CPS 2023: The Right AI Fits for Content Workflows
CULVER CITY, Calif. — For a while now, studios have been attempting to figure out where AI fits into their content workflows. At the Dec. 5 Content Production Summit, Keith Ritlop, co-chair of production security work streams for the Content Delivery & Security Association (CDSA) was on hand to offer some clarity.
“I know AI is fairly new in the area of content production for film and TV, but there are some basic strategies we can use,” he said during his presentation, “Using AI to Drive Efficiencies in Production and Post.”
Ritlop provided an overview of the three areas of AI that need to be considered: artificial narrow intelligence (ANI), which highly specialized, can make basic predictions and identify patterns, and makes up the chatbots, facial recognition tech and virtual assistants of the world; artificial general intelligence (AGI), which is comparable to human intelligence, can problem solve, and is partially represented by the work being done with tech like ChatGPT and IBM Watson; and artificial super intelligence (ASI), which is theoretical, would surpass human intelligence and creativity, which is “the crazy SkyNet, theoretical stuff,” Ritlop quipped. “Worst case scenario is the end of life as we know it.”
“Basically, all AI we’ve seen up to this point falls in [ANI], highly specialized, task-based AI,” Ritlop said. “Where we’re really not at yet is artificial general intelligence, that’s the next theoretical step in the AI journey.”
But the AI tools that are available today have vast applications across content workflows. In development, AI can be used for brainstorming, script breakdowns, even for casting suggestions. There’s storyboarding and location scouting for pre-production. Motion capture, animation and digital makeup and prosthetics during production itself. And in post, de-age and re-age characters, add sound effects, edit and color grade, all with AI tools available right now.
On the distribution and marketing sides, subtitling, predictive analytics, personalized experiences, all can be assisted with AI tools. For legal, think contract analysis and talent deal creation.
“We want to make sure AI doesn’t replace humans … augmentation, not automation,” Ritlop said. “Anything where you’re using huge amounts of data, extremely large data sets, tons of historical data, whether that’s finance or watch history, we’re in a very data-driven world, there’s devices all around us, and they’re all generating data.
“AI is a great application for harnessing all that data, distilling it down to make decisions.”
In all, Ritlop said AI tools can help media and entertainment players do the work faster. But there are drawbacks, and challenges the industry needs to be aware of. The unauthorized use of copyrighted materials, used to train AI, remains a major point of contention, and not all AI platforms commit to keeper user data confidential. The sourcing of data for AI isn’t always accurate, resulting in incorrect outputs. And the ability for some AI to generate realistic images of people and their voices is rife for legal headaches, cybersecurity risks, and outright scams.
“Just because the tool has the capability, we want to make sure we’re being responsible, building in the right safeguards, so that, again, we make it more of an augmentation, and not an automation,” Ritlop said.
Produced by MESA, the Content Production Summit was presented by Fortinet, and sponsored by Convergent Risks, Friend MTS, Amazon Studios Technology, Indee, NAGRA, EIDR, and Eluv.io, in association with the CDSA and the Hollywood IT Society (HITS).