared. She then also uses LLMs to format and edit text to take out pauses and fill words, plus asks AI for notes, scene lists, a synopsis, character descriptions and the like. Instead of spending five days on a synopsis, she uses AI to get the job done based on her own writing. Her advice: “Give it the genre, short storyline and the characters and their roles (hero, bad guy, etc.) to help it make sense of the script.”
Manninen even uses AI to do preliminary research for her. But she warned: “You’ll only get a general idea and not all the details — and some of them are likely wrong,” so always fact check AI research.
“AI is really good at getting me ‘unblinded’ with instant feedback,” especially when other people are not around or would take too much time, she also told the Tallinn session. Since LLMs have learned to please humans, Manninen suggested, though, to “ask for only the most brutally honest feedback” to get at least “semi-honest” feedback. In comparison, “human co-writers have ‘skin in the game’ and can’t be ignored,” she emphasized.
Szymanowski, in his comments, compared the shift coming to the film industry from AI to the change from silent movies to “talkies,” the arrival of CGI, and other technological innovations. He recommended “getting their hands on the tools” to help people better understand them and their implications.
“We’ve seen short films made with AI,” he said. “But can we make a full-length feature film?” His take: “The future of cinema is hybrid.” Pointing to his work on Captive Mind, he cited efficiency, sustainability, for example, thanks to work via the cloud, and innovation as benefits of AI filmmaking.
High fidelity is now “there” in AI, Szymanowski said in discussing the continued technological improvements, mentioning that AI can be a VFX supplement and help with world-building or with character augmentation.
He projected that for a full feature with “several full AI sequences,” the post-production timeline can be reduced by 25-40 percent, while the team size can be cut down by 40-60 percent, leading to $120,000-$300,000 in cost savings.
Joining the session via video conference from Barcelona, Artefacto Studio executives Anna Giralt Gris, CEO and founder, and Jorge Caballero, co-founder, cited AI concerns, such as Guillermo del Toro saying: “I’d rather die than use generative AI.” They highlighted that the age of AI means a “paradigm shift in film production.”
The traditional production process is “linear and sequential,” which changes to “iterative and continuous,” they offered. For example, “boundaries between phases blur,” they explained. “AI tools extend development” via concept generation, visualization and simulation. And they mentioned the chance for “real-time iteration during production,” including the testing of narratives, visuals and sound.
Finally, Zemojcin, who calls himself “the film AI guy,” told the Tallinn event to “trash the past dream!” “If you are not ready for this shift, this will not shift for you,” he warned, encouraging people with the words: “No fear!”
He said the term “film and TV industry” is outdated, with him instead preferring the term “technology storytelling.” His take on the value in focus in this age: “In the fake content times, authenticity is a value.” While IP and ownership are “in flux,” Zemojcin also argued that “emotional connection with the audience is valid.”
A “different decision process and chain” are key to the age of AI, which people with industry experience will have to get used to, he said.
He asked an AI agent that he created using ElevenLabs and that uses the voice of none other than Charlie Chaplin for insight on best practices and potential risks for the use of AI in the film space. The expert said his long prompt for the AI agent included the likes of a description of personality and values, with the voice synthesized from The Great Dictator and Chaplin’s voice.
“The future of film will be decided by which side wins the close-up, studios using AI to cut humans out, or humans using tools carefully with unions, laws and strong norms to protect their faces, voices and stories,” the AI agent offered. “I think cinema survives if we insist on three things: transparency about when AI is used, consent and fair pay for every human whose work or likeness trains or appears in it, and audiences choosing soul over sludge. So the real question is not what the future of film will be, but what you here in this room are willing to refuse and what you are stubborn enough to protect.”