Cristóbal Valenzuela, co-chief executive officer of generative artificial intelligence startup Runway, said AI could eventually take over roles traditionally performed by human actors, arguing that the technology is already beginning to change the way stories are created and presented.
Speaking during a fireside chat titled “Every Pixel Will Be Generated,” Valenzuela responded affirmatively when an audience member asked whether AI could replace human performers.
“Yeah, absolutely. I think it’s already happening in a way,” he replied, describing artificial intelligence as “a new medium of expression” and “a new medium of storytelling.”
His comments highlighted the growing divide between those who view generative AI as an opportunity to expand creative possibilities and those concerned about its implications for performers, filmmakers and other professionals whose livelihoods depend on traditional production methods.
Richard Kerris, Nvidia’s vice-president of developer relations and general manager of media and entertainment, offered a more measured perspective, emphasising the enduring importance of human performance and the unpredictable creative exchanges that shape filmmaking.
During the discussion, Kerris recalled a director’s argument that the spontaneous interaction between a director and an actor could not easily be reproduced through technological prompts. He cited the example of legendary actor Marlon Brando improvising with a cat during the making of The Godfather.
“That craft is something to behold, and I just think that that’s something that I don’t think we’ll ever lose,” Kerris said.
Valenzuela, however, suggested that advances in AI models could eventually make it possible for directors to guide artificial performers in much the same way they work with human actors.
“Not now, but I think models are going to get probably to a point where you can probably direct them in a very similar way,” he responded.
The exchange underscored a fundamental question facing the film industry: whether technological advances will primarily serve as tools for human creativity or gradually assume responsibilities that have historically required actors, directors and other creative professionals.
Kerris also stressed that filmmaking involves more than generating moving images, noting that creators must understand character development, composition and storytelling. The process, he suggested, cannot be reduced to a single automated command.
“There is no ‘make movie button,’” he said, prompting Valenzuela to joke: “I wish.”
The economics of a pixel
Valenzuela’s broader vision for the industry is outlined in his essay, It’s All About the Pixel Economy, where he argues that the cost of producing visual content is steadily declining as generative AI becomes more sophisticated.
“The value of creating pixels is trending towards zero,” he writes, predicting a future in which the time required to generate entertainment could approach the time audiences spend consuming it.
Under the traditional filmmaking model, a production may require hundreds of professionals working over months or years, while millions of viewers spend approximately two hours watching the finished film.
Valenzuela imagines a different arrangement in which audiences could generate personalised entertainment at a pace that closely matches their consumption habits. Such a development could fundamentally alter the relationship between content creators, studios and viewers.
In another essay, The Cost of a Frame, he examines the history of media production through the changing cost of creating a convincing visual frame. He argues that artificial intelligence is rapidly reducing the expense associated with producing high-quality visual content, a category that has traditionally demanded substantial investment in Hollywood and other major production centres.
“This is the last expensive layer,” Valenzuela writes, referring to the high-production-value content that AI tools are increasingly making accessible.
If these costs continue to fall, the consequences could extend beyond production budgets to influence how films are commissioned, how creative teams are assembled and how audiences access entertainment.
However, the extent to which AI-generated productions can consistently reproduce the artistic judgement, emotional depth and collaborative processes associated with human filmmaking remains a subject of debate.
AI adoption and the debate over creative control
Nvidia executives at the conference presented the adoption of generative AI as part of a broader pattern of technological change, drawing comparisons with earlier innovations that initially provoked resistance within the creative industries.
Kerris said the concerns surrounding AI were reminiscent of the anxieties that accompanied previous developments, including the introduction of computer-generated imagery.
“The similarities of fear and rush to judgment on things is so similar to what I’ve seen in the past,” he said.
Jeff Kember, Nvidia’s principal technical evangelist for media, expressed a similarly optimistic view in his presentation, “Creative Control in the Age of AI Agents,” arguing that generative AI use is already widespread across the industry, even where companies are reluctant to discuss it publicly.
“There’s a whole bunch of people talking about how they’re using generative AI and then there’s also a bunch of people who aren’t talking about it and they’re not talking about it primarily because their legal team is not letting them talk about it,” he said.
“But both people on each side of this fence are both using generative AI. I don’t know anyone who isn’t.”
Yet conversations among visual effects artists, video game studio executives and animation directors attending the conference revealed that concerns about the technology extend beyond fears of job displacement. A major source of frustration is the material used to train generative AI systems, particularly when that material includes copyrighted creative works.
Copyright disputes cast a shadow
Responding to questions about training data, Valenzuela challenged what he described as a misunderstanding of how artificial intelligence models operate.
One of the “core misbeliefs” about AI, he said, was the assumption that the technology functioned primarily as a repository of existing material.
“They’re not databases,” he said. “They’re learning from representation.”
The distinction, however, does not resolve the legal and ethical questions surrounding the use of copyrighted material to develop generative models. Those questions are increasingly being tested in courts as authors, artists and other rights holders seek greater control over how their work is used.
In July, a federal judge approved a $1.5 billion settlement involving Anthropic and authors whose books had been downloaded from pirate libraries. An earlier ruling had found that using copyrighted works to train AI models could qualify as fair use in certain circumstances, while distinguishing that issue from the separate question of acquiring and retaining pirated copies.
Meanwhile, several lawsuits involving major AI companies remain before the courts. One proposed class-action case brought by visual artists targets Stability AI, Midjourney, DeviantArt and Runway, alleging that copyrighted artworks were used to develop generative AI systems without permission or compensation.
The allegations underscore a central tension in the AI debate: while technology companies argue that their models learn patterns and representations from training material, creators continue to question whether the use of their work is lawful and whether they should receive compensation.
The outcomes of these disputes could help determine the legal boundaries within which generative AI companies and creative professionals operate.
A changing future for filmmaking
The discussions at the View Conference reflected a vision of filmmaking in which the cost of generating moving images could fall dramatically, potentially reshaping the industry's established production and distribution structures.
For AI developers, the technology promises new forms of expression, faster production and greater access to sophisticated visual storytelling. For actors, artists and other creative workers, however, the same developments raise questions about employment, authorship, consent and the protection of creative contributions.
As generative systems become more capable, the debate is likely to move beyond whether AI can produce convincing images and performances to questions about who controls the technology, whose work is used to develop it and how the benefits of automation are distributed.
Valenzuela’s prediction that the value of generating pixels could approach zero captures the scale of the transformation envisaged by AI proponents. Whether that future expands opportunities for human storytellers or places greater pressure on traditional creative professions remains an open question.
What is becoming increasingly clear is that artificial intelligence is challenging long-established assumptions about how films are made, who makes them and what audiences may eventually expect from the experience of watching them.