Artificial intelligence is increasingly reshaping film production, with Netflix reporting that AI tools have been utilized across approximately 300 of its titles. The disclosure came from Lee Sung-gyun, Netflix’s senior director of APAC production, during a panel discussion at the Busan International Film Festival.
Lee shared the stage with Kimball Thurston, chief technology officer at Wētā FX, and Kang Yoon-sung, director of the film “The Outlaws.” The session, titled “Film Production and AI: New Tools, New Possibilities,” was co-hosted by Netflix and the MPA as part of the Creative Asia initiative for emerging Asian talent.
Kang highlighted the financial impact of these technologies, stating that AI workflows can reduce a feature film’s production budget by between 30% and 40%. He illustrated this point with a test clip for his upcoming project “The Gulf of Aden,” which focuses on military operations at sea. The clip was generated entirely using Kling, a Chinese software platform.
Although the test was rendered at reduced quality, Kang noted that current AI tools are capable of producing 4K output with 10-bit color or higher, meeting cinematic standards. When calculating savings for “The Gulf of Aden” to attract investors, the team found the reduction amounted to roughly 40%.
Kang explained that the savings extended beyond simply replacing computer-generated imagery (CGI). The shoot itself became simpler, and the overall production timeline was shortened. He credited AI with reviving projects that had previously fallen apart due to budget constraints, potentially opening doors for genre and sci-fi films in the Korean industry.
However, the panelists agreed that technology alone cannot replicate the emotional core of storytelling. Kang pointed out that while AI handles much of the creative workload, it can eliminate the collaborative discussions and accumulated experience that filmmaking traditionally requires. This often results in images derived from a single person’s vision rather than a collective effort.
“So the quality of the footage AI makes is very high, but when it comes to delivering emotion and moving audiences, I think we’re in a period where it has actually gotten worse,” Kang said.
Lee echoed this sentiment, acknowledging that while new tools allow for quick results, they still require human refinement. “Some things now come out with a single click, but when it comes to conveying emotion, ultimately a human has to touch it once more. I strongly agree that a human touch is absolutely essential.”
Thurston described the deep learning denoiser behind his recent Academy Scientific and Technical Award as “assistive AI.” He emphasized that such tools are designed not to alter creative intent but merely to correct pixels. He also noted that Wētā has employed machine learning for at least a decade, with roots extending back 20 to 25 years to their Massive system used in “The Lord of the Rings.” h>
Thurston also addressed technical limitations, such as color accuracy issues where AI models trained on video might render a Coca-Cola can with “Netflix red” rather than the brand’s specific shade.
Regarding the pace of technological adoption, Lee stated that Netflix operates under strict internal guidelines and tests new tools before integration, noting they are “not wildly obsessed with speed.”
The discussion also touched on sustainability. Thurston revealed that Wētā’s Wellington data center now uses zero liters of water annually thanks to closed-loop cooling. Lee added that Netflix considers sustainability across its entire operation, aligning with a 2030 goal to cut carbon emissions, rather than studying AI’s environmental impact in isolation.
Kang predicted a bifurcation in AI usage: short-form content will likely prioritize productivity, with some Chinese projects already being made entirely via AI from planning to distribution. Conversely, film and drama will focus on credibility and authenticity.
When asked about the possibility of AI replacing actors, Kang dismissed the idea. He argued that even if AI acting becomes highly sophisticated, it lacks the ability to transmit genuine emotion due to its mechanical nature. Instead, he sees AI playing a role in stunts and stand-in work. Thurston agreed, stating, “The point is not to replace actors, right? No way. We are audiovisual storytellers, and storytelling is studying the human condition.”
Lee concluded the session by remarking that while technology will continue to surprise and shock, humans will remain the writers and the source of meaningful stories.
It’s fascinating that Chinese tools like Kling are leading this. The global film industry landscape is shifting faster than we realize.
I’m skeptical about the sustainability angle. Cloud computing and training those models takes massive energy; the water cooling is a nice bonus but not the whole story.
Wait, so AI handles the pixels but humans still fix the soul? Sounds like we’re just paying for cheaper mistakes now.
The panel’s point about emotional depth resonated with me. Tools help, but they can’t replace the messy collaboration of human filmmaking.
A 40% budget cut is staggering. I wonder how many mid-tier films will now be made with AI instead of practical effects?