The good outweighs the challenges: Trilith Studios CEO talks AI in media at Georgia Senate committee

Caption

Overhead shot of Trilith Studios in Fayetteville, Ga.

Credit: Trilith Studios/Facebook

Trilith Studios is one of the biggest movie studios in North America, where many blockbuster films, including Marvel Studios' Avengers: Infinity War and Avengers: Endgame and, more recently, James Gunn's upcoming Superman, are filmed in its lots.

On Tuesday, Trilith Studios president and CEO Frank Patterson explained how movie studios are using AI and his take on the concerns over AI at the Senate Study Committee on AI's third meeting at the movie studio in Fayetteville, Ga.

Chairman Sen. John Albers said the study committee hearing was dedicated to discussing "all things media, film, music, and all the stuff AI is playing a role in."

Paterson first acknowledged that while the concerns over AI are valid, he thinks "the good is going to largely outweigh the challenges in the future."

"I'm genuinely excited about this technology, but as a technologist in my background, I also recognize the concerns that you, Sen. [Albers], and the committee are raising and the ones we're exploring," he said.

On AI's use in the film business, Patterson focused on digital replicas.

In an AI study in July 2024, the U.S. Copyright Office defined digital replicas as video, image, or audio recordings digitally created or manipulated to realistically but falsely depict an individual.

"The film business is to blame for all this technology," Patterson said about digital replicas.

He explains that the first time studios "faked you out as an audience" was in the 2008 film The Curious Case of Benjamin Button, where Brad Pitt plays a character who ages in reverse from 84 years old to a baby.

"This is a very challenging movie," Patterson said. "It was in development for 30 years because no one could figure out how to make it. The digital domain of LA created the first digital reference, so the first 54 minutes of this movie, [lead actor] Brad Pitt is not even in." 

The Digital Domain team used Pitt in training the AI technology to understand how his performance so the technology could perform.

"A very effective job — the world didn't know it was the beginning of digital replicas," Patterson said.

When discussing AI's usage for digital replicas, Patterson explained the phenomenon of deep fakes. These are videos of a person whose face or body has been digitally altered to appear as someone else, usually used to misrepresent the words or actions of the person involved.

Per Patterson, the concerns about deep fakes stem from the transition from high-tech tools used by movie studios and filmmakers to more accessible tools now available to the public.

"The large language models and what happened with the ChatGPT was a big awakening, but it's the realization — and this committee's concern — is what are we going to be doing with all of this fake stuff and how can we tell what's real and what's not in a world where fake is a common term?" he said.

Regarding the public concern that it will be hard to keep up with AI as it relates to deep fakes and other systems, Patterson said that there are many obstacles to advancing AI technology that fast.

"We haven't figured out a lot of the programming necessary to make these AI tech systems as useful as I think everyone is saying they're going to be," he said. "This is going to happen a lot slower than a lot of people are saying."

Patterson suggests Georgia lawmakers focus on policies related to publicity and privacy when creating laws and policies on AI in Georgia.

"Thoughtful policy around protecting Georgians and their right to their name, image, and likeness and their right to control the publicity of that name, image, and likeness," he shared. "Anything that starts to create policy about protecting AI from us right now, I think it's too early."

"What we can be doing is stifling important innovation that might actually be the protections that we need from the stuff that we're nervous about."

The next meeting of the Senate Study Committee on AI is scheduled for Thursday, Sept. 12, 2024.