At a conference on film and artificial intelligence over the weekend in South Korea’s Bucheon, it was difficult to know whether the new technology was being embraced, normalized or underestimated.
Bucheon, a high-rise city on the outskirts of Seoul, has long harbored both high-tech and cultural industry aspirations. The city turned a WWII bunker into a digital art hub and boasts a comic book museum, a Webtoon convergence center and two film festivals as well as a Philharmonic Orchestra founded in the analog-era.
This year, its long-running Bucheon International Fantastic Film Festival (BiFan) launched a competition for AI-produced short films, co-hosted a conference on the topic and operated a hands-on AI workshop. The embrace of AI felt like a brave response to every mature film festival’s fear of becoming stale.
The Bucheon conference’s impressive lineup of speakers included: Caleb Ward, CEO of Curious Ridge; directors Dave Clark, Kwon Hansl and Piotr Winiewicz; Lee Seungmoo, professor of film at the Korean Academy of Film Arts; and Margarita Grubina and Anna Bulakh of voice cloning company Respeecher.
Maciej Zemojcjin was on hand to explain how he had created the “Murals” art installation, showing in the Art Bunker, that preserved some of the endangered or already destroyed works created by Banksy in Ukraine.
After last year’s writers and actors strikes in Hollywood over the potential threat of AI and the reality that AI is already being used in Korean commercially-released feature films such as “Wonderland” and TV shows such as “Queen of Tears,” a discussion was welcome. (The recently-released “Wonderland” feature film, directed by Kim Tae-yong, used voice cloning to add dialogue by actor Gong Yoo that had not been recorded during principal photography four years ago. It was also used to create a younger iteration of another actor, Lee Eol, who died two years ago. Tvn’s “Queen of Tears” series combined generative AI with virtual production to have actor Kim Ji-won encounter her double in a snowy forest.)
Sten-Kristian Saluveer, strategic adviser, head of Cannes Next and CEO of Storytech, led the debate and analysis. He described the media and entertainment’s current “poly-crisis” as one where the industry is shifting from an era of producers and single creators, to one of user centric creation – with AI accelerating and empowering the users.
Tools such as Unity, Unreal, Ableton and platforms such as Substack, Patreon and YouTube already allow creators to create, distribute and monetize outside of the traditional studio or broadcaster context.
Korea provides one of the best examples of how new tools can outrun traditional media. Netflix’s globally-successful hit show “Squid Game” was seen by 142 million people, Saluveer said. YouTube creator Mr Beast’s “Squid Game in Real Life” has been seen by over 628 million users. In addition to his financial windfall, Mr Beast’s reward was a deal with Amazon Prime Video.
Saluveer also explained two different approaches toward AI emerging within the entertainment industry. A U.S. approach in which experimentation is prioritized and where tangles, such as OpenAI’s clash with Scarlett Johansson over a synthetic voice, will be figured out in court. The European Union, on the other hand, says safety is paramount and that regulation should come ahead of industry use.
Saluveer described an AI-powered future as one in which everything that can be automated will be automated. Like lemmings chewing at the foundation of a house, AI is starting with the simpler things and those nearest to hand, such as design and subtitling. Over a longer time span, perhaps 10 years from now, IP-generated content may have shoved traditional creators to the sidelines. “Author-generated cinema will still exist, but it may be local and niche. Like opera,” he said.
Using a clip called “And I didn’t say a thing,” Respeecher’s Grubina demonstrated the possibility of human redundancy as an imminent possibility. From the smallest of samples, such as a still photo or video clip, it is already possible to clone a voice to a new face, make synched face movements accurately fit the sounds and make the voice speak in a new accent or language – perhaps one the original sample speaker does not know. The kit can be used for resurrection, customization, scaling, ADR, dubbing and localization.
While speaking of a “voice marketplace,” Grubina also touched on the need for permissions and the ethical use of voices, such as the avoidance of sexual or political dialogue. Abuses of voice and image cloning have already occurred in the real world, such as the recent case of a Ukrainian woman who discovered that her digital likeness was being used to spout pro-Russian propaganda in Chinese to online audiences in China.
Winiewicz unveiled a couple of clips from his feature-length film “About a Hero,” which is likely to have its premiere at one of the autumn festivals this year. Among the hybrid project’s themes is AI itself, the work of legendary German filmmaker Werner Herzog and the history of Kaspar Hauser, the 19th century man who had no language experience until age 16, but once he was integrated into society and started learning delivered surprising outcomes. Herzog previously made “The Enigma of Kaspar Hauser” and more recently issued a defiant statement suggesting that AI will need hundreds of years before it can make a film as good as his.
Winiewicz says “About a Hero” was not made to challenge Herzog’s words, but rather to play with some of the concepts of AI, such as obtaining outcomes that depend on the material that is used as input.
BiFan festival director Shin Chul embraced the challenges of the AI-powered, user-generated era. “Filmmaking will cease to be a battle of cash, it will only be one of creative challenges,” Shin said.
He gave AI the materials to come up with both the festival’s poster and its trailer. The poster is a colorful hybrid of previous BiFan posters. The short film asks the question “AI, who are you?” The uncomfortable answer: “I’m just your mirror.”
