AI Apocalypse: Tech CEO Declares Human Actors Obsolete, 'It's Already Happening!'

At Turin’s View Conference, during a fireside chat titled “Every Pixel Will Be Generated,” a significant debate emerged between industry leaders regarding the future of generative AI in media production, specifically its potential to replace human actors and reshape filmmaking economics. Cristóbal Valenzuela, co-CEO of generative AI startup Runway, posited an affirmative view, stating, “Yeah, absolutely. I think it’s already happening in a way,” describing AI as a “new medium of expression” and “new medium of storytelling.” He further predicted that models would reach a point where they could be directed similarly to human actors.

Conversely, Nvidia’s Richard Kerris, VP of developer relations and general manager of media and entertainment, expressed a more traditional perspective. He recalled a director's argument that the unique spontaneity between a director and an actor, exemplified by Marlon Brando improvising with a cat in “The Godfather,” could never be replicated by prompts. Kerris stressed that the craft of filmmaking still demands a deep understanding of character, composition, and story, emphasizing that there is no simple “make movie button.”

Valenzuela’s vision, as outlined in his essay “It’s All About the Pixel Economy,” suggests that “the value of creating pixels is trending towards zero.” He foresees a future where the number of “seconds consumed” converges with “seconds generated,” envisioning audiences creating personalized entertainment as rapidly as they consume it. In another essay, “The Cost of a Frame,” Valenzuela traces the history of media through the cost of producing a credible frame, highlighting high production value content—the domain of Hollywood—as the “last expensive layer” that AI is collapsing.

Kerris frequently framed the resistance to generative AI as a common historical pattern, akin to the fears that accompanied previous technological advancements like computer-generated imagery. He noted, “The similarities of fear and rush to judgment on things is so similar to what I’ve seen in the past.” Jeff Kember, Nvidia’s principal technical evangelist for media, echoed this optimistic outlook, asserting the widespread adoption of generative AI. Kember observed that while many openly discuss their use of AI, others do so privately due to legal team directives, but nearly everyone is employing the technology.

Despite the industry’s embrace, a peculiar and unique concern surrounds generative models: the material used for their training. Many VFX artists, game studio heads, and animation directors expressed anger regarding this issue. When questioned about training data, Valenzuela clarified a “core misbelief,” stating that AI models are not databases but rather “learning from representation.” This concern has already translated into legal challenges. A federal judge approved Anthropic’s $1.5 billion settlement with authors whose copyrighted books were downloaded from pirate libraries. While a prior ruling deemed using copyrighted works for training as potentially fair use, it differentiated this from the acquisition and retention of pirated copies. Numerous lawsuits are currently progressing against major AI companies, including a proposed class action by visual artists against Stability AI, Midjourney, DeviantArt, and Runway itself, alleging the unauthorized use of copyrighted work for model development.

Collectively, the discussions at the View Conference, featuring companies at the forefront of AI, painted a picture of a future where the cost of generating moving images approaches zero. This technological shift is poised to fundamentally reshape the traditional structures and economics of media production, even amidst ongoing legal and ethical debates concerning data sourcing.