While artificial intelligence remains a dirty word in Hollywood and cause for broader consternation over job losses in the U.S., studios and artists in other countries are more readily embracing the technology. It’s a dynamic that could have ramifications for how competitive the next generation of creatives will be on the global stage as AI plays a more central role in media.
That was one of the key takeaways in a panel I moderated at this week’s AI on the Lot conference, which was held on the Amazon MGM Studios Lot in Culver City, Calif. The discussion focused on the differing attitudes of AI around the world, and how that’s affecting adoption of the technology among creatives from different countries.
Participating in the discussion was Richard Chuang, who co-founded Pacific Data Images, which later became Dreamworks, Stephan Vladimir Bugaj, an Emmy-winning creative who earlier this year joined Indian media giant JioStar as its senior VP of GenAI content and technology, Mrinalini Rao, head of research in international markets and growth at Google, and Christian Schussler, CEO of Norwegian AI studio Reimagine Studios.
The panelists each noted how AI had been more readily embraced around the world. In India, Jiostar already released a fully AI-generated show called “Mahabharat: Ek Dharmayudh.” Rao noted that in South Korea, the government is putting together an AI textbook for students. Schussler noted that in Norway, the strong education system has allowed people to take a more nuanced look at the technology.
By contrast, Chuang said that in one Northern California school, students signed a petition seeking a ban of AI. He said there was a lot of misinformation about the technology that has led to this trepidation, and that risks leaving the next generation of creatives behind on the global stage.
“I blame a lot of educators for not educating them properly, because it turns out our educators are not as versed in AI as I can imagine they should be,” he said. “That’s a big challenge for us in the U.S.”
Chuang called for universities to start holding AI classes for their professors. “I think educating our educators is the first step because they need to understand what are the limits, what are the guardrails and how to leverage it for their benefit rather than running away from it,” he said.
AI powering local and niche content
There’s plenty of discussion about AI cutting down on the time and cost to produce content. The upside, the panelists said, is that this opens the door for creatives to produce shows or movies that narrowly target specific interests that would’ve previously been deemed too expensive or unrealistic to produce under traditional means.
“Making niche content affordable is super interesting, and I think there’s a lot of opportunity there,” Bugaj said. “There are all of these sub fandoms, like furries who love Star Trek. You couldn’t make a $100 million feature film for that audience, but you could make things for that audience, and there are thousands of audiences out there that have really devoted, engaged fandoms who will pay for experiences that are in their niche.”
He referenced a feature film he saw last year from a guy in Malaysia that explored LGBTQ+ and domestic violence issues — topics that would’ve gotten him arrested had it been reviewed by local censors — but was generated with AI and released in international film festivals.
“That was his way of being able to speak to this in his own language with people who look like his culture, with settings that look like his culture,” he said. “There’s entire cultures out there who don’t necessarily get their material seen, and some of them can’t even make their material.”
In Korea, one of the K-pop companies is building an AI K-Pop star, and rather than be turned off, Rao said Korean audiences are invested in that process.
“It’s not about broad reach, they’re bringing in the super fans, they’re leveraging them to co-create with them,” she said, noting that the AI is able to have individual personalized chats with those fans to further engage them.
Moving past western models
The problem when relying on models built by U.S. companies is that the photos and video generated come from a database trained on Western-centric imagery. Or, with the rise of Chinese models, the imagery may have a bias towards Chinese imagery.
That’s a problem when trying to generate videos featuring authentic Indians from specific regions. Rao noted that India was building a national model fed with local data and imagery.
It’s an issue Malik Afegbua, a Nigerian artist who rose to fame for his AI art, encountered when using tools like Midjourney. When I talked to him about this for an earlier story, he mentioned building an African LLM that incorporate data, imagery and stories that are locally sourced.
Schussler noted that whenever something is published in Norway, it enters the national library. He said that he is working with the government to tap into that vault of Norwegian culture dating back hundreds of years to train his models.
“It will really help safeguard the culture in a world where everything’s sort of generalizing and washing out,” he said.
The post Hollywood’s AI Aversion Could Have Consequences for the Next Generation appeared first on TheWrap.
