Will artificial intelligence take over my job? These days, that question looms large in media and entertainment, where talk of a “robotic revolution” in which easily taught tasks can be automated strikes fear into the hearts of even skilled professionals such as cinematographers and editors.
Experiments have tested the premise. In 2016, audiences at the Cannes Lions Festival’s New Directors Showcase screened “Eclipse,” a film created using AI tools. The project began as an experiment commissioned by execs from ad agency Saatchi & Saatchi who approached Zoic Labs, a sister company of visual effects house Zoic Studios that focuses on data visualization.
Could Zoic Labs use AI to conceive of, write, direct and edit a short movie? Zoic exec VP Matt Thunell was game.
AI, already used to handle jobs based on routine, repeatable tasks, “becomes really interesting when you try to apply [it] to anything that requires semantic reasoning or flexibility,” Thunell says. “That was the crux of our short film.”
The production used several AI tools, including cognitive tech IBM Watson, Microsoft’s AI chatbot Rinna, Affectiva’s facial recognition software, drones and a proprietary AI-based editing program.
The creators of “Eclipse” deemed it a good first try. Producer-developer Brian Frager says the original goal was to make a movie “based on very little human interaction.” “We tried to stay out of the creative process,” he says. “Yet in the end, a lot of humans worked on it. It was more AI assisted.” As Thunell says, “these are early days for AI.”
Much has been written about how AI and machine learning will take over jobs as disparate as truck driving and radiology. In media and entertainment, tech consultant-author Shelly Palmer singles out location scouting, script polishing and assistant editing as areas likely to be automated. “There are plenty of journeyman tasks done in pre-production, production and post,” he says. “If it’s a binary technical decision, a computer will do it better.”
The most compelling reason for adopting AI, says Palmer, is cost cutting. He points to synthesizers in music, which at first were used as a way to economize.
“Ultimately it became a style of music, but initially one guy and a computer replaced 30 string players.”
Using AI to take on routine tasks, Palmer adds, will devastate the apprenticeship system and trim departments. “An AI-assisted tool set will give the senior art director the capabilities of 10 art directors,” he says. “So you won’t need the other nine on staff.”
DP Curtis Clark, chair of the American Society of Cinematographers’ Technology Committee, says that any time new tech comes on the scene, it creates fear of job loss. But technology could also create enhancement skills, he notes.
AI tools could assist, say, a focus puller, but there are variables. “What if the operator feels the instinct or need to do something more impactful because the actor didn’t hit his mark?” Clark posits. Cinematography has to do with creative choices, he says, which resist any easy mechanization or program, no matter how “intelligent.”
Another area that can benefit from automation is visual effects. “The way we do 3D camera-match moves has been largely automated, as has rotoscoping,” Thunell says. Zoic Studios is exploring machine vision learning to associate metadata to its hundreds of thousands of digital assets. But will AI replace the VFX artist? Thunell believes that rather than replacing jobs, AI will let people focus on new industries.
Of all production jobs, those of assistant editors have been pegged by some as the most vulnerable for replacement by AI, but editor-author Steven Cohen isn’t buying it. “The assistant editor’s job gets more complicated every year,” he says. “It requires a combination of creativity, inventiveness and resourcefulness. Maybe one day, I’ll be able to say, ‘OK, Avid, turn over this reel, create sound effects for that scene, score this one and figure out how to matte out this rig.’ But I don’t see it happening anytime soon.”
USC School of Cinematic Arts professor Norman Hollyn leads discussions on the topic with his editing students. “The assistant’s job won’t go away, but it will change in terms of what it does to support the process,” he says.
As the entertainment industry debates, the companies that make AI tools continue to advance. Technology provider Nvidia is teaching machine learning to about 100,000 developers, says Greg Estes, the firm’s VP of developer marketing. A number of those developers, he adds, are working on applications in media and entertainment — which will further solidify the role of AI in the production of content.