Emily Reigart, Senior Editor of NAB Amplify, moderated a session featuring Yves Bergquist, the Director of USC’s Entertainment Technology Center’s AI, blockchain, and media project, and AI director Pinar Seyhan Demirdag, who spoke about their work in generative AI in media and entertainment. A data scientist, Bergquist heads up ETC’s AI & Neuroscience in Media Project which has helped the M&E industry “accelerate the deployment of next-generation analytics standards and solutions,” including AI. He is also chief executive of AI engineering firm Novamente, “which applies neural-symbolic Artificial General Intelligence to large enterprise problems.”
Demirdag, with Gary Koepke, founded Seyhan Lee, which they dub “a bridge between generative AI and the entertainment industry.” The company produced AI VFX for Descending the Mountain, a feature film, and a brand-sponsored AI film titled Connections/Beko. Last year, they introduced Cuebric, a collection of AI tools to “streamline the production of 2.5-D environments for virtual production stages.” She noted that Cuebric “works with AI from concept to camera.” The company also offers “commercial and ethical Generative AI production.”
Reigart asked the panelists what the immediate applications for Generative AI in the M&E industry are. Demirdag noted that you have to “think outside of your paradigm” to optimize the workflow. She noted pain points in the production process that could benefit from AI tools. “In the VFX process there are many dull repetitive tasks like rotoscoping,” she said. “There’s a barrier of entry to virtual production because it’s a completely new technology and cinematographers are terrified of it. It requires 3D real-time engines with an expert to operate them. Last, 5 percent to 20 percent of a film budget goes to reshoots which can get up to $24 million for a big budget movie.”
Bergquist reported that, right now, generative AI is enabling “really fast, rich and cheap pre-production including look-books, storyboarding and pitching.” “Down the road, we’ll be able to use 2D images to create 3D environments which will really disrupt virtual production,” he added. “A lot of VFX workflows will also become optimized. Post-production enhanced by AI is already happening.”
“The thing it won’t disrupt is development,” he continued. “ChatGPT is a nice party trick but it isn’t capable of writing scripts. Cuebric is a very promising tool and beyond automating workflows, we’re seeing the latest chapter in the macro trend, which is giving individuals extremely powerful tools that used to be reserved for institutions and organizations. We’re heading into a world where influencers — the micro-content creators — will be able to make full use of these tools.”
Reigart polled attendees to find out how many were excited about AI — a lot of people raised their hands — and how many are also afraid for the future of their careers, which prompted many of the same people to raise their hands again.
Bergquist noted that, “no one received more funding than OpenAI to create ChatGPT.” “The thing to remember [about] AI,” he said, “is that most of the time it doesn’t work and you shouldn’t use it.” He gave kudos to everyone behind Generative AI models — “the first time that models work extremely well.” “The excitement is the ability of creatives to take it to the next level,” he added, noting that there is also concern over those who will use it to create “generally harmful content.”
At the same time, putting such tools in the hands of all creators “puts us on the cusp of an explosion in creativity,” he also pointed out that, “there are very few talented indie filmmakers.” In other words, the jury is out on whether the democratization of these tools will foster more great content or just more easily-made bad content.
Demirdag agreed, noting that, “what we call it is the danger of the normalization of mediocrity.” She stated that the “long journey” to understanding how to best use AI is to realize that “most of our decisions are made with our unconsciousness.” “The long way of overcoming AI fear is to realize that we only fear what we put on a pedestal,” she said.
Bergquist raised another question that the new AI tools bring to the fore. “When all the crafting parts of creativity become faster, we’ll find out how much knowing about the craft is material to the artistic quality of what we’re doing,” he said. “I’m curious to see if the quality of creativity is knowing about the craft or the decisions we make.”
When Reigart asked if another element is artists who believe that AI is “cheapening their work,” Bergquist said “We’re seeing a lot of visual artists saying that it seems too easy at times using Generative AI. And it’s cool to hear them say that. If everything has a much higher production quality, where is the space for human artistic talent? I think it will be in the stories you choose to tell and the decisions you make.”
“As long as we place AI on a pedestal, we forget our superiority,” said Demirdag. “You dehumanize yourself if you do that. You have consciousness, a unique perspective, and love that no machine can replicate. Just start your journey.”